Language and the Cognitive Construal of the World (Trends in Linguistics. Studies and Monographs)

Citation preview

=-

Studies and Monographs 82 Language and the Cognitive Construal of the World John R. Taylor .Robert E.MacLaury (Editors)

l\tOUTON DEGKlJY ER

Trends in Linguistics Studies and Monographs 82

Editor

Werner Winter

Mouton de Gruyter Berlin New York

Language and the Cognitive Construal of the World

edited by

John R. Taylor Robert E. MacLaury

Mouton de Gruyter Berlin New York

1995

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter & Co., Berlin.

@l Printed on acid-free paper which falls within the guidelines

of the ANSI to ensure permanence and durability. Library of Congress Cataloging-in-Publication Data

Language and the cognitive construal of the world I edited by John R. Taylor, Robert E. MacLaury. p. em. - (Trends in linguistics. Studies and monographs; 82) Chiefly papers presented at a conference entitled Language, thought, and culture which was held in Broederstroom, South Africa in 1991. Includes bibliographical references and index. ISBN 3-11-014301-1 (cloth: alk. paper) I. Language and culture. 2. Cognitive grammar. 3. Categorization (Linguistics) I. Taylor, John R. II. MacLaury, . III. Series. Robert E., 1944P35.L29 1995 306.4'4-dc20 95-18973 CIP

Deutsche Bibliothek - Cataloging-in-Publication Data Language and the cognitive construal of the world I ed. by John

R. Taylor ; Robert E. MacLaury. - Berlin ; New York : Mouton de Gruyter, 1995 (Trends in linguistics : Studies and monographs ; 82) ISBN 3-11-014301-1 NE: Taylor, John R. [Hrsg.]; Trends in linguistics I Studies and monographs

©

Copyright 1995 by Walter de Gruyter & Co., D-10785 Berlin All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Typesetting and printing: Arthur Collignon GmbH, Berlin. Binding: Liideritz & Bauer GmbH, Berlin. Printed in Germany.

Contents

Preface: Linguistic and anthropological approaches to cognition Robert E. MacLaury Introduction: On construing the world John R. Taylor

vn 1

Seeing it in more than one way Eugene H Casad

23

Possession and possessive constructions Ronald W Langacker

51

What lack needs to have: A study in the cognitive semantics of privation Savas L. Tsohatzidis

81

The construal of cause: The case of cause prepositions Rene Dirven

95

Conceptual grammaticalization and prediction Bernd Heine

119

Metaphors of anger in Japanese Keiko Matsuki

137

Looking back at anger: Cultural traditions and metaphorical patterns Dirk Geeraerts-Stefan Grondelaers 153 Anger: Its language, conceptualization, and physiology in the light of cross-cultural evidence Zolt(m Kovecses 181 The metaphorical conception of mind: "Mental activity is manipulation" Olaf Jakel

197

Vantage theory Robert E. MacLaury

231

VI

Contents

The terror of Montezuma: Aztec history, vantage theory, and the category of "person" Jane H. Hill-Robert E. MacLaury 277 Selection of Japanese categories during social interaction Munekazu H. Aoyagi

331

Genus, species, and vantages Jeff Lansing

365

On construing the world of language Nigel Love

377

Index of names

391

Subject index

396

Contributors

407

Preface: Linguistic and anthropological approaches to cognition Robert E. MacLaury

Traditionally linguists and anthropologists together have sought to fathom the tie between language and cultural construction. Boas (1911) exquisitely documents the indispensability of categorization to language as well as the culture-specific content of categories particular to any language. His statement articulates a knowledge of the language-culture relation that had accrued from the systematic comparison of human societies since it began at least four centuries before him (Hodgen 1964). On Boas's empirical foundation, the intimate bond has been further illuminated from the perspectives afforded by Sapir's (1916) concern with timedepth, Kroeber's (1917) superorganic, Levi-Strauss's (1949) structuralism, the ethnoscientists' formalizing of cognitive models (Tyler 1969), and the arguments supporting universals (Berlin 1992, Brown 1991, Hardin 1995, Heine this volume). The following ch(!.pters foretell that the theoretical nexus of language, categorization, and culture will tighten even more and will yield yet another conceptual advance. Currently, cognitive linguistics, as it has grown out of Space Grammar under Langacker and others (1982, 1987, 1993; Lakoff 1987; Taylor 1989), recognizes not only that reliance on mental imagery and on the figure-ground construction is vital to language but that this practice is seemingly innate. In cognitive anthropology, vantage theory proposes nearly identical innate processes, but specifically in the construal of categories. Here the processes differ from noncategorical constructs in that grounds and figures are arranged as priorities of concentration and must include attention to similarity and attention to distinctiveness among them; the latter define categorization apart from other image schematic and figure-ground cognition. In agreement with the various kinds of perspectivization explored by Langacker and his colleagues, any category is a point of view based on the inborn human proclivity to make sense by formulating analogies with space-time as a source. Linguists (Bickerton 1992) and anthropologists (Falk 1992, Burling 1993) alike offer speculations on cognitive evolution, which at the very

vm

Robert E. MacLaury

least lends metaphysical depth to our ideas about the relation of language and culture. They envision, at the onset of language, an intelligent, fully bipedal creature who shared a savannah with major carnivores, who lived by scavenging their kills, who had to plot their movements meticulously and constantly, and who was too small to fight them, too large to hide from them, and too slow to flee from them. On this prelinguistic horizon, natural selection favored more than sheer stealth and cunning but specifically the yen to manipulate multiple mental maps and alternative spacetime plans, and these logistical conceptions had to include not only one's own position but the purviews of all actors germane to one's survival, two-footed and four-footed; and selection not only favored this parallactic talent but, equally as vital, the drive to apply it incessantly such that reality would be at every moment construed as a mutable scheme from any of various viewing angles. Then, on the basis of a single analogy and likely with the booster of another mutation, hominids extended their capacity and attendant drive to the construction of plastic categories, thus putting in place the conjoint foundation for rudimentary language and culture. Certainly the advance from the Oldowan eolith to the Acheulean handax required premeditated imagery, mental rotation, and plotting of viewpoint in reference to both a fixed axis and the turning of a bifacial blade, all projected into an original blank of spherical stone. Yet, as this technology remained stable for the next million years, it certainly was not the development of lithic tools that impelled our ancestral brain to double its size while the oral cavity contracted and the pharynx deepened. Finally, only 30,000 years ago, homo sapiens emerged in the Upper Paleolithic with a cultural explosion of art and artifacts that rival the aesthetics of contemporary masterpieces. Cultural anthropologists have contributed virtually no hypotheses of instinct to the Human Genome Diversity Project, a proposal to map human chromosomes (Khan 1994). Kroeber, in formulating his superorganic, explained why it is unlikely that any sort of human behavior would be genetically determined (cf. Sperber 1980). However, if certain fundamental strategies of making sense - such as figure-ground projection and space-time analogy - are genetically foreordained, and if they enable all orders of flexible and revisable constructions on their bases, then the search for innate pattern beyond these fundamentals may not pay off. Conversely, genetic determination is likely to occur at the stem of language and culture, as the stem cannot be invented anew by every human group. Although a child may learn from its elders and peers that a particular word should name both blue and green, who teaches the child the

Preface

IX

method of composing this category? Perhaps it is more than coincidence that both cognitive linguistics and vantage theory, each with its separate emphasis, attribute innateness to the same sorts of process. As John Taylor describes, the chapters of this volume go a long distance toward covering the facets of construal. Yet certain companion pieces already in the literature add to the overview. Clark (1990) shows that children face the problem of mastering the particular points of view that are sanctioned by their first language. Each language features a different combination of conventional vantages. Yet the task before any child of commanding specific perspectives is an immense part of language-learning because, in general, the use of language consists largely of adopting, manipulating, and conveying one view after another. Although Clark's intent is to demonstrate how critical to language-acquisition is the learning of viewpoints, she incidentally imputes further claims of global scope: first, points of view are inseparable from language and constantly adopted as it is spoken; second, this part of language has been neglected by linguists, at least in proportion to its importance. To my knowledge there are no studies that, say, compare two or more languages to show how each countenances a different constellation of sentential vantages. Clark's suggestion seems to pertain to what rule-writing linguistic grammarians used to call "transformations," that is, rearrangements of one core of words to convey different slants on a single event in "deep structure." The notion of perspective may supersede this awkward way of expressing the relation between intringuingly similar sentences. Such an account of how arrangement relates to meaning may bring us closer to a comprehensive theory of language. The present gulf between autonomous syntax and cognitive linguistics is unlikely to last much longer. While the former has eschewed semantics, it can, nevertheless, address regularities that we in the cognitive camp have no way of modeling. Is the current Chomskian school really claiming that language names the world objectively? Or are they attributing to language particular processes other than those, such as figure-ground or metaphor, that cognitivists usually analyze? Like proponents of cognitive linguists or vantage theory, autonomous syntacticians attribute innateness to language, although they emphasize its systematic acquisition and the deftness by which utterances of amazing complexity are strung together and innovatively composed. This is quite apart from our own accounts of primordial imagery, primeval figure-ground, and the use of space-time analogy to form categories.

x

Robert E. MacLaury

The homo sapiens who deposited their elegant Upper Paleolithic arti~ facts likely commanded all of the myriad linguistic capacities studied by autonomous syntacticians, whereas our forebears in the Lower Paleo~ lithic are less likely to have controlled all of these faculties. Homo habilis and homo erectus may have only built upon innate properties of concep~ tualization of the like that Langacker and I identify. Cognitive linguistics, cognitive anthropological vantage theory, and autonomous syntax may offer each other a natural division of labor, each dealing with a different part of language that emerged at a distinct epoch in accord with a sepa~ rate mutation. Certainly a speaker of any contemporary language com~ bines the entire endowment to construe any utterance; the construction of viewpoint may integrate it all. Yet the capacities are likely to differ in function, quality, and genetic basis, if only the layers of development could be experimentally isolated for analysis. Jackendoff (1991), like Langacker's disciples or myself, is limited by his point of departure; but unlike some among us, he conjectures the need to supersede his original framework. Cognitive linguistics and cognitive anthropology will ad~ vance substantially when more pioneers of this ilk work toward each other's positions from opposite directions. Most chapters of this auspicious collection result from the conference Language, Thought and Culture: A Cognitive Linguistic Perspective that John Taylor organized at Broederstroom, South Africa, in 1991 (see also Taylor 1992). Those by Hill and MacLaury, Aoyagi, Matsuki, and Lan~ sing were presented at various symposia that I organized at conferences in 1989 and 1990 in Arizona (LSA), Washington DC (AAA), and Cali~ fornia (SWAA). Versions of my "Vantage Theory" were delivered under both auspices. After I developed vantage theory as a model of color cate~ gorization (MacLaury 1986, 1987), the problem of applying it beyond this test domain seemed insurmountable. Yet before I arrived at the Uni~ versity of Arizona for a temporary appointment in 1988, Jane Hill had read my 1987 paper, and, with only this at hand, had prepared the first draft of our collaborative manuscript. Dr. Hill emphasized then that our paper proved cognitive anthropology to be as capable of contribution to discourse analysis as are the hermeneutic approaches of Bakhtin (1981; Hill and Hill 1987; Hill 1995) and others (e. g., Ricoeur 1978) that are widely acclaimed in anthropology. Keiko Matsuki and Munekazu Aoyagi prepared their papers in my seminars while Jeff Lansing joined us with the forerunner his at our LSA symposium. Matsuki analyzes Japanese anger to proffer a cross~cultural angle on Kovecses's work. Aoyagi employs vantage theory to achieve a step-by~step insider's account of socio-

Preface

xi

linguistic decision making, which Brown and Gilman (1960) had rendered algebraically in their prime overview of the same topic. Lansing finds parallels between vantage theory and Aristotle's thoughts on the nature of classification while showing how subsequent schools have successively modified the classical thinking beyond recognition. On the basis of scholarship, he counters a recent impulse to attribute the classical view of the category to Aristotle alone. Lansing underscores our need to know more about our predecessors, and he does so in the same spirit with which Geeraerts and Grondelaers draw attention to our naivety regarding the historical forces that have shaped our raw data. Elsewhere works of vantage theory are cultural and historical but nonlinguistic (e. g., M. McLaury 1989), which supports Langacker's key point that kinds of cognition critical to language may obtain as readily in other realms of thought. In sum, the chapters secured by John Taylor and the contributions from my associates fell together by opportune accident out of our meeting in South Africa, but the coherence of this collection arises from aligned thinking in related fields that was bound to coalesce in short time. References Bakhtin, Mihail M. 1981 The dialogic imagination. Austin: University of Texas Press. Berlin, Brent 1992 Ethnobiological classification: Principles of categorization of plants and animals in traditional societies. Princeton. Princeton University Press. Bickerton, Derek 1992 Language & species. Chicago and London: University of Chicago Press. Boas, Franz (ed.) 1911 "Handbook of American Indian languages". Bureau of American Ethnology Bulletin 40:1. Brown, Donald E. 1991 Human Universals. New York eta!.: McGraw-Hill. Brown, Roger - Albert Gilman 1960 "The pronouns of power and solidarity", in: Thomas Sebeok (ed.), Style in Language. Cambridge, MA: MIT Press, 253-276. Burling, Robins 1993 "Primate calls, human language, and nonverbal communication". Current Anthropology 34:25-53. Clark, Eve V. 1990 "Speaker perspective m language acquisition". Journal of Linguistics 28:1201-1220. Falk, Dean 1992 Braindance. New York: Henry Holt and Company.

xu

Robert E. MacLaury

Hardin, Clyde L. (ed.) 1995 Color Categories in Thought and Language. Cambridge, MA: Cambridge University Press. Heine, Bernd [this volume] "Conceptual grammaticalization and prediction". Hill, Jane H. 1995 "The voices of Don Gabriel: Responsibility and self in a modern Mexicano [in press] narrative", in Bruce Mannheim and Dennis Tedlock (eds.), Dialogical Anthropology. Philadelphia: University of Pennsylvania Press. Hill, Jane H. - Kenneth C. Hill 1987 Speaking Mexicano: Dynamics of Syncretic Languages in Central Mexico. Tucson: University of Arizona Press. Hodgen, Margaret T. 1964 Early Anthropology in the Sixteenth and Seventeenth Centuries. Philadelphia: University of Pennsylvania Press. Jackendoff, Ray 1991 Semantic structures. Cambridge, MA: MIT Press. Kahn, Patricia 1994 "Genetic diversity project tries again". Science 266:720-22. Kroeber, Alfred L. 1917 "The superorganic". American Anthropologist 19:163-213. Lakoff, George 1987 Women, fire, and dangerous things: What categories reveal about the mind. Chicago and London: University of Chicago Press. Langacker, Ronald W. 1982 "Space grammar, analysability, and the English passive". Language 58:22-80. 1987 Foundations of cognitive grammar. Vol. I. Stanford: Stanford University Press. 1991 Foundations of cognitive grammar. Vol. 2. Stanford: Stanford University Press. Levi-Strauss, Claude 1949 The elementary structures of kinship. Boston: Beacon Press. MacLaury, Maria I. 1989 La Placita: Vantages of urban change in historic Tucson. Tucson: University of Arizona, College of Architecture, Master of Arts Thesis. Pp. 169. [Ann Arbor, MI: University Microfilms International, No. 1339280]. MacLaury, Robert E. 1986 Color in Mesoamerica, Vol. 1: A theory of composite categorization. Berkeley: University of California, Department of Anthropology, Doctoral Dissertation. Pp. 435. [Ann Arbor, MI: University Microfilms International, No. 87!8073]. "Coextensive semantic ranges: Different names for distinct vantages of one 1987 category", in: Barbara Need, Eric Schiller, and Anna Bosch (eds.), Papers from the 23rd Annual Regional Meeting of the Chicago Linguistic Society, Part I: The General Session. Chicago: Chicago Linguistic Society, 268-282. Ricoeur, Paul 1978 "The task of hermeneutics", in: Michael Murry (ed.), Heidegger and modern philosophy. New Haven: Yale University Press.

Preface

xiii

Sapir, Edward 1916 "Time perspective in aboriginal American culture: A study in method." Canada, Department of Mines, Geological Survey, Memoir 90 Anthropological Series 13. Ottawa: Government Printing Bureau. [Published 1963 in: David G. Mandelbaum (ed.), Selected writings of Edward Sapir in language, culture, and personality. Berkeley and Los Angeles: University of California Press, 389-462.] Sperber, Dan 1980 "Remarks on the lack of positive contributions from anthropologists on the problem of innateness", in: M. Piatelli-Palmarini (ed.), Language and learning: the debate between Jean Piaget and Noam Chomsky. Cambridge, MA: Harvard University Press, 244-249. Taylor, John R. 1989 Linguistic categorization: Prototypes in linguistic theory. Oxford: Clarendon Press. Taylor, John R. (ed.) "Language, thought, and culture". South African Journal of Linguistics 10(4). 1992 Tyler, Stephen A. (ed.) 1969 Cognitive anthropology. New York et al.: Hold, Rinehart and Winston.

Introduction: On construing the world John R. Taylor

In his essay "How to Talk: Some Simple Ways", 1 J. L. Austin invites us to imagine a simplified world, and a simplified language with which to talk about it. The "world" proposed in the essay consists of a finite set of discrete items, each bearing a unique name. Each item is assignable to one, and only one, of a finite set of types. The types could be, say, kinds of geometrical figures, while the items could be named by numerals. The syntax permits only sentences of the structure "I is a T", where I is the name of an item, and Tis the name of a type. A possible sentence in the language might be 1227 is a rhombus. "How to Talk" is a curious piece, especially when viewed against Austin's ceuvre as a whole. Austin's genius lay in his sensitivity to "ordinary language", and his philosophical method consisted in elucidating the meanings of words and expressions by considering the kinds of circumstances in which they would ordinarily be used. His purpose, in so doing, was not to provide the "last word" on long-standing philosophical problems but rather to provide the "first word" (Austin 1979: 185). Any examination of a problem, Austin seems to be saying, needs to begin by examining the words which people use in stating, and talking about, the problem. In this way, conceptual confusions can be eliminated at the outset. One result of such an examination might be, for example, the discovery that what is being talked about is not a single, clear concept, but rather a family of related concepts. Now, the "language" that is described in "How to Talk" seems rather remote from ordinary language; the "world" described in the paper likewise bears little relation, it would seem, to the world which human beings inhabit. It comes as a surprise, therefore, when Austin, the "ordinary language philosopher", states that his purpose in the essay is to "elucidate some of our ordinary thought and language about the use of speech" (1979: 132). Equally curious, the reader may be thinking, is the fact that I have chosen to introduce a volume entitled Language and the Construal of the World by mentioning this strange essay, devoted to a "language" radically removed from "ordinary language", used of a "world" which is a parody of the "real world".

2

John R. Taylor

The clue to my intention may be found in a little phrase that Austin smuggles into an introductory paragraph. Austin says that we possibly never are actually in a situation exactly like the one described in his paper, although "more probably we sometimes are so, or, more correctly, regard ourselves for current intents and purposes as being so" (1979: 134; emphasis added). In certain circumstances, then, we may indeed talk as ifthe model in "How to Talk" applied. What might these circumstances be? At the risk, perhaps, of reading into Austin's work an interpretation which he did not intend to put there, I would suggest that the circumstances are those presumed to hold in model-theoretic, or truth-conditional, approaches to semantics. A statement to the effect that "Snow is white"- a statement which is said to be "true" only when the stuff referred to by the word snow happens to exhibit the property denoted by the expression is white-is of the same order as a statement to the effect that "1227 is a rhombus"-where 1227 denotes a uniquely identifiable entity in "the world", is a constitutes an "assertive link", and rhombus denotes one of a finite number of types existing "in the world". The rub, of course, is that, as Austin says, we only "regard ourselves", "for current intents and purposes", as being in a situation characterized in the manner described in "How to Talk". 2 Once these special "intents and purposes" are put aside, it is clear that the "real world" is radically different from the world described in "How to Talk" -not just quantitatively different, in the sense that the real world contains many more items, and many more types, than Austin's model world- but qualitatively different, too. Take a trivial example. Suppose I ask: How many "items" are there on my writing desk at the moment? Well, amongst many other things, there is a box of matches. Does this count as one "item" or two "items" (the tray plus the outer box), or three "items" (the whole and its two parts)? What about the matches inside, or the word LION printed on the match box? Is this word an additional item, or four additional items (i.e., the four letters of the word), or five items (the word plus the constituent letters)? Etcetera, etcetera. The point is, the number of "items", even within such a circumscribed portion of the world as the top of my writing desk, can only be indefinite (cf. Putnam 1988). The same goes for the number of "types" to which these items are to be assigned. How many "types" would a Japanese speaker, or a nonJapanese speaker, want to identify in a page of Japanese script? It is evident that what we identify as "items", and as "types" to which the items are assigned, are functions of our interests, concerns, previous knowledge, etc. -in a word, they are functions of our cognition.

Introduction: On construing the world

3

Here we can take our cue, once again, from Austin. Let us look more closely at the word world. Austin spoke of the "trailing clouds of etymology": [A] word never-well, hardly ever-shakes off its etymology and its formation. In spite of all changes in and extensions of and additions to its meanings, and indeed rather pervading and governing these, there will still persist the old idea. (Austin 1979: 201)

In view of its relevance to the concerns of present-day cognitive semantics (and to the topics of the papers in this book), I cannot resist citing the continuation of the above passage: Going back into the history of a word, very often into Latin, we come back pretty commonly to pictures or models of how things happen or are done . . . . We take some very simple action, like shoving a stone, usually as done by and viewed by oneself, and use this, with the features distinguishable in it, as our model in terms of which to talk about other actions and events: and we continue to do so, scarcely realizing it, even when these other actions are pretty remote and perhaps much more interesting to us in their own right than the acts originally used in constructing the model ever were, and even when the model is really distorting the facts rather than helping us to observe them .... "Causing", I suppose, was a notion taken from a man's own experience of doing simple actions, and by primitive man every event was construed in terms of this model: every event has a cause, that is, every event is an action done by somebody-if not by a man, then by a quasi man, a spirit. When, later, events which are not actions are realized to be such, we still say that they must be "caused", and the word snares us: we are struggling to ascribe to it a new, unanthropomorphic meaning, yet constantly, in searching for its analysis, we unearth and incorporate the lineaments of the ancient model. ... Examining such a word historically, we may well find that it has been extended to cases that have by now too tenuous a relation to the model case, that it is a cause of confusion and superstition. (Austin 1979: 202-203)

To return, then, to the word world. What of its etymology? Old English weorold, or worold, is a compound formation, from Germanic *weraz 'man' (cf. Latin vir) and aid 'age, life-span'. In origin, then, world was the "age of a man", "human life", and the circumstances of a life (Lewis 1967: 214-16). Modern uses of the word have retained the subjective, anthropomorphic perspective. In speaking of my world, the world of a child, the world of Walt Disney, I mean by these expressions not some objectively existing set of circumstances, but a subjective reality, i.e., a set of experiences, impressions, or creations of the imagination. I would even suggest that the very adaptability of the word in linguistic semantics-witness the phrase"possible world" -is itself testimony to the word's inherent subjectivity.

4

John R. 'li1ylm·

In hricl'. then, the "world" is not something objectively given, it is something "construed" by human cognition. It is "construals of the world" that are properly regarded as the object of linguistic semantics. Jackcndoff expressed this view as follows: We must take issue with the naive position that the information conveyed by language is about the real world. We have conscious access only to the projected world -the world as unconsciously organized by the mind; and we can talk about things only insofar as they have achieved mental representation through these processes of organization. Hence the information conveyed by language must be about the projected world. (Jackendoff 1983: 29; author's emphasis).

In contrast to Jackendoff's notion of "projected world", the notion of "construal" implies a more active role of the language user in organizing and structuring his or her world. The term is taken from Langacker (1987: 487-488), who defines it as "the relationship between a speaker (or hearer) and a situation that he conceptualizes and portrays". There are several aspects to the construal relationship, as understood by Langacker: 1. Firstly, and rather obviously, a speaker may vary the detail, or specificity, with which a scene is portrayed. We can describe an object as being "red", or, with greater specificity, "bright red", or "pinkish red"; or, less specifically, simply as "colored". We can say of a person that he is "running", or, characterizing his activity more precisely, that he is "jogging", or "sprinting". And, of course, a vast number of components of a scene are simply ignored in our linguistic accounts of it. ii. Related to specificity is the degree of precision with which a situation is characterized. I may say, without intent to deceive, that it is now half past twelve-although, if pressed, I would have to admit that it is not exactly half past twelve, but, say, twelve thirty-three. Lakoff (1972) and Kay (1983) have studied the phenomenon of "loose speaking", as did, of course, Austin, in chapter 11 of How To Do Things With Words. 111. The use of a linguistic form may evoke certain background assumptions. In describing a person as a "bachelor" (or "spinster"), I convey more than just the fact that the designated person is an adult male (or female) who has never married. The words invoke a "theory" (what Lakoff 1987 calls an ICM, or "idealized cognitive model") of "bachelorhood", or "spinsterhood". The model contains such components as the notion of a "marriageable age", and "explanations" of

Introduction: On construing the world

5

why it is that a person passes the marriageable age without marrying (Taylor 1989: 97). iv. The use of certain linguistic expressions may suggest a construal of a situation in terms of something else. Nonliteral expressions (metaphor, metonymy) are, of course, prime examples of this phenomenon. To say that a person "exploded with anger" is to convey, not just that the person was very angry, but also presupposes a conceptualization of anger in terms of a pressurized fluid inside a container. But the borderline between the "metaphorical" and the "literal" is fuzzy. Even the choice of a grammatical construction can impose a construal in terms of something else. A nominalization, such as the crossing of the river by the two men, differs from the two men crossed the river to the extent that in the former case the event is presented as an a temporal "thing", in the latter case as a temporal "process". v. Finally, all linguistic coding incorporates perspective, where "perspective" is being used in a rather broad sense, to cover a number of related notions, such as figure/ground organization, deixis, and viewpoint. Necessarily, in any construal of a scene, certain components are foregrounded whilst others serve as reference points for the characterization of the foreground. The picture above the sofa and the sofa below the picture could well be truth-conditionally equivalent. The difference is that the first expression locates the picture (figure) with . reference to the sofa (ground), while in the second expression the relations are reversed. Deixis is a familiar notion, having to do with the presentation of a scene from the location of an observer, usually, but not necessarily, the speaker. Finally, the notion of "viewpoint" may be defined as the "mental route" that a speaker takes in presenting a scene. The sentence pair The roof slopes steeply upward and The roof slopes steeply downward (Langacker 1990: 157) illustrates the phenomenon. In the first case, it is as if the speaker scans the scene from the bottom of the roof upwards, while in the second he scans from the top downwards. Another example of the same phenomenon is that I can designate one and the same person as John's wife, Jill's daughter, Jeff's mother, and so on. In each case, I designate the person via some other person, who bears a salient relation to the designated person. This last example also shows that the very choice of "type" ("wife", "daughter", "mother") to which an "item" is assigned-i.e., the way in which an entity is categorized -testifies to a certain perspective on the part of the speaker.

6

John R. Taylor

The above by no means exhaust the notion of "construal". (Even from the few examples that have been given, however, it will be clear that the "same" situation may be construed in many different ways, in an indefinite number of different ways, in fact.) Some further aspects of the phenomenon, and the kinds of issues it raises, will be the topics of the papers in this book. While focusing on different aspects of the construal relation, all the papers have this in common: each rejects the "objectivist" picture of language and the world described (parodied?) in Austin's "How to Talk", by showing how "objective" situations are mediated by the speaker's cognitive processing. Casad, basing himself closely on Langacker's work, offers a detailed exemplification of the notion of construal. Casad makes several crosslanguage comparisons which strikingly document how one and the same state of affairs may be construed very differently in different languages. Cross-language differences are touched upon in several papers (Langacker, Dirven, Matsuki). In this connection, Casad draws attention to the conventionality of linguistic resources. Casad is therefore reluctant to equate "semantics" with "conceptualization" tout court. (Such would seem to be the position of Jackendoff 1983: 17, for whom "semantic structures" constitute a subset of "conceptual structures".) Admittedly, Casad does not make a semantic versus conceptual split, as does Bierwisch (e. g., Bierwisch 1981). Casad emphasizes, however, that not every, or any, conceptualization has the status of the meaning of a linguistic expression. Rather, conceptualizations need to be "shaped", or "structured", so as to permit symbolization by the resources made available by a given language. Such a notion, while clearly "Whorfian" in spirit, does not entail that conceptualization is determined by language, since the symbolic resources of a language do not constitute a fixed inventory. On the contrary, speakers can, and do, exercise considerable creativity in extending and adapting the symbolic potential of their language. Langacker deals with the semantics of possessive constructions in English, and some other languages. In spite of the name traditionally given to these constructions, it is evident that possessive constructions may be used to denote a very large number of semantic relations, in addition to a relation of "possession", narrowly understood in the sense of legal ownership. Alongside John's car, we have the dog's tail, the girl's uncle, the plane's departure, and Kennedy's assassination. It has even been suggested that the possessive relation can be "any relation at all" (Williams 1982: 283). Yet there are clearly constraints on the relation between possessor and possessed. Symptomatic is the fact that the nominals in a

Introduction: On construing the world

7

possessive expression may not ordinarily be reversed. We have the relation of a whole to a part (the dog's tail), but not the relation of a part to the whole (?the tail's dog), we have the relation of a participant to an event (the plane's departure), but not the relation of an event to a participant (*the departure's plane). The essential point, Langacker maintains, is that the possessor must be construable as a "reference point entity". This notion encapsulates the insight that in order to conceptualize (or "establish mental contact with") an entity x, we frequently, or typically, need first to conceptualize an entity y; y serves as a "landmark" on the mental path we need to trace in order to "locate" entity x. We do not ordinarily think of, or encounter, detached tails. A tail is a part of an animal, and is generally thought of as such. It is normal, therefore, that in referring to a "tail", we do so via reference to the animal of which the tail is a part. Animals, in contrast, are "conceptually autonomous" entities, in the sense that we can conceptualize an animal without making necessary reference to any other kind of entity. We certainly do not (normally) have to first establish mental contact with a tail, in order then to refer to the animal of which the tail is a part. The notion of reference point is inherently subjective, and anthropocentric. It invokes the mental path taken by a human mind in conceptualizing a given entity. And if human beings had been created differentlyif we were of Lilliputian dimensions, say, or disembodied spirits-the range of potential reference points would no doubt also be very different! Tsohatzidis shows how examination of a single word -the English verb lack-can validate a number of points that have been made so far. Tsohatzidis argues that a statement to the effect that "x lacks y" not only conveys the fact that "x does not have y", the statement also communicates the presupposition that x may be categorized as a peripheral, or untypical member of a category C. The peripheral status of x inC rests, precisely, on the fact that x does not "have" the property, attribute, possession, etc. denoted by y, in contrast to the central, or prototypical members of C, which do have this property, attribute, etc. The central point that Tsohatzidis makes is that the membership of x inC is not something that is independently, or antecedently, given; there is, as he puts it, no "Great Book" in which the allocation of entities to types has been fixed for all times and for all purposes. Categorization is inherently unstable and context-dependent; it is a product, namely, of how a speaker, in harmony with his present concerns and the direction of his present thoughts, chooses to construe a certain state of affairs.

8

John R. 1itylor

Dirven addresses the concept of "cause". Now, it is traditional, in linguistic semantics, to treat cause as a unitary concept, an undifferentiated semantic primitive, that is incorporated into the meanings of many predicates. True, Miller and Johnson-Laird (1976: 489) recognize that cause is "an extremely complex concept"; they even go some way towards explicating people's conceptions of causal relations. Even so, they still persist in treating cause, along with other notions like "act" and "do", as semantic primitives-or if not as primitives, then as functors of constant semantic value. (Jackendoff 1991 proceeds no differently.) Consider, for example, Miller and Johnson-Laird's definitions, or decompositions, of the verbs permit (1976: 511) and eat (1976. 518): PERMIT (x, y, z): Someone x "permits" someone y to do z if: (i) ACT(x, S)

(ii) PERMISSIBLE {y, DO(y, z)) (iii) CAUSE (S, (ii)) EAT (x, y): An animate x "eats" something y if there is a z such that z is the mouth of x and: (i) ACT(x, S)

(ii) CAUSE(S, (INTO (THROUGH (TRAVEL))))(y, z, x)

Note that the proposed definitions entail that permit and eat incorporate semantic components which are exactly the same, viz. "act" and "cause", even though few people, I imagine, would see much commonality between an "act" of permitting, and an "act" of eating. Casad fails to be convinced by Jackendoff's (1983) claim that butter the bread and it's raining both incorporate the notion "go". So, too, we migh,t wonder whether I ate a sandwich really means anything at all like "I acted in such a manner that my action caused a sandwich to travel through my mouth into me". In brief, the above definitions manifestly fail to capture the many-faceted nuances of the concepts "permit" and "eat". The topic of Dirven's chapter is the expression of causal relations by the English prepositions. His study leads him to conclude that what we loosely call "cause" in fact comprises a family of distinct cause concepts. (Austin was making a similar point in the passage cited earlier.) Furthermore, there may not be a single, overarching concept of cause encoded in English (or at least, not by the prepositions). Comparisons between English and two languages closely related to English, viz. Dutch and German, lead to some further conclusions, namely, that even these closely related languages do not share the same cause concepts, and that the "same" objective situation may be construed in terms of different cause

Introduction: On construing the world

9

concepts in the different languages. It may even happen that a situation that in one language is construed in terms of cause, may be construed in another language in terms of a noncausal relation. There is, of course, a much wider issue at stake here, namely the degree to which syntax, i.e., the form of complex expressions, is in general motivated by semantics. Actually, in the debate between those who claim that syntax is semantically motivated, and those who claim that syntax is essentially arbitrary, it has to be admitted that the playing field is by no means level. Proponents of the arbitrariness position apparently feel that they need do no more than cite a syntactic fact, declare that there is no semantic or conceptual reason why the fact should be as it is, and so rest their case. Such claims are remarkably frequent in the literature. To cite but one example, taken from a book selected at random from my bookshelf, Matthei and Roeper, in their introduction to psycholinguistics, state that "there is nothing about the meaning of the verb sleep that would lead us to predict that the sentence *John slept the bed is ungrammatical. It is simply an idiosyncratic fact about the verb sleep in English that it is intransitive and does not appear with a direct object" (1983: 178). I have not cited this example in order to claim that Matthei and Roeper are wrong. They might well be right. But to legitimize this conclusion, one would first have to make some serious attempts to come up with a semantic explanation of the phenomenon in question. (The work of Wierzbicka [especially Wierzbicka 1988] is an outstanding example of how many apparent idiosyncrasies in syntax are, on careful examination, subject to semantic explanations.) Only if such attempts manifestly fail is it justified to say that the intransitive status of sleep is "an idiosyncratic fact" about the verb. Dirven's chapter shows how detailed semantic analysis can counter the widely held belief that preposition selection is essentially arbitrary, and of little semantic significance. But Dirven is concerned not just to motivate preposition selection, he also wishes to derive from the evidence of linguistic usage insights into the nature of the concept(s) of cause. It is an enterprise with which Austin would certainly have been sympathetic. Austin's method of enquiry, he notes somewhat disarmingly, is motivated by a "superstition", namely, that "the forms of words and expressions are highly significant for their meaning" (Austin 1979: 281). Present-day linguists may quibble over some of Austin's semantic analyses. (For these, the reader is referred to Austin's essays.) But Austin's point was simply that if we are to talk about "cause", or whatever, we need first to ascertain what we are talking about. And the linguistic evidence, Dirven suggests, indicates that "cause" covers a large number of different concepts.

I0

John R. Taylor

Heine also deals with the compositional motivation of complex expressions. Specifically, he addresses the diachronic origins of orientational expressions in Mrican and Oceanic languages. By assembling data from a large number of languages, he is able to arrive at conclusions which would not be available from the examination of a single language. His data show a very strong trend for certain body-part terms, such as words for "head", "face", or "buttocks", to be pressed into service to denote such spatial notions as "top"/"on top of', "front"/'in front of', "bottom" /"underneath". The motivation for these semantic extensions, clearly, is a universal tendency for human beings to construe nonhuman entities in terms of the model provided by the human body, or, in some cases, an animal body. Heine also raises the possibility that environmental factors may be relevant to the use of animal body-part terms. It is predominantly the languages of nomadic peoples-for whom animal husbandry is a central activity-which tend to employ a "zoomorphic" rather than a "human" model for the derivation of orientational terms, as revealed by the extension of "back" to denote, not "behind", but "on". Three papers, by Kovecses, Matsuki, and Geeraerts and Grondelaers, deal with emotions, specifically, with the emotion of anger. Like cause, emotion (or more exactly, the possibility of talk about emotions) raises old philosophical problems. The central issue is the "problem of other minds". I may presume to know the emotional state that I am in (by definition, almost, one might say), but I have no direct acquaintance with another person's emotions. How then do I know that when other people talk about "anger", they are referring to the same kind of emotion that I am referring to by the word anger? Indeed, one can even ask, how do I know that other people have emotions at all? And yet, we need to assume that people do have emotions, and do mean more or less the same thing by psychological predicates, otherwise there could be no basis for the learning of these words, or for judging that the learner has successfully acquired the words. I would like to mention just two solutions to this conundrum that have been put forward in the philosophical literature. First, let us turn, once again, to Austin. In his essay "Other minds", Austin claims that "to be angry" is not just to be in a certain emotional state. Anger involves a whole "pattern of events" -what we would nowadays, following Schank and Abelson (1977), call a "script". It seems fair to say that "being angry" is in many respects like "having mumps". It is a description of a whole pattern of events, including occasion,

Introduction: On construing the world

11

symptoms, feeling and manifestation, and possibly other factors besides. It is as silly to ask "What, really, is the anger itself?" as to attempt to fine down "the disease" to some one chosen item ("the functional disorder"). That the man himself feels something which we don't (in the sense that he feels angry and we don't) is, in the absence of ... telepathy, evident enough, and incidentally nothing to complain about as a "predicament": but there is no call to say that "that" ("the feeling") is the anger. (Austin 1979: 109)

Just as, say, the "restaurant script" evokes the series of events which typically occur in going to a restaurent, so the "anger script" activates knowledge of the sequence of events associated with anger (such as provocation, desire for retribution, attempts at suppression, acts of retribution, return of equilibrium; see Kovecses 1986). Knowledge of the script makes it possible for a person to assign "default values" to unobserved episodes. Thus, by observing certain outward signs, a person can infer the entire pattern of events, including the mental state of the angry person: [I]t is our confidence in the general pattern that makes us apt to say we "know" another man is angry when we have only observed parts of the pattern: for the parts of the pattern are related to each other very much more intimately than, for example, newspapermen scurrying in Brighton are related to a fire in Fleet Street. (Austin 1979: 109~110)

Austin concedes that a person who has never himself experienced the script, i.e., has never been angry, is unlikely to be able to recognize the scenario (1979: 104). The script must therefore be learned from experience. But in this respect, recognizing that a person is angry is in principle no different from recognizing that there's a bittern at the bottom of the garden- both activities presuppose that I have "learned to recognize or tell" bitterns (or, as the case may be, anger) (1979: 80). For the second treatment of the "problem of other minds", let us turn to Fodor (1981 ). According to Fodor, we attribute psychological states to a person in much the same way that a physicist infers the path of a charged particle from the observation of tracks in a cloud chamber. The physicist does not see the moving particle, no more than I experience another person's psychological state. Neither has the physicist defined the path of a particle in terms of the track it leaves, such that observation of the track is a criterion for the presence of a moving particle (particles do not have to leave tracks as they move). Likewise, pace Wittgenstein, we do not define psychological states in terms of their behavioral manifestations (whereby a verbal report of the psychological state might also count as a "behavioral manifestation"). Nor, Fodor argues, does the

12

John R. Taylor

physicist infer the movement of a particle on the basis of a presumed correlation between observed tracks and movements of particles. For to suppose such a correlation requires prior knowledge of the particles' movement-the very point that is at issue. By the same token, we cannot attribute mental states to a person in virtue of a correlation between overt behavior and the inner state, since in the absence of prior access to other people's inner states, we have no basis on which to validate the correlation in the first place. Rather, the physicist infers the movement of the particle on the basis of a scientific theory, i.e., in terms of the "simplicity, plausibility, and predictive adequacy of an explanatory system as a whole" (Fodor 1981: 56). In much the same way, Fodor argues, we attribute psychological states to a person on the basis of a theory (or theories) of psychological states: Much everyday conceptualization depends on the exploitation of theories and explanatory models in terms of which experience is integrated and understood. Such pre-scientific theories, far from being mere functionless "pictures," play an essential role in determining the sorts of perceptual and inductive expectations we form and the kind of arguments and explanations we accept. (Fodor 1981: 62)

To say, then, that a person "is angry" is to make a "theoretical inference" from observed behavior (including verbal reports) to "underlying mental occurrences" (1981: 61). The papers by Austin and Fodor are highly programmatic in nature. Austin, for example, makes only schematic reference to the actual contents of the anger script. Fodor does not offer us even the outline of a "conceptual theory"; he merely states that people have such theories. A further point seems relevant. Austin appears to take it for granted that two people, each experiencing anger, will construct on the basis of their experiences the very same script. Likewise, Fodor seems to assume that the theories that people have of psychological states are, if not identical, then at least commensurate. The possibility exists, however, that different people, or groups of people speaking different languages, or people belonging to different cultural traditions, or people living in different historical epochs, may actually turn out to have rather different "theories" about their emotions. Theories of anger are the subject of the three papers in this book. In previous publications, Kovecses examined the expressions that American English speakers use in talking about anger, and other emotions (e. g. Kovecses 1986, 1990). Many of these are clearly metaphorical in nature

Introduction: On construing the world

13

(I was boiling, He exploded, She was seething with rage, etc.). Drawing on the methodology of Lakoff and Johnson (1980), Kovecses related these metaphorical expressions to a small number of underlying "conceptual metaphors". Prominent amongst these is the conceptualization of the body as a "container" for the emotions, and the conceptualization of anger, specifically, as the heat of a fluid in a container (i.e., the body). Significantly, the "logic" inherent in this construal of anger makes it possible to elaborate an "anger script", i.e., a prescription, not only of sequential stages of anger, but of possible courses of action that an angry person can take. Increasing the heat of a liquid in a closed container causes an increase in pressure inside the container. Unless the pressure is released, or otherwise reduced, the container will explode. To prevent the inevitable "explosion", the angry person needs to "cool it", or "simmer down"; he must "let off steam", or "release" his anger in some other way (e. g., through an act of retribution), after which the initial state of equilibrium is regained. Arguably, the anger is heat metaphor has a physiological base. There is, namely, evidence that anger does cause a slight (actually, a very slight) increase in body temperature (Ekman- Levenson- Friesen 1983). This grounding of the metaphor in a physiological process could be the reason why so many different cultures employ (variants of) the anger is heat metaphor. Matsuki documents the presence of this metaphor in Japanese. Her data, however, testify to an additional, and language-specific theory of anger, one which, again, displays a remarkable internal logic. Hara, the stomach and bowel area, is the source of anger, whereby hara can stand metonymically for the emotion itself. As anger increases, hara rises up to the chest/heart region mune, the location of nausea. Finally, anger rises to the head region atama, the seat of rationality. According to the logic of this schema, the angry person goes through a stage of inner conflict and frustration (typified as "nausea"), during which the person still retains rational control over his actions, to a final stage of irrational behavior, when anger has taken over the centre of rationality. Geeraerts and Grondelaers propose a radical reanalysis of Kovecses's original data. They argue that the heat and fluid metaphors in English could well be reflexes of a highly elaborated theory of emotion, one that held sway over popular and scholarly thought in the West for over 2000 years, namely, the theory of the four humors. As Geeraerts and Grondelaers document, this now discarded and largely forgotten theory was not only the basis of premodern medical practice, it was integrated into a

14

John R. Taylor

whole cosmology, with the establishment of correspondences between the humors and such diverse domains as the plant and animal kingdoms, dietary practices, the seasons, and the planets. Thus it was that "choleric" persons, i.e., persons prone to anger, were recommended to eliminate garlic and ginger from their diet. Anger, or a proclivity to anger, was believed to be caused by excess of yellow bile (or "choler") in the body. Choler was taken to be a warm and wet substance. This characterization of choler, according to Geeraerts and Grondelaers, is sufficient to explain the dominance of the hot liquid metaphors in discourse about anger. (The role of blood, more specifically, hot blood, is due to the belief that the humors circulated in the body as admixtures of blood.) With the general demise of the humor theory, a number of things could, and (according to Geeraerts and Grondelaers) did happen: Some expressions, which obviously and explicitly invoke the theory (such as stir one's bile), have begun to fall into disuse. n. Other expressions have been reinterpreted, possibly in terms of the heat in a container metaphor documented by Kovecses. Thus, my blood is boiling (probably) no longer invokes the humor theory, as it once may have done. iii. "Reinterpretation" may involve the metaphorical construal of expressions that once may have been understood quite literally. In terms of the humor theory, a choleric person was quite literally hot-blooded. iv. Yet other expressions might persist, as quaint, uninterpretable relics of intellectual history. Such, for example, is the status of the belief, still current, perhaps, that masturbation can cause blindness. 1.

The more general, methodological point that Geeraerts and Grondelaers make is that theories, being cultural phenomena, are the product of historical processes. Kovecses counters by pointing to the ubiquity of the hot liquid metaphor, not only in languages influenced by classical antiquity and mainstream European thought, but also in Chinese and Japanese. Kovecses also raises the question whether the humor theory itself may not be grounded in the very same range of physiological experiences as the hot liquid metaphor. Such a supposition would mean that an expression like my blood is boiling, at least in earlier centuries, was multiply motivated- by physiological experience, and by the explicit, expert theory of the humors. The topic for further research, as Geeraerts and Grondelaers note, is to trace the origins of the humor theory itself, and its for-

Introduction: On construing the world

15

tunes in popular, as well as in official culture, in the West, from ancient Greek up to modern times. At least in Western culture, emotionality is commonly opposed to rationality. Thus we have the well-known dichotomies of feeling and thinking, the heart and the head. If the papers by Kovecses, Matsuki, and Geeraerts and Grondelaers deal with the conceptualization of emotions, Jakel addresses the construal of mental processes themeslves. Using the technique popularized by Lakoff and Johnson (1980), Jakel examines a broad range of everyday English expressions that have as their subject matter various mental activities and attributes: intelligence, understanding, thinking, problem solving, remembering, and so on. Lakoff and Johnson had identified the vision metaphor as the dominant metaphor of cognition: understanding is seeing. Jakel's detailed investigation suggests that the vision metaphor may in fact be rather peripheral. He perceives in everyday talk of mental processes and attributes another metaphor, or cluster of metaphors, suggesive of a rather elaborate folk theory of the mind. The dominant metaphors construe the mind as a workshop-a place in which ideas are objects, where thought is the manipulation of objects, where intelligence is the sharpness of the tool used to process the objects, memory is a storehold for objec'ts, remembering is retrieving objects from the storehold for use in the workship, and so on. Professional psychologists and cognitive scientists, of course, have their own metaphors, drawn predominantly from computer science. The brain is construed as computer hardware, the mind as computer software, and thinking is the running of the software program on the hardware. (For a presentation, and radical critique of this metaphor, see Searle 1992, especially chapter 9). Jakel discovered a much more "homely" metaphor, or cluster of metaphors, underlying our talk of the mind. Moreover, taking a glance at some other languages, both European and nonEuropean, he surmises that the mind as a workshop metaphor enjoys considerable cross-linguistic, and cross-cultural validity. Looking into the etymology of some cognitive predicates in the modern languages, he further suggests that the mind as a workshop metaphor was dominant in earlier times, too. Lakoff and Johnson (1980) had noted that metaphors are not always fully coherent, that is to say, it is not always the case that every component of the source domain can be projected onto some aspect of the target domain. Take, for example, the rather general spatial metaphor, according to which we construe the future as being in front of us, and the past as being behind us. In one important respect, the metaphor is at

16

John R. 1l1ylor

odds with our experience of time, in that we can see what is (literally) in front of us, but we cannot (metaphorically) see what is in the future. No doubt, in this case, the discrepancy between the logic of the source and target domains is so crass that no speaker who employs the metaphor is likely to be misled by it. But consider some entailments of the workshop metaphor. If the mind is a workshop, and ideas are objects that are processed in the workshop, what corresponds to the workman in the workshop? Who is it that wields the tool ("intelligence") to shape the object ("ideas")? Who places the objects in the storeroom ("memory"), and retrieves them at a later date? If we take the workshop metaphor too seriously (one feels tempted to say: if we take the metaphor too literally), one has to postulate a homunculus in the mind/workshop who performs all these various activities. The computational metaphor of mind raises the same question, of course: If the brain is a digital computer, who is the user? (Searle 1992: 214). Perhaps the "homunculus fallacy" (Searle 1992: 212) crops up at so many turns precisely because of the prevalence of the workshop metaphor, and because the metaphor has indeed been taken "literally"? MacLaury presents a succinct account of a sophisticated and elaborated theory of construal, which he has named "vantage theory". Vantage theory developed out of Mac Laury's work on color categorization. Color terminology has long held a privileged position in semantic research, for a number of reasons. In the first place, the physical properties of a color stimulus are amenable to precise, quantifiable description. In contrast with many other domains of experience, therefore, it is possible to obtain precise correlations between an "objective" property, and the name which a person attaches to it. Three other factors have contributed to the attractiveness, to linguists, psychologists, and anthropologists alike, of color research: 1. The color spectrum is continuous. ii. Languages differ in their categorization of the spectrum. iii. The neurological processing of color stimuli in the retina and optical nerve is fairly well understood. In particular, there are good reasons to suppose that the human visual system is predisposed to perceive a set of "pure colors", the so-called "unique hues".

Factors (i) and (ii), taken together, used to be cited as evidence for the Saussurian doctrine of the arbitrariness of the linguistic sign. How a person divides up the color spectrum- how many color categories he recog-

Introduction: On construing the world

17

nizes-was said to be determined by the color categories conventionally, and arbitrarily, made available by his language. The arbitrariness thesis is in conflict with (iii), i.e., with the fact that the human visual system is predisposed to recognize a number of focal points on the spectrum, i.e., "pure red", "pure green", and so on. Berlin and Kay's (1969) thesis that the languages of world follow the very same path in differentiating up to eleven basic colors, only partially resolves the conflict. For if every person with normal vision "sees" the very same focal colors, why do speakers of different languages not categorize the color spectrum in exactly the same way? MacLaury's answer invokes a further factor in color categorization, over and above properties of the stimulus and properties of the visual system, namely cognition. Categorizing a stimulus involves "placing" the stimulus with respect to a cognitive reference point. In so doing, a person can attend primarily to the similarity between the stimulus and the reference point, or to the distinctiveness of the stimulus vis-a-vis the reference point. Attention to distinctiveness may result in the creation of a new reference point, which itself serves as the locus of judgements of similarity or distinctiveness. In terms of color categorization, the reference points are, primarily, the unique hues. Maximum attendance to similarity divides the spectrum into only two broad categories, the "warm" colors, and the "cool" colors. Increased attendance to distinctiveness gives rise to a progressive differentiation of the color spectrum, essentially along the lines hypothesized by Berlin and Kay. MacLaury speculates that attention to distinctiveness may correlate with technological complexities, with rapid social change, as well as with extreme hardship in eking out an existence. The interplay between attention to similarity and attention to distinctiveness is nicely illustrated on the example of "coextensive" color terms where a person uses two terms to name the members of a single color category. The two terms are not obviously in contrast, nor are they strict synonyms. One of the terms-typically, the older and better established one in the language-tends to have its focus in the center of the category. Colors are named by this term to the extent that the speaker attends to similarity. MacLaury refers to this as the "dominant" term. The other term-typically, a more recent innovation in the language-tends to be focused at the periphery of the category. This "recessive" term testifies to the speaker's attendance to distinctiveness within that portion of the spectrum. Typically, only a restricted segment of the category is named by this term. However, if pressed, speakers are prepared to extend the

18

John R. Taylor

recessive term to encompass more and more of the items named by the dominant term. The spatial metaphors in the above account are not fortuitous. MacLaury draws extensive analogies between color categorization and the tracking of location in space. We typically give the spatial location of a moving object vis-a-vis a fixed reference object. The reference object constitutes a "fixed coordinate", whereas the located object constitutes the "mobile coordinate". But the located object itself may come to serve as a reference point for the location of a further entity, and so on. This same mechanism, MacLaury suggests, underlies the very process of categorization, and in the second part of his paper he suggests some applications of vantage theory outside the domain of color. Vantage theory turns out to offer an insightful account of some otherwise highly puzzling phenomena. For example, similarity, for the logician, is a symmetrical relation; if A resembles B, then B resembles A. It would be incoherent to say that A resembles B "more", or "less", than B resembles A. Yet precisely such an asymmetry holds in the case of degrees of perceived resemblance between a peripheral and a central member of a category. With respect to the category of birds, for example, a duck is reckoned to be more similar to a robin than a robin is to a duck. This asymmetry is predicted by vantage theory. With the central member as reference point, attention to similarity causes the perceived size of the category to contract. With a more peripheral member as reference point, attention is on differences, causing the perceived distance between members to expand. Three further papers exemplify applications of vantage theory. Hill and Mac Laury trace in Aztec accounts of the Spanish conquest the emergence of a distinctiveness vantage in the representation of Aztec rulers. Montezuma, the defeated ruler, is represented not in terms of stereotypical speech and external behavior, but in terms of his unique inner states and his inability to speak and act. The paper also, incidentally, documents the cultural specificity of even such an apparently basic and universal concept as "person". Aoyagi addresses the "speech styles" of Japanese. These, as is well known, reflect the formality of a situation, and the relations of power and distance between speaker and hearer. These parameters, however, are not set by some mechanistic algorithm, but according to a speaker's personal, and inherently flexible, construal of the discourse situation; they constitute a speaker's subjective vantages. Lansing offers a vantage theory account of Aristotle's theory of forms. He claims that the theory was successively "corrupted", first by Porphyry, then by Linnaeus, to survive in latter-day feature theories of structuralist linguistics.

Introduction: On construing the world

19

What was initially a way of delimiting species, became, for Porphyry, a taxonomic tree; the nodes of the tree then became, for Linnaeus, divinely ~.:reated kinds; while for the structuralists, the higher kinds became abstract semantic features. In conclusion, Love gives a final twist to the theme of the volume by turning the notion of construal onto the very subject matter of linguistic enquiry, i.e., "language", or "languages". He suggests that current notions of what (a) language is have been profoundly influenced by specific cultural and sociopolitical developments in the West. Foremost amongst these was the emergence of nationalism, of nation states, and the attendant phenomenon of language standardization. A more general factor, which not only enabled the process of standardization to take place, but which significantly affected our concept of language itself, has been the invention of writing and the spread of literacy. On the one hand, one might want to say that the existence of literacy is a precondition for progress in any scientific or intellectual enquiry. But when the object of study is, precisely, language, the fact of literacy subtly changes our conception of the object of study. Literacy makes it possible for an utterance to be fixed, and contemplated outside of the communicative context in which it was produced. Thus arises the notion that there exists something more abstract and intangible than the utterance, some underlying invariant, of which the utterance is but an instance. Love reminds us that this notion, so central to linguistic theory, is itself the product of culture and history. The papers in this volume have in common a rejection of objectivist semantics that Austin sketched in his essay "How to Talk". But over and above their sharing of this privative feature, the papers, in their different ways, each promote the notion of construal, i.e., the relationship between a speaker (or hearer) and a situation that he conceptualizes and portrays, as an important parameter oflinguistic meaning. Some (e. g., Tsohatzidis, MacLaury) have focused on the role of cognitive events inside the individual language user; others (Casad, Dirven, Matsuki, Jakel, etc.) on the role of patterns of construal made available by the conventional resources of a language; while the influence of sociocultural factors on these linguistic conventions is documented by Geeraerts and Grondelaers, and others. Taking a broader perspective still, Langacker, Heine, and MacLaury touch on presumably universal aspects of construal. I trust that this collection of papers will give the reader an insight into some of the exciting, even if (for some) unorthodox, directions which current semantic research is taking.

20

John R. Taylor

Notes I. References to Austin's essays are to Austin (1979). 2. Austin's attitude to "truth" is ambivalent. Note the opening sentence of How to do things with words: What I shall have to say here is neither difficult nor contentious; the only merit I should like to claim for it is that of being true, at least in parts. (Austin 1989: I) In the final chapter, however, he admits to an inclination "to play Old Harry" with what he calls the "true/false fetish" (1989: !51). With this curious phrase, Austin seems to be saying that while "truth" may well be one important aspect of linguistic semantics, it is not the only one, and that a fixation on truth can cause us to ignore these other aspects.

References Austin, John L. 1979 Philosophical papers. (3rd edition.) Oxford: Clarendon Press. 1980 How to do things with words. Oxford: Oxford University Press. Berlin, Brent-Paul Kay 1969 Basic color terms: Their universality and evolution. Berkeley: University of California Press. Bierwisch, Manfred 1981 "Basic issues in the development of word meaning", in: Werner Deutsch (ed.), The child's construction of language. London: Academic Press, 341-387. Ekman, Paul-Robert W. Levenson-Wallace V. Friesen 1983 "Automatic nervous system activity distinguishes among emotions", Science 221: 1208-1210. Fodor, Jerry A. 1981 Representations: Philosophical essays on the foundations of cognitive science. Brighton: Harvester Press. Jackendoff, Ray 1983 Semantics and cognition. Cambridge, MA: MIT Press. 1991 Semantic structures. Cambridge: MA: MIT Press. Kay, Paul 1983 "Linguistic competence and folk theories of language", Proceedings of the annual meeting of the Berkeley Linguistic Society 9: 128-137. Kovecses, Zoltan 1986 Metaphors of anger, pride, and love. Amsterdam: John Benjamins. 1990 Emotion concepts. New York: Springer. Lakoff, George 1972 "Hedges: A study in meaning criteria and the logic of fuzzy concepts", Papers from the eighth regional meeting of the Chicago Linguistic Society, 183-228. 1987 Women, fire, and dangerous things: What categories reveal about the mind. Chicago: University of Chicago Press. Lakoff, George- Mark Johnson 1980 Metaphors we live by. Chicago: University of Chicago Press. Langacker, Ronald W. 1987 Foundations of cognitive grammar. Vol. 1. Stanford: Stanford University Press. 1990 Concept, image, and symbol: The cognitive basis of grammar. Berlin. Mouton de Gruyter.

Introduction: On construing the world

21

Lewis, Clive S. 1967 Studies in words. (2nd edition.) Cambridge: Cambridge University Press. Matthei, Edward- Thomas Roeper 1983 Understanding and producing speech. London: Fontana Paperbacks. Miller, George- Philip Johnson-Laird 1976 Language and perception. Cambridge, MA: Belknap Press. Putnam, Hilary 1988 Representation and reality. Cambridge, MA: MIT Press. Schank, Roger- Robert Abelson 1977 Scripts, plans, goals and understanding. Hillsdale: Lawrence Erlbaum. Searle, John R. 1992 The rediscovery of mind. Cambridge, MA: MIT Press. Taylor, John R. 1989 Linguistic categorization: Prototypes in linguistic theory. Oxford: Clarendon Press. Wierzbicka, Anna 1988 The semantics of grammar. Amsterdam: John Benjamins. Williams, Edwin 1982 "The NP cycle", Linguistic Inquiry 13: 277-295.

Seeing it in more than one way Eugene H. Casad

1. Introduction Langacker's view of semantics, as well as that of Lakoff and others, is that semantics does not reflect objective reality, but rather is subjective in nature (cf. Lakoff 1987, 1990; Johnson 1987; Casad 1988a; Giv6n 1989). 1 Among others things, semantics incorporates what Langacker terms "conventional imagery" (Langacker 1987, 1988: 6, 1990b, 1991). By this it is meant that semantics takes its form from the whole panoply of alternate ways that speakers have for conceptualizing situations. The speaker's ability to conceptualize situations in a variety of ways is, in fact, the foundation of cognitive semantics (cf. Langacker 1991: 294). Note that this characterization does not specifically refer to visual or perceptual imagery, although of course, in many instances the particular semantic value of an expression is in part determined by these factors. Thus, as both Lakoff and Johnson discuss in detail, much of semantics is experientially based (Lakoff 1987; Johnson 1987). Many sentences that have identical truth values and are therefore "objectively the same" are nonetheless distinct semantically (Tuggy 1981: 68-69; Vandeloise 1985: 40). In this paper, I look at a range of data from English, Spanish and Cora, a Uto-Aztecan language of Mexico, in order to substantiate those points. 2 The analyses are presented within the framework of cognitive grammar as developed by Ronald W. Langacker, exemplified by such recent works as Langacker (1987, 1990a, 1990b, 1991) and by an increasing number of his associates (cf., for example, Achard 1993, to appear; Cook 1988; Hawkins 1984; Lindner 1981; Manney 1993; Rice 1987; Smith 1987, Tuggy 1981; Vandeloise 1985, to appear and van Hoek 1992, to appear). In particular, I focus on certain kinds of data to show that one of the most important facets of semantic structure, one that is consistently overlooked by the analyses and explanations of formal semantics, is the speaker's role in construing entities and interrelationships in particular ways, not always predictable, but almost always motivated by discoverable aspects of particular usages of a grammatical construction (Wierzbicka 1985; Traugott 1985; Langacker 1987, 1990b, 1991; Casad 1988a; Smith 1987; Taylor 1989).

24

Eugene H. Casad

The usual strategy employed by both formal syntax and formal semantics is to consider the speaker's role in framing the form of his utterance to be a part of "pragmatics" and to conclude that information related to such speaker roles is outside the interest of syntax and semantics per se. In the following sections of this paper, I illustrate, first, the kinds of conceptual entities that need to be invoked in accounting for the various ways that speakers have for talking about particular scenarios. I go on to illustrate the kind of formal analysis that results when the speaker's construal of the situation is ignored. Finally, I illustrate the kinds of analyses that emerge from a usage-based view of grammar, in which the notion of construal is given both a rich meaning and its rightful place in the account of the forms that grammars take.

2. Seeing it in more than one way One of our most basic cognitive capacities is the ability to view a situation or scenario from a number of different perspectives, making distinct comments on it, depending on both what we want to say about it and how much detail we want to put into our description. As Giv6n appropriately notes, the number of distinct descriptions of a particular situation that a speaker can contrive is essentially open-ended (Giv6n 1989: 89). Much in the spirit of his discussion of the description someone might give of a man beating a dog in a park (1989: 88-90), let us take a scenario that involves two people, a stream that is running quite swiftly because of heavy rain in the area and a speaker vantage point from the side of the stream opposite the two people. 3 We also know that they have just gone from one side of the stream to the other. The sentences in (1) represent some of the statements that we could make regarding this scenario. (1) a. The man and the boy crossed the stream. b. The struggling pair just barely made it to the other side ot the stream. c. Two people waded the stream. d. The stream was crossed by two people. e. The crossing of the stream was accomplished by two people. f. The crossing of the stream was done with extreme difficulty. g. The crossing of the stream was made difficult by the high water level. Each of these sentences makes some kind of assertion about the event, but does so in its own way, selecting certain entities and interactions for

Seeing it in more than one way

25

comment, while backgrounding other details or leaving them implicit to the particular account given. For example, sentence (la) highlights the individual participants and categorizes them, i.e., a man and boy, at the same time that it describes their activity in global terms, i.e., they crossed the stream. On the other hand (1 b) treats the two individuals as a single multiplex entity without identifying or categorizing either one. The use of the lexical item pair explicitly designates two generic entities. The sentence itself places the highest degree of prominence on the difficulty of the situation by drawing on the adjective struggling and linking it to the pair in question. This tells the hearer that these two people had to work hard in order to cross the stream, as evidenced by the adverbs just and barely, which contribute the implication that the work was so difficult that they almost failed to reach their goal. Finally, the notion of crossing is designated in part by an explicit designation of the goal that the people were following. In this sentence, the speaker implicitly relates him/herself to the location of the goal in the situation with the use of the prepositional phrase to the other side of the stream. The speaker's involvement consists in his location serving as an implicit reference point in terms of which the notion the other side is defined. This low-profile reference-point role correlates with a high degree of subjectivity in the speaker's usage (cf. Langacker 1985, 1987, 1990a, 1991 ). The crossing itself is expressed by the schematic process verb to make. The speaker's attitudes often directly shape the form of an utterance. This is a possible explanation for the appropriate usage of (lc) in the stream-crossing situation. One of the observers, for instance, might be quite insecure and try to cover this up by making a gross understatement of the situation, i.e., the comment in (lc) makes no reference whatsoever to anything unusual in the crossing of the stream. The expression two people directly expresses lexically the number of entities involved, a notion that is sublexical in the use of the expression pair in (1 b) (cf. Talmy 1987: 190; Langacker 1990b: 75). The speaker's use of the verb waded is the real source of the understatement. Ordinarily, wading is a pleasurable activity involving no salient expenditure of energy on the part of the wader. It is also a relatively undirected activity; that is, the wader usually is not interested in going anywhere, in contrast to the traverser, who does have a particular goal in mind. The understatement therefore also downplays the successful crossing of the swollen stream. Whereas the first three sentences of (1) place the highest degree of prominence on the people who crossed the stream, the sentences in (ld-

26

Eugene H. Casad

g) place the highest degree of prominence on either the stream itself (ld) or on the essential interactions within the scene, i.e., the crossing of the stream (le-g). In (ld) the stream is accorded highest prominence semantically, with the joint agents of the process of crossing backgrounded, as seen by their syntactic status as object of the preposition by in postverbal position. This is therefore a kind of passive sentence, but not a prototypical one, since the stream itself was not affected in any way by the event. 4 The passive verb phrase was crossed presents the activity in global perspective without elaborating at all on any degree of expended effort on the part of the agents involved in the activity. The usage of cross illustrated by sentence (la) is basically processual. This means that the notion is being construed by the speaker as a series of successive states through time in which each successive state reflects a change in the configuration of the interrelationships that held in a previous state. All of these changing states are summed up in a complex configuration that has a beginning state, distinct intermediate states and a unique final state. This is what Langacker terms the temporal profile of a process (Langacker 1987: 244, 1990b: 81, 1991: 223). This characterization of the notion process allows us to characterize precisely the difference between the verbal use of cross in (la) and the use of the participial in (ld). In (ld), the speaker is focusing on the completion of the event and construes the most salient subpart of the setting within which it occurred as a highlighted passive subject. The event itself is presented atemporally, a perspective in which the successive states of its temporal profile are presented as a set whose members collectively form a complex configuration. All of this is effected by the use of the perfective participle -ed (cf. Langacker 1987: 220-221, 1990b: 129-131, 1991: 131-134). Not only do speakers have the ability to focus differentially on the beginning and end points of a changing configuration, they can also focus on the entire set of states that comprise the temporal profile of a process. This allows them to treat an event as a distinct entity and manipulate it syntactically like a noun. Thus, in English, a verb can take an -ing suffix that we call a gerund. This is illustrated by the usages of cross in sentences (le-g). The construal of cross as a derived nominal has distinct syntactic consequences (cf. Langacker 1987: 246-247), 1990b: 80, 1991: 52-53). In particular, it takes the definite article, just like ordinary concrete nouns. In addition, whereas the nominal stream is the direct object of cross in its verbal usages, in the nominalized usages of cross, stream is related to crossing as the object of a preposition in a prepositional phrase that specifies the notion of crossing in further detail.

Seeing it in more than one way

27

By according prominence to the event and then construing that event as a discrete entity, the speaker of English frees him/herself to make additional descriptive statements about the stream-crossing scenario. Thus in (le), the speaker highlights the event as involving a significant amount of purposeful, expended energy. This highlighting is done by the selection of the passive verb phrase was accomplished. In addition, the crossing is saliently labeled a success, partly by the meaning of the verb itself and partly by the particular tense-aspect used in this expression, i.e., the perfective use of the past tense morpheme -ed. In (lf), on the other hand, the speaker selects the schematic verb do to signal the successful carrying out of the endeavor, but adds the propositional phrase with extreme difficulty to highlight the concurrent struggle that the pair were involved in. Finally, (lg) makes explicit reference to both the swollen state of the stream at the time of the effort and to the attendant condition of difficulty associated with that endeavor, while specifically stating a causal connection between the swollen state of the stream and the difficulty experienced in the crossing of it. There is no mention in (lg) of a volitional agent who carried out the crossing of the stream; instead this is left implicit as a sublexical part of the meaning of the word crossing.

3. Some simplistic formal analyses In Langacker's view grammatical structure is organized in part by "the content requirement". The only structures that appear in the grammar are: (a) the linguistic structures that are directly attested by the data; (b) the structures that are schematic for the attested linguistic structures; and (c) the categorization relationships that occur both within grammatical constructions and within schematic networks (cf. Langacker 1987: 5354, 1990b: 18-19). The problems with most formal approaches to both syntactic and semantic analysis relate as much to what these analyses do not say as to what they do say. Among the facets of semantic structure that are commonly ignored by formal semanticists are: (a) processing time; (b) event coordination; (c) relative prominence; (d) figure/ground alignment; (e) levels of organization; (f) sequential scanning; (g) degrees of schematicity; (h) scope of predication; and (i) effective homogenity (Langacker 1990: 101). All of these are part and pared of what Langacker calls "the con-

28

Eugene H. Casad

strual relation" and are needed in order to give an observationally accurate and explanatorily adequate account of the semantics of natural language. Strictly formal accounts are often simplistic in both the assumptions that they make about relevant data and the notational devices that they employ for both describing and explaining semantic phenomena. In this section, I illustrate these points with three analyses from Jackendoff (1983). In later sections I present contrastive cognitive-grammar analyses of some English data as well as from my own research on the Cora language of Northwest Mexico. In a few cases in this paper, I have elaborated on Langacker's examples. Intuitively, I find Jackendoff's formal representations of "conceptual structures" to be unconvincing. Basically, these representations frequently fly in the face of this English speaker's intuitions about what is relevant for an adequate description of the data, i.e., what is salient to the representation vis-a-vis what is backgrounded to one degree or another. For example, let us consider the use of the "functor" GO. In order to capture generalizations according to his grammatical constraint, Jackendoff posits classes of events that are defined according to certain semantic functions. Thus, one class of events is a GO class (1983: 171-172). Here Jackendoffis using GO as a functor in the sense of mathematical logic, which ostensibly bears no affinity whatsoever with the English motion verb go (cf. Lakoff 1987: 221-222, 227). Note, for example, that GO has grammaticalized versions in which it signals purposeful action, rather than physical motion toward a point in three dimensional space. A typical example is (2a), which contrasts with the motion verb use of go in (2b ). (2)

a. b.

I am going to build a workshop. I am going to the workshop.

Jackendoff rejects the idea that the function GO can be construed as expressing a change of state from one position to another. His first argument is that GO is not restricted to occurring with bounded paths, but also with directions and routes and that "Go expresses the traversal of every point" along a path (1983: 174). His real concern is that such a construal would eliminate "a primitive spatial function" (1983: 174). Here he seems to equate his GO function with that of [PATH]. His fear, however, is neither logically necessary nor cognitively valid. The notion of [PATH] is certainly not a primitive, but can rather be treated as an

Seeing it in more than one way

29

image schema (cf. Miller-Johnson-Laird 1977; Casad 1982, 1992; Hawkins 1984; Johnson 1987; Lakoff 1987; Lindner 1981). Furthermore, the data do not support Jackendoff's contentions. Thus, various kinds of construals are possible that he does not mention. Four of these are given in (3a-e). (3)

a. b. c.

d.

He went from first to second base on a wild throw. He goes only part way without taking a rest. Her cheeks go from rosy pink to bright red every time she hears that phrase. It goes from glossy to dull with exposure to the sun.

Sentence (3a) expresses the change of location from one position to another within a more complex and more extended pathway. The portion of the path that the player traverses is backgrounded to the endpoints, an aspect of conceptual structure that Jackendoff consistently ignores in his analyses. Example (3b) indicates an indefinite number of shorter bounded areas within a more extensive but backgrounded pathway. Sentence (3c) illustrates the use of go to indicate the change of facial pallor from its normal state to one revealing acute embarrassment, whereas (3d) expresses the expected change in the appearance of a discrete object that is expressed by the polar terms glossy/dull. Jackendoff's claim that the functor GO cannot be construed as expressing a change of state can only be maintained by ignoring examples such as these and it must be abandoned if he hopes to be observationally adequate. Jackendoff's use of functor labels that have the same orthographic content as conventionalized lexical items in English amounts to obfuscation rather than elucidation. His treatment of GO as a functor is also not well-based in several other respects. He finds himself in a quandary as to know how to characterize the relationships between GO as a verb of motion, as in (4a), GO as a verb of temporal extent, as in (4b), and GO as a verb of spatial extent, as in (4c). (4)

a. b. c.

He goes from San Diego to Los Angeles three times a week. He goes nonstop from morning till night staring at his computer screen. The divided road goes all the way from Nogales to Navojoa.

His problem is that he cannot decide how to interpret GO and GOext: are they distinct functions that share a good deal of internal structure or are they the same function with an interpretive rule specifying that you get either a traversal construal or an extent construal, depending on whether

30

Eugene H. Casad

you are dealing with an [EVENT] or a [STATE] (1983: 173)? His alternatives may well be notational variants, but the second one seems to lend itself to a more adequate characterization of the data, whereas the characterization of GO and GOext as discrete abstract functions is misleading. Jackendoff really needs to characterize these notions in terms of figure and ground relationships that are realized within particular conceptual domains. In particular, Jackendoff needs to link GO to the directed path image schema, which is a concept that has a directly understood structure of its own (cf. Casad 1992) and is used metaphorically to structure related concepts (cf. Lakoff 1987: 283; Lindner 1981: 171-173). A second set of examples concerns his conceptual structures related to the events "possess", "receive" and "lose". With respect to the notion "possess", Jackendoff lumps together as synonymous the sentences Beth has the doll, Beth possesses the doll, Beth owns the doll and The doll belongs to Beth (1983: 192). All these sentences are given the single representation shown in (5). (5)

[stateBEPoss([DOLL], [PiaceATPoss([BETH])])]

As it stands, Jackendoff's representation most closely relates to The doll belongs to Beth, which differs from Beth has the doll, Beth posseses the doll, and Beth owns the doll in both its figure-ground organization and its lexical content. In The doll belongs to Beth, the doll is the salient figure

and Beth is the salient landmark or ground of the relationship, whereas the situation is exactly reversed in the other three sentences. The verbparticle construction belongs to lexicalizes this figure-ground reversal. With respect to the other three sentences, the verbs have, possess, and own each structure the possessive relationship in a different way, have indicating a schematic associative relationship, which in some contexts designates a possessive relationship and in other contexts designates other kinds of relationships, including that of being affected by an ailment, e. g., Beth has a cold, and that of obligation, e. g., Beth has to go to school. Both possess and own are more specific with regard to the notion of possession. Possess places the highest degree of salience on the possessorpossessed inanimate object relation, whereas own places the highest degree of salience on the possessor's control over the possessed inanimate object. The differences between all of these verbs become even more evident when we examine the extended usages and paraphrases of each. All four of these verbs, i.e., belong to, have, possess, own, relate to distinct experiential gestalts in Lakoff's terms (cf. also Geeraerts 1988; Taylor 1988, 1989).

Seeing it in more than one way

31

The final pair of Jackendoff's examples that I discuss are Beth received the doll and Beth lost the doll, whose putative conceptual structures are given in (6a) and (6b). (6)

a.

[EventGOp 088 ([DOLL],[pathTOposs([BETH])])]

b.

[EventGOp 0 ss([DOLL],(pathFROMp 088 ([BETH])])]

An examination of the conceptual structures that Jackendoff associates with each shows that formally the only difference between receive and lose is reputed to be the directionality of the path associated with the function GO (cf. 1983: 192). In other words, Jackendoff's notation characterizes Beth received the doll as being the semantic inverse of Beth lost the doll. This would not be so bad if the meaning of the sentence Beth lost the doll could be fully characterized as Now you see it; now you don't (Doris Bartholomew, personal communication February 1991). Obviously, however, there is a lot more to the story. For one, the two sentences differ in the degree of intentionality associated with the subject role. The subject of receive does so intentionally and (usually) consciously, whereas the subject of lose prototypically does so unintentionally. Note in passing that Jackendoff classes both sentences as noncausative (1983: 193), but incorrectly says that the subject exercises no control in either sentence. However, in at least one reading of receive, the subject obviously does exercise some kind of control, one that can be paraphrased as "to take possession of'. In other words, in many of its usages, the conceptual scene associated with "X receives Y" also includes the implicit situation that Z gives or offers to give Y to X and X takes Y (cf. also Langacker 1990b: 226-228). This accounts for certain other usages of receive, such as in Beth received a generous inheritance. In this case, the control is exercised by an unspecified agent who graciously included Beth in the benefits of his or her estate. A similar example is seen in the sentence We mailed a doll to Beth, but she never received it. In this case, the control is attributed to the subject of the initial clause (and perhaps also in the post office) which creates the potential for Beth to receive something. To summarize, there certainly is a class of GO verbs in English, but not every verb that relates to a conceptual scene in which some entity moves or gets moved is a member of that class. Jackendoff's use of functors as schematic class markets (cf. 1983: 204) needs to be linked to psychological reality and to reflect the generalizations that English speakers really do extract from the established patterns of the language. For this English speaker, verb phrases such as butter the bread (Jackendoff 1983: 185) are

32

Eugene H. Casad

definitely not GO verbs, nor is Spanish !lover 'to rain' a GO verb (1983: 185). Here, Lakoff's comment on the need to attend to the linguistic details is especially appropriate (cf. Lakoff 1987: 379), as is Langacker's content requirement for grammars (mentioned earlier), that rules out all arbitrary devices used to make descriptions internally coherent (Langacker 1987: 53-54, 488, 1990b: 18-19).

4. Construal: Conventional imagery In Langacker's terms, all grammatical units are symbolic structures that pair a meaning with phonological form. In this framework, a semantic structure is a conceptual structure that serves as the semantic pole of a linguistic expression (Langacker 1987: 98). Thus, a given semantic structure is a portion of a greater bipolar, symbolic structure, whereas conceptual structure is unipolar, not multifaceted. Grammatical structure more generally is the conventional symbolization of semantic structure. In Langacker's view, therefore, all grammatical structures are inherently symbolic regardless of their internal complexity. Symbolization is largely nonarbitrary because the prototype for complex expressions is for them to be analyzable into smaller, meaningful components; usually a salient relationship exists between a composite structure and the morphemes and lexical items that compose it. Grammatical structure also embodies conventional imagery (Smith 1987: 57-63). There are various aspects to the semantic properties that Langacker subsumes under "conventional imagery". They include the following: (a) alternate construals of scenes; (b) alternate paths of composition; (c) alternate salience of parts; (d) different levels of specificity; and (e) alternate speaker vantage points (Langacker 1987: 51; Lindner 1981: 227; Tuggy 1981: 60-71). To illustrate this, it is clear that different languages code the same conceptual structure by means of different images. In examples (7a-d), from Cora and English, the same conceptual structure corresponds to the same entity or situation in objective reality. (7) a. b. c. d.

Cora: yuhtYiviina ahka'iwa'imi tYa'iika metYenYu

vs. vs. vs. vs.

English: hillbilly a long time ago dust devel It just thundered

Seeing it in more than one way

33

Both the Cora word yuhtYiviina and the English word hillbilly designate a class of people who have moved into the speaker's area from a region characterized by hilly terrain. In both cases, the people are viewed as a distinct social class. Yet semantically they are very different. The Cora word consists of a nominalized morphemically complex topographic adverb. The initial y- 'here' indicates the speaker's location, the following -uh- designates a path straight up the slope of a hill from the speaker's location, the suffix sequence -tYivi designates the goal of that path, which is the point of origin of the designated person, i.e., an uphill area. Finally, -na designates the individual in schematic terms. On the other hand, the English word consists of a noun-noun compound, in which the first noun hill specifies the location from which a person comes and billy designates the person in a patronizing, if not pejorative manner. The Cora use of the term yuhtYiviina, however, ranges from neutral emotive value to strongly antagonistic. In terms of imagery, compositionality and social implications, the apparently functionally equivalent terms are quite different. The terms in (7b) illustrate an even more closely matched concept that is construed in very different ways in Cora and English. Both terms are adverbials, designating a point of time in the distant past, far removed from the time of the speech act in which the terms are employed. The Cora term akha'iwa-'imi is a topographic adverbial phrase consisting of the topographic adverb meaning "off at the side of the hill" and the degree adverb imf which means "far away". The particular usage thus represents the extension of a purely spatial term into the temporal domain. On the other hand, the English phrase a long time ago reifies the concept of temporal succession, lexicalizing it as an unanalyzable nominal form and modifying it with an adjective that normally contributes the notion of spatial extension to the noun it modifies. In this case, the meaning of the adjective accommodates to the nominal, taking on a temporal construal, which is reinforced by the temporal adverb ago. This adverbial functions in the same way as the Cora imf in that it locates an event in the remote past relative to the time of speaking. Once again, both languages have functionally equivalent conventionalized expressions that designate the same concept in the projected world, but they employ very different means for expressing that concept. Example (7c) concerns a meteorological phenomenon, i.e., a small rapidly twisting column of air that picks up dirt from off the ground as it moves along. The Cora term for this is tYa'iika. One common English term is dust devil. The Cora expression is constructed from a locative

34

Eugene H. Casad

prefix tYa- 'in the middle of a flat surface' and a following body part noun 'tika 'foot'. Cora has a common pattern in which a locative prefix or a sequence of prefixes in combination with a following body-part nominal designates a possessive relation. This term may well have originally meant "the one who has a foot on the ground", but it has most likely lost its analyzability for the present-day Cora. On the other hand, the English term is fully analyzable. The initial word in this loose compound designates the entity that gets stirred up into a cloud just above ground level, whereas the second word, which is the head of the compound, is a nominal form that reifies the column of twirling air as a malevolent spiritual force. One way of contrasting the Cora and English expressions is to say that the Coras have lexicalized the effect of the phenomenon in terms of their system of locative prefixes and body part names, whereas English speakers have focused on the presumed causer of the event. The terms in (7d) also relate to meteorological phenomena. To refer to the sound of thunder that follows a bolt of lightning, the Cora use the expression, me-tYe-nYuu 'they responded' (3PL.SUBJ-PERF-respond), whereas English uses the third singular impersonal It thundered. In this case, we English speakers focus on the observed phenomenon without saliently designating any particular cause. On the other hand, the use of the third person plural subject prefix in Cora portends a whole complex network of knowledge about the personages who are responsible for the forces of nature. The use of the verb meaning "to respond" points to another aspect of that cultural knowledge, i.e., the necessity for the religious leaders to perform particular rituals in order to provoke the responsible personages to come to earth with the rains. To summarize, both Cora and English have succinct ways for designating the very same meteorological event in the physical environment, but their respective phrases imply completely distinct foci and networks of cultural knowledge. This potential for differential expression could not occur if it were really true that semantic structure is both universal and is conceptual structure tout court (cf. Lakoff 1987: 311).

5. Construal: Fitting the pieces together Structures that designate the same conceptual entity may have different compositional paths, i.e., they may be composed of different morphemes, or consist of the same morphemes combined in different ways, or simply reflect variable directness in their conventional symbolization of concep-

Seeing it in more than one way

35

I ual structure. For example, Langacker's pair father and male parent may designate the same individual in the objective scene, but they encode information in two different ways and are therefore not synonymous (Langacker 1987: 293-294, 462; 1990b: 10). In particular, the gender of the designated relative and the generational relationship are overtly lexicalized by the component morphemes in male parent, respectively, but are sublexical and are therfore symbolized directly by the word father. 5 A restriction to particular domains is another part of their differential meaning. For one, I would more likely use the term male parent when discussing penguins than I would when discussing humans. In a genetics class, however, the term male parent would be fully appropriate. As a second example, English employs a single unanalyzable morpheme to encode the concept "coffee". Yaqui, of Northern Mexico, and Luisefio, of Southern California, both encode that concept by means of an adjective noun construction involving the concepts "black" and "water", but they combine the grammatical elements in opposite orders (8a--c).

(8)

English: [coFFEE]/[kh;,fiY] Yaqui: [[BLACK][wATER]]/[[cukui][baa'a] ___,. [coFFEE] Luisefio: [[WATER][BLACK]]/[[paala][yuvataat] ---+ [COFFEE)

Structures with the same compositonal paths may also have vastly different meanings. Compare the Luisefio example above with the following from Spanish that almost, but not quite, matches it compositionally. The Spanish example (9) is conventionally expressed as a plural form, indicating a mass noun, but does not conventionally designate "coffee". (9)

Spanish: [[AGUA-s] [NEGRA-s]] Water-PL Black-PL

---+

[SEWAGE]

One would not, therefore, order three cups of coffee using the expression in (10): ( 10)

*Deme tres aguas negras, por favor. *Give me three cups of coffee, please.

Alternate construals of a given situation can easily be built into a single sentence, as in (11), which I concocted while packing my luggage for a trip starting the following day. (11)

Wow! This is either going to be a long night, or a short one.

In (11) my use of "long night" referred to the potential for many hours to be spent packing my bags, whereas the use of "short one" referred to

36

Eugene H. Casad

the potential for my having very few hours to sleep. This is just a typical example that illustrates a basic cognitive ability that we have for quickly shifting from one image to another in the flow of discourse (cf. Langacker 1990b: 12).

6. Construal: The speaker's vantage point Cora locative prefix plus verb stem constructions illustrate in a striking way the need for invoking the relation of construal between the speaker and the situation in which he employs a given utterance. In addition, these constructions vividly illustrate the role of conventional imagery in grammar. Finally, they provide an endless stream of examples that suggest strongly the need to distinguish between conceptual structure per se and its conventionalization as semantic structure. The Cora examples in (12a-e) show the range of responses a Cora speaker might give if asked how to describe a well-lit house at night. 6 (12)

a.

b.

c.

d.

e.

u-wa-nYeeri-'i inside-EXT-illuminate-STAT 'It is all lit up inside the house' U-tY e-nYeeri-'i inside-middle-ill umina te-STAT 'the doorways and the windows of the house are all lit up.' w-ii-ra-nYeeri- 'i inside-toward-face out-illuminate-STAT 'It is all lit up in front of the doorway from light coming this way from inside the house.' a-ii-ra-nY eeri- 'i outside-toward-face out-illuminate-STAT 'It is all lit up on the ground by light coming through the doorway.' a-ii-re'e-nYeeri- 'i outside-toward-around corner-illuminate-STAT 'It is all lit up at the side of the house by a light at the back of the house.'

These five sentences represent very different images of a lit-up house viewed at night. The speaker's vantage point is crucial to an adequate explanation of these data. In all cases, the speaker is viewing the house from an external viewpoint, but both the vantage point and the scope of

Seeing it in more than one way

37

his field of vision differ. Thus, in (12a), the speaker is either close enough to the door that he can see a substantial part of the interior and infer that the illumination extends throughout that interior area or the windows and doorways are numerous and expansive enough for him to reasonably draw that inference. In (12b), the speaker is standing far enough away from the house to take in the whole expanse of the house within his field of vision which allows him to focus more or less simultaneously on the illumination visually accessible through the door and the windows. In (12c), the speaker is prototypically looking straight through the doorway and has within his field of vision an illuminated area that is partly inside the house and partly outside of it so that he perceives a natural directionality of the illumination toward his own position. The speaker's focus of attention in (12d) is confined totally to the outside of the house. This expression would be appropriate either for light coming from a fixture mounted above the doorway and shining out into the front yard or light coming from within the house, the whose interior source is being ignored by the speaker, whose vantage point may well be oblique to the doorway. Finally, in (12e), the speaker is looking along one side of the house, and sees a well-lit area in the back of the house, which has one of its boundaries determined by the corner of the house. This allows the speaker to correctly infer that the unseen source of illumination is around the corner of the house from his own position. To summarize, an explicit characterization of a conceptual scene and its components is needed here in order to give an observationally adequate characterization of these data. This characterization must also note the differing degrees of subjectivity or objectivity with which the speaker presents the scene (cf. Langacker 1990a: 7, 34; 1990b: 12; 1991: 215-). 7

7. Construal: Categorizing nuts and fishes The construal relation allows the speaker to handle figurative language easily and coherently. Often salient similarities will evoke immediate understanding. For example, a walnut is hard shelled and round. We conventionally call it a nut; for some of us, the walnut represents the prototypical member of the class that also includes cashew nuts, coconuts, pecans, brazil nuts and peanuts. (For others of us, however, the peanut is the prototypical one.) Common in industrial societies is a small piece of metal with a threaded hole in its middle which is approximately equidimensional. It usually has a hexagonal cross section, sometimes a square-

38

Eugene H. Casad

shaped one. The threads in the hole in its middle allow it to be attached, by turning it, to a metal shaft with a threaded end, called a bolt, whereas the small piece of metal that attaches to it is called, in English, a nut. The approximately equidimensional shape of the industrial nut, as well as its obvious hardness and hollowness, allow the speaker to apply the term nut, meaning a hard-shelled edible fruit, to the metal object. The speaker's knowledge of the distinct functions and non-shared characteristics of the two entities allows him to keep them apart conceptually. The choice of the lexical item nut to designate an industrial implement reflects the following categorization, in which semantic units are indicated by the use of capital letters, whereas phonological units are marked by lower case letters. The slash mark represents the symbolization relationship and the arrow indicates semantic extension: (13)

[[[WALNUT]/[nut])

-+

[[HEXAGONAL BLOCK]/[nut]]]

Cultures do not necessarily construe these entities in the same way. It is instructive to compare the English and Spanish terms for two particular kinds of this industrial fastening device, e. g., the regular hex nut and the winged nut. (14)

English: nut wingnut

Spanish: tuerca palo mil/a

English views the industrial implement as analogous to a hard-shelled edible fruit; Spanish views it in terms of the required motion that results in its becoming attached to a bolt; this motion in Spanish being called tornillo. Also, note that Spanish treats bolt as a masculine gender noun, but nut as a feminine gender noun. For a wingnut, both English and Spanish focus on the two protuberances at opposite sides of the nut and both languages relate the pair of protuberances to the form of a bird, but English names the body part, whereas Spanish names it after a particular species of bird, i.e., paloma 'dove', and makes prominent the smallness of the nut itself by attaching the diminutive suffix -ilia. As another example of the cultural specificity of semantic structure, I mention the example of English fish and Cora we'i 'fish'. The point is that the English nominal fish designates the entire class of acquatic creatures with scales and gills, without singling out any one single species. English fish, therefore, is a fully generic lexical item. On the other hand, Cora we'i serves both to designate the entire class of fishes, as well as to

Seeing it in more than one way

39

designate what is, for the Cora, the probable prototypical member of the class, the catfish. The differences between English fish and Cora we'i are much more broad than just the categorical structure. To begin, both English and Cora call on nominal stems to serve in predicate roles. Thus, conventional English usage allows us to take the root fish and put it into a sentence such as (15): ( 15)

I am fishing.

This sentence means that the speaker is engaged in a particular activity that is directed toward his capturing a nominal object which is incorporated into a verb phrase of the form shown in (16): (16)

VP

= BE

+ Noun-ing

Cora views the fishing scene in very different terms. Rather than using we'i as an incorporated noun to designate this activity, the Cora either use the incorporated noun kwl:'i¢i'ipwa 'fishhook' as in (17a), or a transitive clause with a verb meaning "to kill", as in (17b ). (17)

a.

nYa-kwe'i¢i'ipwa

!-fishhook 'I am fishing with a hook and line.' b.

nY-ait-ce'e we'i-tYe kuura

1-LOc.Base-still fish-PL kill 'I am fishing.' Example (17b) probably reflects the method of killing fish by throwing a stick of dynamite into the water, resulting in death by concussion. More recently, Coras have also begun spearing fish using primitive spearguns. Cora also uses a conventionalized pattern of forming verbs by prefixing the distributive plural prefix to a noun. Thus, the noun 'i¢ilri 'loom' is incorporated into a verb phrase which means 'to weave', as in (18): (18)

nYe-tYf-'i¢ah

'I -DISTR-loom 'I am weaving.' One might suppose that the distributive prefix plus noun schema would sanction a Cora verb phrase involving the incorporated stem we'i, with the meaning "I'm fishing". However, as the following two examples show, such is not the case. In (19a), the element following the distributive suffix is a homophonous morpheme meaning "to speak falsely". In (19b),

40

Eugene H. Casad

a second schematic pattern comes into play, one that summarizes equational statements of the form 'X is a Y'. Thus, the Cora sentence that prefixes tYf'i- to the noun we'i 'fish' comes out meaning "I am a fish". (19) a. nYe-tYf'i-we'i 1-DISTR-speak:falsely 'I am telling lies.' b. nYe-tYi'i-we'i 1-DISTR-fish 'I am a fish.' In addition to the foregoing, I must also mention that you cannot compute the meaning "I'm fishing" from a nYa-kwe'i¢i'ipwa. This shows that the composite structure has conventionalized meaning beyond that of its component morphemes. Conventional imagery, then, is seen in that English and Cora structure the same conceptual scene in very different ways. English places primary salience upon the person engaged in the activity as the responsible energetic instigator of that activity. It places secondary salience (prominence) upon the object of the endeavor. On the other hand, Cora places primary salience upon the responsible energetic instigator or purposing entity, and then places secondary salience on the instrument by means of which the activity is carried out. The entity being sought is not even lexicalized in nYa-kw e' i¢i'ipw a. 8

8. Kinds of categorizing relationships Categorizing relationships of various sorts and degrees of abstraction relate conventionalized usages to the generalizations that speakers employ in framing their thoughts for linguistic expression. These relationships can consist of either elaborations or specializations and exist at all linguistic levels. One example in phonology is the categorizing relationship between alternate pronunciations of what speakers normally recognize as being instances of the same word. These relationships are oftentimes reciprocal, as (20a) and (20b). [newt] +-+ [rut] route [i:or] ...... [aior] either a typical example of a categorizing relationship involves the of common phrasal patterns. For example, the schematic

Seeing it in more than one way

41

pattern for a propositional phrase is elaborated in a number of distinct ways by the examples in (21a-b) (21)

a. b.

P NP- [[beside][the house]] P NP- [[inside][the barn]]

A common elaboration of the verb phrase is a sequence of a verb plus a noun phrase consisting of some kind of an article or modifier and a following head noun that functions as the direct object of the verb as in (22a-d). These sequences are often enough vehicles for metaphors. (22)

a. b. c. d.

V NPV NPV NPV NP-

[[stub][ one's toe]] [[skin][the cat]] [[break][the bank]] [[water][the lawn]]

A third common categorizing relationship in English syntax involves a verb phrase that not only has a following object-noun phrase, but also has a phrase-final preposition, often referred to as a "particle", as in (23a-c) (cf. Lindner 1981: 2). (23)

a. b. c.

V NP P- [[[wake][the children][up]]] V NP P- [[[bring][the house][down]]] V NP P- [[[pass][the food][around]]]

Spanish has a distinct categorizing pattern for forming attributive noun phrases. In the case illustrated by (24a-b), the pattern is used to indicate that the head noun is composed of the substance that is named by the nominal object of the preposition de. (24)

a. b.

N P N- [[pan][de trigo]] 'wheat bread' N P N- [[ojo][de agua]] 'spring'

Semantic extensions illustrate a different class of categorizing relationships in which the categorizations are not as clear cut as some of those given above. Many semantic extensions are metaphorical and reflect to varying degrees the ease with which they can be likened to literal meanings or specialized meanings that motivated their use. Langacker accounts for this by attributing varying degrees of salience to the categorizing patterns upon which they are based (Langacker 1987: 386). For example, the English word star, which designates a bright object seen in the sky at night (except for the moon, of course), has become extended into the field of entertainment so that it now designates a person who gained fame in the movie industry, e. g., a movie star, with fur-

42

Eugene H. Casad

ther extensions to a person carrying the leading role in a movie, play or television program, e. g., the star of the show. In addition, it has become further extended to designate a person who performs exceptionally in fields such as sports, e. g., a star player, and education, e. g., a star pupil. This set of extensions involves the categorizations shown in (25a-c) (cf. Langacker 1987: 386). (25)

a. b. C.

[[STAR]/[star] [[STAR]/[star] [[STAR]/[star]

~ ~ ~

[[CELEBRITY]/[star]]J [[LEADING PERFORMER]/[star]]] [[EXCEPTIONAL]/[star]J]ADJ

The examples in (25) show that semantic extensions occur both within a given grammatical class as well as across grammatical class borders. Note that the noun star is categorized as an adjective in (25c).

9. How you sanction cats Semantic extension is at the heart oflinguistic creativity (Langacker 1987: 73; Smith 1987: 55). Extensions of the English word cat provide a glimpse of the convoluted paths that such linguistic creativity follows. To begin, one common extension of cat refers to nondomestic felines, i.e., tiger. This categorization is given in (26). (26)

[[CAT]/(cat]

~

((TIGER]/(cat]]]

English has another extended use of cat to refer to a kind of tractor. This categorization is superficially analogous to the extension that we invoke in referring to a tiger as a "cat", and is given as follows in (27): (27)

[(CAT]/(cat]

~

[(TRACTOR]/(cat]]]

It turns out, however, that the "cat" that lies behind the extension to "tractor" is not a four legged feline as is the "cat" that lies behind the extension to "tiger". Instead, the term cat as applied to the tractor is a shortened form of the word caterpillar, which is a fuzzy insect that crawls over the ground aided by numerous pairs of little feet that are hardly perceptible to the observer. This relates directly to the kind of mechanism that a tractor employs for moving across the terrain. In this case we are considering not a "wheel tractor" but a "track tractor", one that has two bands of connected metal plates that are strung over and supported by sets of wheels at each side of the tractor. Tanks, of course, also use this form of locomotion. 9

Seeing it in more than one way

43

In short, the way that a track tractor crawls over the ground was perceived by certain industrialists in the US as looking like the way a caterpillar moves across some surface. This perceived similarity carried over into other aspects of the design of the tractor and the operation of the company that began to manufacture it: the tractors were originally all painted yellow since a particularly common large caterpillar is mostly yellow; the company named itself Caterpillar, Inc and the tractor became known as a Caterpillar tractor. A more adequate categorization schema for all this involves the two steps illustrated in (28): (28)

a. b.

[[CATERPILLAR]/[caterpillar]]-+ [[TRACTOR]/[caterpillar]]] [[TRACTOR]/[caterpillar]]-+ [[TRACTOR]/[cat]]

A whole range of new usages developed from the shortening of caterpillar to cat. Of a person whose work requires him to operate such a tractor, we may describe his role in life as in (29). (29)

He drives cat all day long.

Caterpillar tractors have come to be employed in both construction work and in fighting fires with forestry services, both because of their power and their stability on a hillside. The extension and shortening of caterpillar to cat shows that in certain contexts native speakers of English have forgotten the original source of tractor cat and have equated it with feline cat. In particular, a tractor driver who works for the Forestry Service in the US is often called a "cat skinner". This term refers to the tractor driver's work in bulldozing down brush and overgrowth to form a roadlike open space along the perimeter of a fire zone in order to prevent the fire from burning a wider area. The road-like zone is called a firebreak and the forceful removal of burnable material with the subsequent leaving of an area of bare gound is likened to the operation of removing the skin from a furry animal such as a cat. The coining of the term cat skinner has apparently been mediated by an extension from the literal expression used to describe skinning cats to the metaphorical expression used to talk about solving problems in general. Thus example (30) presents a common expression used to signal that if one strategy does not work, then another one will. (30)

There's always more than one way to skin a cat.

It is likely that the formation of the compound cat skinner involved sanctioning by both the literal meaning of to skin a cat and the metaphorical

expression above. These data, then, illustrate a number of points impor-

44

Eugene H. Casad

tant to the cognitive-grammar characterization of semantics. For one, several domains may be operative at once. For example, the pattern for forming nominalizations on complex noun phrases by means of the agentive suffix -er is one of them. In addition, a new extension may well spawn a further extension that obscures the difference between two otherwise distinct domains. The new extension may also spawn an extension that is not even directly related to the original concept. Finally, the endpoints of these extensions are commonly unpredictable, but very well motivated.

10. Implications Grammar consists of established patterns for combining simple symbolic expressions into more complex ones. Speakers must exploit the potential that conventional symbolic units afford them for forming linguistic expressions and they must also operate within the constraints that are inherent to particular situations. Usage entails the selection of a target structure as the vehicle for conveying the selected expression. The degree to which a target structure conforms to the conventional units of the grammar is the degree to which the grammar sanctions that usage. A clear implication of the analyses that I have presented is that beyond simply providing the speaker with a means for symbolizing and expressing a complex conceptualization, a grammatical construction also allows him to structure that conceptualization in a particular manner (Langacker 1987: 294). Grammar, in its sanctioning role, allows the speaker bountiful resources to draw on for using alternate grammatical constructions to express ideas related to a single conceptualization or scenario, as we saw in the river-crossing examples. It also allows for the expression of contrastive imagery as we saw in the examples drawn from English, Cora and Spanish. Distinct nuances of meaning can be expressed quite precisely through the choice of particular constructions. This was illustrated by the five Cora examples of the house lit up at night. The freedom that the speaker has for construing his conceptualizations in various ways has widespread grammatical consequences and needs to be an integral facet of any theory of language structure. As Giv6n aptly notes, the everpresent interpreter of events accompanies each act of coding a message for linguistic expression (1989: 90). Two kinds of figures and ground relationships, base and profile, and trajector and landmark

Seeing it in more than one way

45

also are central to many alternate construals as the discussion of Jackendoff's analyses showed. This paper has also strongly suggested that semantics is not universal, but is very highly constrained by the cultures within which people exist and interact. This suggestion is not new by any means, but has been a central tenet of such functional approaches as Pike's tagmemics, which has for many years insisted on an "ernie" analysis of linguistic data and has linked itself to a symbolic view of language by its insistence that grammatical units are form-meaning composites (cf. Pike 1967, 1988). To close, an increasing number of cognitive grammar analyses are beginning to relate the speaker's construal of situations to particular syntactic phenomena. For example, Achard (1993, to appear) relates the choice of French indicative and subjunctive inflectional marking in sentential complements to whether or not the speaker conceptualizes the complement as consisting of a proposition. Cook (1988) applies cognitive grammar to both the morphological and clausal structures of Samoan that relate to the notions of case marking and transitivity in Samoan. He finds as being essential to his analyses the use of constructs such as Langacker's idealized cognitive model of the finite clause, the speaker's construal of the situations that he is discussing and the role of prototypes as standards for categorization. Lindner (1981) presents a cognitive analysis of verbparticle constructions in English that nicely illustrates many of the main tenets of cognitive grammar. Smith (1987) is largely concerned with showing that the choice of accusative vs. dative case marking in German reflects a difference in conventional imagery, i.e., a difference in the way that the speaker construes and structures a scene in order to talk about it linguistically (Smith 1987: 51). Tuggy (1981) discusses various aspects of verb morphology in Tetelcingo Nahuatl (Aztec) that relate to transitivity; he provides cognitive analyses of noun incorporation, verb stem categorization and the causative-applicative suffixes. Both Hawkins (1984) and Vandeloise (1985) concern cognitive analyses of prepositions (Hawkin's dissertation those of English and Vandeloise's those of French), and illustrate clearly the extent to which the speaker's vantage point on a scene relates to the meaning of grammatical structures. Rice (1987) examines in detail three classes of syntactic phenomena that are generally left untreated by formal syntactic and semantic analyses of English data. These include a variety of prepositional verbs with ON, AT, OFF, FROM and WITH. In addition, Rice discusses the semantics of imperfective verbs such as accompany, contain, maintain, surround, admire, regret, feel and afford. The third class of data includes cognate ob-

46

Eugene H. Casad

jects, omitted objects, reflexives and reciprocals. She considers these data in relation to both the process of passivization and the sliding scale of transitivity proposed by Hopper and Thompson (1980: 251-253). Rice concludes that transitivity is to a significant degree a matter of the speaker's construal of the situation he or she is describing (Rice 1987: 184, 191, 261). Rubba (to appear) shows how the speaker's construal of mental events as either states or processes, to a large degree determines the choice of dative or accusative marking with German verbs of cognition. Finally, Verhagen (to appear) discusses the question of word order within a sentence in terms of Langacker's notions of subjectivity and objectivity. These are only a few of the studies that are now available, in addition to the papers that follow in this volume. Notes 1. I would like to express my thanks to Maurice Aldridge, Ronald W. Langacker, and John Taylor for their comments on this paper. 2. The Cora data on which this paper is based were collected between February 1971 and the present during the course of field investigations carried out under auspices of the Summer Institute of Linguistics. These data are culled from elicited materials such as word lists, extracted from texts, and gleaned from active conversations with Cora speakers. All the Cora data are from the Jesus Maria dialect. I would like to thank numerous Cora speakers, who continue to be my teachers in their language. 3. For a similar discussion of the usages of English run, see Smith 1987: 70-71. 4. Cook (1988) shows that middle clauses in Samoan are in some ways like transitive clauses, but that they are certainly not prototypical transitive clauses, but rather intransitive (1988: 67). 5. The contrast between pair and two persons discussed earlier illustrates the same point. 6. A more detailed account of these examples is given in Casad (1992). 7. Langacker uses these terms to characterize a maximal asymmetry between the conceptualizer and the object which he is conceptualizing (1990a; 1990b: 12). For example, to the extent that the conceptualizer explicitly includes his role as part of some conceptualized event or situation that he wishes to describe, any statement he makes about that scenario is being construed "objectively"; to the extent that he leaves his role in it implicit, the situation being described is being construed "subjectively" (cf. Langacker 1987: 128-132; 1990b: 151, 223). A second instance of the subjective vs. objective distinction is when a speaker treats time as a thing to talk about, i.e., an object of conceptualization in contrast to invoking conceptualizations within the framework of time, i.e., time is employed as the medium of conceptualization (Langacker 1990b: 168). As is the case with many distinctions, the difference is one of degree and few cases are purely at one extreme or the other of the continuum, although the speaker and the hearer who fall within the scope of grounding predications are maximally subjective (1991: 93).

Seeing it in more than one way

47

8. Another aspect of this analysis is that possibly Cora nYa-kw{:'i¢i'ipwa is a calque from Spanish estoy anzoleando 'I am fishing', which employs the imperfective participial form of the verb anzolear 'to fish with hook and line', cf. also Spanish anzeuelo 'fishhook'. 9. The tracks of World War I tanks were apparently themselves called "caterpillars".

References Achard, Michel 1993 Complementation in French: A cognitive perspective. [Unpublished Ph.D. dissertation, University of California, San Diego.] to appear "Complement construal in French: A cognitive perspective", in: Eugene H. Casad (ed.), Cognitive linguistics in the Redwoods. Berlin- New York: Mouton de Gruyter. Casad, Eugene H. 1982 Cora locationals and structured imagery. [Unpublished Ph.D. dissertation, University of California, San Diego.] 1984 "Cora", in: Ronald W. Langacker (ed.), Southern Uta-Aztecan grammatical sketches. Arlington, TX: The University of Texas and the Summer Institute of Linguistics, 152-459. 1988a "Conventionalization of Cora locationals", in: Rudzka-Ostyn (ed.), 345-378. 1988b "Post-conquest influences on Cora (Uto-Aztecan)", in: William Shipley (ed.), In honor of _Mary Haas: From the Haas festival conference on Native American linguistics. Berlin- New York-Amsterdam: Mouton de Gruyter, 77-136. 1992 "Cognition, history and Cora yee", Cognitive Linguistics 3(2): 151-186. 1993 "Locations, paths and the Cora verb", in: Richard A. Geiger- Brygida Rudzka-Ostyn (eds.), Conceptualizations in natura/language processing. Berlin- New York- Amsterdam: Mouton de Gruyter, 593-645. Casad, Eugene H.- Ronald W. Langacker 1985 '"Inside' and 'outside' in Cora grammar", IJAL 51: 247-281. Cook, Kenneth W. 1988 A cognitive analysis of grammatical relations, case, and transitivity in Samoan. [Unpublished Ph.D. dissertation, University of California, San Diego.] Geeraerts, Dirk 1988 "Where does prototypicality come from?", in: Rudzka-Ostyn (ed.), 207-229. Giv6n, Talmy 1989 Mind, code and context: Essays in pragmatics. Hillsdale, NJ- London: Lawrence Erlbaum Associates. Haiman, John (ed.) 1985 !conicity in syntax. Amsterdam-Philadelphia: John Benjamins. Hawkins, Bruce W. 1984 The semantics of English spatial prepositions. [Unpublished Ph.D. dissertation, University of California, San Diego.] Hoffman, Robert R. 1985 "Some implications of metaphor for philosophy and psychology of science", in: Paprotte- Dirven (eds.), 327-380. Hopper, Paul J.-Sandra A. Thompson 1980 "Transitivity in grammar and discourse", Language 56: 251-299. Jackendoff, Ray 1983 Grammar and cognition. Cambridge, MA: MIT Press.

48

Eugene H. Casad

Johnson, Mark 1987 The body in the mind. Chicago- London: University of Chicago Press. Kosslyn, Stephen Michael 1980 Image and mind. Cambridge, MA- London: Harvard University Press. Lakoff, George 1987 Women, fire. and dangerous things. Chicago: University of Chicago Press. 1990 "The invariance hypothesis: Is abstract reasoning based on image-schemas? Cognitive Linguistics 1: 39-74. Lakoff, George- Mark Johnson 1980 Metaphors we live by. Chicago: University of Chicago Press. Langacker, Ronald W. 1985 "Observations and speculations on subjectivity", in: Haiman (ed.), 109-150. 1987 Foundations of cognitive grammar. Vol. 1. Stanford, CA: Stanford University Press. "An overview of cognitive grammar", in: Rudzka-Ostyn (ed.), 3-48. 1988 "Subjectification", Cognitive Linguistics 1(1): 5-38. 1990a Concept, image. and symbol: The cognitive basis of grammar. (Cognitive Lin1990b guistics Research I.) Berlin- New York: Mouton de Gruyter. 1991 Foundations of cognitive grammar. Vol. 2. Stanford, CA: Stanford University Press. Lindner, Susan 1981 A lexico-semantic analysis of English verb-particle constructions with UP and OUT. [Unpublished Ph.D. dissertation, University of California, San Diego.] Manney, Linda 1993 Middle voice in Modern Greek. [Unpublished Ph.D. dissertation, University of California, San Diego.] Miller, George A.-Philip N. Johnson-Laird 1977 Language and perception. Cambridge, MA: Harvard University Press. Paprotte, Wolf-Rene Dirven (eds.) 1985 The ubiquity of metaphor: Metaphor in language and thought. Amsterdam: John Benjamins. Pike, Kenneth L. 1967 Language in relation to a unified theory of the structure of human behavior. (2nd edition.) The Hague: Mouton. 1988 "Cultural relativism in relation to constraints on world view-an ernie perspective", The Bulletin of History and Philology Academia Sinica 59(2): 385399. Rice, Sally Ann 1987 Towards a cognitive model of transitivity. [Unpublished Ph.D. dissertation, University of California, San Diego.] Rubba, Johanna to appear "The interaction of folk models and syntax: Case choice after prepositional verbs of cognition in German", in: Eugene H. Casad (ed.), Cognitive linguistics in the Redwoods. Berlin- New York: Mouton de Gruyter. Rudzka-Ostyn, Brygida (ed.) 1988 Topics in cognitive linguistics. Amsterdam-Philadelphia: John Benjamins.

Seeing it in more than one way

49

Smith, Michael Brockman 1987 The semantics of dative and accusative in German: An investigation in cognitive grammar. [Unpublished Ph.D. dissertation, University of California, San Diego.] Talmy, Leonard 1987 "Lexicalization patterns: Semantic structure in lexical forms", in: Timothy Shopen (ed.), Language typology and syntactic description. Part 3. Cambridge- New York: Cambridge University Press, 57-149. 1988 "The relation of grammar to cognition", in: Rudzka-Ostyn (ed.), 165-205. Taylor, John R. 1988 "Contrasting prepositional categories: English and Italian", in: Rudzka-Ostyn (ed.), 299-326. 1989 Linguistic categorization: Prototypes in linguistic theory. Oxford: Clarendon Press. Traugott, Elizabeth 1985 "Conditional markers", in: Haiman (ed.), 289-307. Tuggy, David 1981 The transitivity-related morphology ofTetelcingo Nahuatl: An exploration in space grammar. [Unpublished Ph.D. dissertation, University of California, San Diego.] Vandeloise, Claude 1985 Description of space in French. [Ph.D. dissertation, University of California, San Diego. Reproduced by the Linguistic Agency, University of Duisburg, Series A, Paper 150.] to appear "Touching: A minimal transmission of energy", in: Eugene H. Casad (ed.), Cognitive linguistics in the Redwoods. Berlin- New York: Mouton de Gruyter. van Hoek, Karen Ann 1992 Paths through conceptual structure: Constraints on pronominal anaphora. [Unpublished Ph.D. dissertation, University of California, San Diego.] to appear "A cognitive analysis of bound anaphora", in: Eugene H. Casad (ed. ), Cognitive linguistics in the Redwoods. Berlin- New York: Mouton de Gruyter. Verhagen, Arie to appear "Sentential conceptualization and linear order", in: Eugene H. Casad (ed.), Cognitive linguistics in the Redwoods. Berlin- New York: Mouton de Gruyter. Wierzbicka, Anna 1985 '"Oats' and 'wheat': The fallacy of arbitrariness", in: Haiman (ed.), 311-342.

Possession and possessive constructions Ronald W Langacker

1. Introduction The linguistic category of possession is clearly both universal and fundamental. Possessives are found in every language, are of frequent occurrence, and serve a variety of grammatical functions. Yet in many respects they remain quite mysterious. Although the idea of possession would seem to be self-evident, an adequate semantic description of possessive relationships has proved elusive. An affinity between possession and such notions as location, existence, and perfect aspect is readily demonstrated empirically but resistant to satisfactory explanation. Equally recalcitrant is the common use of possessives to express the subject of a clause or nominalization. Our purpose here is to examine these issues from the perspective of cognitive grammar. Following a brief sketch of that framework, proposals will be offered concerning both the semantic characterization of possession and the analysis of various possessive constructions.

2. Some fundamentals of cognitive grammar The theory of cognitive grammar offers naturalness, conceptual unification, and theoretical austerity. 1 Reflecting the semiological function of language-to allow the symbolization of conceptualizations by means of phonological sequences-it posits only semantic structures, phonological structures, and symbolic structures (form-meaning pairings). Lexicon and grammar are seen as forming a continuum whose proper characterization comprises only symbolic structures. It follows that all grammatical elements are attributed some kind of conceptual import (though it may be abstract, redundant, or tenuous). Whether lexical or grammatical, a symbolic element is often polysemous: it has not just one meaning but a family of related senses, usually clustered around a prototype. Some of these senses are schematic relative to others, representing the abstract

52

Ronald W. Langacker

commonality inherent in certain more specific values. There is no reason to assume, however, that a lexical or grammatical element invariably includes among its senses a highly schematic value with respect to which all of its other senses constitute instantiations (elaborations). 2 Meaning is a conceptual phenomenon (in the broadest sense). Any aspect of mental experience has the potential to be invoked as part of the meaning of a linguistic element or expression. The conceptions that serve in this manner-referred to as cognitive domains-can occupy any position along the parameters of complexity and abstractness. 3 They range from such basic notions as the experience of time, color, and spatial extensionality, at one extreme, to higher-order concepts, conceptual complexes, and even entire knowledge systems, at the other. Cognitive domains also run the gamut from rich, detailed conceptualizations to the highly abstract image schemas (e. g., container-content; source-path-goal; center-periphery; force) hypothesized by Lakoff (1987) and Johnson (1987) as being fundamental to cognitive structure and development. Fundamental in another sense are certain complex notions, intermediate in level of specificity, that are so ubiquitous in our experience and so obviously important both linguistically and cognitively that the term conceptual archetype is perhaps not inappropriate: our (schematized) conception of a human face; of the human body as a whole; of discrete physical objects; of canonical transitive events (agent-patient interactions); of a face-to-face verbal exchange; of using an instrument to affect another entity; and so on. Image schemas and conceptual archetypes are each essential to cognition and linguistic structure. I believe (counter to Lakoff and Johnson) that image schemas are best thought of as reflecting innate cognitive abilities essential for the emergence of any kind of structured mental experience. Some conceptual archetypes may also have an innate basis, but they are strongly shaped by experience and incorporate substantial conceptual content representing the commonality inherent in countless everyday bodily experiences made possible by image-schematic abilities. Moreover, certain image schemas and conceptual archetypes bear special relationships to one another. Corresponding to the conceptual archetype of physical motion through space, for instance, are more abstract kinds of motion that essentially reduce to mental scanning (Langacker 1986b). Corresponding to the archetypal conception of a physical container and its contents is the abstract, image-schematic conception (applicable to any domain) of an inclusion relationship (cf. Langacker 1987a: section 6.2).

Possession and possessive constructions

53

At least in initial stages of learning and thought, the schemas presumably cannot be manifested in isolation, but only within the complex archetypes they structure and allow to emerge experientially. At later stages of cognitive development, they are manifested in progressively more rarified bodies of conceptual content, and perhaps ultimately we can manipulate them directly (in certain kinds of abstract thought). Linguistically, I suggest that image schemas and conceptual archetypes combine to account for the universality and significance of certain fundamental linguistic notions, the former affording a schematic characterization applicable to all instances, and the latter serving to characterize the prototype. For example, I believe that every noun designates a region (or thing), defined abstractly as a set of interconnected entities, whose construal as such reflects the image-schematic ability of conceptual reification. The archetypal conception of a physical object provides the category prototype. Similarly, a subject is prototypically an agent, an archetypal semantic role, but if there is anything that all subjects have in commonand I believe there is-the commonality must be highly abstract: I analyze it in terms of our ability to impose figure/ground organization on a conceived situation. A comparable two-level characterization will be offered for possession. A linguistic expression's meaning involves not only conceptual "content" but also a particular way of construing that content. Numerous aspects of construal can be discerned, including the level of specificity at which a situation is characterized; perspective (e. g., vantage point, orientation, and the direction of mental scanning); scope (the array of conceptual content actually invoked); the construal of one structure against the background provided by another (consider metaphor, presupposition, and discourse phenomena); and various kinds of prominence. Two kinds of prominence are especially significant for grammatical purposes. First, every expression profiles (i.e., designates) some substructure within its scope. The noun intermission, for example, evokes as its base the schematic conception of a performance, and within that base it profiles a scheduled pause. An expression profiles either a thing or a relationship, 4 the nature of its profile determining its basic grammatical class. Thus a nominal expression (e. g., a noun, pronoun, or noun phrase) designates a thing, whereas a relational profile is characteristic of classes such as verbs, adverbs, adjectives, and prepositions. Relational expressions usually exhibit a second kind of prominence pertaining to their participants: one participant-termed the trajector-stands out as the pri-

54

Ronald W Langacker

mary figure within the profiled relationship; there may also be a secondary figure, referred to as a landmark (see Langacker 1991, chapter 7). The semantic contrast between above and below, for example, resides in whether the trajector is identified with the higher of two entities aligned along the vertical axis, or with the lower. Abbreviatory notations for the basic grammatical classes are given in Figure 1 (observe that heavy lines indicate profiling). A thing is conveniently represented by a circle, and a relationship by a dashed line connecting the relational participants, notably the trajector (tr) and landmark (lm). A verb is claimed to profile a process, characterized as a relationship followed sequentially in its evolution through time; indicating this sequential scanning is the heavy-line portion of the time arrow. Evolution through time figures less saliently in the relationships profiled by adverbs, adjectives, and prepositions, which are consequently said to be a temporal. 5 (a) thing

0

(b) atemporal relation

9" I

6zm

(c) process

Qu I

6lm

..,....

Figure 1

Grammar consists of patterns for the integration of simpler symbolic structures to form expressions of progressively greater complexity. Two or more component structures combine to form a composite expression, which can in turn serve as one component at a higher level of composition, and so on indefinitely. At each hierarchical level, semantic integration is effected by correspondences established between subparts of the components, and is symbolized by phonological integration effected in a comparable fashion. Specifying these compositional patterns are constructional schemas: skeletal representations of symbolically complex expression, reflecting any commonality inherent in their formation. Semantic integration is exemplified in Figure 2, which represents the prepositional-object construction. In 2(a), the preposition above combines with the nominal expression the table to form the prepositional phrase above the table. 6 Profiled by above is the spatial relationship between two things that are differentially positioned along the vertical axis.

Possession and possessive constructions

55

Table profiles a thing; for diagrammatic convenience, a rough sketch of its shape is used to abbreviate its multifaceted semantic specifications. The semantic integration is effected by a correspondence (represented by the dotted line) between above's landmark and table's profile. Because the former is characterized only schematically, it is termed an elaboration site (marked by hatching); the horizontal arrow indicates that the preposition's schematic landmark is elaborated by the nominal expression. From these two components, the composite structure is formed by superimposing corresponding entities and merging their specifications. The full prepositional phrase thus inherits the specifications of table, which serve to characterize its landmark. It is also usual for the composite structure to inherit the profiling of one component, in this case the preposition. The component structure that bequeathes its profile to the composite expression is called the profile determinant (identified by the heavy-line box enclosing it). (a)

above (the) table

pp

(b)

9tr

11r 9tr --above

--·D (the) table

I

'!1p

............

--0 NP

Figure 2

Sketched in Figure 2(b) is the constructional schema representing the abstract commonality of prepositional phrases in general. In lieu of a specific preposition like above, we find the preposition schema, whose profile is merely characterized as an atemporal relation. The second component structure is likewise schematic, being identified only as a noun phrase (X stands for any additional semantic specifications- but no particular specifications). Observe that this constructional schema, while lacking detailed conceptual content, is precisely parallel to any specific prepositional phrase in regard to such factors as profiles, correspondence,

and profile determinance; it can thus be thought of as a template for assembling novel expressions on the same pattern. This descriptive apparatus allows us to characterize certain basic grammatical notions. The profile determinant at any level of structural organization constitutes the grammatical head at that level. A complement is a component structure that elaborates a salient substructure of the head (hence the table is a complement of above in above the table). Conversely, a modifier is a component structure a salient substructure of which is elaborated by the head. 7 An object is describable as a noun phrase complement that elaborates the landmark of a relational expression. As a special case, a direct object is an object whose profile corresponds to the landmark of a transitive process profiled at the clausal level of organization. A noun phrase elaborating the trajector of the process profiled at that level constitutes the clausal subject (see Langacker 1990b, chapters 4 and 9; 1991, chapter 7).

3. The problem posed by possessives Possessives are problematic because: (i) they are used for so many kinds of relationships; (ii) there are numerous ways of marking possession, even in a single language; and (iii) possession apparently figures in a variety of special grammatical constructions. Our initial concern is the first problem. What sort of meaning (if any) can be attributed to an element that is used for all the kinds of relationships illustrated in (1 ), and many others besides? (1)

a. b. c. d. e. f. g. h. 1.

J. k. 1.

With respect to the possessor, the possessed may be: something owned (his Porsche) a relative (your aunt) a part (my knee) an unowned possession (the baby's crib) something manipulated (her rook) an associated individual (our waiter) a larger assembly (their group) something at one's disposal (my office) a physical quality (his height) a mental quality (her equanimity) a permanent location (our neighborhood) a transient location (my spot)

m. n. o. p. q. r.

a situation (your predicament) an action carried out (Oswald's assassination) an action undergone (Kennedy's assassination) something selected (your candidate [i.e., the one you back]) something fulfilling a certain function (our bus) something hosted (the dog's fleas)

I do not believe that all these uses are adequately or accurately analyzed simply as "ownership" meanings extended metaphorically from examples like (la). Clearly such metaphor is involved for some cases to some degree. Yet certain examples have no metaphorical feel to them whatever (e. g., Kennedy's assassination). They are metaphorical with respect to "ownership" only if the term metaphor is applied to all instances of categorization or extension from a prototype, in which case the term has no special value and we might as well simply speak of categorization or extension. 8 I should note, however, that an analysis based on metaphor is not inherently incompatible with the one I will propose, which emphasizes abstract commonality; in cognitive grammar, these can be seen as alternate facets of, or perspectives on, the same complex phenomenon. An alternative sometimes suggested is that any kind of association between two entities is sufficient to establish a possessive relation between them. 9 With this highly schematic characterization, two entities can participate in such a relationship merely by co-occurring in any cognitive domain (i.e., there need only be some conception involving them both). There is certainly something right about this account, as with the one based on metaphor, but it does not tell the whole story. First of all, it does not capture the fact that examples like (la-c) have the status of prototypes; ownership, kinship, and part/whole (especially body-part) relations are universally expressed by basic possessive structures. Often, in fact, nouns for kin and body parts are obligatorily possessed. Second, mere association fails to explain the striking asymmetries observed in possessor-possessed alignment: (2)

a. b.

the man's watch; the girl's uncle; the dog's tail; the dog's fleas; Kennedy's assassination *the watch's man; *the uncle's girl [meaning "his niece"); *the tail's dog; *the fleas' dog; *the assassination's Kennedy

In particular, an owner is virtually always portrayed linguistically as the possessor in a possessive construction, even though association per se

58

Ronald W. Langacker

is a symmetrical relationship. Similarly, kin terms and body parts virtually always participate as the possessed element. Cases like *the fleas' dog and *the assassination's Kennedy show that such asymmetries are not limited to prototypical instances. Clearly, something beyond mere association (conceptual co-occurrence) is going on: there is something very natural about these possessor-possessed alignments. A further clue is offered by cases where we might resort to the less natural alignment. For instance, the tail's dog might be used if we found a detached tail lying in the road and wondered what dog it belonged to. Likewise, the fleas' dog suggests that we are empathizing with the fleas, perhaps telling a story about a flea family that got lost on its way home and was looking for its host.

4. The reference-point model All these facts fall into place if we adopt a schematic characterization based on what I call the reference-point model. This idealized cognitive model represents such a basic and ubiquitous aspect of our moment-tomoment experience that it is easy to overlook and not so easy to describe. It involves the notion mental contact, which was adopted for independent reasons in cognitive grammar (notably the characterization of definiteness). By definition, to establish mental contact with an entity is to single it out for individual conscious awareness. The reference-point model is simply the idea that we commonly invoke the conception of one entity for purposes of establishing mental contact with another. For example, we typically do not have individual awareness of body parts, such as elbows and tails, except as they relate to whole individuals. We generally do not know individual elbows and tails, nor think of the world as being populated by such entities (which incidentally have people or animals attached to them). Rather, we think of the world as being populated by people and animals, and we become aware of particular elbows and tails only in the context of the individuals with respect to which they constitute parts. In normal circumstances, therefore, the individual as a whole is a natural reference point for establishing mental contact with a specific body part, but not conversely. Hence the dog's tail vs. *the tail's dog. Of course, special circumstances (e. g., encountering a disembodied tail) may result in the reversal of this natural alignment. Consider how this characterization accounts for the properties of possessive locutions. The open-ended variety of relationships coded by pos-

Possession and possessive constructions

59

sessive constructions, as in (1 ), reflects the ubiquity of the reference-point phenomenon. Using one entity to establish mental contact with another is, I suggest, a fundamental aspect of cognitive organization; we can regard it as a basic image-schematic ability. The fact that ownership, kinship, and part/whole relationships are prototypical for possessive constructions is a consequence of the possessor in each case saliently and naturally lending itself to reference-point function. Thus, we know and recognize people as individuals. It is far more natural and efficient to use a person as a "mental address" for locating a cluster of owned items than it would be to organize our "map" of entities around specific possessed objects (many of which would provide mental access to the same individual). Moreover, by its very nature a kinship term describes the designated person in relation to another person (ego) who serves as reference point for that purpose-one is not an uncle pure and simple, but only with respect to a reference individual (for this reason kin terms are often obligatorily possessed). I would further argue that the notion of a part, or of any specific part (tail; elbow) makes inherent reference to the whole that contains it. Thus ownership, kinship, and part/whole relationships are prototypical for possessives because, first, they are salient and ubiquitous in our experience, and second, they lend themselves especially well to reference-point organization. Beyond this, the reference-point analysis captures the asymmetries noted for possessive relationships, and accommodates the reversal of the canonical alignment when special circumstances dictate. When we come across a disembodied tail, or take the viewpoint of a family of fleas, these entities naturally become reference points for establishing mental contact with others that would normally have greater cognitive salience. A further advantage of this reference-point analysis is that it affords a straightforward account of the use of possessives for the periphrastic specification of the participants in a reified process, e. g. Oswald's assassination, Kennedy's assassination. The participants in a process are natural reference points for it: we cannot conceive of a process without in some way conceiving of the entities that participate in it (even if only in schematic terms); an event's location cannot be distinguished from the location of its participants; participants are usually concrete and easily pointed to within a scene, whereas relationships by their very nature are more abstract and not so easily localized; instances of the same event type are distinguished by the identity of their participants (e. g., Kennedy's assassination vs. Lincoln's assassination). In all these ways the conception

60

Ronald W Langacker

of participants anchors the conception of a process and provides a natural reference point for establishing mental contact with it. It also follows that either major participant should be able to serve this function, although the subject (trajector) does so preferentially by virtue of its greater salience as the figure within the scene. The reference-point model is sketched in Figure 3. The abbreviation C is used for the conceptualizer, R for a reference point, and T for the target, i.e., the entity with which mental contact is established. D labels a construct called the dominion, defined as the set of entities (or the region in conceptual space) that a given reference point provides mental access to; the significance of this notion will become apparent later. The dashed arrows represent possible paths through which the conceptualizer can establish mental contact with potential targets, and the heavyline arrows indicate the actual path used to establish contact with a particular target.

Figure 3

My most basic proposal is that this reference-point model constitutes the cognitive domain invoked for the schematic characterization of possessive elements. In terms of actual conceptual content, this domain is highly rarified -it is merely the apprehension of our image-schematic ability to mentally access one entity via another. There is nothing more that all possessive locutions can reasonably be claimed to share. Typically, however, this image-schematic relationship is manifested as a part of a more contentful conception (e. g., of ownership, kinship, etc.), so that the reference-point relationship is not itself individually recognized.

Possession and possessive constructions

61

It is nonetheless inherent and immanent in these more contentful notions, some of which approach the status of conceptual archetypes and provide the prototypical values of possessive elements.

5. Basic possessive constructions Fundamental and universal grammatical categories are seen in cognitive grammar as reflections of innate, image-schematic abilities, which afford the basis for a universally valid schematic characterization. At the same time, these abilities are manifested initially through everyday bodily experience giving rise to conceptual archetypes that function as prototypes of these categories. 10 I take possession-especially as manifested in noun determiners (e. g., my elbow; Sharon's car)-to be a universal category of this sort, and suggest that the reference-point model (involving the use of a reference point to establish mental contact with a target) is always inherent in the meaning of such expressions. Typically, however, this reference-point function coexists with (and is immanent in) other, more contentful conceptions of the relationship between reference point and target. Among these more specific relationships are some, like ownership and kinship, that have the status of conceptual archetypes and serve to characterize the category prototype(s).

5.1. Possessive determiners The purest, most basic and universal manifestation of the reference-point relationship I take as residing in determiner constructions involving a possessor nominal (often pronominal) and a possessed head noun. Whether marked by a special morpheme or inflection, or by mere juxtaposition of possessor and possessed in the modifying construction, I assume that this image-schematic relationship constitutes the abstract commonality shared by all expressions of this kind. At the most schematic level, therefore, I analyze the English possessive morpheme as shown on the lower right in Figure 4: it evokes the reference-point model as its base, and within that base it profiles the reference-point relationship, whereby the reference point affords mental access to the target. Since it is usually the trajector of a modifier that corresponds to the profile of the head noun, I assume that the trajector (relational figure) is properly

62

Ronald W Langacker

identified as the target. Figure 4 diagrams the formation of a typical expression of this sort. At the first level of composition, the possessive morpheme (Pass) combines with a noun phrase, in this case Jill, to form the possessive modifier Jill's. Their integration is effected by a correspondence between the schematic landmark of the possessive morpheme (namely the reference point) and the profile of Jill, whose semantic specifications (e. g., "female" and

0---Figure 4

Possession and possessive constructions

63

"human") are abbreviated as J. Since the possessive morpheme is the profile determinant, the composite expression Jill's also designates a reference-point relationship, one in which R is identified as Jill in particular. The target remains schematic, but at the second level of composition it is rendered specific by knife, whose semantic specifications are given as K. The two component structures at this level, Jill's and knife, are integrated via a correspondence between the former's trajector and the latter's profile. Since knife is both the profile-determinant and the elaborating structure, Jill's is a modifier of knife (as opposed to a complement). The overall composite structure, Jill's knife, designates a thing that is both characterized as a knife and identified as one for which Jill serves as reference point. It is well known that a possessed noun is usually definite. Why should this be? Definiteness is analyzed in cognitive grammar as residing in the presupposition that the speaker and addressee have each established mental contact with the same target entity, or can do so given the content of the noun phrase itself (Langacker 1991, chapter 3). With a definite possessor, therefore, the speaker and hearer already have mental contact with the reference point (this is precisely what makes it suitable for reference-point function); and from there, the possessive construction -as defined -is sufficient to put them in mental contact with the target (the referent of the head noun). In short, the definiteness of a possessed noun (given a definite possessor) is an automatic consequence of both possession and definiteness being characterized in terms of mental contact.

5.2. Possessive verbs As noted, basic determiner constructions like that in Figure 4 represent possession in its simplest, "purest" form. Other possessive constructions-notably those involving possessive main verbs, and those based on prepositions (or their equivalents)-tend to carry additional semantic "baggage" reflecting their historical origin. A possessive verb or preposition evolves by semantic extension and concomitant generalization from a source whose more specific meaning embodies a conceptual archetype. It is known, for example, that possessive main verbs typically develop from verbs with concrete meanings such as "grasp", "hold", and "keep", reflecting our archetypal experience of controlling and manipulating physical objects. As these verbs are extended metaphorically to more and more situations, where the notions of control and energy transfer are

64

Ronald W Langacker

either more abstract or less salient, a semantic characterization consistent with all uses becomes progressively more schematic. The eventual outcome may be a possessive verb like English have, whose uses cover the full spectrum exemplified in (3): from immediate physical control (3a), to ownership implying the possibility of physical access whenever desired (3b), to more abstract kinds of ownership and access (3c), to relationships in which the subject interacts with the object without in any way controlling it (3d), to cases where the subject is little more than a reference point for locating the object (3e). (3)

a. b. c.

d. e.

Watch out- he has a knife! I have a chain saw, in case I should ever need one. She has a substantial income. He often has migraine headaches. I have brown eyes.

Even a fairly bleached-out verb like English have retains certain vestiges of its physical origin. One vestige is an array of specific values involving physical or abstract control, which might be regarded as prototypical (or at least a center of gravity). Another vestige is the very fact that the relationship is expressed as a verb (whereas the possessive morpheme profiles an atemporal relation), which implies in this framework that the profiled relationship is followed in its evolution through conceived time. Yet another vestige of its origin is the trajactor/landmark alignment that is imposed on the relational participants and reflected at the clause level in subject/object choice. In contrast to the possessive morpheme, the possessor (rather than the possessed) is accorded the status of trajector (relational figure) and is thus coded by the clausal subject. Have's semantic value is diagrammed in Figure 5. The heavy-line segment of the time arrow indicates that the profiled relationship is scanned sequentially through its temporal evolution; this is the distinguishing property of verbs. The verb have profiles an imperfective process, i.e., it follows through conceived time a situation -enclosed by the inner rectangle-that is construed as being stable within the expression's temporal scope. 11 At the most schematic level, this situation is nothing more than the reference-point relationship (as in Figure 4). Usually, additional conceptual content (e. g., ownership or control) is invoked as well, and to some degree this content motivates the trajector/landmark alignment. At the extreme, however, all such content may be stripped away, leaving the reference-point function as the sole import of the possessive verb; its

'/:

Possession and possessive constructions

65

trajector/landmark alignment is then just a vestige of its metaphorical origin. I suggest that the reference-point relation is always there, even though it is generally obscured by virtue of being overlaid by more specific conceptual content.

I

•••

.....

Figure 5

One should not be bothered by the fact that have and the basic possessive morpheme impose opposite trajector/landmark alignments on the same configuration. Such alternations are very common and involve nothing more than our manifest ability to make alternate choices of figure within a scene. 12 Nor should one be bothered by the fact that the target rather than the reference point is characterized as the relational landmark. While this might at first seem contradictory, recall that the terms trajector and landmark are defined-at the most schematic levelsimply as the primary and the secondary figure within a profiled relationship. Trajector and landmark status are spotlights of focal prominence that can be directed at various entities within a scene; it is not implied either that the trajector necessarily moves or that the landmark serves in any literal sense to locate the trajector. Moreover, the reference-point model and the specific prominence relations imposed by have represent different levels of organization. The former evokes a fundamental imageschematic ability and functions as the cognitive domain that figures universally in the characterization of possessive constructions of diverse kinds; this domain per se does not imply any particular figure/ground alignment. By contrast, have represents just one possessive element serving to construe a situation in one particular way, in accordance with its origin as a verb of control. It is only in certain extended uses and schema-

66

Ronald W Langacker

tized senses of have that the reference-point relationship stands alone in essential isolation from the kinds of objective asymmetry that motivate the figure/ground organization it imposes.

5. 3. Locative possessives Consider next a possessive modifier formed with a locative preposition, as in these French examples: a.

(4)

b.

Ce livre est a moi. 'This book is mine.' C'est un ami a moi. 'It's a friend of mine.'

The preposition's possessive value is actually very similar to the one sketched for the basic possessive morpheme in Figure 4: it invokes the reference-point model and imposes the same profiling and trajector/landmark alignment as does the possessive morpheme. The difference, as seen in Figure 6, is that prepositional possessive elements construe the reference-point relationship by means of spatial metaphor. 13 In its spatial sense, a situates its trajector at the location of its landmark-either its exact location or its immediate vicinity, i.e., the region that conception of the landmark automatically invokes. This is quite analogous to the more abstract reference-point relationship, as indicated by the correspondences. Note in particular that the reference point's dominion is the abstract analog of the propositional landmark's immediate spatial vicinity.

SOURCE DOMAIN

I TARGET DOMAIN

(Space)

(Reference-Point Model)

Figure6

Possession and possessive constructions

67

Thus, whereas the possessive morpheme evokes the reference-point model by itself, a possessive preposition construes it (or structures it metaphorically) in terms of its spatial locative value. Now metaphors fade, so in many instances (or at a certain historical stage) this metaphorical structuring effectively disappears, yielding a "pure" possessive marker whose locative source is reflected only in its atemporal character and the figure/ground organization it imposes. But this represents the endpoint of a long evolutionary path; I presume that the examples in (4) still retain some degree of spatial motivation. As in the case of verbs, the historical source of a locative possessive represents a conceptual archetype grounded in everyday experience: that of locating objects in space relative to others which we can more easily find or reach. In English, of course, the main prepositional possessive element is of, as in (5a). (5) a. an acquaintance of Bill; the back of my neck; the mayor of San Diego b. most of the guests; a kernel of corn; a lump of coal; a row of trees c. the chirping of birds; the assassination of Kennedy I consider of to be a meaningful element rather than a semantically empty grammatical morpheme (Langacker 1992). Its value is not primarily spatial, however: schematically, of can be described as profiling an intrinsic relationship between its trajector and landmark. Often this is the relationship that a restricted inherent subpart bears to a whole. Many examples considered possessive are of this sort (e. g., the back of my neck), as are cases like most of the guests and a kernel of corn. Another kind of intrinsic relationship, illustrated by a lump of coal and a row of trees, is that which a whole bears to the substance or constitutive entities that form it. Yet another is the relationship between a (reified) process and its participants, as in (5c). Because its value refers to an intrinsic relationship, of is suitable for possessive function, especially when the possessed head noun is itself inherently relational. That is, a noun like acquaintance (also back, mayor, etc.) profiles a thing characterized with respect to the relationship it bears to some other, unprofiled thing; this is shown on the lower left in Figure 7, where the double line indicates an intrinsic relationship. When one entity bears an intrinsic relationship to another, which serves to characterize it, that other entity is naturally taken as a reference point for establishing mental contact with it. Thus, while of-phrases have a broad range of uses, not all of which are always considered possessive, their applicability to reference-point relationships is fully compatible with their meaning.

68

Ronald W Langacker

Indeed, the most abstract construal of of, namely the generalized notion of an intrinsic relationship (in contrast to more specific notions such as a part/whole relationship), implies the reference-point relation and has little if any additional content.

Figure 7

By now Figure 7 should require little explanation. It represents expressions like those in (5a), in which an of-phrase modifies a relational head noun. The noun (e. g., acquaintance) evokes an intrinsic relationship as part of its base, whereas the of-phrase (e. g., of Bill) profiles an intrinsic relationship between a schematic trajector and the landmark specified in the prepositional object (X abbreviates the landmark's specifications). As is usual in a modifying construction, the head elaborates the modifier's schematic trajector, based on a correspondence between that trajector and the head noun's profile. The other correspondences equate the intrinsic relationship evoked by the head with the one profiled by the of-phrase. The composite structure is obtained in the normal way: by superimposing and merging corresponding entities, and adopting the head's profile. Hence the full expression (e. g., acquaintance of Bill) designates a thing characterized in part by the intrinsic relationship it bears to another, specified entity.

6. Additional of-constructions One kind of intrinsic relationship is that which holds between a process and its participants. This makes of a natural choice for the periphrastic

Possession and possessive constructions

69

specification of processual participants when nominalization renders impossible their elaboration by means of the normal (i.e., clausal) subject and direct object constructions. The semantic pole of such an expression, e. g., (the) assassination of Kennedy, is sketched in Figure 8. Represented on the lower left is the head noun derived from a verb stem by nominalization. It evokes the conception of a process (assassinate), but profiles the abstract region obtained by conceptually reifying that process, as indicated by the heavy-line circle; assassination thus profiles an event construed as a thing. The double line indicates the intrinsic relation between such an event overall and one of its participants, in this case its landmark. The construction is exactly parallel to the one in Figure 7 involving a relational noun. N

Figure 8

Like an intrinsic relationship, the reference-point relationship also by nature lends itself to periphrastic use, since an event's participants are natural reference points for it. In Kennedy's assassination, for instance, Kennedy is the reference point that lets us establish mental contact with a specific instance of assassination. This possessive periphrasis construction is quite analogous to the of-periphrasis construction in Figure 8. The only difference is that Kennedy's profiles the reference-point relationship per se, whereas of Kennedy profiles the relationship of intrinsicness (which has a reference-point relationship as a consequence). Ofs "intrinsic relationship" value renders it especially suitable for possessive locutions involving relational nouns, as exemplified in (5a). Now

70

Ronald W. Langacker

jalongside expressions like a friend of Jill, we sometimes encounter phrases which are comparable except that possessive inflection appears on the prepositional object: a friend of Jill's. What should we make of this difference? I consider it important that Jill's occurs as the object of a preposition; this implies that it is nominal rather than relational, according to regular patterns of English. Recall that Jill's was analyzed as profiling a relationship in forms like Jill's knife (Figure 4). I would also attribute it relational value in predicate position following be (e. g., The knife is Jill's), for be takes relational complements (I would even analyze predicate nominatives as relational). But if Jill's is not relational but rather nominal in a friend of Jill's, what does it profile? The answer I propose involves conceptual reification as well as the notion dominion, defined as the region (or set of entities) that a particular reference point allows one to establish mental contact with. Though part of the reference-point model as previously characterized, this notion has not yet figured prominently in our discussion; in the constructions treated up to now, the reference point's dominion has remained latent rather than being singled out for specific attention. Together with our obvious capacity for conceptual reification (as witnessed in nominalization), this construct permits an analysis that is actually quite straightforward: in expressions like a friend of Jill's, the prepositional object profiles the possessor's dominion (an abstract region characterized with respect to the reference-point relationship). Of then expresses the intrinsic relationship between a part and a whole (i.e., between the target and the dominion overall). This analysis is diagrammed in Figure 9. The pivotal element is a semantic variant of the possessive morpheme, labelled Poss', that evokes the reference-point model but profiles the dominion (D) rather than the reference-point relationship. At the first level of constituency, the reference point anchoring that dominion is identified as Jill, and since Poss' functions as profile determinant, the composite structure Jill's profiles Jill's dominion-an abstract thing-and is thus a noun. At the second level of constituency, of assumes its prototypical value: it profiles the relationship that an inherent and restricted subpart bears to a whole. Ofs landmark is elaborated by Jill's in accordance with the regular prepositional-object construction, which hinges on a correspondence between the prepositional landmark and the nominal profile. An additional correspondence (special to this subconstruction) equates ofs trajector with the target implied by the reference-point model. The composite expression of Jill's therefore designates the inherent subpart relationship that an un-

Possession and possessive constructions

71

specified target bears to Jill's dominion. Finally, at the highest level of constituency, that target is identified as a friend (F) by virtue of a correspondence between the head noun's profile and the schematic trajector of the prepositional phrase. A second correspondence links the dominion's reference point (i.e., Jill) with the other individual evoked by friend. As a consequence, the composite expression (a) friend of Jill's is understood as meaning that the co-friend of the profiled individual is none other than Jill, the reference point implied by the possessive element.

Figure 9

72

Ronald W Langacker

Is this a plausible analysis? Can it reasonably be maintained that a form like Jill's usually profiles an atemporal relation but nonetheless, in this particular construction, designates instead an abstract region, normally latent, which is associated with that relation? The answer is clearly affirmative. The very existence of abstract nouns and nominalizations, together with the ubiquity of polysemy, suggest that this should be possible. More strikingly, we find a direct analog of the nominal sense of Jill's in expressions like those in (6), which many English speakers accept unproblematically. (6)

a. b.

Near the fire is warmer. Under the bed is all dusty.

If only on semantic grounds, the expression that appears in subject position has to be analyzed as nominal in character, even though it takes the form of a prepositional phrase. Given certain notions that are independently established in cognitive grammar, an analysis along these lines is easily described. An expression like near the fire or under the bed profiles a region in space. This region can be identified as the search domain associated with the prepositional phrase in its basic locative sense. In his discussion of English prepositions, Hawkins (1984) defined a locative expression's search domain as the region to which it confines its trajector, i.e., the set of the specific trajector locations that will satisfy its specifications. The subjects in (6) thus represent a productive pattern of semantic extension (or if one prefers, of zero-derivation) that is limited to a particular grammatical construction. Starting from a locative expression that profiles a relationship, this pattern yields a derived expression that is unchanged at the phonological pole, but distinct at the semantic pole by virtue of the profile being shifted to the region comprising the locative's search domain. 14 This shift is diagrammed in Figure 10 for the specific case of under the bed. The change in profile from a relation to a region-a type of thing- has the automatic consequence in cognitive grammar of effecting a change in grammatical category. With the profiling shown on the right in Figure 10, under the bed is nominal in character and thus eligible to function as a clausal subject. The nominal sense of Jill's derives from its basic relational sense in an exactly analogous manner. In fact, if we generalize the notion search domain from space in particular to other kinds of conceptions, we can say that a dominion is simply the search domain of a reference point.

Possession and possessive constructions

I SD I

73

--?1

6tr

under the bed (P)

under the bed (N)

Figure 10

7. Additional have-constructions As exemplified in (3), the possessive have ranges in value from a relationship of physical control and manipulation to one in which the subject is little more than a reference point for situating the object. A construction where its reference-point value clearly comes to the fore is the one in (7). (7)

a. b. c.

We have a lot of coyotes around here. They have a lot of armadillos in Texas. Do you have skunks in South Africa?

The only vestige of the original active sense in which the subject exerts control over the object is a vague intimation that the subject and object have at least some potential for interaction-e. g., in (7a) the existence of coyotes in the neighborhood is potentially relevant to us in some way. More salient, however, is the reference-point function itself: the subject serves mainly as reference point for identifying a spatial region within which the object exists. As support for this characterization, observe that the subject pronoun is interpreted as referring generically to all the people in a geographical region, and that the locative complement identifying that region is virtually obligatory. (I have a lot of coyotes would be interpreted as meaning that I own them.) Figure 5, depicting the reference-point value of have, is thus appropriate for sentences like those in (7), provided that we interpret it spatially. Given this semantic characterization, the construction is readily described grammatically. The sentential subject and object respectively elaborate the trajector (reference point) and landmark (target), as they do in general. The locative complement (e. g., around here; in South Africa) is relational in nature, specifying the spatial relationship between the target and the dominion. More specifically, the dominion is equated with the locative's search domain. So here is another construction in

74

Ronald W Langacker

which the notion dominion is not merely latent in the reference-point model but is specifically invoked for semantic and grammatical purposes. Constructions like this one verge on being existential: they express existence in a particular spatial location, with potential relevance to the reference point that is used to identify that location. Now suppose this notion of potential relevance (the last vestige of the original sense of physical control) should fade away entirely. Suppose, moreover, that the locative itself is sufficient to identify the spatial dominion (as with in South Africa, but not around here). The result will be an expression of existence in a location, such that the subject is no longer necessary for purposes of specifying either a locus of relevance or a reference point in space distinct from the dominion itself. I suggest that, in various languages, these adjustments yield locationallexistential have-sentences in which the reference point and dominion are no longer distinguished. That is, the reference-point and dominion functions collapse on a single, locational entity, as seen by comparing Figures 5 and 11.

I

•••

you

...

.

tr

I

• •• ilya

....

Figure 11

If the reference point and dominion are no longer distinct, what takes on the role of trajector (figure within the scene)? There are two possibilities, the first of which is illustrated by a construction in Mandarin Chinese. Here, as sketched in Figure ll(a), the trajector/subject is identified with the reference point/dominion itself. The sentences in question represent a special case of the general pattern exemplified in (8a), whereby the possessive verb you takes the reference point as its subject/trajector. 15 (8)

a. b.

W6 you shu. 'I have a book.' Zhuo-shimg you shu. 'The table has a book [on it].'/'There is a book on the table.'

Possession and possessive constructions

75

The special case that concerns us, and corresponds directly to Figure ll(a), are sentences like (8b), in which the subject names a location rather than a participant (Langacker 1987c). In expressions of this sort, the trajector, dominion, and reference point all coincide. The other possibility is for the trajector/subject to be construed more abstractly, as in the French construction of (9a). (9)

a.

II y a beaucoup de linguistes en Californie.

b.

II y a beaucoup de linguistes.

'There are many linguists in California.' 'There are many linguists.' /'Many linguists exist.' In addition to a voir 'have', the French existential formula il y a contains the adverbial particle y 'there' as well as the so-called "dummy" il 'it', the same element that occurs in impersonal expressions, in cases of "extraposition", with "weather" verbs, etc. It is natural to suggest that y codes the dominion (more precisely, it profiles the locative relationship that the landmark bears to the dominion). And as indicated in Figure ll(b), I analyze the trajector/subject il as profiling a kind of abstract setting, which can perhaps be characterized as a presentational frame invoked for purposes of introducing a participant or a situation. 16 I would not claim to have offered full, motivated descriptions of these constructions. For our purposes, the important point is that the reference-point value of have-type verbs, as sketched in Figure 5, offers a plausible basis for extension to other, related meanings compatible with the grammar of locative/existential constructions, as in (8) and (9a). Observe, moreover, that if the locative dominion is construed in maximally general fashion, thus being left unspecified, the resulting expression is purely existential, as in (9b ). I conclude that the reference-point model of possession offers a natural account of the relation between possession and notions of location and existence. What about perfect aspect, which is also commonly marked by havetype verbs, as in (1 0)? (1 0)

J' ai fini.

'I have finished.' Elsewhere (1990a; 1991: chapter 5) I have described the evolution of havemarked perfects in some detail, so I will not go into its complexities here (see also Carey 1993). It should nonetheless be apparent that the reference-point characterization makes it possible to see how evolution to perfect value comes about. For one thing, the notion of a reference

76

Ronald W Langacker

point-a temporal reference point in particular-is the central feature of perfect aspect. We have also observed, in regard to (7), that the control sense of have can weaken yet be preserved in vestigial form as a relationship of potential relevance between the target and the reference point. This dovetails nicely with the traditionally noted contrast between the perfect and a simple past, namely that the former is said to involve some notion of current relevance to the reference point. 17 Beyond this, suffice it to say that the evolution of a perfect auxiliary from the main verb have involves: (i) transfer to the temporal domain; (ii) taking the target to be a relationship (expressed by the perfect participle) rather than a thing; and (iii) the process of subjectification, a recurrent feature of grammaticization (Langacker 1990a).

8. Conclusion The foregoing discussion merely hints at the wide-ranging grammatical significance of the reference-point model and associated notions like mental contact and dominion (cf. Langacker 1993a). The model figures crucially in various grammatical constructions that have long been considered interesting because of their special properties, for instance those in (11): (II)

a. b.

II a !eve Ia main. 'He raised his hand.' Je lui ai casse le bras. 'I broke his arm.'

For (11a), where no possessive marking occurs in the direct object, I posit a special variant of the direct-object construction in which the verb's trajector (corresponding to the clausal subject) is construed as a reference point with respect to its landmark. In (11 b), similarly, the indirect object is construed as a reference point in whose dominion the direct object is located. With this kind of analysis, no rule of "possessor deletion" or "possessor ascension" need be posited to account for the meanings or the overt grammar of such expressions. The same notions figure in the characterization of the so-called "dative-shift" construction in English, of "double-subject" constructions (Langacker 1993b; 1991, chapter 8), and many other phenomena, even constraints on pronominal anaphora (van Hoek 1992). Exploring their applicability is an essential and exciting direction of future research in cognitive grammar.

Possession and possessive constructions

77

Notes 1. For a comprehensive introduction, see Langacker 1986a, 1987a, 1990b, 1991, and Rudzka-Ostyn 1988: Part 1. 2. For general discussions of polysemy and linguistic categorization, see Lakoff 1987; Taylor 1989a; Tsohatzidis 1990; and Langacker 1987a: chapter 10, 1990b: chapter 10. 3. Terminology varies (the notion "cognitive domain" is comparable to "frame" [Fillmore 1982] or "idealized cognitive model" [Lakoff 1987]). Moreover, not every terminological distinction is necessarily motivated (e. g., it is not evident that image schemas form a homogeneous class distinguishable in a principled manner from other basic conceptions). The essential point is to realize that the conceptions invoked for linguistic purposes are of diverse character. 4. Like thing (described above), the term relationship is used in an abstract technical sense (see Langacker 1987a: Part 2; 1987b). 5. This does not imply that the relation in question has no temporal extension, but only that it is viewed holistically, its temporal evolution being backgrounded. 6. The definite article is ignored to simplify the presentation. (See Langacker 1991: chapter 3.) 7. For example, above the table modifies chandelier in the higher-order construction (the) chandelier above the table. It qualifies as a modifier because chandelier is the head (profile determinant) at this higher level of organization and serves to elaborate the schematic trajector of the prepositional phrase. 8. I suspect, moreover, that a range of uses comparable to (1) would be found for possessives even in a communally-oriented culture that did not exhibit anything quite comparable to our concepts of property and ownership. One might say that "possession" rather than "ownership" per se is the relevant notion, the basis for metaphorical extension, but that does not solve the problem because the term possession is the one we are trying to explicate. 9. Bendix (1966) ascribes such a meaning to have, for example. 10. Thus the image-schematic ability of conceptual reification corresponds to the conceptual archtetype "physical object", which is prototypical for the noun category. Similarly, a subject prototypically instantiates the "agent" role archetype and is describable schematically as always being the figure within the relationship profiled at the clausal level of organization. 11. The sequences of three dots indicate constancy through time and the absence of intrinsic temporal bounding. 12. Other examples include the well-known contrast between like and please, as well as the active/passive alternation (which I analyze as a matter of figure/ground reversal). Note also expressions of the type Who belongs to this jacket?, which are quite normal for many speakers. 13. The contrast between possessive determiners and locative possessives can also be analyzed from a discourse perspective. Insightful discussion along these lines (quite compatible with the present analysis) can be found in Deane 1987 and Taylor 1989b, 1991. 14. All facets of this analysis can be strongly justified on independent grounds. The notion "search domain" is pivotal to a variety of grammatical phenomena (Langacker, 1993b ). Patterns of zero derivation are common, including productive ones that freely apply to complex novel expressions, e. g., the shift from a path-locative to an endpoint-locative sense exhibited by such expresssions as [The pond is j across the field, through the

78

Ronald W Langacker

woods, and over the hill. Semantic extension is commonly a matter of profile shift, as illustrated by the same example. It can also be limited to a particular construction (consider pig as a verb, found only in the verb-particle combination pig out). 15. The data are from Lyons 1967, which raised the issue of how possession, existence, and location are connected. 16. Detailed justification lies far beyond the scope of this paper. For some discussion, see Langacker 1991, chapter 8. 17. The notion of current relevance can be subsequently lost, as in the French passe compose.

References Bendix, Edward Herman 1966 Componential analysis of general vocabulary: The semantic structure of a set of verbs in English, Hindi, and Japanese. (Publication 41.) Bloomington: Indiana University Research Center in Anthropology, Folklore, and Linguistics. [International Journal of American Linguistics 32(2): Part II.] Carey, Kathleen 1993 Pragmatics, subjectivity and the grammaticalization of the English perfect. [Unpublished Ph.D. dissertation, University of California, San Diego.] Deane, Paul 1987 "English possessives, topicality, and the Silverstein hierarchy", Proceedings of the annual meeting of the Berkeley Linguistics Society 13: 65-76. Fillmore, Charles J. 1982 "Frame semantics", in: The Linguistic Society of Korea (ed.), Linguistics in the morning calm. Seoul: Hanshin, 111-137. Hawkins, Bruce 1984 The semantics of English spatial prepositions. [Unpublished Ph.D. dissertation, University of California, San Diego.] Johnson, Mark 1987 The body in the mind: The bodily basis of meaning, imagination, and reason. Chicago- London: University of Chicago Press. Lakoff, George 1987 Women, fire, and dangerous things: What categories reveal about the mind. Chicago- London: University of Chicago Press. Langacker, Ronald W. 1986a "An introduction to cognitive grammar", Cognitive Science 10: 1-40. 1986b "Abstract motion", Proceedings of the annual meeting of the Berkeley Linguistics Society 12: 455-471. 1987a Foundations of cognitive grammar. Vol. I. Stanford: Stanford University Press. 1987b "Nouns and verbs", Language 63: 53-94. 1987c "Grammatical ramifications of the settingfparticipant distinction", Proceedings of the annual meeting of the Berkeley Linguistics Society 13: 383-394. 1990a "Subjectification", Cognitive Linguistics I: 5-38. 1990b Concept, image, and symbol: The cognitive basis of grammar. Berlin- New York: Mouto1;1 de Gruyter. 1991 Foundations of cognitive grammar. Vol. 2. Stanford: Stanford University Press. "The symbolic nature of cognitive grammar: The meaning of of and of of1992 periphrasis", in Martin Piitz (ed.), Thirty years of linguistic evolution: Studies

Possession and possessive constructions

1993a 1993b

79

in honour of Rene Dirven on the occasion of his sixtieth birthday. PhiladelphiaAmsterdam: John Benjamins, 483-502. "Reference-point constructions", Cognitive Linguistics 4: 1-38. "Grammatical traces of some 'invisible' semantic constructs", Language Sciences 15: 323-355.

Lyons, John 1967 "A note on possessive, existential and locative sentences", Foundations of Language 3: 390-396. Rudzka-Ostyn, Brygida (ed.) 1988 Topics in cognitive linguistics. Amsterdam-Philadelphia: John Benjamins. Taylor, John R. Linguistic categorization: Prototypes in linguistic theory. Oxford: Clarendon. 1989a "Possessive genitives in English", Linguistics 27: 663-686. 1989b 1991 "Possessive genitives in English: A discourse perspective", South African Journal of Linguistics 9: 59-63. Tsohatzidis, Savas L. (ed.) 1990 Meanings and prototypes: Studies in linguistic categorization. London-New York: Routledge. van Hoek, Karen 1992 Paths through conceptual structure: Constraints on pronominal anaphora. [Unpublished Ph.D. dissertation, University of California, San Diego.]

What lack needs to have: A study in the cognitive semantics of privation Savas L. Tsohatzidis

1. Introduction George Lakoff (1987) has claimed that the verb lack offers a good opportunity for realizing that lexical meaning is not a matter of relationships holding between words and mind-independent worlds (a view that he describes as characteristic of "objectivist" semantics), but rather a matter of relationships holding between words and certain types of mental models that people use "for understanding the world and for creating theories about the world" (1987: 134)-a view that he regards (and recommends) as the central insight of "cognitive" semantics. My purpose in this paper is twofold. First, to show that the analysis of lack that Lakoff offers in support of this claim is seriously defective and therefore not really serviceable in the context of the "objectivist" vs. "cognitivist" dispute. Second, to specify some elements of an alternative analysis, which, though far removed from "objectivist" assumptions, is equally far removed from the particular brand of "cognitivism" that Lakoff is interested in promoting. Lakoff's argument from lack (1987: 134-135) runs as follows. First, he claims that an objectivist would regard a sentence like (1) as a paradigmatic instance of a sentence expressing an analytic truth,

Someone lacks something if and only if he doesn't have it.

(1)

and would therefore be committed to the view that lack and not have are paradigmatically synonymous expressions. Second, he claims that the fact that, in each of the following pairs of sentences, the second sentence is dubiously acceptable whereas the first it not, clearly shows that "lack and not have are not synonymous":

My bike doesn't have a carburetor. ?My bike lacks a carburetor. (3) The Pope doesn't have a wife. ?The Pope lacks a wife. Third, he claims that the nonsynonymy of lack and not have that these contrasts reveal is not only one that objectivists have in fact overlooked (2)

a. b. a. b.

H2

Stll'tls

L. Tsohatzidis

hut also one that they arc in principle unable to acknowledge: in order fur 1111 objectivist to count two expressions as nonsynonymous, the choice between them should make a difference to the states of the world represented by sentences containing them; but there is, Lakoff insists, not a single difference between the state of the world represented by a sentence contnining lack and the state of the world represented by a corresponding sentence containing not have, the difference being rather that the sentence contuining lack assumes, whereas the sentence containing not have does not ussume, a particular folk theory concerning (not how the world actually is, hut) how the world should be. Fourth, he claims that this distinctive role of lack could easily be captured by a nonobjectivist definition that would make the presence of that word sensitive to the activation of "an idealized cognitive model with a background condition indicating that some person or thing should have something, and a foreground condition indicating that that person or thing does not have it" (Lakoff 1987: 134-135), and that this definition would clearly enable one to explain the oddity of sentences like (2b) and (3b) by appealing to the obvious fact that our folk theories about the world do not tell us that bikes should have carburetors or that the Pope should have wives, but, on the contrary, that "a bike should no more have a carburetor than the Pope should have a wife" (Lakoff 1987: 134-135). The case of lack, Lakoff therefore concludes, provides a good opportunity for recognizing the virtues of cognitivist and the vices of objectivist approaches to the study of word meaning.

2. Amending the proposed analysis Before going any further, it will be important to incorporate into the discussion a qualification without which both the "cognitivist" and the "objectivist" accounts of lack (as Lakoff presents them) would appear to be unworthy of serious attention. To see the need for that qualification, notice that, although it makes sense to say both (4)

I should have this house alone I to myself

and (5)

I do not have this house alone I to myself

it does not make any sense to say, (6)

*I lack this house alone I to myself

What lack needs to have

83

and that, although it makes sense to say both (7)

We should have this house together I in common.

and (8)

We don't have this house together I in common.

it does not make any sense to say (9)

*We lack this house together I in common.

It is clear that strings like (6) and (9) should be counted as perfectly meaningful both by Lakoff's "objectivists" and by his "cognitivists", since, on the view he associates with the former, (6) is synonymous (5) whereas (9) is synonymous with (8), and, on view he associates with the latter, (6) is synonymous with the conjunction of (4) and (5) whereas (9) is synonymous with the conjunction of (7) and (8). Since, however, it is equally clear that (6) and (9) are not meaningful, both the "objectivist" and the "cognitivist" views of lack would appear to have to be immediately rejected. It seems to me that, in order to avoid this result, one should make a distinction between at least two senses of have-one in which it implies possession without necessarily implying ownership, and one in which it implies ownership without necessarily implying possession- and claim that it is the former, and not the latter, of these senses that have is intended to have both in the objectivist and in the cognitivist metalanguage. Since the examples that have just been cited as problematic involve occurrences of have in which it is clearly paraphrasable by means of own, as in (10) and (11) below,

(10)

I do not own I I should own this house alone.

(11)

We do not own I We should own this house together.

and since, conversely, the occurrences of have in examples of the sort appearing in Lakoff's argument do not admit of such paraphrases-since, in other words, (12) is no more a possible paraphrase of (2a) than (13) is a possible paraphrase of (3a)(12)

*My bike doesn't own a carburetor.

(13)

?The Pope doesn't own a wife.

it seems that the proposed amendment is appropriate, and I will from now on assume that it has been adopted (without, of course, pretending

84

Savas L. Tsohatzidis

to know how exactly the difference between possession and ownership could ultimately be elucidated). The problem before us, then, is the following: Can the problems apparently besetting the "objectivist" definition of lack in terms of a single "not have" component (where have implies possession though not ownership) be removed by means of a "cognitive" definition supplying an additional "should have" component (where, again, have implies possession though not ownership)?

3. Why the proposed analysis fails In order to demonstrate that the introduction of the proposed "should have" component would produce an incorrect definition of lack, one could, I presume, do either of two things: show that a statement where an entity is described as lacking a certain property does not lead to absurdity when conjoined with a statement denying that the entity in question should have that property; or show that a statement where an entity is described as lacking a certain property does lead to absurdity when interpreted as implying that the entity in question should have that property. Unfortunately for Lakoff's proposal, both of these things can be demonstrated quite straightforwardly. Consider first the following sentences: (14)

My proposal has many weaknesses, but it fortunately lacks arrogance, and it is precisely that feature that, in my view, no proposal should ever have.

(15)

Your essay isn't very original, but at least it lacks dogmatism, and, so far as I am concerned, that's the feature that no essay should ever have.

Neither of these sentences would strike anyone as absurd, but, on Lakoff's account of lack, they are as absurd as a sentence could be, since their speakers first imply, by virtue of using the verb lack, that certain entities of certain sorts should have certain properties, and then openly assert that no entities of these sorts should ever have these properties. The "should have" component introduced by that account is, therefore, a source of special disadvantages, rather than a source of special advantages, for definitions of lack that would happen to incorporate it.

What lack needs to have

85

Consider next the following sentences: (16)

As its name suggests, the empty set is the set that lacks members.

(17)

An intransitive verb, on the other hand, is a verb that lacks objects.

(18)

Vacuous names, of course, lack referents.

A mathematician, a grammarian and a logician might well utter (16), (17) and (18), respectively, in the course of trying to explain what an empty set, an intransitive verb, or a vacuous name is. But they would certainly find it absurd to add (or to be understood as implying) that empty sets are sets that do not but should have members, that intransitive verbs are verbs that do not but should have objects, or that vacuous names are names that do not but should have referents. And since, on Lakoff's account of lack, these suggestions not only are not absurd but actually constitute part of what (16), (17) and (18) quite literally mean, it is once more apparent that the presence of the "should have" component would increase the inadequacy rather than the adequacy of definitions of lack that would happen to incorporate it. I conclude that no statement of the semantics of lack would be correct if it included reference to a "should have" component, and that if the apparent difficulties of the objectivist definition of lack are to provide an opportunity for interesting cognitivist claims, a different kind of approach to the matter will have to be used. I will now sketch an outline of one such approach.

4. Sketch of an alternative analysis Consider the sentences in (19), (20), (21) and (22): (19)

Guatemala lacks a capital.

(20)

Mary lacks a lover.

(21)

Dogs lack the ability to speak.

(22)

Computers lack emotions.

In uttering them, their speakers do seem to assert what would be asserted by utterances of sentences like (23), (24), (25) and (26), respectively,

86

Savas L. Tsohatzidis

(23)

Guatemala doesn't have a capital.

(24)

Mary doesn't have a lover.

(25)

Dogs do not have the ability to speak.

(26)

Computers do not have emotions.

but this cannot be just what they purport to be doing, otherwise it would be impossible to explain-among other things-why sentences like (28), (30), (32) and (34) are, though false, acceptable, whereas sentences like (27), (29), (31) and (33) are definitely unacceptable: (27)

?Guatemala lacks a capital- indeed, there are no such things as capitals.

(28)

Guatemala doesn't have a capital-indeed, there are no such things as capitals.

(29)

?Mary lacks a lover- indeed, there are no such things as lovers.

(30)

Mary doesn't have a lover-indeed, there are no such things as lovers.

(31)

?Dogs lack the ability to speak- indeed, such an ability is nonexistent.

(32)

Dogs do not have the ability to speak- indeed, such an ability is nonexistent.

(33)

?Computers lack emotions- indeed, there are no such things as emotions.

(34)

Computers do not have emotions- indeed, there are no such things as emotions.

The distinctive purpose that lack-sentences are designed to serve is, I submit, twofold: on the one hand, to suggest that the entities referred to by their subject terms may be viewed as belonging to certain not explicitly mentioned categories, and, on the other hand, to suggest that the prototypical members of t~ose not explicitly mentioned categories do possess the properties that the entities in question are being described as not possessing; in short, their distinctive purpose is to suggest that, by virtue of not having the properties that they are said not to have, the entities referred to by their subject terms become atypical members of certain not explicitly mentioned categories. On this account, then, what the speaker of Guatemala lacks a capital suggests, in addition to saying that Guate-

What lack needs to have

87

mala doesn't have a capital, is that Guatemala belongs to a category of things (namely, the category of states) whose prototypical members do have capitals (hence the oddity of example [27], whose speaker purports to say not only that Guatemala lacks a capital, but that there is nothing else in the world that does have a capital). And what the speaker of Mary lacks a lover suggests, in addition to saying that Mary doesn't have a lover, is that Mary belongs to a category of things (the category of women of a certain kind) whose prototypical members do have lovers (hence the oddity of example [29], whose speaker purports to say that not only Mary lacks a lover, but that there is nothing else in the universe that does have a lover). Similarly, what the speaker of Dogs lack the ability to speak suggests, in addition to saying that dogs do not have the ability to speak, is that dogs belong to a category of things (namely, the category of living organisms) whose prototypical members (namely, human beings) do have the ability to speak (hence the oddity of [31], whose speaker purports to say not only that dogs lack the ability to speak, but that nothing whatsoever has that ability). Finally, what the speaker of Computers lack emotions suggests, in addition to saying that computers do not have emotions, is that computers belong to a category of things (namely, the category of things capable of intelligent operations) whose prototypical members (namely, human beings) do have emotions (hence the oddity of [33], whose speaker purports to say not only that computers lack emotions, but that there is nothing else in the universe that does have emotions). It will now be instructive to consider how the account just sketched can satisfactorily deal with examples that have already been shown to be impossible to analyse within Lakoff's framework, and also how it can offer a superior analysis even of those few examples whose treatment within that framework has appeared to be prima facie plausible. Consider first examples (16), (17) and (18) (repeated below for convenience): As its name suggests, the empty set is the set that lacks mem(16) bers. (17) An intransitive verb, on the other hand, is a verb that lacks objects. (18) Vacuous names, of course, lack referents. Although the "should have" glosses of these examples would be absurd, the ones suggested by the present account are perfectly intelligible. What the speaker of (16) suggests, on this account, is that, though the empty

88

Savas L. Tsohatzidis

set does not have (nor should have) members, it belongs to a category of things (namely, the category of sets) whose prototypical elements do have members. What the speaker of (17) suggests is that, though intransitive verbs do not have (nor should have) objects, they belong to a category of things (namely, the category of verbs) whose prototypical members do have objects. And what the speaker of (18) suggests is that, though vacuous names do not have (nor should have) referents, they belong to a category of things (namely, the category of names) whose prototypical members do have referents. Consider next examples (14) and (15) (again repeated for convenience): (14)

My proposal has many weaknesses, but it fortunately lacks arrogance, and it is precisely that feature that, in my view, no proposal should ever have.

(15)

Your essay isn't very original, but at least it lacks dogmatism, and, so far as I am concerned, that's the feature that no essay should ever have.

Although the fact that these sentences are acceptable makes the proposed "should have" condition on the definition of lack clearly untenable, their acceptability is hardly surprising on the present account. What the speaker of (14) suggests, on this account, is that the fact that the particular proposal under consideration does not show arrogance makes it an atypical member of the category of unsatisfactory proposals (whose prototypical, for him, members are characterized by arrogance)-hence the implication that the proposal under consideration is less than fully unsatisfactory after all. And what the speaker of (15) suggests is that the fact that the particular essay under consideration does not manifest dogmatism makes it an atypical member of the category of unsatisfactory essays (whose prototypical, for him, members are characterized by dogmatism)- hence the implication that the essay in question is less than fully unsatisfactory after all. Consider finally-and for reasons that will soon become apparentexamples (2b) and (3b), Lakoff's primary pieces of evidence in favor in his proposed analysis of lack: (2)

b.

?My bike lacks a carburetor.

(3)

b.

?The Pope lacks a wife.

Given that, as Lakoff puts it, "a bike should no more have a carburetor than the Pope should have a wife", his analysis does predict the acceptability pattern in these particular examples. And it is easy to see that,

What lack needs to have

89

on the analysis presently sketched, the acceptability pattern would be accountable as well: the oddity of (2b) would be attributed to the difficulty in imagining a situation in which the fact that a particular bike doesn't have a carburetor would make it an atypical bike (since prototypical bikes do not have carburetors either), and the oddity of (3b) would be attributed to the difficulty in imagining a situation in which the fact that the Pope doesn't have a wife would make him an atypical member of the Catholic clergy (since no members of the Catholic clergy have wives either). It would appear, then, that, at least as far as these examples are concerned, one would be free to choose either account. In fact, however, the situation is more complicated, and suggests that even here it is the account presently proposed that should be preserved. Notice that since, on this account, lack-sentences do not explicitly commit their speakers (or their interpreters) to a particular category with respect to which the typicality or nontypicality of the denotata of their subject terms is to be assessed, and since any given denotatum may obviously be categorized (and subsequently assessed for typicality or nontypicality) in many different ways, it is quite possible, within certain very broad limits, for the interpretative reactions to the same lack-sentence to differ widely, depending on the contextual availability of different categorization clues. In particular, it is possible, on the account presently proposed, for a given lack-sentence to be judged definitely unacceptable in a context where the only obvious categorization of the denotatum of its subject term makes it difficult to regard that denotatum as atypical, and for the same lack-sentence to be judged definitely acceptable in a context where a different categorization possibility has emerged, within which the denotatum of its subject term can easily be regarded as atypical. Now, Lakoff's account has absolutely no way to deal with sharp variations of this sort in the acceptability of lack-sentences, and, as long as such variations remain a merely theoretical possibility, this might be taken not to matter. But they are not a merely theoretical possibility, as can be shown by reference to the very examples on which he has chosen to base his account: (2b) and (3b) can certainly be definitely unacceptable in many contexts, but they are perfectly acceptable in the following ones: (35)

a. b.

No vehicle without a carburetor could ever do such a thing. My bike lacks a carburetor, but it did it.

(36)

a. b.

Why on earth wouldn't you like to be the Pope? Well, there are many reasons. For one thing, the Pope lacks a wife.

90

Savas L. Tsohatzidis

The participants to these dialogues may know full well that "a bike should no more have a carburetor than the Pope should have a wife", but this, contrary to Lakoff's expectations, hardly prevents them from being intelligible to each other. The reason is that their dialogues offer them enough material for proceeding to other than the usual categorizations of the entities denoted by the subject terms of their lack-sentences. If a particular bike is thought of simply as a member of the category of bikes, then the fact that it doesn't have a carburetor could hardly make it an atypical member of the category of bikes, and the relevant lacksentence will accordingly be felt to be odd. If, however, that particular bike is thought of as a member of the category of vehicles that have accomplished a certain spectacular feat, then the fact that it doesn't have a carburetor might well make it an atypical member of that category, and the very same lack-sentence will accordingly be felt not to be oddwhich is exactly what happens with the lack-sentence in (35). Similarly, if someone thinks of the Pope merely as a member of the Catholic clergy, then the fact that he doesn't have a wife will hardly seem to make him an atypical member of the Catholic clergy, and the relevant lack-sentence will be felt to be odd. If, however, someone thinks of the Pope as a member of the category of lucky occupants of prestigious positions, then the fact that his otherwise enviable position requires him not to have a wife may well be thought to make him an atypical member of that category, and the relevant lack-sentence will accordingly be felt not to be odd -which is exactly what happens with the lack-sentence in (36). In short, the presence of lack in a sentence of the form X lacks Y serves simply as an instruction to look for some category C in which X might plausibly be taken to belong, and which is such that the fact that X does not have the property Y would make it plausible to regard X as an atypical member of C. In the rare cases in which the search for such a category C will prove to be in vain, the lack-sentence will be found to be uninterpretable. But in the many cases in which that search will prove not to be in vain, the lack-sentence will have as many interpretations as there are contextually defensible allocations of things to categories.

5. Conclusions The idea that human categorization is, in most of its instances, an essentially context-dependent (and, hence, inherently unstable) phenomenonin particular, that it depends not so much on retrieving, and then applying to particular situations, invariant representations of concepts from

What lack needs to have

91

long-term memory, but rather on constructing variable representations of them from elements available in working memory-is one that is increasingly gaining recognition in cognitive psychology (see especially Barsalou 1983, 1987; Barsalou- Medin 1986; Medin- Barsalou 1987), and it hasn't escaped the attention of those linguists (for example, DrayMcNei111990) who have studied the subtle interactions of different types of sign vehicles in the construction of meaningful discourse. The fact, then, that, on the proposed account, reference to contextually determinable rather than to antecedently determined, categories is a built-in semantic feature of lack, should hardly be viewed as surprising. But, though not really surprising, it has a potentially surprising (and so noteworthy) implication on certain widespread views concerning the organization of semantic description. On these views (succinctly summarized in somewhat different terms by Atlas [1989: 25-65]), an adequate semantic description of a sentence should satisfy at least two demands. First, it should specify an interpretation that is context-invariant. And second, it should specify an interpretation that is fully determinate. Sentences containing indexical expressions cannot, of course, satisfy either demand, but this is usually regarded as a theoretically manageable problem, merely requiring a slightly more careful reformulation of the demands. On that reformulation, the first demand is that an adequate semantic description of a sentence should specify an interpretation that is context-invariant once the references of the sentence's indexical expressions have been fixed. And the second demand is that it should specify an interpretation that is fully determinate once the references of the sentence's indexical expressions have been fixed. Under these reformulations the demands are considered to be clearly satisfiable, and to supply, in a sense, the raison d'etre of semantic description. But lack-sentences are among clear cases of sentences showing that the two demands cannot simultaneously be satisfied. Even assuming that the references of the indexical elements of a sentence of the form X lacks Y have been fixed, its interpretation cannot be both fully determinate and context-invariant. For, in order to be fully determinate, it would have to specify the category C to which its utterer purports to allocate the entity referred to by X, and with reference to which he purports to claim that the fact that X doesn't have Y makes X an atypical member of C. But the category C cannot in general be fully determined just by looking at the sentence itself, or by perusing an imaginary "Great Book" in which every given entity would have been allocated once and for all to one or more categories, irrespective of particular contexts of thinking and speaking. It has to be determined, rather, by considering

92

Savas L. Tsohatzidis

who the utterer of the lack-sentence is, what the context of his utterance is, and what the probable character and direction of his thought can plausibly be supposed to be within that context. And so, the closer the interpretation of the sentence would come to being determinate, the further would it be removed from being context-invariant. One of the implications of this last point that seems appropriate to mention in concluding concerns the different perspective that it offers on the "objectivism" versus "cognitivism" dispute referred to at the beginning of this paper. The objectivist's claim that, in uttering a sentence of the form X lacks Y, a speaker just means that X does not have Y, cannot, of course, be right, since it cannot explain why the acceptability ranges of X lacks Y and X does not have Y can diverge. It does seem true, however, that the only fully determinate thing that each speaker of a sentence of the form X lacks Y explicitly says is, indeed, that X does not have Y -the remaining (and more interesting) part of what he means being not fully determinate but determinable on the basis of contextual information. So, the objectivist's most important mistake was to produce a description capable of acknowledging only the determinate and not the merely determinable aspects of a sentence's sense. But it would be just as mistaken on the cognitivist's part to try to supply the elements missing from the objectivist account in determinate, rather than in merely determinable, form- and it is this that Lakoff's "idealized cognitive model" for lack tries, and fails, to do, by assuming that there is but one, contextinvariant, answer to the question as to whether any given entity does or does not belong to any given category, and, hence, that there cannot be but one answer to the question as to whether any given entity "should" or "should not" have any given property. In an important sense, then, the "objectivist" and the "cognitivist" accounts of lack appearing in Lakoff's discussion suffer from a common mistake. And the real issue that emerges once that mistake is recognized is not whether mental categories play a role in semantic intepretation, but whether the role they undoubtedly do play in semantic interpretation is or is not best accounted for by conceiving of them as essentially contextual, and hence as inherently unstable, constructs. At least as far as the interpretation of lack is concerned, the answer to that question must, I think, be affirmative. 1 Note I. I am grateful to Brygida Rudzka-Ostyn and John R. Taylor for their reactions to an early version of this paper, and to I. Veloudis and Robert E. MacLaury for their reactions to a more recent one. The usual disclaimers apply.

What lack needs to have

93

References Atlas, Jay David 1989 Philosophy without ambiguity: A logico-linguistic essay. Oxford: Clarendon Press. Barsalou, Lawrence W. 1983 "Ad hoc categories", Memory and Cognition II: 211-227. 1987 "The instability of graded structure: Implications for the nature of concepts", in: Ulrich Neisser (ed.), Concepts and conceptual development: Ecological and intellectual factors in categorisation. Cambridge: Cambridge University Press, 101-140. Barsalou, Lawrence W.- Douglas L. Medin 1986 "Concepts: Fixed definitions or context-dependent representations?", Cahiers de Psychologie Cognitive 6: 187-204. Dray, Nancy L.- David McNeill 1990 "Gestures during discourse: The contextual structuring of thought", in: Savas L. Tsohatzidis (ed.), Meanings and prototypes: Studies in linguistic categorization. London-New York: Routledge, 465-487. Lakoff, George 1987 Women, fire, and dangerous things: What categories reveal about the mind. Chicago: The University of Chicago Press. Medin, Douglas L.- Lawrence W. Barsalou 1987 "Categorization processes and categorical perception", in: Steven Hamad (ed.), Categorical perception: The groundwork of cognition. Cambridge: Cambridge University Press, 455-490.

The construal of cause: The case of cause prepositions Rene Dirven

1. Some preliminaries 1.1 Concepts of cause The concept of "cause" has rarely been seen as ambiguous or complex. On the contrary, linguists, and especially generative linguists, have tended to look upon cause as a semantic primitive, i. e., as a clear, unitary, and undifferentiated concept. Some first signs of a differentiated cause concept made their appearance in discussions over the semantic nonequivalence of kill and cause to die, when the notions of "direct cause" and "indirect cause" were introduced (see, e. g., Wierzbicka 1975). Lakoff and Johnson (1980: chapter 14) have also pointed to the internal complexity of the notion of cause. This chapter expands upon these studies, by examining the concepts of cause that are expressed by the different English prepositions. As will become apparent in the course of the chapter, cause is a fuzzy and many-faceted concept. It may, nevertheless, be advisable to begin with a preliminary definition of the concept. As we proceed to examine the various expressions of cause, this preliminary definition will need to be refined and differentiated. Let us, then, tentatively define cause in terms of a situation (S 1) which triggers another situation (S 2), where the term situation covers states, events, processes and activities. The relation between S 1 and S2 is such that S2 is the result of S1; to put it another way, S2 would not have come about had it not been for S1 • The difference between cause and the absence of cause may be illustrated on the example of the two phrases John is angry at Mary and John is in love with Mary. In the first case, Mary has initiated some situation (S 1) which has had, as one of its results, John's being angry (S 2); were it not for Mary and what she did, John would not now be angry. In the case of John is in love with Mary, Mary has not necessarily triggered anything; the situation could well exist only in the world of John, where Mary is merely the goal, or object, of John's

96

Rene Dirven

emotional state; she has not "caused" John's emotional state. In German and Dutch, the notion of goal receives explicit linguistic expression through the use of the prepositions in and op (verliebt in 'enamored into', verliefd op 'enamored onto'). The English in love with rather incorporates the idea of accompaniment, as will be shown below. From these examples it will be clear that what is seen as a causal relation in one language (angry at) need not be construed in terms of cause in another language. The German and Dutch equivalents of angry at both incorporate a goal relation: bose auf, kwaad op. The above examples suggest that cause need not solely have to do with objectively ascertainable facts. Rather, cause emerges as a consequence of perspectivization, i.e., two events are seen to be linked together in a cause-result relationship.

1.2 Prepositions and the partitioning of physical space A previous paper (Dirven 1993) discussed the spatial and nonspatial meanings of twelve of the more common prepositions in English. The prepositions may be grouped and related to each other as in Figure 1. While the positioning of the prepositions in Figure 1 has a certain iconographic justification, no "deeper" significance is claimed. As Figure 1 tries to show, at, on and in are the most basic and most general of the space prepositions. At denotes location at a zero dimension or point, on denotes location at a one- or two-dimensional line or surface, and in location at a two- or three-dimensional area or volume (see also

(1)

(3)

on

in

-

(4)

I

(6)

by

through

(5)

(7)

with

about /

(8)

/

I I

/

under

c::_ - - -

-

I -- - - - -

(9)

I _j- -

--

-1- - - - -

I

from

off

out of

(10)

(ll)

(12)

Figure 1. Basic spatial prepositions

- over

The case of cause prepositions

97

Dirven 1989, Bennett 1975). Like at, by and with imply the proximity of trajector and landmark. 1 The antonyms of at, on, and in are from, off, and out of, which otherwise have similar spatial specifications. Like in, the prepositions through, about, and over imply a two-dimensional area or three-dimensional volume, but with respect to a path followed by the trajector. Something of an odd man out is under, which alone amongst the prepositions invokes position on the vertical axis.

1.3 Metaphorical senses of the spatial prepositions Table 1 summarizes the nonspatial uses of the twelve prepositions under discussion (Dirven 1993). The four sets of prepositions that were discussed in section 1.2 clearly show different degrees of aptness for metaphorical extension. Thus, the three basic spatial prepositions (at, on, in) can be used metaphorically for all six nonspatial domains. The proximity prepositions by and with may be used in four (or five) nonspatial senses; the path prepositions through, about, and over have fewer metaphorical uses, though each can be used for time; finally, the three source prepositions from, of, and out of exhibit the fewest extensions. These regularities call for an explanation. Is it a purely English phenomenon, or does it hold for other languages as well? In the latter case, we would be permitted to draw the inference that conceptualization processes exploit similar linguistic structures in different languages. Table I. Survey of the meaning extensions of prepositions

at on in by with through about over under from off out of

Time

State

Area

Means/ manner

Circumstance

Cause/ reason

+ + + +

+ + +

+ + + + +

+ + + + + +

+ + + + +

+ + + + + + + + + + + +

+ + +

+ + +

+

+ +

98

Rene Dirven

These issues lie beyond the scope of the present chapter. Here, I want to draw attention to a further significant aspect of Table 1, namely the fact that all twelve spatial prepositions can express cause, or reason. The following questions arise: 1. 11.

111.

Does language create different kinds, or categories, of cause? Are some kinds of cause-linguistically speaking-more prototypical than others? Do languages tend to select, as their protytypes, different kinds of cause?

In discussing these questions, I will try to present a systematic analysis of expressions of cause in English, focusing on the use of three sets of prepositions, viz., proximity prepositions (at, with, by), source prepositions (of, from, out of), and area or volume prepositions (in, about, over).

1.4 Corpus Existing word frequency data do not give semantic frequencies, i.e., frequencies of each meaning of a given item. A small data base was therefore compiled, which, due to its limited nature, can only suggest tendencies which are open to confirmation or falisification by more extensive corpora. The data base consisted of the 130 sentences containing a causedenoting preposition that occurred in the novel Nice Work by the British author David Lodge. A preliminary analysis of the expressions revealed the frequencies presented in Table 2. In spite of the limitations of the data, the figures clearly point to an interesting aspect in the structuring of causality by means of the English prepositions. 1.

Firstly, it is striking that proximity prepositions are so frequently seized upon for the expression of causality; they constitute over 50

Table 2. Frequencies of causal prepositions Proximity

with 36 Totals:

at

by

of

17

17

24

70

Area/volume

Source

from 7 33

out of 2

about 12

over 7

27

in

8

The case of cause prepositions

99

percent of all the instances in the corpus. Given the discrepancy between the proximity prepositions and the two other groups of prepositions, we may perhaps be entitled to assume that this is not just an incidental property of the corpus, but that it may have more general validity for English as a whole. ii. If so, the predominance of proximity prepositions in English contrasts sharply with prototypical prepositional expressions of causality in German and Dutch. German makes very frequent use of the preposition denoting spatial relations on the horizontal axis, viz. vor 'before', while in Dutch the source preposition van 'from' predominates. The question arises, whether the English preference for proximity prepositions, the German preference for a horizontal-axis preposition, and the Dutch preference for a source preposition, have any bearing on the three questions posed at the end of section 1.3? iii. Amongst the proximity prepositions, English shows a marked preference for the accompaniment preposition with; this is, in fact, the most frequently used preposition in the corpus. This is remarkable, for several reasons. Firstly, the German and Dutch equivalents of this preposition (mit, met) are not used with causal meaning, but only in the basic sense of accompaniment. Secondly, with is the least "spatial" of the prepositions under disucssion, in that it cannot be used to denote the mere spatial "togetherness" of inanimate entities. We cannot say *The telephone is with the table, but only The telephone is near the table. We can, however, say John is (standing) with Mary, since the notion of accompaniment is more natural for persons. Thirdly, while the expression of cause in English seems to prefer proximity prepositions, the contact preposition on is rarely used. As shown in Dirven (1993), on is the most spatial of the English prepositions, in that it always retains some of the spatial attributes of contact and support, even in its nonspatial uses. In the corpus, on occurs only three times with the meaning of "cause" (or reason), and, together with out of, is the least apt to express cause. Summarizing these preliminary observations, we may say that causality in English is structured preferentially in terms of two situations which are in such close proximity that the one situation (S 1) is seen to bear a causal relation to the other (S 2), whereby S~o expressed in a prepositional phrase, is the situation which triggers S2 . German, on the other hand, preferentially construes causality as a force (S 1) in front of (vor) some other situation (S 2), which triggers S2 . English for is rarely used in this sense. There were no instances in the corpus, although Martin et al.

I 00

Rene Dirven

(1984) cite the following examples with weep: weep for/over a person and weep for/with joy. The causal sense of for will not be further discussed here. Finally, in Dutch, cause is construed predominantly as a source (S 1) from which the other situation (S 2) emerges (van). Let us now proceed to a more detailed analysis, considering each preposition in turn, and contrasting it with its alternatives.

2. Cause as proximity Cause, construed in English in terms of two situations in close proximity, may be specified with respect to accompaniment (with), orientation (at), or connetion (by).

2.1 Cause as accompaniment (with) English has extended the spatial sense of accompaniment (John is standing with Mary) to the nonspatial sense of instrument (cut with a knife), manner (speak with enthusiasm), and circumstance (work with the radio on). Much the same holds for German and Dutch. All the above meanings can be expressed in these languages by mit/met. Differences between the three languages emerge with respect to the causal meaning of with. Whereas English has tremble with fear, German has zittern vor Angst, and Dutch beven van angst. The notions of "accompaniment" (with), "in-front-ness" (vor), and "source" (van) can be graphically represented in (1 ). (1)

a. b.

tremble with fear s2 o---o s1 zittern vor Angst

s2 c.

sl

beven van angst

s2

s1

At this point, it needs to be made clear that I am not claiming that accompaniment, "in-front-ness", and source constitute three distinct and unrelated concepts in the three languages. Rather, what we are dealing with is three different ways of construing the experience of cause. More specifially, I would claim that there is no overarching, schematic concept of cause in any of the three languages. What we can identify in different languages are a number of specific, and differentiated cause concepts,

The case of cause prepositions

101

such as accompaniment cause (English with), a forward-driving cause (German vor), a source cause (Dutch van, also attested in English die of cancer), and other specific cause concepts, to be discussed in due course. Some of these specific cause concepts-for instance, the concept of accompaniment cause-may be realized in some languages, such as English, but not in others, such as Dutch or German. 2 As far as English is concerned, we need to consider in which contexts the concept of accompaniment cause can and must be used. There are four main contexts (all examples are from the corpus): 1.

11.

Emotions as the accompanying cause of a physical reaction: besides themselves with jealousy; choke with anger and anxiety; explode with laughter; gasp with a shock; glow with well-being; leap with hope; moan with pain; oppressed with a sense of ... ; pale with anger; puffy with sleep; sick with the sense of betrayal; squeal with excitement; swell with certainty; tremble with fear; (grow) wary with resentment; wheeze with laughter. Persons or things as the accompanying cause of a psychological state: disgusted with the sight; irritated with her; obsessed with the place; pissed off with him; pleased with yourself

With respect to the above two categories, the causal meaning of with has, as far as I know, never been in doubt. More controversial is the following category: 111.

Events, inanimate objects, or nonhuman creatures as the accompanying cause of a state or event: gleam with plumbing; glisten with sweat; loud with voices; resonate with chatter; ring with noise; shake with the pounding of machines; swarm with bees.

With regard to the last example, a noncausal interpretation was proposed by Fillmore (1968), who considered pairs of sentences like those in (2) to be transformationally related. (2)

a. b.

The garden swarms with bees. Bees swarm in the garden.

On this account, garden and bees share the same deep cases, namely locative and objective, respectively. This interpretation has come in for repeated criticism, mostly on the grounds of its failure to distinguish between the holistic reading (2a) and the nonholistic reading (2b). But we can see now that the difference goes much deeper. The garden swarms with bees construes the situation in terms of a causal relation between the situation of the garden swarming and the situation of the presence of

I 02

Nl•Jii• /)irven

bees. In Bees swarm in the garden there is a mere locative relation, nothing more. In (2a), the causal relation entails a holistic interpretation, i.e., the swarming takes place everywhere in the garden. Characteristically, Dutch must here make use of source cause: De tuin gonst van de bijen 'The garden hums from the bees'. 3 The fourth context is perhaps even more debatable: IV.

Entities as the accompanying cause of a state: awash with litter; crammed with relics of yesterday evening's meal; cluttered with broken appliances; crowded with students; defaced with graffiti; neat and trim with goldfish ponds.

Indeed, even in Dutch one could, in the translation of many of these examples, use met instead of van. Here, we are dealing with a borderline category, where the accompaniment phrase can be interpreted either as cause or as circumstance. Even so, these examples are clearly very different from expressions denoting mere circumstance: (3)

It's not fair on the students, with the examinations coming. (Lodge 1989: 332)

Here, accompaniment as circumstance is reinforced by the comma intonation, and by the full propositional content of the examinations coming. The examples in (iv), however, go beyond mere accompaniment, and imply that S2 (awash, crammed, etc.) is a result of S1 (litter, relics). In summary, we can say that with human participants in events or states, with expresses accompaniment cause unconditionally; with nonhuman participants, accompaniment cause is unproblematic in the case of events, but somewhat marginal in the case of states. This should come as no surprise, given that with originated as a preposition denoting the accompaniment of humans (or animates), and cannot be used to denote the accompaniment of inanimates. To some extent, this restriction has been relaxed in figurative uses of the word, though not without leaving the traces that we have pointed out. 2.2 Cause as target (at)

Like with-phrases, an at-phrase can denote mere circumstances without explicitly suggesting a causal relationship: (4)

At this flattering paraphrase of her argument, Helen Lorimer's expression brightened (Lodge 1989: 338)

The case of cause prepositions

103

As in (3), the circumstance interpretation is favored by the loose syntactic linkage (comma intonation), as well as by the preposed position of the at-phrase. And if the evaluative participle flattering were omitted from (4), the at-phrase could even be interpreted as a mere time specification. The concepts of time and circumstance thus turn out to be more closely related than either is to the concept of cause. This is also reflected in the iconic organization of the sentence. When used in a causal sense, the propositional phrase typically follows the statement of the caused situation. (5)

She bridled a little at that. (Lodge 1989: 362)

Here, the situation of bridling ("throwing the head backwards in the neck") is construed, not only as occurring simultaneously with another situation (denoted by that), but also as occurring because of it. In fact, the at-phrase suggests a double movement. The at-event (or entity) triggers a reaction, and this reaction is directed towards the at-event (or entity). We can represent this graphically as follows: (6)

She bridled at his remark. s1~s2

The following groups of expressions exhibit the concept of "target cause". 1.

Events or entities as the target cause of a physical reaction: bridle at a remark; brighten at the prospect of; flinch at the word fuck; frown at a plan; guffaw at the scenario; (go) weak at the sound of ... ; weep for hours at the result.

In these examples, the temporal aspect and the causal relation are so intimately linked that the caused situation almost coincides with the causing situation. This is not necessarily the case with the following group: n. Events or entities as the target cause of a psychological reaction: angry at the teachers; displeased at the interruption; elated at having solved the mystery; (believe in) love at the first fuck; resentment at the desertions; self-righteous at being at her post before the boss; (look) sick at that; (feel a sense of) triumph at her conquest. Apart from the temporal aspect, there is no fundamental difference between the construal of cause in (i) and (ii). In both groups of expressions, an external situation causes a reaction in a human being, who directs his/ her reaction towards the very situation causing the reaction. Comparing at-causes and with-causes, we see that the latter denote, especially with

I 04

Renl: Dirven

humans, an internal cause, as in tremble with fear. It is not surprising, therefore, that with- and at-causes can be combined: (7)

She bridled with anger at his remarks (Martin et al. 1984: s3 s2 s1 181)

Here, S2 (anger) denotes the psychological, or internal cause of the physical response S3 (bridle), whilst S1 (remarks) denotes the external cause of S3 . That both types of cause can be combined in the same sentence is ample proof that the linguistic distinction (use of with vs. use of at) correlates with a conceptual difference. We may also take this as evidence in support of the claim made earlier, that the English prepositions do not denote a general, schematic concept of cause, but only specific construals of causal relations. In fact, the linguistic evidence is somewhat more complex, since alongside bridle at, we also have anger at. Example (8) summarizes the three possibilities, to which we must add a fourth, (8d)-though (8e) shows the limits to the use of at. (8)

a. b. c.

d. e.

She bridled at his remarks. She bridled with anger. She felt angry at his remarks. His anger at the teachers ... *She bridled at the teachers.

Clearly, the verbal predicates occurring with an at-phrase in group (i) require an event noun; they are incompatible with nonevent nouns, or nouns denoting human beings. The reason is clear. Given the temporal character of the at-phrase in group (i), we need event nouns, or nouns such as plan, or scenario that presuppose a psychological activity. On the other hand, the noun or adjective predicates in group (ii) denote a psychological state, and here the external cause can be a human being, or an event. The third proximity preposition that may denote cause is by. We shall now examine how by completes the conceptual field of proximity causes in English.

2.3 Cause as connection and path (by) The basic spatial sense of by is proximity and connection (sit by the fire, park the car by the kerbside, read by candlelight). The proximity of the two entities is not mere nearness, but a functional nearness, i.e., "connec-

The case of cause prepositions

105

tion". In dynamic contexts, this concept undergoes extension to that of path, whereby the trajector passes somewhere in the vicinity of the landmark object (go by the post office). In nonspatial contexts, the path concept may undergo metaphorization into a cause concept. Instantly, by some perverse chemistry of his body or nervous system, he feels tired and drowsy. (Lodge 1989: 15)

(9)

Interestingly, German and Dutch have also exploited a path preposition, viz. durch and door 'through', to express the concept of cause. Moreover, all three languages use the path prepositions, not only for cause, but also to express the concept of agent. 4 A remarkable difference between English on the one hand, and German and Dutch on the other, however, is that whereas English construes cause in terms of a connection path, German and Dutch construe cause as a path through a tunnel. In order to get a better overview of the differences between the causal senses of at and by, we can distinguish three groups of predicates, viz. those that can only take at, those that can only take by, and those that can take either at or by (or, in some cases, even with). (1 0)

a. ashamed at, overengaged at, thrilled at b. fascinated by, impressed by, oppressed by, put off by, scandalized by, transported by [figurative sense] c. alarmed at/by, amazed at/by, amused at/by, annoyed at/by, appalled at/by, disgusted at/by/with, disturbed at/by, excited at/by, irritated at/by/with, moved at/by, shocked at/by, surprised at/by, vexed at/by/with

In order to allow by, the predicate has to be the participle of a transitive verb; this explains why by does not occur in (lOa). In order to allow both at and by, the predicate must be able to function not only as a participle, but also as an adjective. This does not hold for the items in (1 Ob), which are still fully verbal, not adjectival in nature. Thus, at is used with nonverbal forms, by with verbal forms, 5 while the items in (lOc), which share verbal and adjectival features, can take either preposition. This neat formal pattern is matched by semantic differences, as may be seen from the following examples. (11)

a. b. c.

He was irritated at her request. He felt irritated with her and wished she would go. He was irritated still further by her tone.

Whereas adjectival irritated in (11 a) and (11 b) denotes a state, verbal irritated in (llc) denotes a process. As already pointed out, at and with

106

Rene Dirven

can be combined (irritated with her at her request). We see here that the psychological state of irritation can be caused both by an event (at her request) and by a human being (with her). If we compare these two external causes, it turns out that the event, being the act of a human being, is the "more external" of the two, while the human cause is a more permanent source of irritation. In (lib), we could even substitute her presence for her, thus showing more clearly the permanent character of the cause of the irritation. In contrast, causal by (llc) is needed if the predicate denotes a process rather than a state. That various languages tend to choose a path preposition to denote this kind of "process cause" is not surprising; a path is a natural link between two distinct states of affairs. That English prefers a connection path for this purpose should come as no surprise, either, given the general preference of English for proximity prepositions for the conceptualization of cause. 6

3. Cause as source The three source prepositions in English are zero-dimensional from (source as a point), one- or two-dimensional of(f) (source as a line or surface), and two- or three-dimensional out of (source as an area or volume). Whereas source prepositions are a major means for the expression of cause in Dutch (and for the expression of agent in German), source prepositions in English occupy second place, after proximity prepositions. In discussing the possibilities of the three source prepositions in the shaping of cause concepts, we will look both at the differences between the three specific concepts of cause as source, and at the more general differences between cause as source and cause as proximity.

3.1 Cause as separation from contact ( of[JJ) Of is a reduced form of off and "shares with this the basic notion of separation from a former contact situation" (Radden 1989: 565). On is used if contact is present, e. g., on the pill; giving up this contact is expressed by off (off the pill), or, in more figurative contexts, by of (deliver us of evil). In discussing of, one must of course bear in mind that English through the centuries has undergone considerable influence from French, in which

The case of cause prepositions

I 07

language a source preposition, i.e., de 'of', plays an even greater role in the expression of agents and causes than in the Germanic languages. But even taking the possibility of French influence into account, one can still see the major function of of as that of shaping a specific cause concept, viz. an internal, immediate, and "inherent" cause. Whereas the proximity prepositions at, with and by are used to differentiate the nearness of the cause (internal vs. external cause), source prepositions are used to differentiate the inherent nature of the cause. The difference between this specific cause concept and those expressed by at and with may be illustrated from the use of these prepositions with mad. With mad at we have a target as cause. The cause is external, and mad is taken, not in its literal sense, but as somewhat hyperbolic, i.e., in the sense "extremely angry". He was mad at me.

(12)

In mad with we have an internal, psychological cause, and again mad is not taken in its literal sense. Unlike mad at, mad with can be used with positively valued causes as well as with negative ones. Here, in (l3b), mad with has the sense "beside oneself with" (cf. section 2.1 ). (13)

a.

b.

mad with joy mad with pain (Martin et al. 1984: 769)

Finally, in (14), mad keeps its literal sense, and the cause is seen as an immediate and unavoidable source of the state: If we all taught the same thing over and over again we should all go mad or die of boredom, and so would our students. (Lodge 1989: 352)

(14)

Admittedly, go mad of is not a usual collocation. Nevertheless, the expression is conjoined with, and thus of the same semantic nature, as die of (boredom). Turning to the expression die of, we find that not only illnesses, but also situations of extreme hardship, such as cold, exhaustion, hunger, thirst, etc., are seen as internal, immediate and inherent causes of death: ( 15)

a.

die of anemia, of anorexia, of cancer, of a heart attack, of hunger, of migraine, ofpoisoning, of thirst

Psychological states may also count as internal and immediate causes of death (though not, perhaps, without a touch of hyperbole). (15)

b.

die of anxiety, of boredom, of jealousy, of loneliness

I 08

Rene Dirven

The English construal encoded in these expressions becomes more transparent when compared with equivalent expressions in Dutch. The Dutch equivalent of die of is either sterven aan 'die on' or sterven van 'die from'. Sterven aan is used only with the names of illnesses, never with cases of physical or psychological hardship, hence sterven van dorst 'die from thirst', not *sterven aan dorst. In fact, Dutch does not really have an equivalent of causal of(f); the closest cognate is af, the opposite of op 'up'. In spite of the above differences between English and Dutch, it should be noted that both languages do exploit the same "contact" metaphor to express cause of death. Dutch, namely, uses the "positive contact" preposition aan 'on', while English uses the "negative contact" term of(f). This difference may also go some way towards explaining why Dutch makes only limited use of the "inherent cause" category. One can construe people as being "in contact" with their visible illnesses, but not with invisible psychological states such as boredom, loneliness, hunger, or thirst. The example illustrates very clearly how a language can structure categorization, and hence conceptualization. English has not only categorized a number of phenomena as typifying immediate and inherent cause, as in (15a) and (15b), it has also developed a systematic contrast between these cases and a more external, circumstantial cause. With respect to die, external, or circumstantial cause is denoted by the preposition from.

3.2 Cause as separation from a point (from) We have seen that die of can only be used with illnesses or extreme physical or psychological hardship. In contrast, die from may denote all kinds of concrete phenomena, either entities or activities, which are construed as more remote causes, almost incidental to the main situation. die from alcohol, from drugs, from jogging, from making love, from overeating, from poison, from running too fast, from water in one's lungs, etc.

(16)

The inherent character of of-causes is also suggested by the greater degree of idiomaticity of the predicate plus of-phrase construction, as in (17). ( 17)

a. b.

afraid of, apprehensive of, ashamed of, complain of, frightened of, glad of, jealous of, proud of, scared of, sick of, tired of reek of, smell of, taste of

The case of cause prepositions

109

The contrast between die from and die of may be illustrated by the following quasi-minimal pair. (18)

a. b.

die from poison die of poisoning

Die from poison points to a substance which may have been around for some time, incidental, as it were, and detached from the activity of dying. The expression therefore suggests a mediate, or remote cause. Die of poisoning, in contrast, points to a process occurring in the body of the victim, whose effect follows directly and inevitably from the very nature of the process. A similar contrast is exhibited by the following quasi-minimal pair:

(19)

a. b.

She was shivering from the cold. She died of a cold.

Crucial to the interpretation of these examples are the articles. The cold denotes the external climatic conditions, and there is no suggestion that the cold was the direct or immediate cause. A cold, on the other hand, denotes an illness, which was internal to the sick person, and which directly triggered her death. Given the concept of indirect external cause, it is natural that from may be used with a wide range of predicate types; neither is it surprising that from typically does not constitute an idiomatic unit with the main predicate, as of typically does. Some striking examples of this greater variety and freedom are the following: (20)

His hair looked as light and fluffy as a baby's from all the washing and drying. (Lodge 1989: 283)

(21)

He wept at that, whether from mortification or gratitude, or a mixture of both, she couldn't be sure. (Lodge 1989: 296)

(22)

Philip Swallow swayed slightly on his feet, whether from inebriation or fatigue she couldn't tell. (Lodge 1989: 326)

(23)

... her cheeks, still shiny from last night's application of face cream. (Lodge 1989: 23)

Some of the above predicates would be acceptable with with (weep with joy, shiny with cream). These latter expressions typically imply that the cause (i.e., the joy or the cream) is currently present, as befits the basic sense of with, i.e., accompaniment. From-cause implies no such presence. On the contrary, examples (21) and (22) explicitly raise the question of

110

Rene Dirven

what the possible cause might be, while in (23) the cause (last night's application of face cream) is explicitly located at a prior time, such that only the effect of the process, but not the process itself, is directly present. The above characterization of from as a freestanding cause preposition seems to be contradicted by the high frequency of the idiomatic collocation suffer from: (24)

Prendergast suffered from Irritable Bowel Syndrome. (Lodge 1989: 117)

(25)

She suffers from migraines. (Lodge 1989: 261)

One might even ask: Why does English have die ofbut suffer from, while Dutch and German in both cases employ the contact preposition aan/an 'on': sterven aan/sterben an 'die of', lijden aan/leiden an 'suffer from'? For the sake of clarity, the contrasts between English and Dutch are set out in Table 3. If the picture looks puzzling, this is because one's first impulse is to expect the same extralinguistic facts to be categorized in the same way in different languages. But languages do not work on the basis of preexisting categories, which cut up reality into a narrow range of logical possibilities. Seen intralinguistically, however, the above data do exhibit a great deal of "logic", such "intralinguistic logic" being the very point I am trying to make in this paper. Let us first take Dutch aan 'on'. Aan presupposes a tangible contact situation. This fact explains why sterven takes aan only for illnesses; for the very same reason lijden also takes aan only for illnesses: the illness (or abnormality) is tangibly present and the contact preposition denotes the place where the suffering is located. But psychological states such as sorrow are construed as intangible causes, hence Dutch has only lijden van verdriet 'suffer from sorrow', not *lijden aan verdriet 'suffer on sorrow'. In other words, Dutch creates two categories, i.e., illness vs. all other sufferings. English employs the antonym of the contact preposition on, viz. of(f). This makes all the difference. Table 3. Prepositions with die and suffer, English vs. Dutch

Direct cause illness physical hardship psychological hardship Indirect cause

die

suffer

sterven

lijden

of of of from

from from from from

a an van van van

aan van van van

The case of cause prepositions

111

Die of is used for all direct, or inherent causes of death, whether these be illnesses or hardships, while processes or entities indirectly causing death are expressed with die from. Suffer can never be said to denote an inherent process like die. Although a number of ailments can lead to both suffering and dying, the relationships do not have to be seen as the same, as the Dutch use of the contact preposition aan testifies. The most obvious real-world difference between the situations denoted by die and suffer is that you can suffer from many more things (e. g., night blindness) than you can die ofl This is reflected linguistically in the fact that die of and similar expressions with causal of constitute a closed class of idiomatic collocations, whereas die from is one of an open-ended class of expressions.

3. 3 Cause as separation from a volume (out of) The corpus contained only two instances of out of as an expression of cause; the other low frequency prepositions were from (four instances), in (eight instances), and over (seven instances). It remains to be seen whether any more general significance attaches to these low frequencies. The two instances of out of represent strikingly different uses, namely an outer cause (25) and an inner cause (26). (25)

He knew his father got a kick out of being collected in the Jag. (Lodge 1989: 172)

(26)

She would probably drink too much sherry out of nervousness and burn the dinner or drop the plates. (Lodge 1989: 227)

As discussed in the previous section, outer causes can be expressed by from. Indeed, out of in (25) could be replaced by from, although from does not seem to work so well in (26). In (26), the person is in a nervous state which is seen as enveloping her and thus triggering her action. Also, out of expresses a less inherent, or less regular, link between cause and effect, or at any rate the link is less inherent than with from. On the other hand, the cause denoted by out of in (26) is clearly an internal cause. One can therefore "act out of love", but not "act from love". 7 In fact, you can even "be severe out of love", but you certainly cannot "be severe from love". These contrasts show that the out of-cause, though not frequent, may nevertheless be indispensable. But does out of really denote a unique concept of cause? The question needs to be asked, since from one point of view the construal of cause by means of out of is comparable to the construal of cause by means of the accompaniment preposition with.

I 12

Rene Dirven

The existence of the competing item with, which, moreover, is of extremely high frequency, could be the explanation for the low frequency of out of The question why out of remains in use at all is still pertinent, however. Perhaps the answer is the greater degree of vagueness which attaches to out of You can do practically anything out of practically any state or other, whereas the predicates and causes linked by with are subject to many more restrictions (section 2.1 ).

4. Cause as volume The discussion of out of has already introduced the notion of cause as volume. The analysis needs to be completed by considering its antonym in. Causal in is also a relatively low frequency item. Rather more frequent amongst the volume prepositions are about, denoting dispersion, and over, denoting back-and-forth movement.

4.1 Cause as enveloping volume (in) Like out oj; causal in may denote an outer cause (27) or an inner cause (28). (27)

But I really do not find much pleasure in going over manufactories. (Lodge 1989: 91)

(28)

If they had heard of them and their ideas, they would probably have recoiled in horror. (Lodge 1989: 82)

Predicates which, like find pleasure in, presuppose an outer cause include delight in, exult in, indulge in, rejoice in, revel in, triumph in, and a number of other idiomatic collocations. With regard to examples like (28), which denote an inner cause, we need to distinguish three successive stages in the event, viz.: (i) the existence of an outer state of affairs, often expressed by at; (ii) a psychological reaction to the outer state of affairs, expressible in a number of ways, e. g., by with, or out of; and (ii) a physical reaction, denoted in (28) by recoil (in). The difference between out of and in resides in the fact that the source preposition out of expresses a much clearer cause relation, whereas an in-phrase is typically ambiguous between several interpretations such as manner and circumstance, as well as cause. Whereas drink out of nervousness (26) expresses a clear causal relation, we cannot say the same of recoil in horror, since the horror is part and parcel of the recoiling activity, both causing it and constituting an essential feature of

The case of cause prepositions

113

it. The recoiling and the horror are really two aspects of one reality, whereas with out of the cause is clearly distinct from, and precedes, the effect. It is worth noting also that almost every instance of in-cause is ambiguous between cause, manner, and circumstance: (29)

'You British', said Penny Black, shaking her head in despair. (Lodge 1989: 94)

(30)

She punched her head with her fist in self-reproach. (Lodge 1989: 110)

(31)

Everthorpe fell back, in exaggerated astonishment. (Lodge 1989: 138)

(32)

Vic stared in wonderment. (Lodge 1989: 271)

Even so, these ambiguous examples are clearly distinct from the following examples of "pure manner" and "pure circumstance". Here, there is no causal relation between the two situations, they just happen to occur simultaneously. (33)

He smoked in silence.

(34)

He drove round to her house in a coldfury. (Lodge 1989: 165)

4.2 Cause as a dispersion path (about) In its spatial sense, about denotes motion along a dispersed path within an area or volume (heat about the bush). In its nonspatial uses, the preposition has only two senses, viz. area, or topic, and cause (Table 1). Of these, the area or topic sense is by far the more frequent. If we examine the distribution of prepositions with regard to the two domains of area and cause, we note an interesting continuum ranging from the separation preposition of (dream of), via the contact preposition on (lecture on), to the more general dispersion preposition about and the back-and-forth motion preposition over (Figure 2). AREA of dream of

on lecture on

CAUSE about talk about

I

excited about

--------t

I

over argue over Figure 2. Difference between area and cause

fight over

114

Rene Dirven

Mental activity predicates (dream) and communication predicates (lecture, talk) denote a goal-oriented activity, where the goal is realized as the area, or topic (indicated by the rightward-pointing arrows). Excited is an emotional predicate, and the combination with about triggers a causal interpretation (indicated by the leftward-pointing arrow), in that the about-phrase denotes the cause of the excitement. Also fight over can have a causal interpretation, as in The dogs fought over a bone. Often, fight over, like argue over, denotes a "verbal" fight only, in which case over can be ambiguous between topic and cause. In many contexts, causal about contrasts with causal at: (35)

a. b.

I am delighted about my promotion, but I am less delighted at the idea of living in the US.

The about-cause is seen as something that has been present for some time, and through which the mind or psyche moves as along a dispersed path; this underlying image may contribute to the notion of a more lasting, or more encompassing, situation. The at-cause suggests a single occurrence of an idea, or prospect, at a given point in time. Some predicates which allow both prepositions are the following. (36)

annoyed at/about, delighted at/about, excited at/about, guilty at/about, shocked at/about, vexed at/about

The above expressions denote purely emotional states; there is no component of mental activity or communication. This may, in fact, be seen as a characteristic of a prototypical about-cause. With the following predicates, only about is allowed, not at. (37)

a. b.

Verbs of caring: Care about, bother about, worry about. Predicates denoting feelings: apprehensive about, in despair about, enthusiastic about, obsessive about, queasy about, sceptical about, sentimental about, uneasiness about.

The predicates taking both at and about in (36) are mostly past participles that can readily take on a process interpretation and hence imply a point in time at which a causal interpretation can hold. 8 In contrast, the predicates in (37) are either verbs that express more permanent states, or adjectives and nouns that by their very nature also denote more permanent states of affairs. · A contrast of a different nature may hold between an about-cause and an of-cause, as in:

)'!,.{,.

The case of cause prepositions

(38)

a. b.

115

Half of Vic's apprehension about the Sunday lunch ... (Lodge 1989: 235) He was apprehensive for his son and of the future. (Martin et al. 1984: 84)

Although in both cases we are dealing with a more permanent state, the use of about with apprehension lends a nuance of caring, bothering, worrying, whereas the use of the separation preposition ofwith apprehensive induces the interpretation "scared", "afraid". In apprehension about the notion of area goes hand in hand with cause, but in apprehensive of the notion of source as cause prevails. In this way, the cause preposition helps determine the semantic interpretation of the item it is linked to, i.e., apprehension/apprehensive.

4.3 Cause as back-and-forth motion (over) Brugman (1989) has identified no less than 25 different spatial meanings of the preposition over. Its figurative uses, like those of about, are restricted to the domains of area (topic) and cause. A striking difference is that in its nonspatial meanings, over is far less frequent than about. The lower frequency goes hand in hand with a weaker distinction between the two senses than is the case with about. As evidence for this claim, consider the following examples: (39)

A few minutes later, Vic, evidently encouraged by the success of his intervention over "grooves", put his hand up again. (Lodge 1989: 338)

(40)

It you don't believe in love, why do you take such care over your students? (Lodge 1989: 362)

(41)

Even without alcohol, Wilcox became relaxed, almost expansive over the meal. (Lodge 1989: 200)

(42)

He was corrected several times by his financial director over figures. (Lodge 1989: 246)

(43)

He's been extraordinarily helpful to me over the article.

(44)

Denmark has rejected a protest over an agreement with the Baltic republics.

It will be readily apparent that the causal interpretation does not predominate, but neither, for that matter, does the area interpretation. Observe

116

Rene Dirven

that in none of the above examples could we replace over with about and get a satisfying result. It may be instructive to compare the above data with some dictionary examples. The following examples (from Martinet al. 1984) tend to have a more outspoken causal character. (45)

He fell in disgrace over debts; they got soft over Jane; cheat someone over a transaction; all this fuss over a trifle; weep over one's failure.

Also in these examples, about does not seem to be possible-or if it is, it gives rise to a different, usually area interpretation. Thus, make a fuss about nothing strongly suggests a communicative situation, only weakly present, if at all, in make a fuss over nothing. All in all, over as a cause preposition is very vague, but so too are about and in. We suggested earlier that these three propositions have a somewhat marginal status as prepositions of cause. This turns out to be justified, not only by their low frequencies as expressions of cause, but also by their twilight existence between circumstance and cause (in), and between area and cause (about, over).

5. Conclusions The above analysis has shown that most of the prepositions under discussion (out of seems to be an exception) allow different, albeit closely related interpretations. Thus at may waver between time, circumstance and cause, with between circumstance and cause, in between manner, circumstance and cause, about and over between area and cause. This fact has the important consequence that for each preposition, common meaning attributes are at least as important as the commonalities of the broader semantic roles (time, area, manner, circumstance, cause) for which they can be used. In particular, the specific attributes of each preposition make it possible for each to be used for a different construal of cause. Without wishing to subscribe to a componential, or feature semantics, we can summarize the different notions of cause expressed by the prepositions as in Table 4. As Table 4 shows, each of the eight prepositions-with the exception of about and over-exhibits a unique combination of attributes. The prepositions, in other words, encode at least seven different concepts of cause.

The case of cause prepositions

117

Table 4. Types of cause

of with in out of from at about over

Inherent

Immediate

Simultaneous

Internal

+

+ +

+ + +

+ + + +

+ + +

These have their origins in the spatial configurations denoted by these prepositions. Ultimately, then, it is the linguistic construal of space in a given language which has also structured the construal of the abstract conceptual domain of cause. Notes I. The terms trajector and landmark are from Langacker (1987). The landmark is a "refer-

2.

3.

4. 5.

6.

7.

ence point entity" (generally realized as the object of the preposition), while the trajector is the entity whose location is specified with reference to the landmark. Accompaniment cause may be a fairly recent development in English. According to the Shorter Oxford English Dictionary, the earliest example is in Shakespeare: Went they not quickly, I should lie with laughter. In fact, neither (2a) nor (2b) can be translated into Dutch by means of the equivalent of swarm, i.e., zwermen, In (2a), the sound-denoting word gonzen 'hum' has to be used, while a translation of (2b) employs the nominal form een zwerm: Er is een zwerm bijen in de tuin 'There is a swarm of beens in the garden'. The Dutch equivalent of (2a) contains not the slightest notion of locative, while the notion of locative is expressed twice, even three times, in the equivalent of (2b ), first by the existential/locative phrase er is 'there is', next by the nominal een zwerm 'a swarm', which, by denoting a physical entity implies location at some point in space, and finally by the locative adjunct in de tuin 'in the garden'. In contrast to English and Dutch, however, German does not employ a path metaphor for a human agent in the passive, but rather the source preposition von 'from'. It is not being claimed that the by-phrase in a passive sentence always expresses cause, rather that by is polysemous and that cause and agent are closely related. This is supported by a comparison with equivalents in other languages. The relation between by expressing cause and by expressing agent deserves closer attention. What is especially intriguing is that by can have the meanings of cause and agent even with nonanimate nouns: He was disturbed by a telephone call. Here, the physical telephone call may disturb him, i.e., interfere with his activities, or he may be vexed by what he was told in the call. The former is agent, the latter cause. Note the similar situation in Dutch: handelen uit liefde 'act out of love', but not handelen van liefde 'act from love'.

118

Rene Dirven

8. Guilty, though not a past participle, can readily be used in a temporal context: I felt a little guilty at having refused the invitation. This contrasts with the durative character of I was feeling frightfully guilty about missing my classes (Lodge 1989: 297). The adverbs of degree are also important in the choice of and difference between the two prepositions.

References Bennett, David C. 1975 Spatial and temporal uses of English prepositions: An essay in stratificational semantics. London: Longman. Brugman, Claudia 1989 The story of over: Polysemy, semantics, and the structure of the lexicon. New York: Garland. Dirven, Rene 1989 "Space prepositions", in: Rene Dirven (ed.), A user's grammar of English: Word, sentence, text, interaction. Frankfurt/M: Lang, 519-550. 1993 "Dividing up physical and mental space into conceptual categories by means of English prepositions, in: Cornelia Zelinski-Wibbelt (ed.), The semantics of prepositions in natura/language processing. Berlin- New York: Mouton de Gruyter, 73-97. Fillmore, Charles J. 1986 "The case for case", in: Emmon Bach-Robert T. Harms (eds.), Universals in linguistic theory. New York: Holt, Rinehart and Winston, 1-88. Lakoff, George- Mark Johnson 1980 Metaphors we live by. Chicago: University of Chicago Press. Langacker, Ronald W. 1987 Foundations of cognitive grammar. Vol. I. Stanford: Stanford University Press. Lodge, David 1989 Nice Work. London: Penguin Books. Martin, Willy eta!. (eds.) 1984 Groot Woordenboek Enge!s-Nederlands [Major dictionary English-Dutch]. Utrecht/Antwerp: Van Dale Lexicografie. Radden, Gunter 1989 "Figurative uses of prepositions", in: Rene Dirven (ed.), A user's grammar of English: Word, sentence, text, interaction. Frankfurt/M: Lang, 551-576. Wierzbicka, Anna 1975 "Why 'kill' does not mean 'cause to die"', Foundations of Language !3: 491528.

Conceptual grammaticalization and prediction Bernd Heine

1. Introduction One of the basic findings made in recent studies on grammaticalization is that there is a limited range of source concepts that are pressed into service for the expression of grammatical functions. This finding raises a number of questions, such as the following: a. What is the nature of source concepts? b. What is the relation between source concept and target concept? c. What are the factors determining the choice of a source concept as against other possible source concepts? d. Are these factors the same across languages and cultures? e. How does the nature of source concepts affect the linguistic structure of the resulting grammatical categories? All these questions have been discussed in some way or other in the relevant literature and a number of answers have been volunteered. We know, for example, that source concepts tend to be derived from domains of cognition that are basic to human experience, that they tend to be referred to in linguistic discourse more frequently than most other concepts, that they are more concrete in content than the grammatical functions derived from them, that one main parameter for choosing source concepts is to express abstract/grammatical notions in terms of "corresponding" concrete/lexical notions, that in spite of all the variation found in the languages of the world there are some striking similarities in the way source concepts are recruited, and that the linguistic characteristics of a source item are likely to survive to some extent even when the transition from source category to grammatical target category has been concluded. So far, however, none of the questions raised above has been answered satisfactorily (see Heine-Claudi-Hiinnemeyer 1991; Traugott- Heine 1991 for references). In the present paper, question (d) will be looked at in more detail. Our concern will not only be with similarities but also with the differences that exist across cultures in the choice of source concepts. The domain

120

Bernd Heine

chosen is that of basic spatial orientation. The discussion will be confined to what Clark (1973) refers to as "P-space", i.e., space as it is cognitively structured or perceived, rather than to L-space, the language of spatial relations (cf. Tanz 1980: 32), although our evidence is exclusively linguistic. A number of recent studies have been devoted to the transfer from concrete concepts to spatial concepts (Brugman 1983; Brugman- Macaulay 1986; Svorou 1986, 1987, 1994; Heine 1989; MacLaury 1989; Bowden 1991; Stolz 1991), some of them with a specifically cross-linguistic perspective (see especially Svorou 1994). The main purpose of the present paper is to look at the quantitative evidence that has become available on languages in two different parts of the world and to propose some generalizations on certain patterns of cognitive transfer. It will be argued that these generalizations are such that they enable us to predict linguistic evolution within limits.

2. The data In 1989, a survey of 125 Mrican languages was carried out to study the way in which some basic reference points of spatial orientation are developed and encoded in these languages (Heine 1989). Shortly thereafter, John Bowden (1991) conducted a similar survey in 104 Oceanic languages. In the following paragraphs, the findings of the two surveys are compared with a view to establishing whether, or to what extent, cognitive transfer patterns as they are manifested in language structure are similar or even identical across cultures and larger geographical areas. In the survey of adpositions in 125 Mrican languages (Heine 1989), five spatial reference points were chosen that can be assumed to be consistently distinguished across languages since each of them receives a distinct linguistic coding in the languages studied. These reference points, which are typically encoded linguisticially as locative nouns, adverbs or adpositions (prepositions, postpositions), are listed in Table 1. Table 1. Five spatial reference points and their linguistic equivalents

Reference point

Spatial relation

Typical linguistic expressions

ON

Superior, top Inferior, base Anterior Posterior Interior

up, above, on, on top of down, below, under, bottom before, in front (of) behind, back inside, in, within

UNDER FRONT

BACK IN

Conceptual grammaticalization and prediction

121

Two years later, John Bowden (1991) carried out a similar survey on 104 Oceanic languages. 1 The main concern of this paper will be with the question as to what kind of sources are employed for the linguistic expression of the five concepts listed in Table 1. Previous research (Svorou 1986, 1994; Heine 1989; Heine-Claudi- Hiinnemeyer 1991: chapter 5; Bowden 1991) has established that there are three main source domains: body parts ("back", "head", etc.), environmental landmarks ("ground", "sky", etc.), and static relational concepts ("bottom", "top", etc.; see below). 2 What appears to form the physical base for the body part model is the location of body parts of a human being in upright position. There is, however, an alternative model: one which is derived from the animal body. The zoomorphic model, as it has been called (Svorou 1994; HeineClaudi-Hiinnemeyer 1991), is said to be present when the relation between a spatial concept and the location of a given body part cannot be accounted for in terms of the human body but rather is suggestive of a model that has the body of a four-legged animal as its basis. We will assume that this is the case in particular when: is metaphorically derived from "back"; from "head"; and BACK from "buttock" or "anus".

1. ON

ii. iii.

FRONT

The occurrence of the zoomorphic model is limited; in Mrica, it appears to be largely confined to a few pastoralist societies of Eastern Africa, i.e., to ethnic groups typically leading a nomadic life whose survival depends on animal husbandry. These include peoples speaking Western Nilotic, Eastern Nilotic and Eastern Cushitic languages, inhabiting the semiarid and arid lands of East and Northeast Mrica. Another instance of it has been reported by Brugman (1983) and Brugman and Macaulay (1986) from Chalcatongo Mixtec, which has different nouns for human back and animal back. A table, for example, is conceptualized as an animal whose back is the top and whose belly is the underside, and the top of a wall is expressed with the noun for animal back, at the same time a wall also has a human back. It should be emphasized however that while there are languages which derive concepts of spatial orientation exclusively from human body parts, no language has been found to rely exclusively on the animal body. The domain of (static) relational concepts consists of items like "top", "front", "bottom", "backside", and 'inside" which show no distinct phys-

122

Bernd Heine

ical contours but exclusively refer to spatial relations. 3 Items belonging to this group are not treated here as source concepts on the same level with body parts and landmarks, in particular for the following reason: wherever there is sufficient historical information available these concepts turn out to be etymologically derived from either body parts or landmarks (Heine 1989). The Swahili lexemes nyuma 'backside' or chini 'bottom', which also occur as adverbs and prepositions, are examples of such relational terms. They are historically derived from the Proto-Bantu lexemes *(n)itma 'back' and *-ci 'earth, ground' (plus the locative suffix *-ni), respectively; on the synchronic level, however, they do not show any resemblance to other lexemes. We therefore will not treat them here as source concepts on the same level with body parts and landmarks, and they will remain out of consideration in the remainder of this paper. Similarly, instances where other sources are involved or where either no etymology or no consistent data is available will also be ignored here (but see Heine 1989; Bowden 1991). The relative significance of these domains is revealed by the available quantitative data for African languages in Table 2, and for Oceanic languages in Table 3. Table 2. The major source domains for the concepts

ON, UNDER, FRONT, BACK,

and

IN

in

125 African languages (Heine 1989)

Body parts Landmarks Relational concepts Other sources No information

ON

UNDER

FRONT

BACK

IN

Total

46 34 28

26 50 24 4 30

83

103 0

63

I

18 7 25

2 28

321 86 101 17 138

25

Table 3. The major source domains for the concepts

30 3 30

ON, UNDER, FRONT, BACK,

and

IN

in

I 04 Oceanic languages (Bowden 1991)

Body parts Landmarks Locative Other sources No information

ON

UNDER

FRONT

BACK

IN

Total

41 17 17 14 62

25 34 22 10 13

68 2 5 28 20

60 0 29 7 26

45 7 69 2 43

239 60 142 61 164

Conceptual grammaticalization and prediction

123

3. Some generalizations The main question that will be looked into now is: Is there any common denominator between African and Oceanic languages in the way the spatial concepts ON, UNDER, FRONT, BACK, and IN are derived from more concrete domains of cognition? In order to answer this question, the quantitative evidence provided by Heine (1989) and Bowden (1991) are analyzed in more detail. By comparing the data presented in Tables 2 and 3, for example, it would seem that the following generalization can be proposed: (1) If in a given language a lexical item is recruited for the expression of the spatial concepts ON, UNDER, FRONT, BACK, or IN, then the first choice will be a body part term, the second choice being a term denoting an environmental landmark. While this generalization summarizes the quantitative evidence presented in Tables 2 and 3, it is not very informative considering the enormous variation to be observed between individual spatial concepts and languages. Each of the five concepts will therefore be looked at in turn. The following generalizations would seem to suggest themselves, again based on the quantitative data of Tables 2 and 3: (2) If in a given language a lexical item is recruited for the expression of the spatial concept ON, then the first choice will be a body part term, the second choice being an environmental landmark term. (3) If in a given language a lexical item is recruited for the expression of the spatial concept UNDER, then the first choice is an environmental landmark and the second choice a body part term. (4) If in a given language a lexical item is recruited for the expression of any of the spatial concepts FRONT, BACK, or IN, then the first choice is to draw on a body part term, all other domains being largely irrelevant. These generalizations are still rather unspecific since they are based on the lumping together of a number of individual concepts. We will therefore look at one particular domain in more detail in order to qualify these generalizations. The domain chosen is that of parts of the body. Tables 4 and 5 provide a split up for the various body parts serving as lexical sources in Mrican and Oceanic languages, respectively.

124

Bernd Heine

Table 4. The main source concepts of the body part domain for five concepts of spatial orientation in 125 African languages (Heine 1989)

Body part "back" "belly/stomach" "face"

'head' "buttock/anus"

ON

UNDER

2

Total

BACK

IN

80 58

2 40

47 6 22

22 14 8 6 6

"eye"

"forehead" "mouth" "breast" "foot" "palm of hand" "heart" "shoulder" "chest"

FRONT

4 3 2 2 2 46

26

89

103

63

Total 82 58 49 46 44 14 8 6 6 5 3 2 2 2 327

Table 5. The main source concepts of the body part domain for five concepts of spatial orientation in 104 Oceanic languages (Bowden 1991)

Body part

ON

UNDER

FRONT

"back" "face"

"head" "belly/stomach" "tooth" "feet, legs" "breast" "heart" "shoulder" "liver" "bowels" "forehead" "'waist" "tongue" "thigh" "hair"

BACK

57

2 2 45

232

8 12

5

10 8 6 4 5 5 2

3 2

2 3 3 2

"forearm" "throat" All others

2

41

3

2

18

68

60

Total 57 55 25 13 12 10 8 6 5 5 5 5 4 3 3 3 2 2 2 7

49

6 25

"'mouth"

Total

IN

Conceptual grammaticalization and prediction

125

On the basis of the quantitative data presented in Tables 4 and 5, the following additional generalizations can be formulated: (5) If in a given language a lexical item is recruited for the expression of the concept ON, then the first choice is the noun "head". While a number of other body part terms are used as well, such as "face", "shoulder", ''forehead", "hair", or "back", none of them appears to constitute a quantitatively significant alternative to "head". (6) If in a given language a lexical item is recruited for the expression of the concept FRONT, then the first choice is the noun "face". Again, there are a number of other body parts that have been employed for the expression of FRONT, like "forehead", "breast", "chest", "belly/ stomach", "mouth", and "forearm", but none of them appears to seriously compete with "face". Note that in various Mrican languages, "eye" serves as a source for FRONT; conceivably, this is due to the fact that the concept "face" is etymologically derived from "eye" in many of these languages, which suggests that there appears to be a certain cognitive relation between these two body parts (see Heine-Claudi- Hiinnemeyer 1991: chapter 5). (7) If in a given language a lexical item is recruited for the expression of the spatial concept BACK, then virtually the only source is the body part term "back". This generalization does not seem to be invalidated by the fact that in a number of Mrican languages, the noun "buttock/anus" constitutes an alternative source model; we will return to this point in section 5. These observations suggest that, of the three major subparts of the body, namely head, trunk and extremities, the latter are rather insignificant as a source for the spatial concepts considered: as Table 6 shows, there is only a minority of terms for extremities that have been used for the expression of these concepts, both in Mrican and Oceanic languages. In addition to these positive generalizations, there are also a number of negative ones. One example is provided by the virtual insignificance of extremities such as "hand", "arm", or "leg" as cognitive models for the spatial concepts in question; 4 apart from the link between "foot/leg"

126

Bernd Heine

Table 6. The relative contribution of the head, the trunk, and the extremities as sources for the concepts ON, UNDER, FRONT, BACK, and IN in 125 African and 104 Oceanic languages African languages

Oceanic languages

Number of instances

Number Percentages of instances

Percentages

Head Trunk Extremities

123 196 8

38% 60% 2%

108 107 17

47% 46% 7%

Total

327

100%

232

100%

and UNDER, no other significant link could be established. Furthermore, while certain parts of the body are consistently selected for the expression of spatial concepts, others tend to be equally consistently avoided. Why, for example: is "face" or "forehead" or "breast" commonly exploited for FRONT, but apparently never "nose", "cheek", "chin", or "navel"? ii. is "back", but hardly ever "spine", employed for the expression of BACK? m. are "stomach" or "heart", but not "lung", "liver" or "bile" commonly employed for IN? 5 IV. are the genitals consistently avoided as a conceptual pool for spatial concepts? 1.

Further research is required on these questions.

4. Geo-cultural differences While the extent of agreement in the conceptualization of spatial notions between the two continents considered is remarkable, there are also a few conspicuous divergencies, as will now be demonstrated. One of these divergencies concerns the kinds of body part that have been recruited. As Table 7 suggests, of all body part items, the ones most frequently employed are "back", "face", and "head". As we saw above, these three form the primary sources, respectively, for BACK, FRONT, and ON in both

Conceptual grammaticalization and prediction

127

African and Oceanic languages. There are, however, two more body parts which are of primary importance in Mrica but insignificant in Oceania, namely "belly/stomach" and "buttocks/anus", while "tooth" is a relatively important source notion in Oceania but none at all in Mrica. This finding would seem to account for an observation that can be made on the basis of Table 6, according to which in Mrica the human trunk constitutes the primary source domain, while in Oceania, the head and the trunk are of roughly equal importance. Table 7. The most important body parts serving as a source for ON, UNDER, FRONT, BACK, and IN in 125 African and 104 Oceanic languages, respectively (Heine 1989; Bowden 1991)

Body part

"Back" "Face"

"Head" "Belly/stomach" "Buttock/anus" "Tooth" All others Total

African languages

Oceanic languages

Total

Total

Percentage

Percentage

82 49 46 58 44 0 48

25% 15% 14% 18% 13% 0% 15%

72

25% 22% 11% 5% 0% 5% 32%

327

100%

232

100%

59 49 25 13 13

That the human body as a pool for spatial orientation is conceived differently in the two continents can also be derived from another observation. There is a remarkable difference in the conceptualization of uNDER and IN: whereas in Mrica the primary body part source for UNDER is "buttock/anus", it is "foot/leg" in Oceania, and for IN, clearly the primary source is the body part "belly/stomach" in Mrica, while in Oceania there does not appear to be any primary body part model for IN, rather a number of secondary sources, such as "tooth", "belly/stomach", "heart", "liver", and "bowels" are made use of. The primary body part sources employed in these two continents are schematically presented in Figure 1. Note that the size of the schematized figures is represented in proportion to the relative contribution of the body part areas concerned, in an attempt to provide some rough visual guidance as to the respective proportions.

128

Bernd Heine

t

t

BACK - -

UNDER

---' AFRICA

+

OCEANIA

Figure 1. The primary body part sources for oN, UNDER, FRONT, BACK, and IN in African

and Oceanic languages (Heine 1989; Bowden 1991)

5. Competing body part models While, on the whole, there is a clear association between a given spatial concept and the body part that serves as a primary source for its expression, some body parts have been exploited for contrasting spatial reference points. In a number of instances this is due, at least in Mrica, to the presence of two competing models of conceptualization, namely the anthropomorphic and the zoomorphic models, which have, respectively, the body of a human being in an upright position and of a four-legged animal as their cognitive base; we have alluded to this fact in section 2 above. In addition, however, there are a few body parts that are intrinsically ambivalent with regard to the potential they provide for the conceptualization of spatial orientation, in particular the following: a. In both Mrica and Oceania, "face" provides the primary source for the concept FRONT; at the same time it also serves as a secondary model for ON

Conceptual grammaticalization and prediction

129

b. Similarly, "forehead" is employed in Oceania for both ON and FRONT c. The body part "belly/stomach" serves as a secondary model for both IN and FRONT in Oceania, though not in Africa d. In Africa, on the other hand, "buttock/anus" is employed to the same extent for UNDER and BACK Leaving aside the concept IN, what these associations seem to imply is that there exists some closer relation between ON and FRONT on the one hand, and between UNDER and BACK on the other. While there are two seeming counterexamples (see "back" in African and "shoulder" in Oceanic languages), such a relation appears to be statistically significant, as Table 8 suggests. Table 8. Number of instances where different spatial concepts are expressed by the same

body part term in African and Oceanic languages (not considered: Same expression for: and

UNDER ON ON ON

and and and

UNDER FRONT

BACK

FRONT BACK UNDER

and FRONT and BACK

IN)

Number of instances 24 16 3 0 0 0

As Table 8 indicates, there are only two conceptual relations that ap-

pear to be statistically significant, namely UNDER and BACK, and ON and respectively. This observation might suggest that there is a closer cognitive relation between what is in front and above on the one hand, and between what is behind and below on the other, roughly as sketched in Figure 2. On the basis of this observation we may add the following to the list of generalizations proposed in section 3:

FRONT,

(8) If a given body part serves as a model for the expression of more than one spatial reference point, then this is likely to involve either the concepts UNDER and BACK or ON and FRONT. It remains to be investigated how (8) is to be explained. One possibility would be to argue that the cognitive basis of the body part model is a human being not in a standing but rather in a moving position, with the

130

Bernd Heine

ON -

Figure 2. Some possible major cognitive links of spatial orientation (on the basis of data from 125 African and 104 Oceanic languages)

upper part of the body leaning slightly forwards. In addition, one might also consider the fact that in general, "head/face" on the one hand and "buttock" on the other are the most favored source regions for the conceptualization of the spatial reference points considered here, and due to their relative location, these two regions have the potential for being exploited each for two alternative reference points, viz. ON/FRONT and UNDER/BACK, respectively. There are no comparable body part sources of equal perceptual salience in the case of either the ON/BACK or the UNDER/ FRONT regions; potential regions such as the back of the head are generally insignificant, and the region of the private parts is avoided as a source domain for spatial orientation.

6. Discussion The example discussed in the preceding paragraphs is not the only case of a cross-linguistic study of the cognitive transfer patterns underlying grammaticalization; the reader is referred, e. g., to the "gramcat" project of Bybee and associates which provides various comparable cases (see Bybee-Pagliuca-Perkins 1989, 1991). The quantitative data available so

Conceptual grammaticalization and prediction

131

far on grammaticalization processes in the languages of the world suggests a number of tentative generalizations, such as the following: 6 a. If a given language develops a new category of tense, aspect, or modality it is most likely to select from a limited range of concrete propositions or event schemas for this purpose. These propositions or schemas take the form of periphrastic constructions. 7 b. If a language develops a (new) progressive aspect, the most probable choice is a locational proposition of the form "X is at Y", other source structures being less likely to be recruited for this purpose (Heine 1993). c. If a language develops a (new) perfect aspect category, the most likely source structure is a proposition involving the verb "finish" or "end". d. If a language develops (new) future tense category, it is most likely to select a proposition involving either of the motion verbs "go" or "come" or else a verb of volition ("want", "desire") as the predicate (Bybee-Pagliuca-Perkins 1991). Compared to these sources, alternative structures turn out to be statistically insignificant. The German werden future, for example, which derives from an event structure having a change-of-state verb ("become") as a propositional predicate, constitutes a more exotic case which turns out to be statistically insignificant among the languages of the world. e. If a language develops a (new) relative clause marker, then the most likely candidate is a demonstrative pronoun. In addition to such generalizations of universal import there are other observations that appear to be regionally defined. For example, while (b) applies to the majority of the languages of the world, there are certain areas where it does not apply: in parts of central and south-central Africa, many Niger-Congo languages use a proposition of the form "X is with Y", rather than the locational proposition, as a source for expressing progressives, and languages in Europe have drawn primarily on the use of a possessive proposition of the form "X has Y", rather than a verb for "finish" or "end" to develop new perfect categories, or to use interrogative pronouns ("who?", "which?", "what?") as an important alternative to demonstratives (e. g., "that") for developing relative clause markers. 8 To summarize, the same principles that we observed in the conceptualization of spatial orientation are also operative in other areas of grammaticalization: there is usually a limited range of concrete conceptual structures that are recruited for the development of a given grammatical

132

Bernd Heine

category and in most cases, these structures can be arranged in accordance with the statistical probability with which they are recruited; hence we are able to determine primary sources of grammaticalization, secondary sources, etc. While the generalizations presented above are based on diachronic findings and, hence, can be interpreted as probabilistic diachronic universals, they may also be said to constitute predictive statements as they allow us to predict future linguistic evolutions. The notion of prediction is closely associated with that of explanation, although there is no necessary one-to-one correspondence between the two. 9 Grammaticalization studies are not only a means of relating present language states to past situations, rather, by proposing generalizations on past development, they also allow us to predict future developments. Thus, Traugott remarks: The evidence is substantial that the process of semantic change outlined for the semantics of grammaticalization belongs to a larger set of crosslinguistic processes of semantic change that are in general quite regular. Indeed, they are so regular that it is possible to develop predictive hypotheses that can be tested against historical data. They are sufficiently predictive that one can take synchronic polysemies from any period in any language and project change back into the past. (1989: 31)

Prediction is based on the one hand on the unidirectionality principle that applies to virtually all instances of grammaticalization: concrete entities are grammaticalized to more abstract ones which again are grammaticalized to even more abstract ones. 10 On the other hand, it is based on observable instances of diachronic facts, and like virtually all facts in the humanities, these facts are based on statistical generalizations. For example, since in the majority of languages studied so far, nouns denoting the body part "back" have been grammaticalized to adverbs and/or adpositions referring to the concept BACK, it is likely on statistical grounds that the same will happen in the future when a given language develops a new designation for this relational concept. It goes without saying that such predictions are probabilistic rather than mechanistic or deductivenomological (cf. Anttila 1989: 401). Prediction concerns two different kinds of observation. First, it immediately relates to the kind of quantitative generalizations made in the preceding paragraphs, which can be understood as predictive statements. Second, it relates to the further process of grammatical development once the transfer from concrete concept to a more abstract concept has been concluded.

Conceptual grammaticalization and prediction

133

7. Conclusion The observations made in this paper suggest that people, irrespective of whether they live in Nigeria or Fiji, use essentially the same strategies to conceptualize spatial orientation. While there are a number of regionally defined divergencies, the generalizations proposed might turn out to be of universal significance; this at least is suggested by the evidence available from other continents and genetic stocks (see, e. g., Brugman 1983; Svorou 1994; MacLaury 1989). These generalizations may be said to be diachronic in the true sense of the word: they are retrospective/historical in that they are based on grammaticalization as a diachronic process.U At the same time, however, they are also prospective in that they allow us to predict within limits what is going to happen in the future development of a given language. Notes 1. In addition to the five concepts analyzed in our survey, Bowden studied

OUT, SEA, and as further referent points which he found were equally relevant in Oceanic languages. To secure comparability, the last three concepts will not be considered here. There is a fourth source domain, namely verbal concepts such as "precede", "follow", etc., which is very rarely encountered and will not be further considered here. It goes without saying that they can be transferred to other domains of cognition; for more details, see Heine-Claudi-Hiinnemeyer (1991: 65-97). This, however, does not apply to other kinds of spatial concepts where, e. g., "left hand", "right hand" form key notions of cognitive orientation; see Needham 1973 for references. Note, however, that in Ngbandi, a Niger-Congo language spoken in the Central African Republic, the word for "liver" (be') has been grammaticalized to a relational noun for "(in the) middle", e. g., ndo be' da 'in the middle of the house' (lit. 'place liver house'); cf. also be' ti 'palm of hand' (lit. 'liver hand') (Lekens 1958; Helma Pasch, personal communication). These generalizations are based on quantitative data collected within the project "Lexicon of Grammaticalization" which is being carried out by Thomas Stolz and the present author. The main propositions employed are described by Heine (1993). While interrogative pronouns do form an alternative source to demonstratives even outside Europe, they are clearly a less frequent source domain for relative clause markers in the language of the world. Lass (1980: 13), for example, argues that there is a certain asymmetry between them since any (correct) explanation involves correct prediction, while not every correct prediction involves explanation. Concerning some counterexamples, see Heine-Claudi- Hiinnemeyer (1991: chapter I). Grammaticalization has been defined as a panchronic process which has both a diachronic and a synchronic dimension (Heine-Claudi- Hiinnemeyer (1991: chapter 9)); the present paper, however, is based on findings that are confined essentially to the diachronic dimension of this process. LAND

2. 3. 4.

5.

6.

7. 8.

9.

10. II.

134

Bernd Heine

References Anttila, Raimo 1989 Historical and comparative linguistics. (2nd revised edition.) (Amsterdam Studies in the Theory and History of Linguistic Science, 6.) AmsterdamPhiladelphia: John Benjamins. Bowden, John 1991 Behind the preposition: Grammaticalization of locatives in Oceanic languages. [Unpublished M.A. thesis, University of Auckland.] Brugman, Claudia 1983 "The use of body-part terms as locatives in Chalcatongo Mixtec", (Report No. 4) Survey of California and Other Indian Languages 4: 235-290. University of California, Berkeley. Brugman, Claudia- Monica Macaulay "Interacting semantic systems: Mixtec expressions of location", Proceedings 1986 of the annual meeting of the Berkeley Linguistic Society 12: 315-327. Bybee, Joan L.- William Pagliuca- Revere D. Perkins 1989 On the asymmetries in the affixation of grammatical material. [Typescript, Department of Linguistics, University of Buffalo.] 1991 "Back to the future", in: Elizabeth C. Traugott- Bernd Heine (eds.), 17-58. Clark, Herbert 1973 "Space, time, semantics and the child", in: Moore 1973, 28-63. Heine, Bernd 1989 "Adpositions in African languages", Linguistique Africaine 2: 77-127. Auxiliaries. New York, Oxford: Oxford University Press. 1993 to appear "Grammaticalization as an explanatory parameter". [Paper presented at the symposium on Explanation in Historical Linguistics, Milwaukee, WI, April 20-22, 1990. To appear in the symposium proceedings.] Heine, Bernd- Ulrike Claudi- Friederike Hiinnemeyer 1991 Grammaticalization: A conceptual framework. Chicago: University of Chicago Press. Lass, Roger 1980 On explaining language change. Cambridge: Cambridge University Press. Lekens, Benjamin Ngbandi idioticon. Vol. 2. (Annales du Musee Royal du Congo Beige, Serie 1958 in-8°, Science de !'Homme, Linguistique 3, II.) Tervuren: Commission de Linguistique Africaine. MacLaury, Robert E. "Zapotec body-part locatives: Prototypes and metaphoric extensions", In1989 ternational Journal of American Linguistics 55(2): 119-154. Moore, T. (ed.) 1973 Cognitive development and the acquisition of language. New York: Academic Press. Needham, Rodney (ed.) 1973 Right and left: Essays on dual symbolic classification. Chicago- London: University of Chicago Press. Stolz, Thomas 1991 Von der Grammatikalisierbarkeit des Korpers. Part 1: Vorbereitung. (Prinzipien des Sprachwandels 2.) Essen: University of Essen.

Conceptual grammaticalization and prediction

135

Svorou, Soteria 1986 "On the evolutionary paths of locative expressions", Berkeley Linguistics Society 12: 515-527. 1987 "The semantics of spatial extension terms in Modern Greek", Buffalo Working Papers in Linguistics (University of Buffalo) 87-01: 56-122. 1994 The grammar of space. Amsterdam, Philadelphia: John Benjamins. Tanz, Christine 1980 Studies in the acquisition ofdeictic terms. Cambridge: Cambridge University Press. Traugott, Elizabeth C.- Bernd Heine (eds.) 1991 Approaches to grammaticalization. Vol. 2. Amsterdam-Philadelphia: John Benjamins.

Metaphors of anger in Japanese 1 Keiko Matsuki

1. Introduction Evidence from the Japanese language shows that emotions have structure; it lends crosscultural support to Kovecses's (1987) analysis of emotion in American English, which draws the same conclusion. The present study illustrates in detail the concept of anger in Japanese; it compares the ways that anger is conceptualized in Japanese and in American English, highlighting culturally unique as well as shared aspects. The cross-linguistic perspective points to sociocultural factors underlying the linguistically encoded framework of emotion in Japanese. A consideration of social and cultural contexts contributes to the examination of a cognitive model. Lakoff and Johnson (1980) demonstrate the importance of metaphor to language and thought. Kovecses (1987) argues further that human emotions are highly structured rather than amorphous. His rich range of evidence from emotive expressions in American English underscores the critical role of metaphors and metonymies in organizing the concepts of anger, pride, and love. Kovecses focuses exclusively on language, and he is successful in extracting a systematic structure of metaphors and metonymies in American English. However, the study of the Japanese concept of anger and its variability requires a consideration of the sociocultural context in combination with linguistically expressed metaphor. In a study of Ifaluk ethnotheory of emotions, Lutz (1987) explains emotive discourse in terms of a schema and a socially contextualized proposition. The proposition instantiates a schema in a specific social context, and Ifaluk discourse on emotions is meaningful only after such instantiation occurs. Lutz writes, "emotion concepts have in themselves cultural propositions and in turn are nested in larger networks of knowledge about persons, roles and goals" (1987: 307). In an extensive discussion of language and affect, Besnier (1990) demonstrates the embeddedness of affective language in social situation, social category, and social structure. My Japanese data derive from Nakamura (1979), whose sources are a variety of writers in Japanese modern literature. He focuses especially upon "anger", "happiness", and "sadness". 2 Some of the expressions

138

Keiko Matsuki

seem to be idiosyncratic of particular authors, although they extend the elaborate Japanese system of metaphors and metonymies. In addition, I conducted my own search for unreported expressions of anger. American English and Japanese share some metaphors and metonymies that structure the conceptual framework of anger; however, it is impossible to decide which ones are fully universal on the basis of Japanese and English alone. Kovecses (1987) detects the following metaphors and metonymies in American English, which are also found in Japanese. Metaphors: "Anger is heat" (1987: 13) "The body is a container for the emotions" "Anger is the heat of a fluid in a container" (1987: 14) "Anger is insanity" (1987: 20) "Anger is an opponent (in a struggle)" (1987: 22) "Anger is a dangerous animal" (1987: 23) "Angry behavior is aggressive animal behavior" (1987: 24) "The cause of anger is a physical annoyance" "Causing anger is trespassing" (1987: 26) "Anger is a burden" (1987: 27) Metonymies: "The physiological effects of an emotion stand for the emotion" (1987: 21) "Insane behavior stands for anger" "Violent frustrated behavior stands for anger" "Aggressive verbal behavior stands for anger" "Aggressive visual behavior stands for anger" (1987: 25) Section 2 elaborates on these shared elements of the model. In section 3, culturally unique metaphoric systems found in Japanese examples are discussed. Sections 4 and 5 focus on unique metaphors involved in the prototypical scenario of anger.

2. Japanese metaphors shared with American English 2.1 Folk model of physiological effects The folk theory positing physiological effects of anger provides the starting point for Kovecses's analysis. He argues that an elaborated system of metaphors and metonymies for anger is based on a folk theory: 3

Metaphors of anger in Japanese

139

The [presumed] physiological effects are increased body heat, increased internal pressure (blood pressure, muscular pressure), agitation, and the interference with accurate perception: (Kovecses 1987: 12)

The model can also be applied to the metonymies in the following Japanese anger expressiOns. [body heat] Mune ga atsuku naru hodo no ikari o oboeru. 'To experience anger to the degree that the chest becomes hot.'

(l)

(2) a. b. c.

d.

[internal pressure] Ikari ga chooten nimade tassuru. 'Anger reaches the top.' Fukureru. 'To swell up.' Fukureta kao a suru/Future-Usura o suru. 'To have a swelling face.' Aosuji o tatete okoru. 'To get angry with blue streaks standing out.'

Note that in (2d) aosuji means "blue streaks", a reference to veins standing out because of internal pressure. Kovecses develops the metonymic association of body heat and internal pressure with redness of the face and neck, agitation, and impeded perception. Equivalent expressions appear in Japanese. (3) a. b. (4) a. b. c. (5)

[redness in face and neck] Kao a makka ni shite okoru. 'To get angry with a red/scarlet face.' Makka ni natte okoru. 'To get red/scarlet and angry.' [agitation] Ikari de furueru. 'To shake with anger.' Koe ga ikari de furueru no o osaekirenai. 'To be unable to control the voice shaken by anger.' Kobushi o furuwasete okoru. 'To get angry with shaking fists.' [impeded perception] Ikari de zengo no misakai ga tsukanai. 'To be unable to tell which side is front or back because of anger.'

140

Keiko Matsuki

2.2 Metaphors and metonymies of contained hot fluid Kovecses introduces the general metaphor "the body is a container for the emotions" (1987: 14). Underlying this container metaphor are further metaphors: "the emotions are fluid", and "the emotions are fluid in a container". Japanese expressions of anger as well as of other emotions also appear to be structured by this "fluid in a container" metaphor. (6)

[the emotions are fluid in a container] Yorokobi de mune ga afureru. 'The chest is filled with joy.' b. Kanashimi de mune ga ajureru. 'The chest is filled with sadness.' c. Ikari ga jojoni wakidasu. 'Anger gradually flows out.' d. Ikari ga komiageru. 'Anger rises up.' e. Ikari ga hageshii nami no yooni zenshin ni hirogaru. 'Anger spreads all over the body like violent waves.' a.

Kovecses emphasizes that the heat metaphor accompanies the fluid metaphor. He focuses on "anger is heat" (1987: 13), and the combination of this metaphor with "the body is a container for the emotions" generates "anger is the heat of a fluid in a container" (1987: 14). The metaphor of heat for the emotion of anger is explicit in Japanese, resulting in an equally rich system. (7)

a. b. c.

(8)

[anger is the heat of a fluid in a container] Harawata ga niekurikaeru. 'The intestines are boiling.' Ikari ga karada no naka de tagiru. 'Anger seethes inside the body.' Ikari ga hara no soko o guragura, saseru. 4 'Anger boils the bottom of the belly.' [intense anger produces steam] Atama kara yuge ga tatsu. 'Steam rises up from the head.'

(9) a. b.

[intense anger produces pressure on the container] Ikari no kimochi o osaekirenai. 'To be unable to suppress the feeling of anger.' Atama ni chi ga noboru. 'Blood rises up to the head.'

Metaphors of anger in Japanese

141

(10)

[when anger becomes too intense, the person explodes] Haha wa toutou bakuhatsu shita. 'My mother finally exploded.'

(11)

[when a person explodes, what was inside them comes out] Ikari ga bakuhatsu suru. 'Anger explodes.' Ikari ga fukiageru. 'Anger blows up.'

a. b. (12) a. b. c. d.

[anger is fire] Ikari no hi o kesu. 'To put out the fire of anger.' Ikari ga moedasu. 'Anger starts burning.' Ikari ga moetatsu. 'Anger starts burning.' Ikari ga moeagaru. 'Anger flares up.'

Examples from Japanese of other metaphoric and metonymic concepts in Kovecses's model include: (13) a.

b.

[anger is insanity] Ikari de ware o wasureru. 'To lose oneself because of anger.' Ikarikuruu. 'To get angry and crazy.'

(14)

[insane behavior stands for anger] Yokottsura o haritaoshite yatta. 'To strike a person down, in the side of the face.'

(15)

[violent frustrated behavior stands for anger] Jidanda o fumo. 'to stamp on a pair of bellows.' . .. to o pishari to shimeta. ' ... slammed the door shut.'

a. b. (16)

[anger is an opponent (in a struggle)] Komiagetekuru ikari to tatakau. 'To fight against the rising anger.'

142

Keiko Matsuki

(17)

[anger is a dangerous animal] Suzamajii ikari ga mayu no atari ni hau. 'Terrible anger crawls around the eyebrows.'

(18)

[angry behavior is aggressive animal behavior] lkari de me o giragira saseru. 'To have glaring eyes.'

(19)

[aggressive verbal behavior stands for anger] Sonnani gamigami iwanaide yo. 5 'Don't snap at me'

(20)

[aggressive visual behavior stands for anger] Haha wa watashi wo jitto niranda. 'Mother glared sharply at me.'

(21)

[anger is a burden] Okottara kimochi ga karuku natta. 'I feel light after having expressed my anger.'

3. Japanese metaphors not shared with American English Hara is the area surrounding the navel, corresponding in meaning to the English word belly. It metaphorically contains the emotions in Japanese. (22)

a. b. c.

[hara is the container of the emotions] Hara no naka de hidoku okoru. 'To get terribly angry in(side) one's hara.' Hara ni suekaneru. 'To be unable to keep it/anger in the hara.' Kimochi wa wakaru keredo hara ni osamete kudasai. 'I understand how you feel, but keep it inside your hara.'

The container hara can stand for its content. This metonymy spawns a wide range of hara expressions in radial polysemy. 6 According to Koojien (Shinmura 1991), one of the most popular dictionaries of Japanese, hara is summarized as denoting: ( 1) the belly; stomach; center; (2) the inside of the mother's womb; the child born from the mother's womb; (3) the heart; real intention; and (4) courage and nerves. Metaphoric and metonymic principles crosscut the expressions listed below.

Metaphors of anger in Japanese

143

The primary meaning of hara, 'belly', has become a metaphor for the womb. (23) a. b.

[hara is the womb] Harachigai no kyoodai. 'Siblings of a different mother.' (lit. ' ... of a different hara.') Hara o itameta kodomo. 'The child who is born with the mother's pains of hara, i.e., the biological child.'

When hara is used to refer to the contents of the container, it is the word for something real but hidden. When it is used for the container itself, it is the container for some invisible, hidden truth. Here, the underlying association is "something hidden and invisible is the truth", and this unites with the principal metaphor "hara is the container of the emotions" and the metonymy "hara stands for the content of the container" to generate the metaphors "hara is the container of real intention and emotion" and "hara is real intention and emotion". (24)

a. b.

c. d. e. f.

g.

h.

(25)

a.

Hara ga kuroi. 'Hara is black, i.e., not fair/wicked.' Hara o saguru. 'To search hara, i.e., seek to discover a person's real intention.' Hara o minuku. 'To see through hara, i.e., to discern a person's real intention.' Hara o yomu. 'To read hara, i.e., interpret a person's real intention.' Hara o waru. 'To split hara, i.e., open one's heart.' Kuchi to hara ga hantai da. 'The mouth is the opposite of hara, i.e., say the opposite of one's real intention.' Hara ni ichimotsu aru. 'To have a secret intention in hara, i.e., to have a secret, hidden intention.' Hara o awaseru. 'To match one's hara with the other's, i.e., one conspires with another to deceive a third person.'

[hara is courage and nerves] Hara o kimeru. 'To determine hara, i.e., be determined to tackle a task.'

144

Keiko Matsuki

b. c.

d.

Hara ga suwaru. 'Hara is set, i.e., be calm and without anxiety.' Hara ga dekiteiru. 'Hara is done, i.e., be ready to tackle a task.' Futoppara (jutoi + hara) no hito. 'A person of thick hara, i.e., a person with "a lot of nerve"; a broad-minded person.'

The invisible, truthful content of hara is called honne 'private self'. This word is usually used in contrast with tatemae 'social face'. According to Lebra, tatemae refers to the "standard principle, or rule by which one is bound at least outwardly" (1976: 136). When a person's inner mind is not appropriate to be outwardly expressed according to accepted standards or common sense, a separation between honne and tatemae is encouraged. Bonne and tatemae are attended to at the same time by a person who hides, or backgrounds honne and manifests, or foregrounds tatamae. What Lebra calls the "cultural pressure for situational discrimination" (1976: 136) is embedded in the intricacy of social roles and positions, and the metaphor of hara is located within such Japanese social milieu. The consideration of the sociocultural context in which these notions function is fundamental to understanding the prototypical scenario of Japanese anger. Even when a person gets angry, his honne, or anger may be kept inside; he may smile while fighting increasing anger. Hara, honne, and tatemae are parts of the Japanese scenario of anger, structuring such emotions in conflict.

3.1 "Hara rises up" When a person is offended and gets angry, hara is said to "rise up". (26)

Hara ga tatsu. 'Hara rises up.'

Hara does not rise up when a person becomes sad or happy. Sadness and/or happiness may fill mune 'chest', but only anger makes hara rise. What is the metaphoric structure of hara in the phrase hara ga tatsu? What rises up? Does hara, as a metaphor for the container of anger emotions, rise up? Or does hara, as a metaphor for the contents, rise up? The boundary is ambiguous, 7 but the underlying metaphoric and metonymic principles are clear: the metaphor of "hara is the container

Metaphors of anger in Japanese

145

of anger" and the metonymical principle "the container stands for the contents" (anger) generate the metaphors "hara is anger that rises", and "hara can rise".

4. The prototypical scenario Kovecses incorporates a temporal dimension in his model of American anger by constructing a prototypic scenario of five stages (1987: 28): Stage Stage Stage Stage Stage

1: 2: 3: 4: 5:

Offending event Anger Attempt at control Loss of control Act of retribution

The scenario applies to anger in Japanese, although stage 3 is more elaborate than in English. As mentioned, when a Japanese person is offended and gets angry, hara rises up. Further, when a person attempts to control anger, he tries to keep it in hara: Hara ni osameteoku. 'Hold it in hara.' Hara ni shimatteoku. 'Keep it in hara.'

(27)

If he cannot or does not have to control it, the act of retribution is reached:

(28)

Hara ni suekaneru. 'Cannot lay it in hara.' Anmari hara ga tatta node hon o nagetsuketa. 'I threw a book because hara rose up so much.'

When a person needs to control his anger in order to show tatemae, but still experiences the increasing anger, conflicts go beyond the container of hara and move to mune, the chest. (29)

a. b.

Haradatashisa ni mune o shimetsukerareru. 'To feel strangled by mune because of the rise of hara.' Mune ga mukatsuku. 'To feel nauseated by mune.'

146

Keiko Matsuki

Mune never rises up in Japanese. Mune is the container for anger overflowing from hara. Mune is the seat of nausea; conflict and frustration caused by efforts to control growing anger provoke nausea. When a person is about to lose control, increasing anger comes to atama. (30)

Toutou atama ni kita. 'Finally, (it) has come to atama.'

Atama 'head' does not rise up. Nor is atama responsible for nausea. Atama is the place that anger reaches after extreme internal conflict; it undermines mental faculties. When anger is in hara and/or mune, a person is still able to control it by rationalization. However, when anger reaches atama, one loses rationality. Here, the adverb toutou 'finally' indexes this prototypical scenario: Toutou atama ni kita 'the anger finally has come to the head'. The word toutou can be used once the anger has risen through two bodily zones of hara and mune, and, finally, reached atama. 8 See Figure 1.

otomo

mune

horo

'01

(~8

u

Figure 1. Zones of rising anger in Japanese: hara, mune and atama

Metaphors of anger in Japanese

147

5. Nonprototypical anger Kovecses describes nonprototypical scenarios, deviations from his prototypical model. I will explore one nonprototypical example in Japanese. In the scenario "hara rises", anger moves toward mune and finally reaches atama. However, sometimes people respond immediately with atama. One of Kovecses's unanswered questions regards the relationship between "the range of offenses that cause anger" and "the corresponding range of appropriate responses" (1987: 36). The relationship might be complicated by individual differences. Although Japanese examples do not offer a solution either, one general correspondence is noteworthy. One of my Japanese informants made a distinction between direct and indirect causes of anger. She defines a "direct cause" as that which makes a person angry directly after an offense has occurred and makes them lose control instantly. An "indirect cause" takes a while to make a person angry. And, even if the person does become angry, the indirect nature of the offense may prevent them from expressing their anger outwardly, because the person may not be able to pinpoint the overt offense. Here, whether a cause is direct or indirect depends on individuals and contexts, but I consider the correspondence between direct and indirect causes and responses involving body zones such as hara and atama. It is not normal to say "hara finally rises up". It may not be totally impossible, but it requires a particular context. The Japanese language elaborates stage 3 of Kovecses's model to fit the cultural anger scenario. The Japanese stage 3 has subdivided stages, hara, mune, and atama. The ontology of Japanese anger includes an ascending scale based on three body zones, metaphors not found in American English. When a person gets angry directly following another person's offensive act, anger might come to atama, the head, directly, instead of making hara rise up. (31)

Atama ni kachin to kita. ' ... came to atama with a click.'

Kachin is the onomatopoetic sound of sudden collision, the symbol of instantaneous response to a cause of anger. The responding body zone for such a collision is atama. Kachin denotes a sudden collision between an offense and atama. Hara never makes such a click. Kachin implies that the anger has risen up from hara to atama so fast as to result in collision against the external offense. It is important to note that the elevation scenario proceeds quickly but is not skipped. A person is about

148

Keiko Matsuki

to lose his control when he responds to an offense with atama. He has no time for rationalization. Another informant (IF) talks about direct and indirect responses in the following interview excerpt. This example evidences differentiation of metaphors regarding hara and atama, one beginning in the belly and rising gradually, and the other occurring in the head and involving the sound of collision. KM: Okoru to shitara douiu koto ni okroi-masu ka? IF: Yappari bujoku-sareta toki toka ne. Bujoku-sareta to yuu ka, aite ga sonna fuuni wa omotte-nakute mo jissai ni wa bujoku-datta to. Sorede atode okoru-n-desu. Atode omoidashite. Aa iwareta kedo jitsuwa koo-yuu koto dakara aa okoru-n-datte. Demo hara ga tatsu-n-desu. KM: Hara ga tatta toki doo-yuu-fuu ni nari-masu ka? IF: Iroiro nee kangaeru-n-desu yo. Sono toki no jookyoo o. Soshite aa koo itte-yarreba yokatta toka ne. Jookyoo a moo ikkai setteishite yarikaesu-n-desu aite ni. KM: Atama ni wa konai? IF: Atama ni kuru. Kaa-tto natta toki wa watashi. Ikkai ikkai dake oboeteru-n-desu-kedo ne. Chuugakkoo no toki ni "baka" tte iwarete ne '"baka" to wa nani yo' tte. Na ka aite ni tsuzukezama ni iwarete nee. Ikari no amari koe ga denaku-naru-n-desu. [KM: At what kind of things are you angry when you get angry? Well, when I am humiliated. Even though the other person does IF: not mean it, it can actually be a humiliation. Then, I get angry, later. Later, I remember it. I remember to get angry, realizing what I was told earlier. But, hara rises up. KM: What is the condition when hara rises up? IF: I think about many things, the situation. Then, "I should have said this". I imagine myself responding to the other person, remembering the situation again. KM: It doesn't come to atama? IF: It comes to atama. When I become so hot. Once, only once, I remember. I was called "fool" when I was at junior high school, and I said "What do you mean by 'fool'?" This person kept telling me many things and I could not speak, I was so angry.] In the usual course of events, it is not until the informant remembers the situation and then later realizes the humiliation that her hara rises up.

Metaphors of anger in Japanese

149

However, in the second experience of anger the informant describes, she got angry immediately, responding with atama when somebody called her a fool.

6. Contextual. variation: alteration of scenario In this section it is argued that we must look at social context in order to better understand culture-specific structures of anger, in particular at the social relationships of people. 9 A third informant told me that, as a rule, he does not express his anger and that he controls it especially in the public domain where he is part of a social hierarchy. He strives so hard to control his anger that the blood in his atama becomes congested. In the prototypical scenario, one suffers loss of control after anger reaches atama. But, this informant continues to fight against his anger, making the blood in atama congested: "Atama no naka de chi ga katamaru-n-desu" (The blood gets congested in my head]. Another informant said that she used to be patient, controlling her anger when she was angry with her boss in the workplace. She expressed her dilemma, saying "Hakeguchi ga nai-n-desu" [there is no vent]. These individual variations can be accounted for by social factors such as status, role, and sex. The female informant strenuously controls her anger when at work, but she might easily blow her top in other contexts. Her social status of being hierarchically inferior to her boss forces her to control her anger. The distinction between the gender roles in the Japanese workplace is even clearer than it is in the American context. This informant also told me about an infuriating experience in which she was asked to go out and buy a pack of cigarettes just because she was a woman. College-educated working women may still encounter such conflicts in today's Japanese society; and usually they have no means of retribution, even when they get angry. Social conventions governing public life provide women with limited ways of communicating their anger. Although the legal rights of women have improved, gender roles have stayed relatively traditional. "There is no vent" for angry feelings. The prototypical model is an ideal, and in the real world nonprototypical cases cluster around this model. These nonprototypical cases differ from one another, depending on where a person is located within their social milieu. Kovecses writes that "study of the language as a whole gives us no guide to individual variation" (1987: 37). This is true in that the language itself does not foretell individual idiosyncrasies. Social

150

Keiko Matsuki

factors account for some variation in the choice of metaphor, although systematic correlations between social factors and use of specific metaphors remain unexplored.

7. Conclusion The study of anger in the Japanese language supports Kovecses's claim that emotions are highly structured. It invites discussion of cross-cultural difference in cognitive models of anger. Although the cognitive model of American anger is partially applicable to Japanese anger, the concepts of hara, mune, and atama are unique to the latter. The metaphor of hara is further linked with notions of honne and tatemae, which are embedded in the Japanese sociocultural context. Notes I. I am grateful to Robert E. MacLaury for his critical comments on the present work. 2. Lakoff and Turner (1989) discuss metaphoric structure which is coherent in creative literature such as poetry as well as in ordinary language. 3. Davitz (1969) studied how American people define a certain label of emotion. One of his results reveals that the pattern of anger description clusters around hyperactivity, in which physiological conditions such as blood pressure, heartbeat and pulse are activated. Davitz's study supports the physiological folk theory of Kovecses. 4. The sound-symbolic word guragura describes the condition of boiling. 5. Gamigami is a sound-symbolic word for snapping. 6. Lakoff (1987) develops the concept of radial structure in categorization. A radially structured category is composed of a central core, and nonprototypical variants that extend from it on linkages of diverse metaphors and metonymies. 7. The intransitive verb tatsu is used for the motion in which an object stands up/rises up generally, while it also describes the appearance of something phenomenal. One etymological hypothesis (Origuchi 1929, cited in Yamanaka 1976: 323) is that this word was originally used for the appearance of supernatural entities. 8. The aspectual category of the verb kita is past, and it is also used for the perfective sense. The other encoded structure is the spatial direction of kita, and atama constitutes the goal of the verb. These temporal and spatial dimensions fit the scenario of anger. 9. Kovecses refers to Rosaldo's study of an anger concept, liget in Ilongot. Rosaldo (1980) found "a set of principles and connections with elaborate ramifications for I! on got social life" in the concept of liget.

References Besnier, Niko 1990 "Language and affect", Annual Review of Anthropology 19: 419-451. Davitz, Joel R. 1969 The language of emotion. New York: Academic Press.

Metaphors of anger in Japanese

151

Kiivecses, Zoltan 1987 Metaphors of anger, pride and love. Philadelphia: John Benjamins. Lakoff, George 1987 Women, fire, and dangerous things. Chicago: University of Chicago Press. Lakoff, George- Mark Johnson 1980 Metaphors we live by. Chicago: University of Chicago Press. Lakoff, George- Mark Turner 1989 More than cool reason: A field guide to poetic metaphor. Chicago: University of Chicago Press. Lebra, Takie Sugiyama Japanese patterns of behavior. Honolulu: University of Hawaii Press. 1976 Lutz, Catherine 1987 "Goals and understanding in Haluk emotion theory", in: Dorothy HollandNaomi Quinn (eds.), Language and cultural knowledge. Cambridge: Cambridge University Press. ·• Nakamura, Akira 1979 Kanjoo hyoogen jiten [The dictionary of emotive expressions]. Tokyo: Rokkoo Shuppan. Origuchi, Shinobu 1929 Kodai kenkyuu [Study of ancient times]. Tokyo: Ookayama Shoten. Rosaldo, Michelle 1980 Knowledge and passion. Cambridge: Cambridge University Press. Shinmura, Izuru (ed.) 1991 Koojien. Tokyo: Iwanami Shoten. Yamanaka, Joota 1976 Kokugo gogenjiten (Japanese etymological dictionary]. Tokyo: Azekura Shoboo.

Looking back at anger: Cultural traditions and metaphorical patterns 1 Dirk Geeraerts-Stefan Grondelaers

1. The mysteries of masturbation In the course of 1989, the Belgian Department of Education started a school campaign against truancy. The major slogan of the campaign read Van spijbelen wordje doof'playing truant makes you deaf'-ajocular (but probably ineffective) reference to the old belief that excessive masturbation could cause deafness. It is not likely that this belief itself is still very much alive in our post-sexual revolution, sex education era, but the very fact that knowledge of it was assumed in the campaign seems to indicate that it is still around, and that it was being handed down from educators to pupils not too long ago. But what was the origin of that belief (which, incidentally, came in a number of variants, in the sense that next to deafness, blindness and deterioration of the spinal marrow were cited as the sinful results of promiscuous self-indulgence)? Was it just a conspiratorial invention of priests and parents, intended to keep personal frustration and public morality up? Or was there an actual basis for it? Let us turn to a specialist for an answer. In 1772, the honourable doctor Tissot (member of learned societies in London, Basel, Bern, and Rotterdam) published a lengthy treatise entitled L'Onanisme: Dissertation sur les maladies produites par la masturbation (Grasset, Lausanne), in which all is revealed about this crime obscene. After an extensive treatment of the detrimental influence of masturbation, he asks the question "Comment une trop grand emission de semence produit-elle tous les maux que je viens de decrire?" [How does an excessive emission of sperm produce all the evils that I have just described?] On page 69, he begins his answer with a reference to the father of medicine, Hippocrates of Kos (approximately 460-377 B. C.). (The relevant passages are in the Hippocratean treatise known as De Genitura.) Hippocrate a cru qu'elle [la semence] se separoit de tout le corps, mais surtout de la tete. La semence de l'homme vient, dit-il, de toutes des humeurs de son corps, elle en est la partie la plus importante .... Il y a des veines & des nerfs qui de toutes les parties du corps vont se rendre aux parties geni-

!54

Dirk Geeraerts~ Stefan Grondelaers

tales; quand celles-ci se trouvent remplies & echauffees, elles eprouvent un prurit, qui se communiquant dant tout le corps, y porte une impression de chaleur & de plaisir; Ies humeurs entrent dans une espece de fermentation, qui en separe ce qu'il y a de plus precieux & de plus balsamique, & cette partie, ainsi separee du reste, est porte par Ia moe!e de I'epine aux organes genitaux. [Hippocrates thought that semen secreted itself from the entire body, but specifically from the head. A man's semen, he says, comes from all the humors of his body, of which it is the most important component. ... There are veins and nerves that go towards the genital organs from all over the body, and when these organs are filled up and warmed up, they experience an urge that communicates itself through the entire body, producing an impression of warmth and pleasure. The humors then enter into some sort of fermentation that separates out the most precious and balsamic substance they contain, and this part, when it is separated from the rest, is carried to the genital organs by the spinal marrow.]

If, in other words, semen is produced by a process of fermentation and distillation of the "humors", it is plausible that an overproduction of sperm weakens the body, given the vital importance of those "humors". Specifically, the role of the head and the spinal cord in this process explains why, in particular, the functions of the head (such as seeing and hearing), and the spinal marrow may suffer the detrimental effects of excessive sperm production. In this sense, it all falls into place- but it only does so against the background of the doctrine of the four humors. In Classical and Medieval physiology, the human body was thought to contain four kinds of fluids, or humors, which regulated the body's functioning and whose disproportionate presence could cause illness. Admittedly, conflicting views were expressed within the humoral framework about the origins of sperm, and the humoral doctrine as a whole was no longer valid in its original Hippocratic form by the end of the eighteenth century. But, although Tissot hardly follows Hippocrates in detail, he quotes him approvingly, and like most of his contemporaries does retain the basic idea that the production of semen (and the harmful effects of overproduction) involves the extraction of the seminal substance from the vital bodily humors. Against the background of this historical link, extending from classical antiquity well into the modem era, the lingering belief in the negative effects of masturbation appears to be a recently deceased (or at least moribund) remnant of what was once solid science. And to be sure, it is not the only relic. In the Dutch-speaking part of Belgium, for instance, the concept "cold, inflammation of the mucous membrane of nose and throat" is expressed in a majority of the dialects by the word valling.

Cultural traditions and metaphorical patterns

155

While valling is morphologically complex (being a nominalization of the verb val/en 'to fall'), it is not semantically transparent to the majority of speakers. Historically speaking, however, its formal complexity makes perfect sense in the framework of the theory of humors: when the nose runs or sputum is expectorated, what "falls" is nothing else then phlegm, one of the four humors. Given that the head was considered to be the major locus of phlegm (and taking into account that phlegm was described as a cold humor, in contrast with, for instance, blood as one of the warm humors), a valling or a cold is nothing but a precipitation of the brain's fluid. And of course, in nonhistorical parlance, the English word phlegm now refers precisely to the thick semifluid secretion of the mucous membranes of the respiratory passages. Similarly, we speak of catarrh (derived from the Greek katarrheo 'to flow down'), and -in the case of another disease attributed to an excess of phlegm- of rheumatism, in which the Greek verb rhea 'to flow' can be discerned (cf. Siegell968: 323).

Faced with examples such as these, we would like to address the question of the influence of the humoral doctrine on our contemporary vocabulary more systematically. What other relics of the old beliefs can we find? Specifically, given the psychological part of the humoral theory, is there any way in which the influence of the theory can still be felt in the way we talk about emotions? Concentrating on the concept of anger, we will try to show that there is. By taking a closer historical look at Kovecses's analysis of emotional expressions in terms of generalized metaphors (1989), we shall argue that his ahistorical method obscures the possible role of cultural traditions as a source of emotion concepts. Our purpose, in other words, is factual to the extent that we will try to establish the importance of the old humoral theory for our contemporary emotional vocabulary, critical to the extent that we will try to qualify Kovecses's analysis, and methodological to the extent that we will stress the methodological importance of a diachronic perspective for linguistic studies with a cultural orientation. Before dealing with the specific linguistic part of the investigation, we will first give some additional information on the theory of humors and its historical importance.

2. The history of the humors In this section, we will briefly (and simplifyingly) present the humoral doctrine, and sketch its historical development. On various aspects of the history of medicine at large and the humoral theory in particular, more

156

Dirk Geeraerts-Stefan Grondelaers

information may be found in Lindeboom (1985), Godderis (1988), Beek (1969), Irwin (1947), Siegel (1968), Major (1954), Schafer (1966), and Diepgen (1955). Klibansky, Panofsky and Saxl (1964) deserve to be mentioned separately for their detailed history of the humoral doctrine up to the seventeenth century (with special emphasis on the concept of melancholy). The foundations of the humoral doctrine were laid by Hippocrates of Kos. Three aspects of this approach should be mentioned: the physiological, the psychological, and the medical. Physiologically, the four humoral fluids regulate the vital processes within the human body; the secretion of the humors underlies the dynamic operation of our anatomy. Psychologically, on the other hand, they define four prototypical temperaments, i.e., a person's character is thought to be determined by the preponderance of one of the four vital fluids in his body. Thus, the choleric temperament (given to anger and irascibility) is determined by a preponderance of yellow bile, while the melancholic, gloomy and fearful, suffers from a constitutional excess of black bile. The phlegmatic personality is typically placid and unmoved, while the sanguine temperament (defined in correlation with blood, the fourth humor) is passionate, optimistic, and brave. The singular combination of physiological and psychological concepts that characterizes the theory of humors also shows up in the fact that a disequilibrium of the fluids not only characterizes constitutional temperaments, but also causes temporary diseases-which are then typically described in bodily, biological terms as well as in psychic terms. For instance, an overproduction of yellow bile may be signaled by the patient's vomiting bile, but also by his dreaming of fire. In the same line, an excess of blood shows up in the redness of the skin and swollen veins, but also in carelessness and a certain degree of recalcitrance. In this sense, the humoral theory is a medical doctrine: it identifies diseases and their symptoms, and defines a therapy. Obviously, the basic therapeutic rule will be to restore the balance of the humors, given that a disturbance of their well-balanced proportion is the basic cause of the pathological situation. The long-lasting popularity of bloodletting, for instance (a standard medical practice that continued well into the nineteenth century) has its historical origins in the theory of humors. The connection between yellow bile and fire that was mentioned a moment ago is not accidental. It is part of a systematic correlation between the human, anatomical microcosm and the macrocosm, thought to be built up from four basic elements. Thus, yellow bile, black bile,

Cultural traditions and metaphorical patterns

157

phlegm, and blood corresponded with fire, earth, water, and air respectively. In the Aristotelian elaboration of the Hippocratic doctrine, a "componential analysis' was added to these correlating sets of microcosmical and macrocosmical basic elements. They were defined, in fact, as combinations of four basic features: cold, warm, wet, and dry. (Needless to say, these four features are themselves related along two dimensions.) Blood was thought to be warm and wet, phlegm cold and wet, yellow bile warm and dry, and black bile cold and dry. The classical humoral doctrine received the form in which it was to dominate the Middle Ages in the work of Galen (129-199 A. D.). His incorporation of the humoral approach into an encompassing theory of the human digestive system is of particular interest. Galen distinguishes between three successive "digestions". In the first digestive process, food is transformed into chyle in the stomach; the residue of this first digestion is feces. In the second step, the humoral fluids enter the picture. For instance, by the transformation of chyle in the spleen, black bile is produced, while the liver refines chyle into blood; the residue of the second digestion is urine. The third step takes the blood and carries it through the body, sustaining the growth of the body; the residue of this third digestion is perspiration. But while the substance that ensures the growth and maintenance of the body is known as the nutrimental spirit, there are also two other spirits to be taken into account in this third step. In a continuing and cumulative refining process, the heart produces the vital spirit (which regulates the temperature of the body and controls the passions), and the brain produces the animal spirit (Aristotle's pneuma psychikon, which commands the movement of the body, but also feeling and the workings of the mind). Further, Galen's digestive anatomy leads to a dietary pharmacology. All plants (and foodstuffs in general) could be characterized by one of four degrees of warmth, cold, wetness, and dryness. Given that diseases are caused by an excess of one of the four humors, and given that these are themselves characterized by the four features just mentioned, the basic therapeutic rule is to put the patient on a diet that will ensure a decrease of the superfluous humor. For instance, because yellow bile is hot and dry, patients suffering from choleric diseases should avoid plants such as garlic and ginger, which are both warm in the fourth degree and dry in the fourth degree. Rather, they should resort to plants such as opium (wet in the fourth degree) and henbane (cold in the fourth degree).

I 58

Dirk

Geeraerts~ Stefan

Grondelaers

In the course of the Middle Ages, the Galenic framework was further developed into a large-scale system of signs and symbols. In a typically medieval analogical way of thinking, widely divergent phenomena (ranging from the ages of man to astrological notions such as the system of the planets and the signs of the zodiac) were fitted into the fourfold schema presented by the medical theory. In Table 1, an overview is given of a number of those correlations. It should be mentioned, however, that the system was not entirely without unclarities (which is not surprising for a system that was to a large extent devised independently of empirical observation). For instance, while there was general agreement on the core of the system, authors would differ as to the more peripheral elements (such as the question which planet correlates with which humor; in particular, the associated animals are highly unstable across authors). Also, the system, so to speak, contained its own sources of confusion. There is, for instance, a marked ambiguity in the use of the concept "blood", which was not only considered to be one of the four basic fluids, but which was also thought to transport the other humors, and which could hence also be used to refer to the mixture of the humors that was carried through the body (cf. Schafer 1966: 4). And while the basic color associated with yellow bile is obviously yellow, it was believed that yellow bile turned red when heated (and black when it was entirely burned up), so that the color "red" could receive multiple interpretations within the system. The humoral edifice began to be undermined as soon as the Renaissance introduced renewed empirical medical investigations. Harvey's discovery of the circulation of the blood, for instance, was in direct contradiction to the traditional position of the blood in the Galenic "digestive" Table I. A system of humoral correspondences

Characteristic Element Temperament Organ Color Taste Season Wind Planet Animal

Phlegm

Black bile

Yellow bile

Blood

cold and moist water phlegmatic brain/bladder white salty winter North moon turtle

cold and dry earth melancholic spleen black sour autumn West Saturn sparrow

warm and dry fire choleric liver/stomach yellow bitter summer South Mars lion

warm and moist air sanguine heart red sweet spring East Jupiter goat

Cultural traditions and metaphorical patterns

159

system. However, the disappearance of the theory from the medical scene was only very gradual, and it took approximately another three centuries before the last vestiges of the humoral framework were finally removed. The standard view of the historians of medicine is, in fact, that it was only in the middle of the nineteenth century (and more particularly, with the publication of Rudolf Virchow's Die Cellularpathologie of 1858) that the humoral pathological conception received its final blow. This "final" character only holds, of course, for the official medical science: we have already seen in the introductory section about masturbation that traces of the old doctrine continued to exist for much longer in popular belief. Along the same lines, it could probably be shown that the contemporary revival of herbalist medicine at the fringe of official medicine has direct links with the Galenic dietary pharmacology.

3. Anger in art As we have seen, the humoral doctrine had developed into a full-fledged semiotic system in the course of the Middle Ages: an ordered set of signs for medical and psychological interpretation. As a first indication of the fact that this semiotic system was not confined to the field of medicine, let us see how it influenced the artistic production of the Renaissance. We shall give two examples, one from the pictorial arts, and one from the dramatic arts. Cesare Ripa's Iconologia of 1593 was undoubtedly one of the major reference works for the seventeenth-century graphic artist. It contains a thematic inventory of the emblematic subject matter of art, that is to say, of the topics, motifs, and symbols that could be used in paintings, drawings, engravings, and the like. There is a separate section in Ripa devoted to the four temperaments, with a detailed enumeration of the iconography associatd with each of the four types. This is how Ripa introduces the choleric temperament (Figure 1 reproduces the original woodcut from Ripa 1593). Vn giouane magro di color gialliccio, & con sguardo fiero, che essendo quasi nudo tenghi con la destra mano vna spada nuda, stando con prontezza di voler combattere. Da vn lato (cioe per terra) sara vno scudo in mezo del qual sia dipinta vna gran fiamma di fuoco, & dall'altro lato vn feroce leone. [A thin young man in yellow, with a ferocious face, almost naked, holding a drawn weapon in his right hand, ready to fight. From one side, a shield

160

Dirk Geeraerts-Stefan Grondelaers

will be placed on the ground, with a flame of fire painted in the middle, and from the other side, a ferocious lion.]

Each of these characteristics is then further explained and elaborated in the course of Ripa's expose, which is interspersed with references to and quotations from authorities such as Galen, Ovid, S~neca, and Avicenna. In general, the attributes mentioned in Ripa's description can be easily related to the characteristics mentioned in Table 1. Basically, the irascibility of the choleric person is symbolized by depicting him as a battle-prone warrior. Note that each of the details subtly contributes to the meaning of the whole; in particular, the fact that the young man is naked, and the

co:M PLESSIONI C 0 L E R J C 0

PER J L

J.'

r

0 C O.

Figure I. The emblematic representation of the choleric in Ripa 1593

Cultural traditions and metaphorical patterns

161

fact that he is not carrying his shield but that it is merely lying on the ground, indicate the impulsiveness of his hot-tempered nature: in his fits of rage, he does not even think about his own protection. If this impulsiveness is the negative side of his personality, the braveness symbolized by the lion is its positive side. Further elements that can be traced easily are the fire, and the yellow color (corresponding, of course, with the yellow bile that is the physiological basis of this type). Less clear perhaps is the leanness of the young man's body, but this is an expression of the consuming character of the dry heat that is typical of the choleric physiology. It is worthwhile noticing that Ripa's description contains only the basic iconography of the four temperaments. It suffices to have a look at Klibansky, Panofsky and Saxl's (1964) magisterial monograph on Durer's well-known woodcut Melancolia I to get an idea of the intricacies and subtleties that arise when the humoral iconology is used and transformed by a truly creative artist. But the influence of the humoral semiotic system was not confined to the graphic arts. For instance, it has been described by various authors (Campbell 1930; Cruttwell 1951; Draper 1965; Schafer 1966; Pope 1985; Kail1986) how the psychology of Shakespeare's dramatic characters unmistakingly refers to the theory of humors. Just a few quotations from The Taming of the Shrew suffice to demonstrate this. (1)

Now, were I not a little pot and soon hot [IV.l.5]

(2)

Is she so hot a shrew [IV.1.18]

(3)

I tell thee, Kate, 'twas burnt and dried away, and I expressly am forbid to touch it, for it engenders choler, planteth anger; and better 'twere that both of us did fast, since, of ourselves, ourselves are choleric [IV.1.157 -161]

(4)

Gru. What say you to a neat's foot? Kath. 'Tis passing good, I prithee let me have it. Gru. I fear it is too choleric a meat. How say you to a fat tripe finely broil'd? Kath. I like it well. Good Grumio, fetch it me. Gru. I cannot tell. I fear 'tis choleric. What say you to a piece of beef and mustard? Kath. A dish that I do love to feed upon. Gru. Ay, but the mustard is too hot a little [IV.3.17-25]

162

Dirk Geeraerts- Stefan Grondelaers

The conceptualization of anger in these quotations conforms to the model furnished by the theory of humors: anger is caused by choler (3), the production of which may be stimulated by certain kinds of food (3), (4); while a choleric temperament is a permanent personality trait (3), the main attribute of the choleric personality is hotness (1 ), (2). (This is not to say, by the way, that Shakespeare's use of the humoral doctrine is unoriginal: see Pope 1985 on the vivid and original way in which he handles the humoral concepts.) The fact that passages such as the ones quoted above can be multiplied from the works of Webster, Marlowe, or Johnson, leads Schafer (1966) to the conclusion that the humoral conception of physiology and psychology was something of a true fashion in Elizabethan drama. He attributes this to the fact that it was only in the middle of the sixteenth century that the doctrine became known to a wider audience than that of learned men who could read the medical authorities in their Latin and Greek originals. It was only, in other words, after the invention of printing that works such as Thomas Elyot's Castel of He/the (1539), Andrew Boorde's A Breuyary of H elth (c. 1542) and A Compendyous Regyment or A Dyetary of Helth (c. 1542), or Thomas Vicary's A Profitable Treatise of the Anatomie of Mans Body (1548) could be widely distributed, and that they could contribute to the spreading of the humoral doctrine to the community at large. (In Chapman 1979: 277, the wide distribution of almanacs is mentioned as a specific factor contributing to its popularity.) But if this dissemination of the doctrine of humors from the realm of learned knowledge to that of popular belief implies that it is technically a piece of gesunkenes Kulturgut, the question arises how far it actually sank. In particular, how deep did it become entrenched in the language itself?

4. The lexical legacy We have already seen, in the case of valling, catarrh, rheumatism, and even cold, that single lexical items that are current today may be traced back to the humoral doctrine. These items are not isolated cases. In Table 2, we have systematically brought together a number of items and expressions in three European languages (English, French, and Dutch) that can be considered a part of the legacy of the theory of humors. It will be noticed that the items exhibit various kinds of etymological or semantic relationships with regard to the older medical vocabulary. To begin with, there are items like me/ancolie, co/ere, and flegme that refer directly to the original Latin denominations of the four basic fluids or types of personality. Next, there are items such as bilious and zwartgallig,

Cultural traditions and metaphorical patterns

163

which are based on a synonym (bile) or a translation in the vernacular (gal) of the technical term for the humor in question. Finally, there are items that have a more indirect relationship with the humors, in the sense of being metonymically related with them. Thus, spleen and valling are not formed on the basis of the name of one of the basic fluids, but rather refer, respectively, to the organ typically associated with black bile (and hence to the associated temperament), and to a physiological effect thought to involve phlegm. Not included in the figure but equally revealing are items such as French humeur 'temperament; mood', that involve the generic term for the four fluids. If we zoom in on one of the cells of Table 2, still further examples may be found. According to Roget's Thesaurus, the items listed under (5) all refer to anger or related concepts (the glosses are our own). (5)

choler 'anger' gall 'anger' rouse one's choler 'to elicit anger' stir one's bile 'to elicit anger' galling, 'vexing, causing anger' choleric 'irascible' liverish 'irascible' splenetic 'irascible' hot-blooded 'irascible' fiery 'irascible' hot-headed 'irascible'

Table 2. Lexical relics of the humoral doctrine

English

French

Dutch

Phlegm

phlegmatic 'calm, cool, apathetic'

avoir un flegme imperturbable 'to be imperturbable'

valling (dial.) 'cold'

Black bile

spleen JO 'organ filtering the blood' 2° 'sadness'

me!ancolie 'sadness, moroseness'

zwartgallig 'sad, depressed' (lit. 'black-bilious')

Yellow bile

bilious 'angry, irascible'

co/ere 'anger'

z'n gal spuwen 'to vent (lit. 'spit out') one's gall'

Blood

full-blooded 'vigorous, hearty, sensua!'

avoir du sang dans les veines 'to have spunk, pluck, spirit'

warmbloedig 'passionate' (lit. 'warm-blooded')

164

Dirk Geeraerts-Stefan Grondelaers

Although we will return to methodological problems more systematically further on in the paper, a brief methodological remark may be useful at this point. The basis for quoting a particular item as evidence for the influence of the humoral doctrine is the degree of etymological or semantic motivation that may be attributed to the item in question when it is interpreted in that historical light. Because items such as gall, liverish, choler(ic), and to stir one's bile would simply remain etymological puzzles if the historical medical background were not taken into account, a humoral interpretation has explanatory value for them. But not all of the cases mentioned under (5) are equally clear. Two kinds of more or less problematic cases can be distinguished. In the first place, there are items whose global motivation in terms of the theory of humors is plausible, but whose local motivation within the theory is not entirely clear. Take the case of hot-blooded: if the typical fluid associated with anger is yellow bile, how come this expression contains a reference to the warming up of blood rather than bile? However, we have already seen that blood had a highly specific position in the whole doctrine: it is not only a humor in itself, but also carries the other humors through the body. If, then, blood can also refer to the mixture of the four humors as it circulates through the body, it is not surprising that the warming up that causes anger may be metonymically said to involve the entire mixture. A similar but less easily explainable case of a possible local lack of motivation is splenetic: although this expression supports our general point that the influence of the humoral theory in our contemporary emotional vocabulary can be demonstrated, there seems to be a contradiction in the fact that items referring to the spleen may be either related to the melancholic temperament (see Table 2) or to the choleric temperament (as in [5]). Given that the link between the spleen and melancholy is the orthodox one in the framework of the humoral approach, there are at least two ways in which the association between the spleen and anger could be explained. First, the association between the spleen and anger could be an effect of the confusion about aspects of the humoral doctrine that we have already drawn attention to. Second, the association could be motivated by specific subtleties and refinements of the theory that we have not yet dealt with. In fact, Klibansky, Panofsky and Saxl (1964: 88) draw attention to a passage in Avicenna where a distinction is made between the natural, primary form of melancholy, caused by an overproduction of black bile, and a secondary form of melancholy caused by a combustion of one of the other humors; thus, there is a specifically "choleric" form of melancholy, which typically ex-

Cultural traditions and metaphorical patterns

165

presses itself as a state of frenzy (compare Starobinsky 1962; Jackson 1986 for the history of melancholy). At stake here is the notion of "adust melancholy", which was thought to lead to more aggressive behavior and less fearfulness and sorrow than the natural melancholy that was engendered in a straightforward manner by an overabundance of black bile. (On the distinction between natural melancholy, adust melancholy, and choler-and on the confusion it leads to among scholars-see Soufas's 1990 argumentation that Don Quixote is an adust melancholic rather than the choleric type he has been made out to be in earlier humoral interpretations of Cervantes's work.) Our intention here is not to choose between these alternatives, but to make the methodological point that settling the question requires a detailed diachronic analysis of the development of the humoral theory and of its influence on our emotional vocabulary. If it is in general clear that the historical motivation behind the meaning "irascible" of splenetic has to be sought in the older physiological-psychological conceptions of the theory of humors, a close historical look at the development of that theory would yield valuable information about the specific history of splenetic. If, on the one hand, an explanation is sought in the less central aspects of the humoral theory, the historical analysis should be able to show how, for instance, the dissemination of Avicenna's view that was mentioned above led to the lexical association between the spleen and anger. If, on the other hand, that association is the result of a confusion, it is probably a later development, caused by impurities in the dissemination of the theory from its learned origins to the common people (or, perhaps, by the fact that the theory became less transparent when it gradually lost its scientific and medical respectability). If, then, the cases that are characterized by an apparent local lack of motivation merely establish the need for more detailed historical research, the items that raise global motivational questions are potentially more damaging for the humoral hypothesis. Consider an example like fiery: the (metaphorical) reference to heat could be attributed to the lingering influence of the humoral doctrine, but it could also be motivated on entirely different grounds. Suppose, in fact, that increased body heat is a physiological effect of being in a state of anger, and that anger is metonymically conceptualized in terms of its physiological effects. Rather than a historical motivation as a relic of a now abandoned medical theory, an expression such as fiery would then have an ahistorical physiological motivation. At this point, we can include another set of expressions for the concept "anger" into the discussion. In an analysis that has been

166

Dirk Geeraerts- Stefan Grondelaers

published in several places (Kovecses 1986; Lakoff- Kovecses 1987; Lakoff 1987; Kovecses 1989), conventionalized phrases such as those in (6) have been subsumed by Kovecses and Lakoffunder the general metaphor "anger is heat", which is further specified into "anger is the heat of a fluid in a container" when the heat applies to fluids, and into "anger is fire" when the heat is applied to solids. (We will base our discussion on Kovecses 1989; there are only minimal differences in any case between the four published versions of the analysis). (6)

I had reached the boiling point. She was seething with rage. He lost his cool. You make my blood boil. He was foaming at the mouth. He's just letting off steam. Don't get hot under the collar. Billy's a hothead. They were having a heated argument. When I found out, I almost burst a blood vessel. He got red with anger. She was scarlet with rage. I was fuming. When I told him, he just exploded. Smoke was pouring out of his ears. He was breathing fire. Those are inflammatory remarks. That kindled my ire. He was consumed by his anger.

At a still lower level of analysis, these and many similar expressions are grouped together under labels such as "when the intensity of anger increases, the fluid rises" (his pent-up anger welled up inside him), "intense anger produces steam" (I was fuming), and "when anger becomes too intense, the person explodes" (when I told him, he just exploded). Next to the basic general metaphor "anger is heat", less elaborate metaphorical patterns such as "anger is insanity, anger is an opponent, anger is a dangerous animal", and "causing anger is trespassing" are identified. It will be obvious that the general metaphor "anger is the heat of a f1uid in a container" neatly fits into the humoral views: the body is the container of the four cardinal fluids, and anger involves the heating up of specific fluids (either yellow bile as the direct source of ire, or blood

as n< tl a 0

t

Cultural lrriditions and n/Ciflf'h'" ,, ,,/ l'"li·,,

1'.!

the mixture of the four humors). HowcVl'l. l'.iven the alternativ~: ~:t. pl.1 1tion of the general metaphor in terms or a physio1ogical metonymy, is tere any way in which we can say that the hun1u1 ;d hypothesis provides better explanation of the motivation behind tl11s p:11linllar subset of •ur emotional vocabulary? Apart from the general mcllt1 •dulo1•icd point hat a humoral explanation achieves greater generality hy hl-1111' ;1hk lo ;ombine an explanation of the cases under (5) with an cxpi;III:IIJoll ol :hose under (6), we have to consider two specific reasons for prckJIIIIJ' it over a purely physiological explanation. First, it seems better able to motivate the reference to fluids in the expressions. Kovecses explains these references in the following terms: 'The fluid version [of the basic metaphor] is much more highly elaborated. The reason for this, we surmise, is that in our overall conceptual system we have the general metaphor 'the body is a container for the emotions'" (1989: 53). The latter is illustrated by expressions such as he was filled with anger and she could not contain her joy. However, it is not clear how this metaphor combines with the basic "anger is heat" metaphor to yield the application to fluids (as Kovecses claims it does): the fact that the body is a container for the emotions does not predispose the interpretation towards a conception of the emotions as fluids; after all, the contained emotions could just as well be solids or gases as far as the container metaphor is concerned. We will presently have more to say about the "solids" version of the basic metaphor, but it can already be remarked here that in Kovecses's view of the matter, one would not expect the fluid version to be more elaborate than the solid version, because the container metaphor that is invoked as an explanation does not seem to favor the one over the other. Second, the humoral interpretation may help us to make sense of cases that are beyond the reach of a physiological explanation. In general, one is tempted to argue that a physiological interpretation entails that like physiological effects lead to like patterns of lexicalization. In this sense, it would be a counterargument for the physiological approach that an emotion such as shame, which is no less characterized by redness in the face (flushing) and a subjective impression of increased body temperature than anger, is not lexicalized by the same set of expressions as anger. It would make no sense, for instance, to say that one's blood boils with shame, or that someone is fuming with shame. However, Kovecses has rightly pointed out that there need not be a simple correlation between physiological effects and linguistic patterns, and that motivation does not equal prediction (1989: 85). The physiological effects of anger motivate

168

Dirk Geeraerts- Stefan Grondelaers

our anger vocabulary, but because they do not predict the linguistic situation, emotions with similar physiological effects may be differently conceptualized. This element of caution does not, however, work as easily in the other direction: similar physiological effects need not have similar lexical reflections, yet similar patterns of lexicalization had better correlate with similar physiological effects if the physiological explanation is to have any generality. To take up an example, there exists a rather hackneyed set of expressions to the effect that love is a fire: you can let the flame of your love die out, you can have a steadily burning devotion for someone, and you can feel warm towards that person. On the one hand, this accords well with the humoral belief that love is one of the "hot" emotions. On the other hand, it is physiologically unlikely that persons in love have a permanently raised skin temperature (we, at least, are not aware of physiological research to that effect). Granting, in other words, that the methodological key element is motivation rather than prediction, it does seem to be the case that taking into account the historical humoral background may lead to better motivational success. (A related point that will only be mentioned in passing here concerns the subjective experiential prominence of the physiological effects thought to underlie our emotional vocabulary. Kovecses refers to the experimental results of Ekman, Levenson and Friesen (1983) to prove that anger indeed correlates with higher skin temperature, whereas fear correlates with a decrease. However, independent evidence is needed to show that these objective increases and decreases correlate with subjective experiences of warm and cold. This question is relevant because the changes that were measured were rather small: an increase of 0.15 degrees in the case of anger, and a decrease of 0.01 degrees in the case of fear. Are these changes noticed at all by the individuals concerned? Are they sufficient to cause the subjective experiences that could influence our vocabulary?) On the other hand, let us now play the devil's advocate. A possible objection against the hypothesis that the "anger is heat" metaphor is a legacy of the humoral theory could be based on those cases in which the basic metaphor is applied to solids rather than to body fluids: there is no reference to solids, after all, in the original medical doctrine. Notice, however, that most of the expressions cited by Kovecses as evidence for the "solids" interpretation, as listed unter (7), refer to fire as the source of combustion rather than to a solid substance that is being warmed up; this is aptly rendered by Kovecses's labeling of this metaphorical subpattern as "anger is fire". By contrasting heat as applied to solids with heat

Cultural traditions and metaphorical patterns

169

as applied to fluids, Kovecses's formulations suggest that in the former case, the solids have the same function within the metaphorical image as the fluids in the latter case. But while the fluids are the object of the process of warming up, there is no reference to solids as things that are warmed up in most of the expressions in (7). On the contrary, we mainly find references to fire as the source of the process of warming up. Therefore, while there would be an incompatibility in the images referring to fluids and to solids as the object of the heating process, there is merely a complementarity between the expressions referrring to fluids as the object of the heating up and the expressions referring to fire as the source of the combustion process. And of course, these complementary images dovetail with the hypothesis that the expressions historically have humoral origins. (7)

Those are inflammatory remarks. She was doing a slow burn. He was breathing fire. Your insincere apology just added fuel to the fire. After the argument, Dave was smoldering for days. That kindled my ire. Boy, am I burned up! He was consumed by his anger.

But what about the last two expressions in (7)? In the metaphorical image, there is an unmistakable reference to the person's body as a solid substance being consumed. But either in the presupposition that the fire of anger naturally takes its fuel from the body, or in the presupposition that it may detrimentally spread to the whole body and consume substances that are not its natural source of fuel, there is again no contradiction with the humoral conception of anger. On the contrary, we have already seen in Ripa's description of the choleric that a consumptive burning up of the body, resulting in leanness and thinness, is part and parcel of the original views. We see no reason, in short, to argue that the "solids" interpretation of the "anger is heat" metaphor endangers a humoral interpretation of the motivation behind that metaphor.

5. Attenuating the analysis The foregoing does not imply, to be sure, that the humoral interpretation of our emotional vocabulary is without problems. In order to avoid misunderstanding, we would now like to specify a number of views that we

170

Dirk Geeraerts-Stefan Grondelaers

explicitly do not intend to propagate. First and foremost, we do not think that our analysis could not be further corroborated. Specifically, because we claim that a sizeable portion of our contemporary anger vocabulary is part of the lexical legacy of the theory of the four humors, we are convinced that a historical analysis of the development of our emotional vocabulary is necessary to supplement the foregoing remarks. It would have to be shown, in this respect, that the conceptual model of anger that we attribute to the humoral doctrine has indeed entered the language under the influence of the popular dissemination of the latter, and further, that there is a continuous tradition from that period to ours. Because the (ahistorical) physiological model does not impose such restrictions on the historical development of the language (assuming at least that the physiological correlates of our emotions are historically stable), the ultimate test for the humoral hypothesis consists of a diachronic lexicological analysis. It is not the purpose of this paper to carry out this diachronic analysis, but merely to show how it follows in an obvious manner from the humoral hypothesis. Further, we do not want to create the impression that the whole of our emotional vocabulary can be motivated in humoral terms. We are well aware that various patterns of conceptualization can be discerned in our emotional vocabulary; the fundamental importance of Kovecses's research is precisely that it takes a .major step towards the identification of those patterns. As a consequence, we do not claim that the humoral doctrine has had the same amount of impact on every possible emotion concept. Our basic claim so far has merely been that such an influence cannot be disregarded if a proper insight is to be obtained into the motivation behind our contemporary emotional vocabulary. Precisely how far the humoral influence goes is another matter, and one that can only be solved by further research. Specifically, we do not wish to imply that physiological factors are unimportant for the structure of our emotional vocabulary, nor that they could not interact with the historical humoral influences. Such an interaction could take various forms, one of which is that the physiological factors have a marked influence on the reinterpretation process that expressions with a humoral origin undergo in the course of time. The fact that a number of contemporary emotional expressions have their historical origin in the theory of humors does not imply, to be sure, that the theory synchronically determines the interpretation of those expressions: though our vocabulary for the concept of anger may still bear the imprint of ancient medical theories, we no longer believe in the theory as such

Cultural traditions and metaphorical patterns

171

(notwithstanding isolated relics like the masturbation beliefs mentioned above). This implies that the expressions have gone through a process of reinterpretation. Pope (1985: 179) correctly identifies this reinterpretation process as one in which expressions that were once taken literally acquire a figurative interpretation: Though it [the humoral doctrine] now may be dead in our minds it is far from dead on our tongues. We ·have been taking each other's temperatures for over a hundred years and finding them steady at around 98.4° F, but we still use and understand the language of humoral psychology. The only difference is that when we describe somebody as having hot blood or a cold heart or a dry wit we realize that we are talking metaphorically, whereas in the past we would have believed ourselves to have been talking about physical qualities.

We would suggest, then, that the physiological factors that Kovecses concentrates on could be a crucial factor in this reinterpretation process. As the original literal motivation gradually disappears, the elements of our emotional vocabulary could receive a new interpretation as figurative expressions of the physiological effects of particular emotions. Such a physiological reinterpretation would not be automatic, however; in some cases, the expression could simply lose all transparency (following the valling model), while in others, the new figurative meaning could be purely metaphorical rather than metonymical along the "physiological effects" line. For instance, taking for granted that the origins of fiery in the sense "irascible" are humoral, and also taking for granted that the expression has not become totally opaque in the way in which valling has, its contemporary reinterpreted meaning could be based on the physiological metonymy that anger causes body heat (as Kovecses would suggest), but it could also be the case that fiery is synchronically interpreted on the basis of a metaphorical image; the propensity of the irascible person to burst out abruptly could be compared with the fire's tendency to flare up suddenly. Methodologically, what is required here is an investigation of the way in which the expressions in our contemporary emotional vocabulary are actually interpreted: what kind of interpretation (if any) do people associate with them? Again, we wish to emphasize that a closer scrutiny of the reinterpretation process also naturally includes a historical analysis-if only because the reinterpretation process is a historical phenomenon. By following the historical development of our emotional vocabulary on a step-by-step basis, information about the reinterpretation process can be obtained. In particular, it can be hypothesized that those lexical items that are not easily reinterpreted will sooner disappear from the language

172

Dirk Geeraerts- Stefan Grondelaers

than others. On the basis of this assumption, the relevance of more specific hypotheses can be determined. Is it correct, for instance, that expressions that are easily reinterpreted along the physiological lines set out by Kovecses, are more resistant to a process of lexical loss? The obsolescence of a number of the expressions mentioned under (5) could be an illustration of the same reinterpretation process: if the expressions under (5) are synchronically less lively than those under (6), this may very well signal a historical shift from a humoral to a physiological interpretative framework. It is beyond the scope of this paper to deal with the question in detail, but we hope to have made clear that historical questions such as these follow logically from a consideration of the possible humoral origins of our contemporary emotional vocabulary. Finally, it should be pointed out that we have not tried to answer the question where the humoral theory itself comes from, conceptually speaking. Obviously, it is based on anatomical observations concerning the body fluids, but is there any reason why, within the theory, the concept of anger should be specifically linked to the yellow bile, and to fire? Within a physiological conception of our emotion vocabulary, it seems attractive to postulate that the humoral theory itself draws on a pretheoretical physiological experience of the emotions. At the time of its conception, the humoral theory would then be a literalization of a preexisting, physiologically motivated metaphorical understanding: the conceptualization of anger as fire, for instance, would then primarily be a physiological metaphor that is later turned into a literal statement in the framework of the medical theory of humors. This is a position that is implicit in Kovecses's contribution to this volume: accepting the possible influence of culture-specific influences on the emotion vocabulary available in a specific language, he argues that there exist cross-culturally uniform factors of a physiological nature that constrain and stabilize the cultural conceptions. Kovecses argues that such a conception contradicts the suggestions made in the present paper, but this is a conclusion that we explicitly have to oppose: as should have become clear from the foregoing remarks, we do not claim_ that only cultural factors are important, and that physiological factors could not play a role in the development of our emotion vocabulary. It is important, in this respect, to distinguish between the methodological and the substantive part of the present paper. The paper purports to do two things: to argue for the importance of culture-specific historical research when present-day emotion vocabularies are being considered, and to put forward a particular hypothesis (the humoral one) within such a perspective. In his contribution to the present volume,

Cultural traditions and metaphorical patterns

173

Kovecses accepts the methodological point to the extent that he explicitly recognizes the potential influence of cultural factors on the development of our emotion vocabulary, but he exaggerates the weight we would like to attach to the humoral theory. In particular, we claim that including the humoral doctrine into the picture is important for accounting for our present emotion vocabulary, but we have made no statement about the origins of the doctrine itself. In general, we would therefore like to leave open the possibility that Kovecses is right when he suggests that the humoral doctrine is a culture-specific rationalization of a universal physiologically-based metaphorical understanding of the emotions. However, we would like to stress that there is once again a methodological point to be made: if the physiological conceptualization of anger (as typically embodied in physiological metaphors) precedes the humoral theory, the only way to establish this is by doing historical research. If the suggestion implicit in Kovecses's contribution to this volume is correct, the preHippocratic conceptualization of anger in classical Greek should be based on physiological metaphors. It is beyond the scope of this article to test the hypothesis, but is it methodologically important to see that it is an empirical hypothesis that can be tested through historical research. Here again, our conclusions on the methodological level are more important than those on the substantive level: regardless of whether the origins of the humoral theory are indeed physiologically metaphorical or not, the very question about the origins of the humoral doctrine calls for historical research. To summarize, the present paper is to a large extent hypothesis-forming: we claim that it is necessary to take into account the historical background of our emotion concepts to get a clear picture of the present-day situation, and in particular, that it is necessary to include the humoral doctrine into the investigation. Also, this investigation naturally entails a longitudinal historical analysis of the development of our emotional vocabulary. But if we hope to have established the necessity and the attraction of such a research programme, we certainly do not pretend that we have already carried it out.

6. Methodological musings In the previous sections, we have presented an alternative to Kovecses's analysis of the general metaphor "anger is the heat of a fluid in a container". Instead of a straightforward physiological interpretation, we sug-

174

Dirk Geeraerts-Stefan Grondelaers

gest that it has undergone the influence of the humoral doctrine, but that the original set of humoral expressions has been subjected to a process of reinterpretation and obsolescence. We are now in a position to make some further methodological remarks. Two related topics will be discussed: the use of conventionalized language, and the relationship between folk models and scientific knowledge. Kovecses explicitly takes the conventionalized way in which a particular culture talks about the emotions as an indication of the way in which that culture conceptualizes the emotions; the conventionalized language under scrutiny includes idioms, cliches, sayings, proverbs, collocations, and set expressions in general (1989: 43). Now, while Kovecses states with some emphasis that "each and every expression related to a concept has to be examined if we wish to uncover the minute details of the concept" (1989: 44), the question arises why expressions such as those mentioned under (5) are not included in the observational basis of his treatment of the concept "anger". Why has Kovecses picked out for consideration the particular set of expressions that he actually concentrates on? It could be hypothesized that Kovecses has explicitly restricted his analysis to those expressions that are the most transparent ones for a contemporary audience, i.e., those expressions whose metaphorical nature is still a live one, or, more generally, those expressions that are most readily considered to be motivated by today's speakers of English. The question of motivation can be illustrated by comparing an expression such as valling with an item such as to make one's blood boil. Although they have a common historical motivation in terms of the theory of humors, valling is entirely fossilized and opaque for the contemporary language user, whereas to make one's blood boil could possibly receive a motivation along the physiological lines set out in the previous section. The problem that is at stake here is the same as the one mentioned in that section. Further, it has been identified several times in connection with the "generalized metaphors" approach of Lakoff and Johnson (1980), on which Kovecses's identification of metaphorical patterns is based: if generalized metaphors are cited as evidence for our contemporary way of conceptualizing the world, it does not suffice to identify the metaphor, but it has to be shown on independent grounds that the metaphors are not just dead ones (see, e. g., Traugott 1985; Geeraerts 1981). Those "independent grounds" could be the researcher's intuition, but also, for instance, psycholinguistic experiments in the line of Gibbs (1990). So, what would have to be shown before expressions such as those in (6) are cited as evidence for our present-day way of conceptualiz-

Cultural traditions and metaphorical patterns

175

ing emotions is not just that they are not dead metaphors (in the sense in which valling is opaque and fossilized), but also that the motivation that they actually receive is in terms of the physiological effects of anger. (On related questions, see also Ortony 1988.) Now, it would be unfair to claim that Kovecses ignores the question whether the conventionalized language we use to talk about the emotions actually reflects our current beliefs. He discusses the problem by making a comparison with our astronomical model of the world: expressions such as the sun came up and the sun went down cannot be used as evidence for a geocentric folk model, basically because formalized education has profoundly influenced our way of thinking. This is, says Kovecses, in marked contrast with the situation in the emotional field. As a result of certain scientific discoveries, our educational system has spent several centuries on changing our geocentric view of the relationship between the earth and the sun. The consequence is that, despite our language use, anyone with at least some elementary education would refuse the geocentric view as his or her folk model of the earth/sun relationship. Nothing like this has been the case with the emotions. No such large-scale attempts have been made to change our thinking about them . . . . As a result, we pretty much believe what we say about them .... It seems then that, as far as the emotions go, we still live by and think in terms of a geocentric emotional universe. (1989: 45-46)

There are various things to be said about this view. To begin with, it does not invalidate the methodological problem identified above. Because Kovecses reaffirms the necessity to take into account all expressionsand to take them at face value-the fact remains that he does not follow his own methodological dictum. But if he had followed it, the unmistakable presence of items derived from the theory of humors might have led him to the conclusion that we indeed still (partially) adhere to a "geocentric" view of emotions (i.e., to a prescientific, medieval theory), but that the theory is a humoral one rather than the physiological one he suggests. So we are faced with a dilemma: either Kovecses is right in affirming that our present-day views about the emotions have not been influenced by scientific discoveries (but then his method of taking expressions at face value would lead to the conclusion that we still have a humoral conception of anger), or he would have to reconsider his statements about the influence of scientific theories on our emotional vocabulary (together, in fact, with his belief that emotional expressions can be taken at face value, i.e., without considering the possibility of reinterpretations). The fact, on the other hand, that we clearly no longer take the humoral expressions literally (the fact, that is, that we no longer believe the theory)

176

Dirk Geeraerts-Stefan Grondelaers

can only be attributed to the same kind of dissemination of scientific theories that led to the downfall of the geocentric view in astronomy. Because we have learned about the new anatomical and physiological discoveries, we have abandoned our earlier folk models (by and large, i.e., not counting relic beliefs like those about masturbation mentioned at the beginning of this paper). And even this older folk model itself was not a pure folk model. As we have seen, it was a piece of high, Latinate culture that was gradually incorporated into the common culture through the intermediary of popularizing publications. There is no reason, in short, to believe that our emotional vocabulary is free of scientific influences, and there is no ground for a methodological exploitation of such a conception of the specificity of our emotional vocabulary.

7. Culture and cognition To sum up, we have tried to establish the following points. First, the medieval physiological-psychological theory of the four humors and the four temperaments has left its traces on our emotional vocabulary. Second, the "anger is the heat of a fluid in a container" metaphor identified by Kovecses (1989) can be seen as one of those traces. It is then not motivated directly by the physiological effects of anger, as Kovecses suggests, but it is part of the historical (and reinterpreted) legacy of the humoral theory. Third, Kovecses's neglect of the historical background of our emotional vocabulary prevents him from appreciating the possible impact of the humoral theory; once this possible impact is taken into account, inconsistencies in Kovecses's methodology become apparent. Fourth, further corroboration of the historical-humoral hypothesis requires a longitudinal scrutiny of the historical development of our emotional vocabulary. Because they are the most wide-ranging, we consider the methodological consequences of our investigation to be of primary importance. The basic point as we see it is this: an adequate analysis of the motivation behind cultural phenomena in general and language in particular has to take into account the diachronic dimension. Cultural models, i.e., the more or less coherent sets of concepts that cultures use to structure experience and make sense of the world are not reinvented afresh with every new period in the culture's development. Rather, it is by definition part of their cultural nature that they have a historical dimension. They can only fulfill their role of shaping a community's life if they have a histori-

Cultural traditions and metaphorical patterns

177

cal permanence, that is, if they can be transmitted from generation to generation, assuring continuity over and above an individual's and an individual generation's activities (though not, to be sure, unaffected by them). If cognitive models are cultural models, they are also cultural institutions, and as such, they carry their history along with them: their institutional nature implies their historical continuity. It is only by investigating their historical origins and their gradual transformation that their contemporary form can be properly understood. Now, while one of the major steps forward taken by cognitive semantics has been to put the study of meaning back into its cultural and experiential context, it would seem that the natural consequence of including the diachronic dimension into the investigation has perhaps not yet been fully appreciated. There is an instructive parallel to be drawn here between the cognitive-semantic study of single lexical concepts (as in prototype theory) and cognitive-semantic research into supra-lexical structures such as the cultural models of emotion that Kovecses concentrates on. In the case of purely lexical research, the emphasis on the mechanisms of semantic flexibility that underlie the structure of polysemy (such as metaphor and metonymy) naturally entails a renewed interest in diachronic semantics (see Geeraerts 1988): to a large extent, the synchronic polysemy of lexical items is a reflection of their diachronic development. The point that we are trying to bring home here is that an awareness of the synchronic reflection of diachronic patterns is just as natural and just as important in the case of supralexical cognitive structures as in the case of lexical concepts. If cultures are only cultures because they have a tradition, and if, therefore, cognitive models are only cultural models if they have a chronological continuity and a historical permanence, an awareness of the history of ideas is methodologically indispensable for cognitive semantics.

Note 1. We thank Gaston Vandendriessche of the Department of Psychology at the University of Leuven for his invaluable suggestions. Without his knowledge of the history of psychology and psychiatry, this paper could not have been written. The completion of the paper was made possible by FKFO-grant 2.0078.90 and by grant OT90/7 of the Onderzoeksraad of the University of Leuven. A preliminary version of the paper appeared as Preprint number 133 (1991) of the Department of Linguistics at the University of Leuven. We are grateful to Zolt:in Kovecses for his detailed reactions to the preliminary version, and for putting his contribution to the present volume at our disposal.

178

Dirk Geeraerts-Stefan Grondelaers

References Beek. H. H. 1969

Waanzin in de Middeleeuwen [Insanity in the Middle Ages]. Haarlem: De Toorts. Campbell, Lily B. 1930 Shakespeare's tragic heroes: Slaves of passion. Cambridge: Cambridge University Press. Chapman, Allan 1979 "Astrological medicine", in: Charles Webster (ed.), Health, medicine, and mortality in the sixteenth century. Cambridge: Cambridge University Press, 275300. Cruttwell, Patrick 1951 "Physiology and psychology in Shakespeare's age", Journal of the History of Ideas 12: 75-89. Diepgen, Paul 1955 Geschichte der Medizin. Berlin: Walter de Gruyter. Draper, John W. 1965 The humors and Shakespeare's characters. New York: AMS Press. Ekman, Paul-Robert W. Levenson-Wallace V. Friesen 1983 "Autonomic nervous system activity distinguishes among emotions", Science 221: 1208-1210. Geeraerts, Dirk 1981 Review of George Lakoff-Mark Johnson, Metaphors we live by, Quaderni di Semantica 2: 389-396. 1988 "Cognitive grammar and the history of lexical semantics", in: Brygida Rudzka-Ostyn (ed.), Topics in cognitive linguistics. Amsterdam: Benjamins, 647-677. Gibbs, Raymond W. 1990 "Psycho linguistic studies on the conceptual basis of idiomaticity", Cognitive Linguistics 1: 417-451. Godderis, Jan 1988 Galenos van Pergamon over Psychische Stoornissen [Galenos of Pergamon on psychic disorders]. Leuven: Acco. Irwin, James R. 1947 "Galen on the temperaments", Journal of General Psychology 36: 45-64. Jackson, Stanley W. 1986 Melancholia and depression: From Hippocratic times to modern times. New Haven: Yale University Press. Kail, Aubrey C. 1986 The medical mind of Shakespeare. Balgowla: Williams and Wilkins. Klibansky, Raymond- Erwin Panofsky- Fritz Sax! 1964 Saturn and melancholy: Studies in the history of natural philosophy, religion, and art. London: Nelson. Kiivecses, Zoltan 1986 Metaphors of anger, pride, and love. Amsterdam: Benjamins. 1989 Emotion concepts. New York: Springer. Lakoff, George 1987 Women, fire, and dangerous things. Chicago: University of Chicago Press.

Cultural traditions and metaphorical patterns

179

Lakoff, George- Mark Johnson 1980 Metaphors we live by. Chicago: University of Chicago Press. Lakoff, George-Zoltan Kovecses 1987 "The cognitive model of anger inherent in American English", in: Dorothy Holland- Naomi Quinn (eds.), Cultural models in language and thought. Cambridge: Cambridge University Press, 195-221. Lindeboom, Gerrit A. 1985 Inleiding tot de Geschiedenis der Geneeskunde [Introduction to the history of medicine]. Amsterdam: Rodopi. Major, Ralph H. 1954 A history of medicine. Springfield, IL: Thomas. Ortony, Andrew 1988 "Are emotions metaphors conceptual or lexical?", Cognition and Emotion 2: 95-103. Pope, Maurice 1985 "Shakespeare's medical imagination", in: Stanley Wells (ed.), Shakespeare survey, Vol. 38. Cambridge: Cambridge University Press, 175-186. Schiifer, Jiirgen 1966 Wort und Begriff "Humour" in der Elisabethanischen Komodie. Munster: Aschendorff Verlag. Siegel, Rudolph E. 1968 Galen's system of physiology and medicine. Basel: Karger. Soufas, Teresa Scott 1990 Melancholy and the secular mind in Spanish Golden Age literature, Columbia: University of Missouri Press. Starobinsky, Jean 1962 History of the treatment of melancholy from the earliest times to 1900. Basel: Geigy. Traugott, Elisabeth C. 1985 "'Conventional' and 'dead' metaphors revisited", in: W. Paprotte- Rene Dirven (eds.), The ubiquity of metaphor. Amsterdam: Benjamins, 17-56.

Anger: Its language, conceptualization, and physiology in the light of cross-cultural evidence 1 Zoltim Kovecses

1. The problem What explains or motivates the existence of a particular set of expressions in a given language for talking about a particular emotion? Or, to put the same question slightly differently, why don't we have a completely different set of expressions for an emotion in a given language? There appear to be three things involved in any attempt to deal with the issue: first, we have an emotion terminology (consisting of the words and phrases used to talk about the emotion); second, we have certain ways of conceptualizing the emotion (folk or expert theories of the emotion); third, we have the experience of the emotion. In the case of many emotions, a large part of the experience of emotion is physiological experience. Given this triad, we get two theoretical candidates for an answer to our question. What explains or motivates the existence of a particular set of emotion expressions is either the conceptualization of the emotion or the physiology of the emotion. That is, emotion language is motivated either by conceptualization (i.e., the theories that underlie it) or physiological experience. This is of course a crude and greatly oversimplified answer because, for example, it leaves out of consideration possibilities that the two factors may work in combination or that both conceptualization and physiology may be informed by further factors, such as the broader cultural context. Nevertheless, it gives us a good starting point for the discussion to follow. To relate all this to anger, it can be suggested that, at least as Geeraerts and Grondelaers (this volume) interpret Lakoff and Kovecses's (1987) work on anger, Lakoff and Kovecses's account of anger expressions is closer to a physiology-based explanation. According to Geeraerts and Grondelaers, Lakoff and Kovecses explain the existence of a large number of anger-related expressions in terms of body heat (cf. the general metaphor "anger is heat") which is a physiological response associated with anger. By contrast, it can be suggested that Geeraerts and Grondelaers's account of anger expressions is closer to the first position in that they explain the existence of a set of anger-related expressions in English

182

Zoltan Kovecses

in terms of the medieval theory of humors-a theory that involves a particular way of conceptualizing anger that emerged as a cultural product. The two extreme and apparently diametrically-opposed positions make very different predictions concerning emotion terminologies in different cultures. On the view that conceptualization as a cultural product motivates an emotion terminology, we would expect to find different conceptualizations of emotion, and hence different emotion terminologies, in different cultures. The more the cultures differ, the bigger the differences in conceptualization, and hence terminology, can be expected to be. By contrast, the second view would maintain that, due to the (real or assumed) sameness of human beings and their physiological functioning in emotion, both the physiological experience of an emotion and the emotion terminologies will be the same across cultures. To test these possibilities, I will examine data from several cultures. I will suggest that neither of the above proposals is a valid way of understanding the intricate relationships between language, conceptualization, the human body, and the cultural context in the case of anger. I will argue that language is determined by conceptualization, which in turn is influenced by both culture and the human body and its physiology. But before we can begin, we have to clarify what is meant by "explanation" or "motivation" when we ask what explains or motivates the existence of a particular terminology and conceptualization.

2. The notion of explanation We can use the roughly synonymous terms "explanation" and "motivation" in several distinct senses. One sense is when we link present-day linguistic expressions relating to a domain with a particular conception or theory in a systematic way. For example, it can be claimed that a particular conceptualization (for example, the humoral theory of anger) produces a set of linguistic expressions to talk about anger. That is, in this sense of explanation or motivation a certain conceptualization can be said to cause to exist, or to create a number of linguistic expressions. We might call this the existential use or sense of explanation or motivation. This is basically the use of the word motivation that Geeraerts and Grondelaers employ. Explanation or motivation can mean several further things. One is that the use of a particular expression to talk about a particular area of experience is natural for speakers of a language because they are able to find a conceptual link between the expressions and the area of experience. This conceptual link can be said to motivate the ex-

Anger: Its language, conceptualization, and physiology

183

pression. Very often, folk etymologies function like this. This sense of motivation can be extended to various conceptual structures, such as (conceptual) metaphors. A metaphor like "more is up" can be cognitively motivated because we perceive a certain correlation between, for example, pouring more fluid into a container and the level of the fluid. I would like to reserve the term conceptual (or cognitive) motivation for explanations of this kind. Another sense of explanation, or motivation is that the meaning of an expression or concept can be embodied (see Lakoff 1987; Johnson 1987). For example, the experience of increased body heat in connection with anger may make our concept of anger embodied. The conceptualization of anger as a hot fluid in a container seems to be (partially) based on the experience of increased body heat when in a state of intense anger. It can thus be claimed that a certain bodily response associated with anger makes the concept embodied, rather than a disembodied abstraction. We can use the term bodily motivation for this kind of explanation. The connection between existential (or historical) motivation on the one hand and conceptual and bodily motivation on the other is that the latter can influence the former, as will be argued below. This might provide us with a possible fourth sense of explanation. It has to do with the effect of bodily and conceptual motivation on conceptualization. Specifically, I will suggest that bodily and conceptual motivation places certain constraints on the range of possible ways of understanding and conceptualizing anger. For want of a better term, this could be called motivation through constraint. There is also an interesting connection between conceptual or cognitive motivation and embodiment (or bodily motivation): embodiment leads to conceptual or cognitive motivation, but conceptual or cognitive motivation does not necessarily involve embodiment (unless, of course, we regard everything conceptual/cognitive as bodily phenomena as well). In short, in existential motivation X creates, or produces Y; in conceptual or cognitive motivation X (which can be embodiment, or bodily motivation) makes Y natural to have and use; in bodily motivation X makes Y embodied; and in motivation through constraint X (which can be embodiment, or bodily motivation) narrows down the ways in which Y can be conceptualized.

3. Anger in a cross-cultural perspective As was pointed out above, if conceptualization is simply a matter of the particular historical development of cultures, then widely different cul-

184

Zoltan Kovecses

tures should produce widely different conceptualizations (and hence terminologies) of anger. In recent years, several studies have been made that investigate the conceptualization and language of anger in cultures that can be considered significantly different from what can be called EuroAmerican cultures. These studies provide important evidence concerning the issue of whether it is cultural history or physiology that determines conceptualization and terminology. Lakoff and Kovecses (1987) suggest that the central metaphor for anger in English is the container metaphor. This metaphor is a complex one: "anger is the heat of a fluid in a container". It consists of three submetaphors, two of which were explicitly mentioned in that paper and one which was only assumed: "the body is a container for the emotions", "anger is heat", and "emotions are fluids". These three general conceptual metaphors taken together point clearly to the humoral theory of emotion. They provide a fairly clear characterization of the essence of that theory. The body is the container for anger, the fluid can be taken to correspond to either yellow bile or blood, the pressure inside the container is the pressure caused by two much blood or yellow bile, the heat of the fluid can be taken to correspond to the heating up of blood, etc. Given this characterization of the container metaphor in English, I think Geeraerts and Grondelaers make a mistake when they assume that Lakoff and Kovecses do not regard the "anger is a hot fluid in a container" metaphor as a variant of the humoral view. We do, but at the time of writing the paper we were not so much interested in the cultural-historical background or antecedents of the metaphor as in its motivation -in the bodily motivation sense. Therefore, I believe it is a mistake to identify and globally characterize our account of the container metaphor for anger in English as a physiology-based account. It is a physiology-based account only to the degree to which the issue of embodiment plays a role in our understanding of the concept of anger. But that role does not exclude regarding the metaphor as a variant and a later cultural development of the humoral theory. Matsuki (this volume) studies the concept of anger in Japanese. The Japanese term corresponding to anger is ikari. She found examples like the following: (1)

Ikari ga karada no naka de tagiru. 'Anger seethes inside the body.'

(2)

Ikari ga hara no soko wo guragura saseru. 'Anger boils the bottom of stomach.'

Anger: Its language, conceptualization, and physiology

185

These examples provide clear evidence of the existence of the container metaphor for anger in Japanese. We have the idea of the container (the body, stomach), the fluid, and the heat of the fluid. The similarity to the humoral view as captured by the "anger is a hot fluid inside a container" metaphor is unmistakable. In another study, King (1989) examined the conception of anger in Chinese. He observed that the Chinese conceptualize anger as excess qi in the body or parts of the body. Qi is energy and it is viewed as a fluid that flows through the body. Here are some examples from King (1989): (3)

qi man xiong tang. qi full breast 'To have one's breast full of qi.'

(4)

qi yong ru shan. qi well up like mountain 'One's qi wells up like a mountain.'

(5)

hie yi duzi qi. hold back one stomach qi 'To hold back a stomach full of qi.'

(6)

bu shi pi qi fa zuo. make spleen qi start make 'To keep qi in one's spleen.'

NEG

In this case also, we have several elements that are reminiscent of the humoral view. We have the containers, the fluid, and the pressure inside the container. What is missing, however, is the property of heat. It does not seem to play a major role in the Chinese version of the container metaphor, as will be discussed below. Hungarian, a non-Indo-European language, also abounds in expressions that are based on the container metaphor. Three examples will suffice: (7)

Fortyogott a duhtol. was seething-he the anger-from 'He was seething with anger.'

(8)

Forr a vere. is boiling the blood-his 'His blood is boiling.'

(9)

Majd szetvetette a harag. almost burst-him the anger 'His anger almost burst him.'

186

Zo/t(m Kovecses

Here we also have the idea of a container, heat, pressure, etc. However, Hungarian culture has for over a thousand years been a part of Western civilization and thus cannot be considered a culture far removed from other Euro-American cultures. Tahitian can serve as a further illustration of a culture where anger is conceptualized as a force inside a container. For example, Levi (1973; quoted in Solomon 1984: 238) quotes a Tahitian informant as saying: "The Tahitians say that an angry man is like a bottle. When he gets filled up he will begin to spill over." In Wolof, an African language spoken in Senegal and Gambia, the word bax means "to boil" in a literal sense. It is also used metaphorically in the sense of "to be really angry" (Munro 1991 ). The existence of this metaphor indicates that Wolof has something like the container metaphor as a possible conceptualization of anger. Examples of cultures where the container metaphor is the predominant view of anger (and of emotion in general) could be easily multiplied. But perhaps the examples given above will suffice to make the point that is relevant to our discussion. It seems to be the case that many cultures that are widely dissimilar produce roughly similar conceptualizations of anger. The notion that anger is a force (of a fluid), that the force is in a pressurized container, that it exerts pressure on the walls of the container, that the fluid is hot (at least in many of them), that there is a danger of explosion, etc., are all elements of the conception of anger that are present in a wide variety of cultures. What this tells us is that conceptualizations or theories of anger cannot simply be the products of the historical development of cultures. If they were, the conceptualizations or theories would have to be very dissimilar-as dissimilar as are the cultures in which they are embedded. It is clear that there must be a constraining factor that produces this relatively high degree of similarity in the conception of anger in these cultures.

4. Physiology of anger I would like to suggest that the factor that constrains the conceptualization of anger in many cultures is the human body and its physiological functioning in emotion. That is, my claim is that the metaphorical understandings of anger as discussed above are in part based on shared ideas about the human body and certain physiological processes that are asso-

Anger: Its language, conceptualization, and physiology

187

cia ted with anger. Consider the following linguistic descriptions of physiological responses of anger in the cultures we have discussed above: (10)

[body heat] English: (examples from Lakoff- Kovecses 1987) Don't get hot under the collar. Billy's a hothead. They were having a heated argument. When the cop gave her a ticket, she got all hot and bothered. b. Chinese: King (1989) provides examples only for a fire/heat metaphor, but not metonymies that refer directly to body heat. c. Japanese: (data obtained from Noriko Ikegami and Kyoko Okabe) ( Watashi-no) atama-ga katto atsuku-natta. my head get hot 'My head got hot.' Karera-wa atsui giron-o tatakwasete-ita. they heated argument were having 'They were having a heated argument.' Atama o hiyashita hoo ga ii. head cool should 'You should cool down.' d. Hungarian: forr6fejii hotheaded felheviilt vita heated argument e. Tahitian: Solomon (1984) does not contain data for heat. f. Wolof: (data obtained from Munro 1991) tang to be hot to be bad-tempered Tangal na sama xol. He heated my heart. 'He upset me, made me angry.' a.

(11)

a.

[internal pressure] English: (examples from Lakoff- Kovecses 1987) Don't get a hernial When I found out, I almost burst a blood vessel. He almost had a hemorrhage.

188

Zo/t(m Kovecses

b.

c.

Chinese: (data obtained from King 1989) Qi de naomen chong xue. qi DE brain full blood 'To have so much qi that one's brain is full of blood.' Qi po du pi. break stomach skin 'To break the stomach skin from qi.' fei dou qi wha le. lungs all explode LE 'One's lungs explode from too much qi.' Japanese: (data obtained from Noriko Ikegami and Kyoko Okabe) Kare no okage de ketsuatsu ga agarippanashi da. he due to blood pressure to keep going up 'My blood pressure keeps going up because of him.' sonna ni ikiri tattcha ketsuatsu ga agaru yo. like that get angry blood pressure to go up 'Don't get so angry, your blood pressure will go up.' aosuji o tatete okoru. a blue vein to stand to get angry 'To turn blue with rage/anger.'

d.

Kare-wa chinoke-ga ooi. he blood a lot of/much 'He has a short temper.' Hungarian: Agyw?rzest kap. cerebral hemorrhage get 'Have a herorrhage.' Felmegy benne a pumpa. up-goes in-him pump 'Pressure rises in him.' Pel tudna robbanni. up could burst 'He could burst.' Kidagadnak a homlokim az erek. out-swell the forehead the veins 'The veins come out on his forehead.'

Anger: Its language, conceptualization, and physiology

e. f.

(12) a.

b.

c.

d.

189

Felment a w?rnyomasa. up-went the blood pressure 'His blood pressure went up.' majd kipattantak az erei. almost out-burst the veins 'His veins almost burst.' Tahitian: no data. Wolof: no data.

[redness in face and neck area] English: She was scarlet with rage. He got red with anger. He was flushed with anger. Chinese: (data obtained from King 1989) Ta /ian quan hong le yanjing mao huo lai. he face all red LE eyes emit fire come 'His face turned red and his eyes blazed. Qi de /ian dou zi le. qi DE face all purple LE "To have so much qi that one turns purple.' Japanese: (data obtained from Moriko Ikegami and Kyoko Okabe) Karewa makka ni natte okotta. he red to be get angry 'He turned red with anger.' makka ni natte okoru red become get angry 'get mad/angry' Kare wa ikari-de akaku-natta. he with anger got red 'He got red with anger.' Hungarian: Voros lett a feje. red became the head 'His head turned red.' Lila lett a feje. violet became the head 'His head turned voilet.'

190

Zoltan Kovecses

e. f.

Elvorosodott diiheben. reddened in-his-anger 'He reddened with anger.' Voros volt, mint a pulyka. red was like the turkey 'He was red like a turkey.' Tahitian: no data. Wolof: no data.

Lakoff and Kovecses (1987) call linguistic expressions like the above metonymies and distinguish them from metaphors. The metonymies describe physiological responses that are associated with anger and stand in a part-whole relationship to anger. As can be seen, body heat is clearly present in English, Japanese, Hungarian, and Wolof. King (1989) does not indicate the presence of a separate body heat metonymy for Chinese, but he does indicate a fire/heat metaphor. We do not have data for Tahitian. Internal pressure is present in English, Chinese, Japanese, Hungarian. We do not have data for internal pressure in Tahitian and Wolof. The physiological response "redness in the face and neck area" can be taken to be the result of both heat and internal pressure. This response seems to characterize English, Chinese, Japanese, and Hungarian. There is no data for Tahitian and Wolof, although the Wolof word boy that means 'to be red hot" (of charcoal) also means "to be really angry" (Munro 1991). The presence of these conventionalized (lexicalized) physiological responses in these languages seems to provide very strong (bodily) motivation for the particular (metaphorical) conceptualizations of anger in the cultures discussed, that is, for the complex container metaphor. The lexicalized responses of body heat and redness serve as a basis for the heat component in some of the metaphors (cf. the English, Japanese, Hungarian, and Wolof metaphors), while the lexicalized response of internal pressure and redness provide the experiential basis for the pressure component in the various metaphorical conceptualizations (cf. the English, Chinese, Japanese, and Hungarian metaphors). Unfortunately, we do not have relevant data for Tahitian that would enable us to see whether Tahitians use any internal pressure metonymies that would motivate their apparently presssure-dominated metaphorical conception of anger. It appears that the metaphors used are all specific forms of a pressurized container image. Some of them work with heat, some of them do not. As we have just seen, this image of an angry person is strongly motivated by

Anger: Its language, conceptualization, and physiology

191

bodily experience. This motivation can be viewed as a constraining factor that delimits the possible metaphorical systems of anger. The metaphors seem to conform to the patterns defined by the physiology-based metonymies. In short, physiology can be seen as setting boundaries to the range of possible conceptualizations.

5. Language, conceptualization, physiology We have seen that a large part of conceptualizing anger consists of metaphorical understanding. The metaphorical understanding of anger is characterized by the various container metaphors in different cultures. The container metaphor accounts for a large number of expressions related to anger. In other words, this particular image of anger and human beings can be viewed as determining the kinds of language we use to talk about anger. But the container metaphor does not exhaust the ways in which anger is conceptualized. Another, and obviously related way of conceptualizing anger occurs when we select, categorize, and name certain physiological experiences accompanying anger. In language, these usually take the form of conceptual metonymies (like body heat, internal pressure, etc.). The particular instances of these categories for physiological responses are linguistic metonymies, just like the particular instances of the container metaphor are linguistic metaphors. In short, we have both a metaphorical and a metonymical conceptualization of anger. What this implies is that the language (or the terminology) of anger is the result of both metaphorical and metonymical understanding. Some languages appear to place more emphasis on metonymical than on metaphorical conceptualization. It is an interesting question whether there are languages where there is more metaphorical than metonymical understanding. We can ask further what is the relationship between metaphorical understanding (i.e., the container metaphor) and metonymical understanding (i.e., the various conceptual metonymies) as regards anger. The conceptual metonymies provide motivation for the metaphorical conceptualization of the angry person as a pressurized container. Specifically, they make this particular conceptualization natural for people. If conceptualized physiological responses include an increase in internal pressure as a major response in a given culture, people in this culture will find the use of the pressurized container metaphor natural. In the case of anger, naturalness arises out of embodiment. Embodiment occurs when it is

192

Zo/tan Kovecses

really the case that my temperature and blood pressure rise. This is what makes studies of human physiology during emotional states crucially relevant for cognitive approaches to the study of the language and conceptual systems of emotion. As a result of these studies, we have evidence that anger does indeed go together with objectively measurable bodily changes such as increase in skin temperature, blood pressure, pulse rate, and more intense respiration (see, for example, Ekman- Levenson- Friesen 1983). In addition to giving rise to motivation (i.e., naturalness), embodiment has another function in the conceptualization of anger. It puts certain limitations (either directly or through the mediation of naturalness) on the possible ways in which anger is conceptualized (including conceptualizations expressed in expert theories-many of which are "hydraulic" in nature, and thus container metaphors (see Solomon 1984). However, it is not suggested that embodiment actually produces the container metaphor, but that it makes a large number of other possible metaphorical conceptualizations either incompatible (in the case of direct influence) or unnatural (in the case of indirect influence). It would be odd to conceptualize anger as, say, softly falling snow, an image completely incompatible with what our body is like and what our physiology does in anger. It is in this sense that the particular embodiment of anger is seen as limiting the choice of available metaphors for anger. A major implication of this state of affairs is that the embodiment of anger appears to constrain, in the sense above, the kinds of metaphors that can emerge as viable conceptualizations of anger. This seems to be the reason why very similar metaphors have emerged for the concept in a variety of different cultures. It is on the basis of this similarity that the metaphors in different cultures can be viewed as forming a category of metaphors, a category that we have called that of the container metaphor. Without the constraining effect of embodiment, it is difficult to see how such a surprisingly uniform category (of metaphors) could have emerged for the conceptualization of anger. The widely different cultures we have examined should have produced a great deal more diversity in (metaphorical) conceptualization than what appears to be the case on the basis of the data we have worked with. But it is equally important to realize that the container metaphors in the different cultures are not exactly the same; there is some variation in them. This variation appears to arise from two sources. One source has to do with culture. When it is suggested that the tendency of cultures to diversify metaphorical conceptualization is greatly constrained by em-

Anger: Its language, conceptualization, and physiology

193

bodiment, it is not implied that culture does not play any role in conceptualization. It does, and a major role at that. Despite their generic-level similarities, the specific container metaphors are composed of greatly differing ingredients. For example, yellow bile as a fluid is a component of the humoral view alone, and focus on the stomach/bowels area (hara) as a pressurized container is a predominantly Japanese characteristic of the metaphor (for this and other culturally-determined differences, sees Matsuki, this volume). This means that the conceptualization of anger is influenced by the broader cultural context, the cultural-social meaning system and physiology. However, these influences seem to operate on two different levels of abstraction. A generic-level container schema seems to be motivated by physiology and the details of the schema at the specific level seem to be filled out by the cultural system. For example, the generic-level component of energy or force in the container is filled out by the specific-level and culture-specific notions of qi in Chinese, yellow bile in the humoral theory, and hot fluid in the more recent version of the humoral view. The other source of variation is physiology itself. The physiological responses associated with anger also seem to vary. As we have seen, Chinese culture appears to place a great deal more emphasis on the increase in internal pressure than on body heat. King's (1989) data suggest that Chinese abounds in metonymies relating to pressure, but not to heat. This seems to result in a particular kind of container metaphor, one in which the component of pressure is emphasized to the exclusion of heat. In general, there may be differences between cultures in both conceptualized and real physiology. That we find many cases where cultures differ with respect to conceptualized physiology (i.e., with respect to what their folk theories of physiological effects are) is not surprising. But it is also possible that cultures do not produce the same physiological symptoms in anger (see, for example, Solomon 1984). These differences in physiology may also contribute to the subtle differences in the metaphorical conceptualization of anger. At the same time, cultures seem to share enough physiological responses on the basis of which they produce roughly similar conceptualizations. This is what gives us the (slightly different versions of the) container metaphor in a variety of very dissimilar cultures. It is important to see in this discussion what is not claimed. It is not claimed that wherever physiological responses that have to do with anger are perceived and named, the container metaphor for anger will also exist. That is, physiological responses do not produce the metaphor (in

194

Zolt{m Kovecses

the sense of existential motivation). For example, we know of cultures where anger is talked about in terms of body heat. In Chickasaw (alanguage in the Muskogean family), the expression sa-palli means "I am hot" and it can also mean "I am angry" (Munro 1991). However, as Munro observes, Chickasaw does not seem to have a hot fluid in a container metaphor for anger. What is claimed then is that if a language does have this metaphor, it is likely to also have expressions that describe physiological responses, which in turn provide bodily (and cognitive) motivation for this way of conceptualizing anger. Furthermore, there are cultures in the world where the container metaphor for anger plays an insignificant role in comparison with folk conceptions that are very different from it; for example, in Ifaluk, a Micronesian atoll, the folk conception of anger emphasizes the prosocial, moral, and ideological aspects of anger (Lutz 1988)-as opposed to the antisocial, individualistic, and physical aspects that the pressurized container metaphor emphasizes in Western cultures.

6. Conclusions Given the provisos expressed in the preceding paragraphs and in the light of the currently available (incomplete) linguistic evidence, I believe the following tentative conclusions can be drawn. If the conceptualization of anger were only a matter of culture, we would have to have radically different conceptualizations in the case of radically different cultures. The conceptions of anger we have looked at indicate that this is not the case, despite the fact that the cultures that were examined were significantly different. Thus the conceptualization of anger must be influenced by factors over and above the particular historical development of culture. The constraining or stabilizing factor seems to be human physiology in anger, as well as the particular image we have of the physical structure of human beings across cultures. This result contradicts the suggestions made by Geeraerts and Grondelaers (this volume). It tells us that the human body and its physiology are not things that people make use of to reinterpret the humoral theory of anger (as a cultural product), but that the humoral theory of anger itself and all the other cultural variants of the container metaphor are based on the body and its physiology (in the sense of naturalness, embodiment, and motivation through constraint).

Anger: Its language, conceptualization, and physiology

195

The conceptualization of anger is influenced by both culture and physiology. A large part of the terminology of anger in a number of cultures is determined by the container metaphor and the various conceptual metonymies for anger. The conceptualization of anger and the angry person seem to be informed by culture (at a specific level) and constrained by certain shared physiological processes in anger and the probably universal image of human beings as containers (at a generic level). The image of anger and the angry person as a force inside a container has emerged because this is compatible with the assumed structure of the human body and its physiology in anger. This image is felt to be natural among ordinary people in many widely different cultures, but also among many psychologists and anthropologists who study emotions professionally (see, for example, Solomon 1984; Lutz 1988). This naturalness of the container metaphor for anger seems to arise from the embodiment of our conceptualization of anger. Note 1. In writing this paper I have benefited greatly from conversations with George Lakoff. Many people have made valuable comments on previous versions of the paper: Catherine Allen, Susan Gal, Tony Graybosch, Andras Sandor, John Taylor, Linda Thornburg. I am grateful to them all. Needless to say, all responsibility is mine. I am also indebted to Dirk Geeraerts and Stef Grondelaers for their useful criticism of my work on emotion concepts. Their criticism has helped me put many of my ideas in sharper focus.

References Ekman, Paul-Robert W. Levenson- Wallace V. Friesen 1983 "Autonomic nervous system activity distinguishes among emotions", Science 221: 1208-1210. Geeraerts, Dirk-Stefan Grondelaers this vol. Looking back at anger: Cultural traditions and metaphorical patterns. Johnson, Mark 1987 The body in the mind. Chicago: University of Chicago Press. King, Brian 1989 The conceptual structure of emotional experience in Chinese. [Unpublished Ph.D. dissertation, Ohio State University.] Kovecses, Zoltan 1990 Emotion concepts. New York: Springer. Lakoff, George 1987 Women, fire, and dangerous things: What categories reveal about the mind. Chicago: University of Chicago Press. Lakoff, George- Zoltan Kovecses 1987 "The cognitive model of anger inherent in American English", in: Dorothy Holland- Naomi Quinn (eds.), Cultural models in language and thought. Cambridge: Cambridge University Press, 195-221.

I 96

Zolt(m Kovecses

Lutz, Catherine A. 1988 Unnatural emotions, Chicago: University of Chicago Press. Matsuki, Keiko this vol. Metaphors of anger in Japanese. Munro, Pamela 1991 "Anger is heat": Some data for a cross-linguistic study. [Unpublished study, Department of Linguistics, University College of Los Angeles.] Solomon, Robert C. 1984 "Getting angry: The Jamesian theory of emotion in anthropology", in: Richard A. Schweder- Robert A. LeVine (eds.), Culture Theory. Cambridge: Cambridge University Press, 238-254.

The metaphorical concept of mind: "Mental activity is manipulation" 1 Olaf Jakel

Socrates: Well now, what names are needed for catching any piece of knowledge one wants, and having taken, for holding it, and letting it go again? Plato, Theaetetus

1. Introduction: Construing a world within What kind of terms does everyday language hold in store for discourse about the mind? In an application of the cognitive theory of metaphor as developed by George Lakoff and Mark Johnson, 2 the abstract domain of mental activity can be profitably investigated. 3 The results of this investigation show that, to a large extent, this domain is conceptualized metaphorically in terms of the physical manipulation of solid objects. The purpose of this paper is to describe the conceptual metaphor "mental activity is manipulation" as a complex cognitive model in which several submetaphors form a coherent system. The explanatory value of this metaphorical idealized cognitive model lies in its accounting for the semantic motivation of a host of conventional English expressions that give voice to cognitive processes such as learning and understanding, problem solving and judging, remembering and forgetting. As a whole, this cognitive model, having been reconstructed from linguistic evidence, might be regarded as representing a kind offolk theory of mental activity, i.e., a folk epistemology. 4 This model is described in section 2. The language data presented consist of only a fraction of the corpus analyzed in the original study. 5 A more finegrained analysis of some of the linguistic material as well as of the invariance issue (cf. Lakoff 1990) will be provided in a more comprehensive study (Jakel, to appear). The aim here is to introduce the folk theoretical model "mental activity is manipulation" as a highly productive option for construing a world within. Following a short summary of results (section 3), the motivation of the metaphorical idealized cognitive model will receive some attention

198

Olaf Jakel

(section 4). Alternative models of construing the intellectual domain are then discussed briefly (section 5), before a section on etymological evidence takes us back to our main idealized cognitive model. The paper concludes with some remarks on the assessment of competing models and on cross-linguistic evidence.

2. "Mental activity is manipulation": The idealized cognitive model of a folk epistemology Our reconstruction of the idealized cognitive model of a folk epistemology will not yield the folk theory of all mental phenomena. 6 Apart from the fact that there are alternative models (see below, section 5), the idealized cognitive model "mental activity is manipulation" does not include passions, affections, or intuitions, which are conceptualized as "passivity". Furthermore, the interpersonal realm, though it also contains "mental" phenomena such as conversation, will be excluded. We will be looking at the conceptualization of active, intrasubjective thought processes. Our examination of "mental activity is manipulation" will reveal a rather complex model made up of a large number of systematically linked conceptual metaphors. 7 The idealized cognitive model is subdivided into eight model-integrated components or M!Cs, basic conceptual metaphors each of which structures a whole segment of the overall model. Although these segments could be seen as forming submodels of the idealized cognitive model, there is no real generic-specific hierarchy (cf. Lakoff 1990: 68-72 and Lakoff- Turner 1989: 80-81) involved. The system of the folk model is that of a loose sequence in which each component focuses on a different aspect of mental activity. For example, one component may have a stronger accent on learning processes, while another will focus on judging. However, such clear distinctions as those of a scientific theory should not reasonably be expected of the folk theory. The overall metaphor "mental activity is manipulation" supplies the conceptual linkage between the segments, each of which utilizes a different kind of physical object manipulation. The eight model-integrated components will be treated one by one in the following sections.

2.1 The stuff that thoughts are made of The conceptualization of mental activity as manipulation starts from the basis of a reification of abstract problems, issues, and ideas as concrete,

The metaphorical concept of mind

199

solid objects: an ontological metaphor par excellence in Lakoff's and Johnson's (1980: 25) terminology. Indeed, the metaphorical concept (MIC 1)

"Ideas are solid objects"

is substantiated by weighty linguistic evidence: 8 (1)

What's the matter?

Here the concrete term for matter is used most naturally to refer to abstract issues. These can be given some of the essential qualities of solid objects: abstract idea objects can be multidimensional as in (2), difficulty can figure as hardness (3) or toughness (4), and importance as physical weight (5): (2)

It is difficult for unilateralists to admit there is another side to the missile question.

(3)

That is a very hard question to answer.

(4)

It's a tough problem.

(5)

Let us turn to less weighty matters.

These and similar expressions are motivated by conceptual metaphors which can be phrased as "complexity is dimensionality", "difficulty is hardness", and "importance is weight". We will encounter these metaphors again in our treatment of the various aspects of mental activity.

2.2 Gathering material Once ideas and problems are conceptualized as solid objects, mental activity can start with the thinking subject's attempts at spatially approaching an idea object. Within the general structural metaphor (Lakoff- Johnson 1980: 14) (MIC 2)

"Understanding an idea is establishing physical closeness"

four phases can be distinguished by which the metaphorical distance between thinker and object is gradually reduced. From searching and hunting (MIC 2a) to seizing (MIC 2b) and picking up (MIC 2c) to taking in (MIC 2d), physical actions are utilized in the transfer to the domain of learning and understanding. We will follow this chain of action step by step in the next four subsections. The idea objects become goals to be reached by the thinker: "ideas and problems are targets". With this, the metaphorical search can begin:

200

Olaf Jiikel

2.2.1 "Understanding starts with attempts at finding and hunting idea objects" ( MIC 2a)

An example of this model-integrated component (MIC 2a) is (6). (6)

Sally searched for an idea all day.

If still unsure, the thinker gropes his way in metaphorical space: (7)

Economists started to grope around with increasing desperation for explanations of the recession.

These and the following examples suggest a conceptualization of difficulty: "the objects can be hidden or out of reach". If lucky, the groping person in search of an idea object can all of a sudden hit upon it: (8)

I've struck on a plan.

(9)

He hit on the idea of cutting a hole in the door to allow the cat to get in and out.

Both mental closeness, as in examples (1 0) and (11 ), and distance, examples (12) to (15), are set in spatial dimensions: (10)

The solution is close at hand now.

(11)

I think the problem is within my grasp.

(12)

That concept was beyond my grasp.

(13)

Nothing was further from my mind.

(14)

That theory is out of my reach.

(15)

That concept was above me.

These expressions are based on the orientational metaphor "mental closeness is spatial closeness I mental distance is spatial distance". The sentence pair (11) and (12) in particular suggests a conceptualization of the thinker's capacity to understand: "intellectual quality is physical range of action". Example (15) shows that cognitive inaccessibility can not only be expressed in terms of horizontal, but also of vertical distance. Difficulty may even be due to objects in motion: "problems can be mobile entities". 9 (16)

It went over my head.

(17)

It flew by me.

(18)

It went by me. 10

The metaphorical concept of mind

201

If problems can act so animal- or birdlike, it is only logical for understanding to be described as catching, and attempts at problem solving as hunting: (19)

I don't quite catch the idea.

(20)

He took a stab at the answer.

(21)

He took aim at the problem.

(22)

Have a shot at solving the problem!

(23)

It's a long shot but I think John must have known about the murder.

(24)

It was a complete shot in the dark but it turned out to be the right answer.

Apart from knives (example [20]), guns ([21] to [24]) are used, especially by somebody just guessing (23), (24), to get hold of a solution. In case the target is assumed to be hidden in metaphorical depths, 11 we find metaphors relating to the domains of fishing, examples (25) to (27), or farming and gold digging, examples (28) to (32): (25)

He cast about for ideas.

(26)

He'd been fishing for the answer for weeks.

(27)

I couldn't fathom his meaning.

(28)

Much of the information he gleaned was of no practical use.

(29)

He unearthed the answer.

(30)

It's hard to sift out the truth from the lies in this case.

(31)

The reporter had raked out some interesting facts.

(32)

We'll have to go over it with a fine tooth-comb.

The fact that it is possible to wrestle with a poblem as with some kind of living opponent, as in examples (33) to (35), takes us back to the image of wild animals suggested by the hunting metaphors: (33)

I grappled with this moral dilemma.

(34)

For decades, mathematicians have wrestled with this problem.

(35)

He struggled with the subject.

Once the idea objects have been caught, the handling of the bag is important:

202

Olaf Jakel

2.2.2 "Understanding is seizing idea objects firmly" ( MIC 2b) Understanding is described by a number of expressions from the domain of grasping at bodies: (36)

The concepts were difficult to grasp.

(37)

We have not yet come to grips with it.

(38)

It's not that easy to seize upon an idea.

(39)

I take your meaning.

(40)

Once one has got hold of certain basic facts the rest is comparatively easy.

In a complementary fashion, misunderstanding is conceptualized as reaching off the mark and getting hold of an idea object different from the one intended: (41)

You've got it wrong.

(42)

Don't get the wrong idea: I really like her.

If the thinker does not know for sure or has forgotten, the object slips away from his grip:

(43)

He lost his grip and things got away from him.

(44)

The idea slipped through my fingers.

(45)

It's a slippery concept.

Thus, we have yet another way of conceptualizing difficulty: "the idea objects can be slippery". Further, "manual handling" stands for the mental treating of idea objects: (46)

That's an interesting idea, but not relevant to the matter in hand.

(47)

Abortion was an issue too explosive for him to handle.

To handle a matter properly, both manual dexterity and physical strength are required- "intellectual quality is physical strength and dexterity": (48)

I'm all thumbs at algebra.

(49)

His mind is strong and supple.

The metaphorical concept of mind

203

Due to the orientational metaphor "conscious is up" (see Lakoff- Johnson 1980: 15), understanding and conscious attention can further be conceptualized as regular picking up. 2.2.3 "Understanding is picking idea objects up off the ground" (MIC 2c) 12

This model-integrated component can be illustrated by the following examples. (50)

This has not been raised as an issue by the West.

(51)

Where did you pick up such ideas?

(52)

I take up one problem at a time.

As a metaphorical entailment of MIC 2c, "speed of understanding" cor-

responds to "speed of picking up". Thus, it is no wonder that in our culture the speed of this performance should serve as a measure of intelligence or intellectual grasp: "intelligence is speed on the uptake". (53)

I tried to explain it to him, but he's rather slow on the uptake.

(54)

My little brother is quick on the uptake.

Moreover the thinker can reject an idea by moving it downward again, the image of a forceful movement towards the ground being intensified by indicating the resulting state of "flatness": (55)

She turned the idea down (flat).

At this stage, the important image-schematic metaphor "the mind (consciousness) is a container" 13 needs to be introduced as it provides the orientational basis for the next as well as for all the remaining components of the idealized cognitive model of mental activity. (56)

I'll put up the shelves if you tell me exactly what you have in mind.

(57)

Who put that idea in your head?

Stupidity is expressed as emptiness of this container, 9-s in examples (58) and (59), and the sensible owner of the mind container is seen as being inside this container (60), from which he can depart only into madness (61). (58)

That boy hasn't a thought in his head.

(59)

He was putting on an act to impress an empty-headed girl.

(60)

Nobody in their right mind would enjoy this show.

(61)

He's gone out of his mind.

204

Olaf Jakel

With this prerequisite, the sequence of conceptualizations of a thinker's learning and understanding can be resumed. The spatially closest relation of thinker and idea object is reached in the following conceptual metaphor, MIC 2d. 2.2.4 "Understanding is taking idea objects into the mind container" (MIC2d)

Examples include: (62)

They listened to my lecture, but how much did they take in, I wonder!

(63)

I can't get this Latin grammar into my head.

(64)

Taking everything into consideration, the result is better than I expected.

Thus it seems logical that intellectual receptivity requires a thinker to open his mind container, as in examples (65) and (66), rather than closing it, (67). For the same reason, a spacious container (68) serves better for the purpose of taking in new idea objects than a narrow one (69): "intellectual receptivity is openness and spaciousness of the mind container". (65)

Open your mind to some new thoughts.

(66)

She tried to keep an open mind on such subjects.

(67)

... a closed mind.

(68)

She assured me that her parents were broadminded.

(69)

How stupid and bigoted and narrow-minded he had become.

2.3 Interior view of the mental workshop The linguistic evidence of the remaining sections suggests a more specific characterization of the consciousness container: "the mind is a workshop". This workshop has certain internal furnishings, which metaphorically represent the structure of our conscious mind. Thus, the idea objects taken in to be worked on can be placed in different parts of the workshop: (MIC 3)

"Within the mental workshop, the idea objects are stowed away according to urgency"

The metaphorical concept of mind

205

The general orientational metaphor "importance is centrality" (cf. Johnson 1987: 124) and its opposite, "unimportant is peripheral", probably account for the conceptualization: "unimportant issues are given peripheral positions", as in: (70)

Put aside for a moment the fact that the man has been in prison.

Problems may even be put away completely, as in (71), even to the remotest corner of the workshop (72), or be placed on shelves put up especially for this purpose (73): (71)

He's had to put away all ideas of becoming a concert pianist.

(72)

She put the idea to the back of her mind.

(73)

I had simply shelved this awkward problem.

(74)

These crucial issues tend to get pushed aside and forgotten.

(75)

She brushed the thought away.

Irrespective of whether the idea objects are removed in a rather rough (74) or in a more gentle (75) manner, the workshop must be kept clear to leave room for the thinker's real mental efforts.

2. 4 The problem-solving scenario Before we can describe the processing of problem and idea objects inside the mental workshop, it is necessary to introduce another ontological metaphor: "the mind (intellect) is a tool" . 14 This gives rise to a conceptualization of alertness, in that the mind tool ought to be carried about by the mental workman: (76)

It he hadn't had his wits about him, he might have been drowned.

Within this frame, we find yet another way of conceptualizing intellectual qualities: "intelligence is sharpness of the mind tool": (77)

He has a razor wit.

(78)

He has a very sharp mind.

(79)

He has a keen mind.

There is even a metaphorical honing of the tool: (80)

The intrigues of court had sharpened her wits.

(81)

He used extensive reading to hone his intellect.

206

Olaf Jakel

Further, "self-control is holding the mind tool": Lack of self-control can be pictured as temporarily letting go of the tool's handle, as in (82). While the total loss of the tool means real insanity (83), more harmless cases may just be in some need of repair (84): (82)

He flies off the handle quite easily.

(83)

Perhaps I was losing my mind.

(84)

He has a screw loose.

But simply possessing an intact mind tool is not enough -it must be put to use: (85)

She could do it if she tried- the trouble is she just doesn't use her mind half the time.

These preliminaries yield the basis for the general structural metaphor: (MIC 4)

"Thinking is working on problem objects with the mind tool"

Within this metaphorical processing of problems, by which the workman hopes to finally produce solutions, a sequence of four phases can be established. Whereas the four subsections of MIC 2 provided four ways of metaphorically construing learning and understanding activities, the nature of problem-solving as investigated in MIC 4 can justly be described as a metaphorical scenario (see Lakoff 1987: 285-286) consisting of four steps in temporal order. We will follow this problem-solving scenario step by step in the next four subsections. 2.4.1 "Thinking starts with applying the mind tool to the problem object" (MIC4a)

The application of the mind tool to a workpiece presupposes a metaphorical conceptualization of attention: "intellectual concentration is turning the mind tool in the object's direction", as in: (86)

Please turn your attention to something more important.

(87)

Have you turned your mind to the question?

(88)

He couldn't bend his mind to his studies.

Then follows the application, the tool is put into position: (MIC 4a) "Thinking starts with applying the mind tool to the problem object"

The metaphorical concept of mind

207

Examples include: (89)

... to apply one's mind to a problem.

(90)

The members of the committee set their wits to work.

(91)

If he would only put his mind to his studies, he could be a brilliant scholar.

Sometimes, the tool gets fixed on the object: (92)

He fixed his mind on the question.

There can also be lack of concentration, when the mind tool is not on the workpiece processed: (93)

Her mind was not on the announcements she was making.

(94)

My mind was on other things.

It is up to the workman to decide if he wants to keep the instrument on the workpiece, as in (95), or rather take it off (96), and set it at rest (97), in order to relax from his mental efforts: (95)

He kept his mind on the problem.

(96)

There is nothing like a good book to take one's mind off one's troubles.

(97)

The children are safe so you can set your mind at rest.

2.4.2 "To solve a problem, the problem container has to be opened" (MIC4b)

Before embarking on the second leg of the metaphorical problem-solving scenario, we must specify the ontological metaphor MIC 1: "problems are solid objects". The material presented in this and the following subsections suggests that problems and their solutions are set in a concrete physical, image-schematic relation to each other: "problems are cqntainers, solutions are hidden inside". Thus, in order to get at the solution contents, the mental workman is faced with the task of opening the problem container: (MIC 4b) "To solve a problem, the problem container has to be opened" Difficulty is conceptualized here in a further specification of the "difficulty is hardness" metaphor introduced above (cf. section 2.1): "the con-

208

Olaf Jakel

tainer puts up some degree of resistance". Hence, "opening the container requires some force", the thinker has to crack a kind of shell: (98)

This one is a hard nut to crack.

As the resistance of a container has to be broken, superficial attempts will not do, as in (99). Instead, intensive thinking demands tenacity and hardness (100), and a difficult problem may even damage the mind tool (101). (99)

So far you have only scratched the surface of the issue.

(100)

I thought long and hard about the problem.

(101)

... mind-bending.

A whole range of instruments can be used in this violent attempt at opening the problem container: teeth (102), a hammer (103), pliers (104), a grinder (105), a spade (106), or a knife (107) are metaphorically applied. (102)

Now there's a theory you can really sink your teeth into.

(103)

I hammered away at the problem all afternoon.

(104)

Now we're dealing with the nuts and bolts of the issue.

(105)

We're still trying to grind out the solution to this equation.

(106)

He dug into the problem.

(107)

He always cuts to the heart of the problem.

Yet despite all efforts, a workpiece may prove too hard to conquer: (108)

I found his explanation impenetrable.

(109)

He got (dead) stuck on the problem.

But in case the thinker succeeds in cracking the problem shell, as in (110), so that he can tear it apart (111), he will have come a good deal closer to the solution: (110)

He has cracked one of the crucial problems.

(111)

He tore the problem apart looking for its solution.

Eventually, he may find that all his drudgery was in vain because there is a more elegant and civilized way of opening the problem container: (112)

He finally found the key to the problem.

The metaphorical concept of mind

209

2.4.3 "The open problem container has to be searched for its solution contents" ( MIC 4c)

At the third stage in our scenario, the real investigations can begin: (MIC 4c) "The open problem container has to be searched for its solution contents" The inside of the container lies open to inspection (113), or access (114); but as the desired contents may be hidden deep down (115), the search may have to go right to the bottom of the problem container (116): (113)

We have to look deeply into this problem for its solution. 15

(114)

Now we're really getting into the problem.

(115)

The problem has a buried solution.

(116)

Now I want to get to the bottom of this problem.

On the background of this idealized cognitive model component MIC 4c, difficulty metaphorically figures as depth of the problem container (117); and since the intellectual task is to reach deep into the problem container, this quality is also transferred to the thinker (118) and his probing tool (119): "intellectual quality is depth of thinking": (117)

There's a deeper issue here.

( 118)

... a deep thinker.

(119)

He had a probing mind.

2.4.4 "Solving a problem is taking the solution contents out of the problem container" ( MIC 4d)

With the problem container successfully opened and the hidden solution contents located, the latter can now be brought out into the open: (MIC 4d) "Solving a problem is taking the solution contents out of the problem container" (120) a. b.

Have you found a solution? !won't tell you-you'll have to find outfor yourself!

This final stage completes the problem-solving scenario: Mter his descent into the depths of the problem container (example 116), the thinker now comes up to light again, bringing with him the solution contents: (121)

He finally came up with a solution.

(122)

The solutionfinally was brought to light.

210

Olaf Jiikel

Facts can be extremely difficult to extract from the container; quite often, · some kind of salvage equipment is needed in this attempt: (123)

We fished out a number of unpleasant facts.

(124)

We dredged up a load of sordid facts about her.

In comparison, conclusions are simply drawn from the container into the open: (125)

Only one conclusion can be drawn from that.

(126)

What did you gather from his statement?

In matters of opinion, different thinkers passing judgment on an idea object will not necessarily take the same item from the container: (127)

He took a different view of it.

Finally, in case the thinker is not content with the results, he can repeat the investigation: (128)

If you are not satisfied with the answer, look into it again.

2.5 Mental craftsmanship Whereas so far ideas were pictured as a kind of natural object, they can also be conceptualized metaphorically as products of a manufacturing process: (129)

He produces new ideas at an astounding rate.

(130)

He was always making grandiose plans to sell or mortgage his house.

Mental activity and its imaginative aspects in producing ideas are more specifically metaphorized as the work of a craftsman or sculptor who brings a kind of idea "slug" to perfection by bringing it into shape: (MIC 5)

"Forming ideas is shaping raw material"

For example: (131)

It's not easy to form an idea of centuries past.

(132)

He has some difficulty in giving shape to his ideas.

The metaphorical concept of mind

211

Part of a craftsman's work is constantly turning the unfinished workpiece: (133)

I have turned the matter over and over in my mind.

The use of turning over is motivated by the "complexity is dimensionality" metaphor introduced above (see section 2.1): the mental craftsman wants to examine a complex object from all sides. The end product is now worked out of the slug (134), sometimes with the help of a hammer (135), or a grinder (136): (134)

I haven't really worked it out in my own mind yet.

(135)

We've got to get together and try to hammer out a solution.

(136)

My brother is grinding away at his studies.

Whereas an attempt at intellectual unravelling takes a metaphor from the realm of forging (137), the kitchen provides the critical image of cakes taken from the oven too early (138): (137)

Let me straighten this out.

(138)

Another of her half-baked schemes!

As a variation on these serious activities, in a craftsman's studio there is also room for playfully toying with idea objects without consequences: (139)

She's been playing with the idea of starting her own business.

(140)

I've been toying with the idea for some time.

2.6 The scales of judgment An essential feature of cognition is the activity of judging on issues and opinions and deciding between contrary arguments and alternative ideas. These abstract thought processes are metaphorized as concrete, manual activities:

(MIC 6)

"Judging diverse arguments is weighing up idea objects"

Examples include: (141)

He weighed the ideas in his mind.

(142)

You have to weigh up in your mind whether to pursue the matter or not.

(143)

You have to weigh the costs of the new system against the benefit it will bring.

212

Olaf Jakel

As was established above (see section 2.1 ), importance is metaphorically understood in terms of weight. In a comparison, the better argument just weighs more. Example (143) shows especially that opposing arguments are properly weighed up against one another. What tips the balance of the decision scale (144) is the amount of importance which the thinker attaches as particular weights (145) to the objects in question: (144)

The scale just tipped in favor of my voting yes.

(145)

I don't attach any weight to these rumours.

Scales do not necessarily appear in the visualization of this process, as the following expression shows: (146)

On the one hand ... , on the other hand ...

This idiomatic phrase places the alternative idea objects to be judged in the open palms of the thinker. To achieve a balanced judgment, one has to weigh the alternatives together (147), carefully balancing them (148): (147)

Taking one thing with another, we have to be content.

(148)

You have to balance the advantages of living in a big city against the disadvantages.

2. 7 Stockkeeping What would cognition be without memory? This must be regarded as one of the most important elements among the inside furnishings of our mental workshop. The folk theory conceptualizes it as yet another imageschematic container, or more specifically: "the memory is a store to keep idea objects in". Examples of this metaphor include: (149)

I've kept this in my memory for years.

(150)

Keep in mind that she is a friend of the director.

Objects are worth keeping because, after such extensive processes of collecting and refining and the subsequent selection by weight, they represent immense values: "ideas and knowledge are valuables". For example: (151)

... treasure something up in one's memory.

The metaphorical concept of mind

213

The linguistic evidence presented in this section substantiates the overall structural metaphor: (MIC 7)

"The valuables of knowledge are kept in the memory store for 'further use"

Within this general conceptualization of memory, the two aspects of memorizing and remembering can clearly be distinguished. Forming a kind of minimal memory scenario, these twin "phases" are treated in two separate subsections. 2. 7.1 "Memorizing is putting the valuables of knowledge into the memory store" ( MIC 7a) The first subcomponent of the memory concept (MIC 7) describes a metaphorical movement towards the memory store: (MIC 7a) "Memorizing is putting the valuables of knowledge into the memory store" Thus, items worth knowing are delivered to the memory store (152) and stored there (153): (152)

You'll need these figures so often that you must commit them to memory.

(153)

He stored the knowledge somewhere in his fearsome memory.

Lasting memories are achieved by means of various storage and fixation techniques: (154)

I fixed the name firmly in my brain.

(155)

He had the car number stamped in his memory.

(156)

The incident was engraved in his memory.

(157)

This terrible event is etched for ever in my memory.

Or, in case no long-term memory is intended, the smart thinker writes a little memo: (158)

When she mentioned her birthday casually he made a mental note of it.

The capacity of the memory container is limited, as expression (159) shows; and all efforts will be pointless with a leaking container (160): (159)

He was cramming for his finals.

(160)

I've got a memory like a sieve!

214

Olaf Jakel

By contrast, in a richly filled store (161), some knowledge items may even get stuck unintentionally (162), or stand out against the bulk of memories (163): (161)

... a head stuffed with facts.

(162)

For some reason, her name did stick in my mind.

(163)

A few things stand out in my memory.

Whereas this sticking out metaphorically marks a high level of consciousness, the subjective stockkeeper deposits subconscious items right at the back (164) and at the bottom (165) of the store; unwanted objects are also pushed into this remotest corner (166): (164)

It was at the back of my mind that I had to phone you, but I completely forgot.

(165)

Deep in his mind he knew he was at fault.

(166)

She's been trying to repress the horrible experience.

2. 7.2 "Remembering is bringing knowledge items back from the memory store to the mental workshop" ( MIC 7b) It seems only logical that, with memorizing activity being described as a

transport of valuables towards the memory container, remembering should figure as the corresponding movement of fetching these items back: (MIC 7b) "Remembering is bringing knowledge items back from the memory store to the mental workshop" The direction of remembering-back to the mental workshop (167)-is as precisely voiced by everyday language, as is the source of the fetching movement-from the memory store (168): (167)

Can't you bring this to mind?

(168)

I quote from memory.

As the store may have been chaotically crammed with knowledge items, the thinker often-has to collect the objects wanted from the depths of the store (169), at times doing a desperate search (170):

(169)

Before you begin to make a speech, you should collect your thoughts and ideas.

(170)

I've searched my memory but can't remember that man's name.

The metaphorical concept of mind

215

Though heavy equipment (171) is not always needed in these salvaging attempts, for some it remains a real fumbling job (172): (171)

... all these old verses that we seem to dredge up from early memories.

(172)

He's not a very good public speaker; he often has to fumble for the right word.

And rounding off the stockkeeping activities, knowledge items that have not been used for long and therefore gathered dust are given a proper polishing with a brush: (173)

If you are going to France you'd better brush up your French.

2.8 Waste disposal Having finished all processes of gathering and refining material as well as storing selected products, the last remaining job to be done is waste disposal. What is there to dispose of? Left in the mental workshop are empty problem containers as well as those idea objects that were discarded in the selection process. As there is no further use for either of them, they are unceremoniously dumped: (MIC 8)

"Abandoning ideas is dumping discarded objects and empty issue containers at the end of the manufacturing process"

A useless idea object must be removed from the mental workshop: (174)

You'd better put the idea of marriage out of your head.

Simply letting go (175) will not always do, as some awkward thought objects can hardly be removed at all (176): (175)

You'll have to let go of your belief in Santa Claus.

(176)

I have made a terrible mistake and now I can't get it out of my mind.

That these attempts at dismissal are conceptualized as a real throwing away can be seen from the concluding evidence: a useless idea is turned down (177) and cast away from the workshop (178); and a thinker, having finished his mental efforts, lets the thought object drop (179): (177)

You'd better drop the idea of beating him.

(178)

Cast all thoughts of freedom from your mind.

(179)

After thinking about it for a while, I let the matter drop.

216

Olaf Jiikel

3. Summary of results The following diagram is meant to provide an overview of the investigated idealized cognitive model and its systematicity. Next, the conceptual metaphors involved in the model are presented in context. In both, the numbering of the model-integrated components (MICs) is maintained. 1.

"Ideas are solid objects" Complexity is dimensionality. Difficulty is hardness. Importance is weight.

2.

"Understanding an idea is establishing physical closeness" Ideas and problems are targets. 2a. "Understanding starts with attempts at finding and hunting idea objects" Source of difficulty: the objects can be hidden, out of reach, and mobile. Mental closeness is spatial closeness/mental distance is spatial distance. Intellectual quality is physical range of action.

~~------~~----_j~--------~--------~ Gathering of material

Processing

Product management

Figure 1. "The mental workshop": Overview of the ICM "Mental activity is manipulation"

The metaphorical concept of mind

217

2b. "Understanding is seizing idea objects firmly" Source of difficulty: the objects can be slippery. Mental control is physical control. Intellectual quality is physical strength and dexterity. Conscious is up: 2c. "Understanding is picking up idea objects from the ground" Intelligence is speed on the uptake. The mind (consciousness) is a container: 2d. "Understanding is taking idea objects into the mind container" Intellectual receptivity is openness and spaciousness of the mind container. The mind is a workshop: 3. "Within the mental workshop, the idea objects are stowed away according to urgency" Importance is centrality: unimportant issues are given peripheral positions. The mind (intellect) is a tool: 4. "Thinking is working on problem objects with the mind tool" Intelligence is sharpness of the mind tool. Self control is holding the mind tool. 4a. "Thinking starts with applying the mind tool to a problem object" Intellectual concentration is turning_ the mind tool into the object's direction. Stopping thinking about a problem is taking the tool off the object. Problems are containers, solutions are hidden inside: 4b. "To solve a problem, the problem container has to be opened" Source of difficulty: the container puts up some degree of resistance. Opening the container requires some force. Mental investigation is physical investigation: 4c. "The open problem container has to be searched for its solution contents" Source of difficulty: the solution contents may be hidden deep down at the bottom of the problem container. Intellectual quality is depth of thinking. 4d. "Solving a problem is taking the solution contents out of the problem container" Source of difficulty: some force may be needed to draw the contents from the container.

218

OlafJiikel

5.

"Forming ideas is shaping raw material" Quality of imagination is artistic creativity.

6.

"Judging diverse arguments is weighing up idea objects" Quality of judgement is being good at balancing.

The memory is a store to keep idea objects in. Ideas and knowledge are valuables: 7. "The valuables of knowledge are kept in the memory store for further use" 7a. "Memorizing is putting the valuables of knowledge into the memory store" Quality of memory is storage capacity. 7b. "Remembering is bringing back knowledge items from the memory store to the mental workshop" 8.

"Abandoning ideas is dumping discarded objects and empty issue containers at the end of the manufacturing process"

To conclude our description of the idealized cognitive model "mental activity is manipulation", we can outline some of the construal options which occur even within the limits of this one model. Apart from the different conceptualizations of understanding provided by the four stages of the model-integrated component MIC 2 (see section 2.2 above), there are quite a few epistemic correspondences concerning some of the key concepts featured in the intellectual domain: importance, difficulty, and intellectual qualities. There are at least two completely different ways of conceptualizing importance: as weight (cf. MIC 1, MIC 2, and MIC 6) or as centrality (cf. MIC 3 and MIC 4). There are even more ways of construing conceptual difficulty by allotting various metaphorical sources to it: idea objects can be multidimensional, hidden, out of reach, mobile, slippery, or hard; the problem container may be hard to open, the solution contents may be hidden deep down at the bottom, and some force may be needed to draw the contents from the container. Consequently, there are also alternatives galore for the conceptualization of intellectual qualities: as a wide physical range of action, physical strength and dexterity, or speed on the uptake; as openness and spaciousness of the mind container, sharpness of the mind tool, or depth of thinking; as artistic creativity, balancing ability, or storage capacity of the memory container.

The metaphorical concept of mind

219

4. Motivation matters In his influential monograph The Concept of Mind, the British philosopher Gilbert Ryle criticizes the "dogma of the Ghost in the Machine" (1949: 15-25) predominant in western culture, and brands some of our most ordinary language expressions as "nothing but a contagion from that dogma" (1949: 35). He even in all seriousness gives this piece of advice: "The phrase 'in the mind' can and should always be dispensed with. Its use habituates its employers to the view that minds are queer 'places', the occupants of which are special-status phantasms". (Ryle 1949: 40) Though we do not need to share this philosopher's prescriptive zeal, we ought to remain sceptical as regards the question whether our conceptualizations of mental activity have any real equivalent outside of our folk theoretical model. As cognitive linguists, we have empirically described the metaphorical concept of mind in a cognitive model and can now add a few words concerning its motivation. Because abstract cognition is not open to direct physical experience (cf. Lakoff-Johnson 1980: 117), cognitive processes are conceptualized in terms of the physical manipulation of concrete objects: "mental activity" becomes the target domain of a complex metaphorical mapping of structures from the source domain of "manipulation". These structures can serve as a possible basis of a conceptual metaphorization because of their image-schematic nature. 16 As a matter of fact, most of the structure of our model of mental activity is supplied by two extremely productive image schemata: the "container" schema and the "manipulation" schema. As the container schema has been sufficiently described by both Johnson (1987: 21-23, 30-40) and Lakoff (1987: 272-273), 17 we can simply recapitulate the ways in which it is utilized in our idealized cognitive model. First, there is the mind, in the sense of consciousness: anything we think of is inside this container, everything else remains outside. What we concentrate on is in its center, while less important matters are only of peripheral interest, and issues of no interest are kicked out completely. A closed mind may protect its proprietor while at the same time shielding him from useful information from the outside. Then there is the memory: as a store it has a mere protective function. Its valuable contents, the possessions of the proprietor, must not be lost.

220

Olaf Jakel

This store is most useful if richly filled. Moreover, as a container within a container it serves for the purpose of selection: not everything that passes through the mind container is memorized and so conveyed into the memory container. And last not least, there is the problem container: protecting its contents from curious intruders, its boundary presents an obstacle to be overcome, a case to be penetrated and torn apart, an outer shell which in its centre embeds the essentials in which we are interested. While the multiple applications of the container schema thus provide much of the basic ontology of our idealized cognitive model, the model owes its structural coherence to the manifold variants of the manipulation schema. Our hand, the most prominent organ in our intentional interaction with the physical environment, can be regarded as the most versatile instrument imaginable. We can use it to grasp at objects, hold them, move them, let them go. The use of both hands doubles the potential of our fine and coarse motor activities: objects can be formed or taken apart as well as weighed against one another. The application of diverse tools eventually multiplies our efficiency in catching, forming and dismantling, enabling almost inexhaustible possibilities of both creativity and destruction. Our description of the folk model in the main section has shown how this enormous spectrum of manipulation variants (see Figure 2) is applied in the metaphorical conceptualization of mental activities (for details see section 2). ' Both the container schema and the manipulation schema more than meet Lakoff's (1987: 278) and Johnson's (1987: 116) demands for image schemata to count as emergent in our experience and function as fertile source domains of conceptual metaphoric mappings.

Figure 2. Richness of the "manipulation" schema (only variants utilized in the idealized cognitive model)

The metaphorical concept of mind

221

Now why is the target domain of mental activity to such a large extent conceptualized on the basis of the source domain of manipulation? The simple answer is: there is an experiential correlation (cf. Lakoff 1987: 278 and Johnson 1987: 116) between mental activity and manipulation. At both levels, there is an intentional gathering of knowledge about the world: through tactile contact with objects as well as through learning and realization. Active investigations are made-by concrete dismantling of solid objects as well as by intellectual analysis. Problem-solving tasks can take the shape of a coconut-useless in its natural state-as well as of a complex piece of arithmetic. This simple answer can be corroborated by relevant evidence from two scientific theories. Piaget's (1975) Genetic Epistemology states for ontogenesis what Evolutionary Epistemology (e. g., Vollmer 1987) confirms for phylogenesis: abstract thinking is an internalized version of manipulative sensorimotor activity. 18 To sum up: rooted so deeply not only in everyday experience but also in the history of mankind as well as of each individual, the conceptual metaphor "mental activity is manipulation" could hardly be better motivated.

5. Alternative models Yes, it's true: mental processes can be talked about in English without recourse to the source domain of manipulation! We will digress briefly on some alternative models of metaphorically construing the intellectual domain. (180)

a. b. c. d. e.

f.

g.

"ideas are self-propelled entities" The idea just came to him. Scotland springs to mind as an example. It never entered his head to help me. The thought had crossed my mind. Suddenly it hit me: my diary had probably been read by everyone in the office. Does anything strike you? I know the rule, but the moment you asked it left my head.

Here ideas figure as ominous beings that come and go on their own resources, sometimes even using some force (180 e, f) to get the thinker's attention. As the subject is reduced to a mere passive role, this model

222

Olaf Jakel

conceptualizes inspiration by lucky coincidence (or by the muses?) rather than mental activity. The model can be partially integrated into our idealized cognitive model within MIC 2a (the idea objects can be mobile). (181) a. b. c. d. e. f. g.

"thinking is a journey" He went over this in his mind. The question of overtime pay proved to be an insurmountable stumbling block to agreement. We reason from premise to conclusion. The evidence pointed in all different directions. Reason leads me to that conclusion. This assumption is what's misleading us. We're going around in circles.

The "journey" metaphor emphasizes the linear nature of purposeful logical reasoning: the straight line counts as the optimal path towards a given goal, obstacles (18lb) and detours (18lf, g) only hold up the thinker. This one-sided orientation towards logic offers no conceptualization of creative thinking or problem solving, memorizing or recall processes. (182) a. b. c. d. e. f. g. h.

"understanding is eating" They gobbled up the ideas. Let me chew on that for a while. That class gave me food for thought. It'll take some time to digest that information. I've been ruminating on that topic for a while. He has a thirst for knowledge. He has an appetite for learning. She has an insatiable curiosity.

Here, not only the container schema is transferred from body to mind, but the complete process of taking in food. One single component (MIC 2d) of our idealized cognitive model is made the sole model of learning and understanding processes. While this model highlights the idea of a natural need to understand (182 f-h), it lacks differentiation to conceptualize mental activities such as problem solving, judging, memorizing, or remembering. (183) a. b. c.

"understanding is seeing" The instructions on the packet weren't very clear. He is very bright. I always thought he was a little dim.

The metaphorical concept of mind

d. e. f. g. h. 1.

j.

223

I see!

He saw through her lies. He wanted to see all sides of the issue. Let's take a closer look at that proposal. It looks different from my point of view. I view it differently. That's an insightful idea.

Our visual perception serves here as a source domain for the conceptualization of understanding processes. Ideas figure as light sources, and intelligence counts as brightness. The fact that this metaphor can be integrated into our model at several stages is hardly surprising if we recall that our concrete source domain of manipulation is not cut off from vision either. The amazing efficiency of our manipulation activities in part results from our ability to watch and thereby coordinate the performance of our hands. The hunting of idea objects (MIC 2a) as well as the search of the problem container (MIC 4c) followed by the bringing to light of its solution contents (MIC 4d) are components of our model hardly conceivable without the aid of visual perception. The view that real thinking is a kind of vision takes after Plato's theory of ideas. Accordingly, this model if applied exclusively condemns the thinker to passivity again. This conceptualization seems to have its eye on the rather contemplative aspects of the mind. Nonetheless, this model must be regarded as the strongest alternative to our idealized cognitive model. Therefore, we will come back to it once more in the final section.

6. Etymological evidence Looking for ways of expressing mental processes beyond our model, we encountered some alternative conceptual metaphors. But what about nonmetaphorical descriptions of the mental domain? Quite a number of expressions which at first sight might appear as candidates in this enterprise turn out ditfetently if inspected through the etymological telescope. The "genealogy" of some of the supposedly most abstract terms from the mental domain reveals their origin in the most concrete of image schemata. As in the synchronic data, more than anything else we find variants of the manipulation schema, which obviously was productive already in

224

Olaf Jakel

those ancient languages. Thus, a simple list can be arranged according to the components of our idealized cognitive model. 19 "Understanding is establishing physical closeness": understand < AS understandan, literally: 'to stand among'; cf. MLG understan 'step under' "Aiming at problem targets": guess: 'take aim' < base of get

"Seizing problem objects": get < IE base *ghed- 'seize' forget < WG *fer+ *getan 'miss or Jose one's hold' apprehend, apprehension < L apprehendere: ad 'to, towards, in the direction of' + prehendere 'seize' comprehend, comprehension < L comprehendere: cum + prehendere 'seize' "Picking up I taking": intellect, intelligence < L intellegere 'choose among': inter + Iegere 'pick up, gather' perceive, perception < L percipere 'seize, obtain, collect': per + capere 'take' conceive, concept < L concipere 'take to oneself': cum + capere 'take' anticipate, anticipation: 'take up beforehand' < L anticipare: ante + capere 'take' "Applying the mind tool": intend, intention < L intendere: in 'into, against, toward's'

+

tendere

'stretch' "Opening the problem container": discover: 'un-cover' < L discooperire: dis + cooperire 'cover' reveal < L reve/are 'unveil': re + velum 'veil' "Searching for the solution": research: 'intensive searching', re 'again'

+ search

"Drawing contents from container": deduce, deduction < L deducere: de 'away from, off'

+ ducere 'lead, draw'

"Weighing I measuring": ponder < L ponderarelpendere 'weigh' meditate < L meditari 'measure' "Bringing back": recollect, recollection < L recol/igere: re 'again'

+

col/igere 'collect'

The metaphorical concept of mind

"Dumping discarded objects": reject < L reicere 'throw back, discard': re

225

+ jacere 'throw'

The alternative model "understanding is seeing" also has a share in this metaphorical ancestry: idea < Gr idea 'look, form' < *id- 'see' wit, wise < L videre 'see'

This list of "dead" metaphors from the domain of mental activity confirms Elisabeth Traugott's (1985) claim that diachronic evidence ought to be included by a cognitive theory of metaphor that focuses on image schemata. Moreover, our etymological investigation corroborates Eve Sweetser's (1990: 23) hypothesis that, in general, "the historical and synchronic data point to one and the same cognitively based analysis of the relevant domain". And finally, we have dug out some weighty etymological evidence for the linguistic productivity and the explanatory value of the idealized cognitive model "mental activity is manipulation".

7. Conclusion: Why "manipulation" beats everything Paying tribute to the conceptual metaphor "understanding is seeing", Mark Johnson (1987: 109) makes the following claim: "There are, of course, other experiential bases for knowledge metaphors (such as touching, hearing, and tasting), but none of these is as dominant as vision" (my italics). And he still (cf. Johnson 1992: 353-355) seems to be slightly preoccupied with "vision" as the privileged source domain for the conceptualization of the mind. To refute this unwarranted claim, we can draw up a set of arguments which result from our empirical investigation. This may also serve as a summary of some noteworthy conclusions. i. Although quantification is neither possible nor useful, the language data presented in section 2 show that the "manipulation" model accounts for far more everyday language expressions than all the alternatives taken together. n. In contrast to the other models (cf. section 5), "manipulation" does not focus on just one aspect of cognition. As has been shown (see section 2), the model elaborates the whole realm of mental activities in great detail and differentiation. iii. If there is coherence between the alternative metaphors (see Lakoff-Johnson 1980: 86) for the target domain of mental activity,

226

Olaf Jakel

this is to a large part due to the idealized cognitive model "mental activity is manipulation". This model can be seen as a kind of coordinating instance between the different source domains (cf. section 5). IV. There is also systematic coherence across target domains (cf. Lakoff-Johnson 1980: 96). Thus with its component (MIC 2d) "understanding is taking idea objects into the mind container", our model connects smoothly to Reddy's (1979) interpersonal communication "conduit". In fact, it can be viewed as the intrasubjective continuation of that conduit. 20 v. As has been shown above (see section 4), the manipulation model is highly motivated. Vl. Compared with Johnson's favourite "vision", "manipulation" seems to have a deeper experiential grounding, as we find some unidirectional dependence between the two concepts (cf. Sweetser 1990: 45). So, while vision can function as source domain for understanding, it remains at the same time a target domain dependent on the conceptualization on the basis of the source domain manipulation: "seeing is touching", or "vision is manipulation" (see Lakoff 1987: 437 and Johnson 1987: 108). Vll. As has been shown (cf. section 6), the diachronic data mass~ely support the manipulation metaphor, which in its turn offers a systematic account of historical language change. Vlll. As a final point, we can only hint at the fact that thete is a so cross-linguistic evidence for our idealized cognitive model. In both German and Italian, basically the same conceptual model could be established with only minor differences in linguistic detail. Tentative investigations (see Jakell990) of Danish, Hebrew, Indonesian, Japanese, Malayalam, Swahili, Tamil, and Turkish have revealed some general tendencies: in almost all of these languages, thinkers grasp at idea objects, the mind is conceptualized as a container, information is picked up, problems are worked on, alternatives are weighed up, and fixation techniqu.es are used for memory storage. To sum up the argument: the conceptualization of mental activity in terms of manipulation seems to be at least a cross-cultural strategy, if not a cognitive universal. As regards the English language, no other conceptualization of the intellectual domain even comes close to the metaphorical concept of mind as construed by the idealized cognitive model "mental activity is manipulation".

The metaphorical concept of mind

227

Notes 1. I thank Giinter Radden for in-depth discussions of a previous version of this paper. All remaining flaws, of course, are mine. I am also grateful to John Taylor for checking my English. 2. See Lakoff-Johnson 1980; Johnson 1987; and Lakoff 1987, 1990. 3. This has been done in Jakell990, and in much more detail than can be presented here. 4. Lakoff (1987: 121 ): "Each theory, whether folk or expert, involves some idealized cognitive model with a corresponding vocabulary." For explanations of the technical terms "idealized cognitive model" (ICM) and "folk theory", see Lakoff 1987, chapters 4 and 8, respectively. 5. Jakell990 comprises more than 700 example sentences taken chiefly from the everyday language corpora of two contemporary English dictionaries: Longman's dictionary of contemporary English (Summers- Rundell 1987) and Collins cobuild English language dictionary (Sinclair 1987). A number of examples were also taken from the Master metaphor list (MML) compiled by the Berkeley group: George Lakoff, Jane Espenson and Adele Goldberg (Lakoff- Espenson- Goldberg 1989). 6. This is one reason why a comparison of my findings with those of D'Andrade (1987) would go beyond the scope of this paper. In an anthropological approach, D' Andrade treats the same conceptual domain, and derives "a folk model of the mind" from narrative interviews with American informants. Though his model neither covers memory nor problem-solving activities while including emotions and desires, there seems to be some convergence in our results concerning such issues as perception and controlled thinking. There is even more apparent concurrence with the conceptual metaphors reconstructed by Anne Salmond (1982) from examples of theoretical texts in anthropology. Though she focuses on the metaphorizations of theoretical knowledge and expert argument rather than on intrasubjective thought processes, the conceptual linkage between these aspects of intellectual activity is obviously very strong. For a treatment, though not linguistic, of some broad metaphorical underpinnings of various psychological expert theories of mind see Sternberg (1990). Roediger (1980) offers a comprehensive survey of psychological expert metaphors of mind, focusing on memory. 7. Brugman (1990: 259) regards it at least as "logically possible for there to be domains of knowledge which are structured entirely by one or more metaphorical mappings". Mental activity is exactly such a domain. Cf. also Lakoff- Johnson (1980: 177). 8. Here "ideas" stand for "problems" and "issues" as well. In the following I will speak of "idea objects" unless further specification is considered appropriate (e. g., "problem objects" in section 2.4). 9. Here is a point of contact with the alternative model "ideas are self-propelled entities" (cf. section 5 below). 10. Examples ( 16) to (18) come from the Master metaphor list (Lakoff- Espenson- Goldberg 1989: 81), where unfortunately they are listed bare of context. II. See below, section 2.4: solutions may be hidden somewhere in the depths of problem containers. 12. Here "understanding" is meant to represent "perceiving", "realizing", "learning". All these mental activities share the metaphorical concept of a lifting up of idea objects onto the level of consciousness. The folk theory is liberal in this respect. 13. As the language data presented in this paper show, the term mind is used with a variety of meanings in everyday English. In the metaphorical ICM (idealized cognitive model),

228

14. 15.

16. 17. 18. 19.

20.

Olaf Jiikel at least the conceptual fields of consciousness (MIC 2d) and intellect (MIC 4) can be distinguished (cf. section 2.4 below). Compare note 13. This is a good example for the coherence between the idealized cognitive model "mental activity is manipulation" and the alternative metaphor "understanding is seeing". See sections 5 and 7 below. For a definition of "image schemata", see Johnson (1987: XIV passim) and Lakoff (1987: 276 passim). For a critical discussion of the "container" schema, see Jakel (1990: 46-49, to appear). For a more detailed account of these issues, see Jakel (1990, to appear). I draw upon the etymologies of Onions (1985) and Weekley (1967). The following symbols and abbreviations are used: < : formed from; * : hypothetical form; AS: AngloSaxon; Gr: Greek; IE: Indo-European; L: Latin; MLG: Middle Low German; WG: West Germanic. Cf. MIC 4, where it may also be word containers that are opened in search of meaning contents. Thus, it is no accident that there should be some overlaps with Reddy's (1979: 311-324) language data.

References Brugman, Claudia 1990 "What is the invariance hypothesis?", Cognitive Linguistics 1(2): 257-266. D' Andrade, Roy 1987 "A folk model of the mind", in: Dorothy Holland- Naomi Quinn (eds.), Cultural models in language and thought. Cambridge: Cambridge University Press, 112-148. Jakel, Olaf 1989 "'Der handgreifliche Intellekt': Zur Metaphorik geistiger Tiitigkeiten", Grazer Linguistische Studien 32: 5-19. 'Der handgreifliche Intellekt', oder "mental activity is manipulatmn". Kog1990 nitiv-linguistische Untersuchung einer grundlegenden konzeptuellen Metapher der englischen Alltagssprache. [Thesis (Staatsarbeit), published in the paper series of the Graduiertenkolleg Kognitionswissenschaft, University of Hamburg.] 'Economic growth' versus 'pushing up the GNP': Metaphors of quantity from 1993 the economic domain. Duisburg: Linguistic Agency University Duisburg (L.A.U.D.). to appear Konzeptuelle Metaphern und die schematische Fundierung kognitiver Madelle. [Unpublished manuscript, University of Hamburg.] Johnson, Mark 1987 The body in the mind: The bodily basis of meaning, imagination, and reason. Chicago- London: University of Chicago Press. 1992 "Philosophical implications of cognitive semantics", Cognitive Linguistics 3(4): 345-366. Lakoff, George 1987 Women, fire, and dangerous things: What categories reveal about the mind. Chicago- London: University of Chicago Press. 1990 "The invariance hypothesis: Is abstract reason based on image-schemas?", Cognitive Linguistics 1(1): 39-74.

The metaphorical concept of mind

229

Lakoff, George- Jane Espenson- Adele Goldberg 1989 Master metaphor list ( MML). [Draft copy, University of California, Berkeley.] Lakoff, George- Mark Johnson 1980 Metaphors we live by. Chicago- London: University of Chicago Press. Lakoff, George- Mark Turner 1989 More than cool reason: A field guide to poetic metaphor. Chicago- London: University of Chicago Press. Onions, C. T. (ed.) 1985 The Oxford dictionary of English etymology. (12th edition.) Oxford: Oxford University Press. Piaget, Jean 1975 Die Entwicklung des Erkennens 1: Das mathematische Denken. Stuttgart: Ernst Klett. Reddy, Michael J. 1979 "The conduit metaphor-a case of frame conflict in our language about language", in: Andrew Ortony (ed.), Metaphor and thought. Cambridge: Cambridge University Press, 284-324. Roediger, Henry L. 1980 "Memory metaphors in cognitive psychology", Memory and Cognition 8(3): 231-246. Ryle, Gilbert 1949 [1990] The concept of mind. Harmondsworth: Penguin. Salmond, Anne 1982 "Theoretical landscapes: On cross cultural conceptions of knowledge", in: David Parkin (ed.), Semantic Anthropology. London: Harcourt Brace Jovanovich, 65-87. Sinclair, John M. (ed.) 1987 Collins co build English language dictionary ( ELD). London- Glasgow- Stuttgart: Collins. Sternberg, Robert J. 1990 Metaphors of mind: Conceptions of the nature of intelligence. Cambridge: Cambridge University Press. Summers, Della- Michael Rundell (eds.) 1987 Longman dictionary of contemporary English ( DCE). Berlin- Munich: Longman- Langenscheidt. Sweetser, Eve E. 1990 From etymology to pragmatics: Metaphorical and cultural aspects of semantic structure. Cambridge: Cambridge University Press. Traugott, Elisabeth Closs 1985 "'Conventional' and 'dead' metaphors revisited", in: Wolf Paprotte- Rene Dirven (eds.), The ubiquity of metaphor: Metaphor in language and thought. Amsterdam- Philadelphia: Benjamins, 17-56. Vollmer, Gerhard 1987 Evolutioniire Erkenntnistheorie. (4th edition.) Stuttgart: S. Hirzel. Weekley, Ernest 1967 An etymological dictionary of Modern English. New York: Dover.

Vantage theory Robert E. MacLaury

1. Introduction Much of abstract thinking proceeds by analogy to physical space. Moreover, we commonly construct our thoughts in terms of figure-ground relations, even though these do not inhere in the world per se; rather, we project them from a standpoint that we may actually occupy or solely imagine. Vantage theory is an attempt to tie together diverse aspects of such spatial reasoning, which is not an original program. However, this theory holds that the analogy between space and thought is formed at the specific level of coordinates by which we construct a point of view. In the physical realm, coordinates consist of perpendicular dimensions and time as a function of relative motion whereas, in the mind, the analogous coordinates consist of specific sensations and cognitions. While some of the mental coordinates are treated as constants, others are held in high relief, allowed to move or change, or otherwise are manipulated in relation to a fixed entity. Relating fixed and mobile points of reference produces an analogous vantage point. In this way, we are directly engaged with our concepts and categories so as to be capable of creating, revising, and dissolving them. We know the world through these constructed perspectives. When we speak, we name and discuss these points of view rather than the world by itself detached from an observer. A discussion of color categorization will illustrate how vantage theory was devised to account for regularities in one well-studied and measurable domain. Then its performance regarding color is compared with that of the prototype model. Finally, I use the theory to address questions in linguistic semantics and cognitive psychology.

2. Color categories Many languages throughout the world name specific color categories with two terms whose best examples or "foci" are at opposite sides but whose ranges coextend to include each other's focus; while their area of

232

Robert E. MacLaury

overlap covers at least 75 percent of all colors named by both, each range exclusively names a negligible remainder of colors on the margin outside its focus. The terms name one category from opposite emphases or slants. In 1979, as I surveyed color categorization in 100-plus Mesoamerican languages with aid of Munsell chips, this semantic relation appeared so commonly that it prompted me to suspect that people were constructing and naming color categories as points of view rather than categorizing color impartially. The Munsell instrument provided an advantage, because it allowed detailed, quantifiable measurement of categories and their semantics by three independent procedures, and it enabled categories of the color domain to be studied as an interlocking system. The measurements yielded further observations of coextensive ranges that suggested how the alleged viewpoints might differ within each pair; the detail was sufficient to allow statistical tests. For example, one range of a pair is characteristically broader and more centrally focused while the other is slightly restricted and skewed toward a more marginal focus. A full account of coextensivity in Mesoamerican color semantics appears in MacLaury (1986, 1987b, 1992, in press), where I develop the supporting numbers and argue that color categories are personal vantages constructed by analogy to spatial viewpoints. My purpose here is only to illustrate the argumentation without introducing yet again the complete evidence, which is necessarily massive, quantitative, and already published. My sole example, collected in South Africa, illustrates descriptive techniques as well. 1

2.1 Description The Munsell system used in fieldwork includes 330 chips, ten that range from white through eight greys to black and 320 covering the rainbow hues saturated to the greatest vividness attainable by pigment technology and ranging in brightness from the palest pastels through intense middleband red, yellow, green, and blue to the darkest maroons, browns, forest greens, indigos, and deep purples. A dim paper substitute appears in Berlin-Kay (1991), Kay-Berlin-Merrifield (1991a), and MacLaury (in press), although the aesthetic impact of the Munsell colors cannot be conveyed without experiencing the real chips. All we have to go by here is Figure la. An English speaker named the 330 chips one by one in random order, then focused each of his eleven terms on an array of the

233

Vantage theory

same chips by picking out the "best example" of each. The figure shows his foci and his 330 derandomized naming responses in the format of the array. The field equipment consists of the 330 loose chips in fixed random order and two arrays, one of the order shown in Figure 1 and another with red-pink at center. As color composes a circular band, any break is arbitrary. The equipment and the rationale behind the Munsell system is explained under color in major dictionaries and encyclopedias and in u

if!!l!!!':t 1 2 3 4

s s

11111111112222222222:S:S333333334 1 8 9 o 1 2 :s 4 s 6 7 e 11 o 1 2 :s 4 s 6 1 e 11 o 1 2 3 4 s & 7 8 9 o

b

hlaza kosazana mhlophe (±) ~

~ mpofu ~ tsagatso lllllJ bomvu

mdaka oranj aubu

Iii inamono El buvende I=(±) mnyame

ra 0 t() • e foci

Figure I. Color terms and foci volunteered in response to 330 Munsell chips: (a) English, USA, male, age 35, 1980; (b-e) Zulu, Escort, South Africa, male, 53, 1991

234

Robert E. MacLaury

D'Andrade- Egan (1984) and MacLaury (1987a: 112-116); the latter relates row letters and column numbers to technical Munsell specifications of pigments. On the array and in the figures, white-grey-black are at the left, lettered rows represent brightness levels, and numbered columns represent increments of hue. Each intersection represents a chip. Figure 1 allows a comparison between the ways that individual speakers of English and Zulu have responded to the chips and procedures, which were identical for both. The Zulu data give a misimpression of scrambled disorder. But this is because the Zulu speaker constructs and names color categories differently than his English speaking counterpart. For example, Figure 1c extracts a Zulu category named with two terms, mpofu and tsagatso, which seem to be near-synonyms; their ranges are almost the same and they are focused close together at B10 and B13. (ffi means the focus of mpofu occurred on a color named with that term in the independent chip-naming procedure; 0 means the focus of tsagatso did not coincide with its name on chip B 13; noncoincidence of focus and name is common when color terms share a range.) Many Zulu speakers apply mpofu to a category equivalent to English yellow, but some do not. This Zulu applies both mpofu and tsagatso to a category of mainly high brightness, but also bluish purple (H31, 33). His categories recognize both brightness and hue while, further, some involve internal semantic relations. This is why his system appears cluttered in comparison to the simple English hue system of one term per category. Figure 2 displays a coextensive relation between hlaza and kosazana, which both name the "cool" category of green-with-blue. These data demonstrate how the argument of vantage points has been built. As a preliminary, there is an absolutely pure green at Fl7 and an absolutely pure blue at F29, technically called unique green and unique blue; their purity and spectral locations are determined physiologically (cf. MacLaury 1987 a: 112-116). They are useful here as guides to specifying the center and margin of the cool category: the center is between F17 and F29, the margins are to the left of F17 or the right of F29, and very light or dark colors outside of rows D, E, F and G are also somewhat marginal. Figures 2a-c extract the ranges of hlaza and kosazana from the system and from each other. Hlaza is focused in blue at G28, kosazana at Cl7 in very light green; as an afterthought, the speaker gave kosazana a second focus at F20. Figures 2d-e show the result of a third procedure called "mapping", which is different than chip-naming or focusing. The speaker was given

Vantage theory

235

a handful of ordinary rice and asked to place one grain on each color of the array that he could name hlaza. When he stopped mapping, he was asked to put rice on more of hlaza. This prompting went through four steps until he insisted he had covered the range. In Figure 2d, the mapping steps are numbered and represented by different shades. Then he mapped other categories for a sufficient time to make him forget his treatment of hlaza until we took up kosazana, whose three-step mapping appears in Figure 2e. This mapping falls short of the first focus choice at C17; its first step does not cover green but is confined to blue where it fails to match either focus of kosazana. Figures 2f-g show outlines of the naming ranges of hlaza and kosazana, representing with "k" and "h" postposed uses of each term to qualify the other. 2

11111111112222222222:S:S:S:S:S:S:S:S:S:S+ 1234S678901234S6789012:S+S678!101234S67890

b

11 1 111 11112 2 2 2 2 22 22 2 :s:s 3 3 :s:s :s :s :s :s + 1234S678!101234S678901234S678901234S67890

F G H I

'-------L---t....:...:..::.::.::.::.=:..:=.J

Figure 2. Zulu names of the cool category: ranges and foci

236 B

c D

Robert E. MacLaury

d

:I hlaza I

E F G

H

I II 1 1 1 1 1 1 1 1 2 222222 222 33 3 3 3 3 3 3 3 3 4 1234S6?8901234S6?8901234S6?8901234S6?890

8~--~------------~----~--------~~------~~------~~

c D

Ikosazana I

e

E F G

H

I

~------------------~--------~----I I I I I I I I I I 2 222222 2 2 2 3 3 3 3 3 3 3 3 3 3 4 I 2 3 4 S 6? 8 9 0 I 2 3 4 s 6? 8 9 0 1234S6 ? 8 9 0 1 2 3 4 S 6 ? 8 9 0

B

c D

k

:I hlaza 1:

E F

=hlaza-kosazana

f

0

I I,

G

H I I I I I I I I I I I 2 2 2 2 2 2 2 22233333333334 I 2 3 4 S 6? 89 0 I 2 3 4 s 6 ? 8 9 0 I 2 3 4 s 6 ? 8 9 0 1 2 3 4 S 6 ? 8 9 0

er-======~------~~~~--~~~~~~~~ g

c

Ikosazana I

D

E F G H

I

~------------------~~----~~--~--~------~~--~

Figure 2. (continued). Zulu color terms: (d---e) mappings; (f-g) naming-range outlines with qualifiers

Vantage theory

237

2.2 Theory

After data from 900 Mesoamerica interviews were collected and organized as demonstrated above with a Zulu case, various patterns emerged. For example, many languages named a "cool" category and a few named a "warm" category, which combines red with yellow; but every individual who maintained a warm category also maintained a cool category such that presence of the former predicted the latter. (Berlin- Berlin [1975] and Kay [1975] observed the same in languages elsewhere, which led them to refine their seven-stage sequence of universal color-category evolution.) Both warm and cool categories showed various phases of division within individual systems. The division process was manifest in the marginalizing of foci and first mapping steps and in the skewing of naming ranges. By comparing categories from many individuals, I constructed an overview of regular change. When a category is named with two terms, it involves a shift in semantic relations which develop from near-synonymy to coextension. Further along in the division, the smallest coextensive range retracts toward its focus leaving only the largest range to cover the entire category; the retraction produces a relation of inclusion, which is well enough known to semanticists that it need not be exemplified here. Finally, the largest range too retracts toward its focus, leaving two separately named categories that may only overlap at their edges; this is a relation of complementation, which is also well known. Data substantiating the sequential development of near-synonymy, coextension, inclusion, and complementation appear in MacLaury (1986, 1987b, 1992, especially in press). The four semantic types have been observed elsewhere and independently among individual speakers of Binumarien in Papua New Guinea: ... if two categories have the same range but differ in focus they are regarded as the same category. Thus if focal GREEN covers GREEN and BLUE and focal BLUE COVers BLUE and GREEN, they are both BLUE-GREEN or in Berlin's term GRUE .... Similarly, if focal BLUE covers BLUE, GREEN, and BLACK, and focal BLACK COVers BLACK, GREEN, and BLUE they are both BLACK or the "darkcool" category [i.e., coextension] .... An examination of individual protocols reveals ... steps in the transition of one phase to the next ... one group ... had a stable focus ... but not bounded GRUE, i.e., the GRUE category matches the dark-cool or BLACK category in range [coextension]. (Informant 1 lacks even a GRUE focus [near-synonymy]). A second group ... has a bounded GRUE, which is subordinate to a large BLACK category [inclusion]. A third group ... has a bounded GRUE which contrasts with a restricted or

238

Robert E. MacLaury

"shrunken" BLACK [complementation] ... a more advanced system results from a redefinition of the taxonomic relations between old categories rather than the ... sudden introduction of a new category. (Hage- Hawkes 1975: 297-299)

Finally, for physiological reasons (considered in MacLaury 1987a, 1992, in press), green and blue look more alike than red and yellow (which perhaps is why green and blue are not both used in traffic signals and rarely combined in national flags). This difference is likely to be part of the reason that warm universally divides before cool. However, for such a sensory difference to influence change, people must pick it out from among other sensations. To account for (1) the universal order of warm dividing before cool, (2) the general process of category division, (3) continuous shift in semantic types, and (4) skewing, polarization, and marginalization of responses to color chips, I have proposed that similarity and distinctiveness are subject to selective emphasis. All people recognize both, but some attend strongly to similarity and weakly to distinctiveness, some do the reverse, and some strike an equal balance. Shift in the balance drives the changes (1) to (4). It is important that the recognitions of similarity and distinctiveness are reciprocal and relative, like hot and cold or heavy and light; emphasis on distinctiveness requires deemphasis of similarity. Both cannot be strongly or weakly attended to at once by the same person. Probably people favor the analytical view as their social and physical environs become complex, varied, and novel. A world that differs from one day to the next would instigate people to routinely focus on detail as a strategy of coping; they would generally shift their basic level of categorization toward greater specificity and differentiation, not only in specialized domains or areas of economic interest but across all domains, color included. Kay (1977) notes the proliferation of vocabulary in world languages. Douglas (1966: 77) proposes that "differentiation" is the pervasive engine of societal change from primitive to modern: The right basis for comparison is to insist on the unity of human experience and at the same time to insist on its variety, on the differences that make comparison worth while. The only way to do this is to recognize the nature of historical progress and the nature of the primitive and modern society. Progress means differentiation. Thus primitive means undifferentiated; modern means differentiated. Advance in technology involves differentiation in every sphere, in techniques and materials, in productive and political roles.

Rivers, a pioneer in color ethnography, noted the same of color-category evolution, but in correlation with environmental demand. He found that

Vantage theory

239

the Toda of Southern India named broad brightness categories rather than singling out specific hues, and he added: They lead a simple existence, their lives being devoted chiefly to the care of their buffaloes, and during a long period of isolation on the Nilgiris they have had no keen struggle for existence. At one time they seem to have lived largely on roots and berries, but this source of food-supply has for a long time been replaced by grain and rice procured from other tribes; and even when they had to seek out their food it was probably easy to find. They have had neither the arduous search for food nor the necessity of being continually on guard against enemies, human and animal, which have done so much to develop the powers of observation possessed by such races as the Australians and American Indians. (Rivers 1905: 321-322)

In a less hospitable setting, Rivers notes this: My previous experience of very defective colour nomenclature has been derived from races inhabiting the tropics and it seemed somewhat unnatural to find a far more highly developed language for colours in the inhabitants of a subarctic country such as Labrador. The Eskimo, however, told me that in the autumn they could see all the colours that I had shown them in the hills and it is possible that when colour is only a transient occurrence in the year's experience, it may excite more attention and therefore receive more definite nomenclature than in those parts of the world where luxuriance of colour is so familiar that it receives little notice. So far as I can gather from reading accounts of Eskimo life, colour does not appear to be largely used in the dress or decorations of these people .. . those whom I examined were working with beads of various colours ... . (Rivers 1902: 149)

The environmental challenges and extremes of climate seem to be a more convincing explanation of Eskimo color categorization than Rivers's localized account, because it would apply globally to people in harsh surroundings. The Labrador Eskimo exploit diverse ecological niches during the months of sunlight. People form very broad color categories both in small societies under benign living conditions and in complex societies that, nevertheless, change very slowly, such as that of Ancient Egypt in which four basic color categories persisted through 3500 years without addition of a fifth (Baines 1985: 283). Probably the enhancement of social and material demand had been building through the millennia by increments imperceptible during any individual's lifetime, but it has become noticeable in recent centuries. Now the build up of demand for analytical thinking has expounded manyfold through the main corridors of the world and is reach-

240

Robert E. MacLaury

ing remote, formerly isolated people, who will also add vocabulary and divide broad categories into narrow ones, whether they pertain to color or to anything else. On top of the above formulations, I proposed that the various sensations and cognitions-such as the unique hues and the emphases on similarity and on distinctiveness-are recruited as the fixed and mobile coordinates by which people construct a color category by analogy to the way they keep their balance or plot their positions in a spatial terrain. In physical space, we have the three perpendiculars of up-down, in-out, and right-left collapsible into a single fixed reference body and we have the mobile coordinate of time as a function of relative motion. Thus, in the classic example, the trajectory of a rock dropped from a moving railway carriage is straight when viewed from the train but parabolic when seen from the ground. The experience shifts with change of the mobile coordinate that constitutes a viewpoint. Likewise a color category changes size and shape when constructed from different coordinate arrangements. When construed primarily in reference to one unique hue and attention to similarity, it will be broad, evenly organized, and centrally focused. Attention to similarity will encourage a focus that represents all colors of the category as equally as possible, and it will contract psychological distance between colors such that many will fit within the purview of the vantage. Conversely, when a color category is constructed mainly in reference to another unique hue and attention to difference, it will be narrow, skewed, and marginally focused. Attention to distinctiveness will encourage contrast of one side of the category against the other and it will protract psychological distance between colors such that fewer will fit within the purview. This prediction can be matched numerically against any sample of coextensively named color categories collected with the 330 Munsell chips. For each term of a coextensive pair, an analyst can count: (a) the size of the naming range; (b) the centrality of focus; (c) the size of the mapping; and (d) the average size of mapping steps. In most cases, one term will show higher numbers on all four counts, even though (a), (b) and (c) are independently elicited by chip-naming, focus-selection, and mapping. Count (d) is thrown in to allow for the null possibility of a two-for-two mixture among high and low numbers. This measure is admissible, because the average size of mapping steps (d) is empirically shown to be least predictable and, in many real cases, unrelated to the size of a complete mapping; many small mappings are executed in a single step, many

Vantage theory

241

large mappings in smaller but multiple steps (MacLaury in press, especially on Tzotzil). For convenience, the coextensive range showing correlation among high numbers is called the "dominant" range, that showing low numbers the "recessive" range. Their predictable relation is called the "dominant-recessive pattern of coextension". It is as diagnostic of the coextensive relation as are the overlapping ranges themselves. The first test of this prediction yielded a significant correspondence of 81 percent (p < .01, chi-square) (MacLaury 1987b: Table 2; in press: Table 6.1; cf. 1992: 141-144). The test was performed upon warm categories of the Mesoamerican sample. Since red and yellow appear markedly distinct, the characteristics of each coextensive term tend to be associated consistently with that term in this sensory environment. However, since green and blue look more alike, some of their characteristics tend to cross over between terms; for example, both terms may be focused on one side of the cool category or one term may be focused on both sides or a first mapping step and a focus will be placed on opposite sides (e. g. Figure 2d). Yet coextensively named Mesoamerican cool categories yield the same statistic, p < · 01 (MacLaury in press: table 11.2). Tests were performed across all coextensively named warm and cool categories of the World Color Survey sample (MacLaury to appear [a]; Kay- Berlin- Merrifield 1991 b); these tests promised to determine whether the dominant-recessive pattern occurred worldwide. Although the World Color Survey data include only naming ranges and foci and exclude mappings, they are adequate to assess the predicted pattern of the most central focus corresponding with the most abundantly named range and the least central focus corresponding with the range named least. Applying strict criteria among the World Color Survey data to identify coextension and to measure centrality, 42 speakers of 23 languages coextensively name the warm category with a positive versus negative result of 32:12 (p < .01, chi-square) while 229 speakers of 45 languages coextensively name the cool category with a result of 155:106 (p < .01 ). The pattern significantly recurs over the four continents where the languages are dispersed. A critical quality of both the Mesoamerican and World Color Survey data from warm categories is that some individuals construct the dominant range in reference to red and others in reference to yellow. Nothing inherent in these sensations predetermines the. dominant-recessive pattern. Moreover, in various communities, speakers of the same language who had known each other all their lives showed opposite patterns

242

Robert E. MacLaury

of red or yellow dominance. The pattern will be practiced even if it is not prescribed by culture or learned socially. The same quality everywhere pertains to data from cool categories. These facts leave explanation to individual cognition. The dominant-recessive pattern results from the method by which each person privately constructs a category. Although culture might prescribe what categories to construct, it does not prescribe the method. Each person privately forms an analogy between fixed and mobile coordinates of space and substitute coordinates of mind to construct a category as though it were a point of view. Implications of this model are reviewed below. 2.3 An example

The Zulu case presented in Figures 1 and 2 affords an example of the dominant-recessive pattern, even though data pertain to a cool category and show minor signs of crossover. Table 1 summarizes the main characteristics. Table 1. Coextensive ranges of a Zulu cool category

Characteristics

Dominant range

Recessive range

Name (a) Size of naming range (b) Centrality of focus (c) Size of mapping (d) Average size of mappings steps

hlaza 62 chips most (028) 100 25

kosazana 56 chips least (Cl7) 71

24

All four counts show that hlaza names the dominant range, kosazana the recessive range. Characteristics (a) and (c) can be confirmed easily from Figures 2b to 2e. Centrality of focus (c) is judged on the basis of first choices, which are probably more intuitive than the second choice at F20. The first and only choice for hlaza at G28 is one column to the inside of unique blue (F29), whereas the first choice for kosazana at C17 is on the column of unique green (F17) and on the very light margin outside of middle brightness rows D to G. The second choice will be discussed below. The average size of mapping steps (d) barely favors the dominant range; this characteristic behaves in accord with the widely observed tendency of (d) to show least reliability. The average size of steps by which hlaza was mapped is reduced by the smallness of steps 3 and 4, Figure 2d.

Vantage theory

243

Other characteristics coincide with the dominant-recessive pattern. Dominant hlaza is the older term, shared throughout Zulu-speaking regions. But this speaker's use of kosazana is the only instance that emerged among 40 interviews. Dominance matches commonality whereas recessiveness coincides with rarity. The dominant naming range is darker and the recessive naming range lighter: hlaza is named 20 times above row F and 48 times on row F or below; kosazana is distributed 28 above and 28 on or below. This difference coincides with the slightly dark focus of hlaza on row G and the very light focus of kosazana in row C. The first mapping step of hlaza covers both unique hues, F29 and F 17, whereas the first step of kosazana covers only unique blue. This difference coincides with the greater emphasis of similarity by which the dominant range is composed. All data regarding hlaza are consistent with one another; however, those pertaining to kosazana show inconsistencies and signs of crossover. For example, the foci of kosazana, both the first and the second, are in green, but the first mapping step is in blue, suggesting that the Zulu speaker took opposite perspectives when he focused and mapped the recessive term. Since the recessive vantage is constructed in reference to distinctiveness, the more analytical view may encourage the observer to consider alternatives; the same may pertain to his selection of two foci. Further, in Figures 2f and 2g, the qualifiers of hlaza are consistently opposite its focus, whereas the qualifiers of kosazana are on both sides of the category. Although kosazana is used less than hlaza, it shows more outlying uses, such as those in columns 32 to 36. The widely dispersed outliers are consistent with the greater polarization that attention to distinctiveness fosters.

2. 4 Formalism Coextensive vantages can be formally represented as "levels of concentration" (Figure 3). An individual can keep foremost in mind only one fixed and one mobile coordinate at a time but can "zoom in" and "zoom out" through the hierarchy while maintaining awareness of the other levels as presuppositions. The zooming process is analogous to any spatial narrowing of scope, as in "the newspaper is on the living room table". To find the newspaper, one must locate the living room in reference to the house, the table in reference to the room, and the newspaper in reference to the table. People constantly zoom in and out during the waking day when they enter and leave structures, confine or diffuse their attention,

244

Robert E. MacLaury

locate objects, or wend their way from place to place. In a hierarchically ordered vantage, one zooms in by converting a mobile coordinate to a fixed status and concentrating upon a new mobile coordinate; one zooms out in the reverse order. The double formalism (Figure 3) accounts for the greater size of the dominant vantage by attributing different top-level coordinates to each point of view: on the dominant side, similarity is included on levels I and II before distinctiveness is inducted at level IV; whereas, on the recessive side, distinctiveness is singled out twice before similarity is brought to the forefront once. This differs from a single formalism that could be considered from either end, because the balance of strength between similarity and distinctiveness would be the same from either view and, thus, would produce a semantic range of equal proportion in either case. Only the double formalism addresses the dominant-recessive pattern, SSD vs. DDS. The formalism adds coordinates of "dark" to one vantage and "light" to the other in keeping with the Zulu emphasis on brightness in the cool category, but these brightness coordinates reside at lower levels because their emphasis is not as strong in Zulu as in certain other languages (MacLaury 1992: Figures 11 to 18). On level I of the dominant vantage, coordinating blue with similarity broadens the category; on level II, the breadth is established but the category is broadened specifically to green (instead of to black or purple); on level III darkness is added, which lowers the focus and naming range, and on level IV the coordinate of distinctiveness confines the category to Dominant

Recessive Entailments

Entailments Broad Centre! foc:us Even Distri button Frequent Naming

Blue

Sim

Green

~

Dark Fixed

Stm

~ ~ ~

I

Green

Green

II

Dtst

Dark

Ill

Blue

Dist

IV

Light

t1a•n•

fixed

~ ~ ~

Dist

Nerro1o1 Mergi nel focus

Blue

Sparse Naming

Ske~o~ed

Light

Sim

t1a•ne

Figure 3. Formal representation of coextensive vantages on the Zulu cool category

.b.

Vantage theory

245

blue and green while preventing it from extending farther, as do some brightness categories (MacLaury 1992). On level I of the recessive vantage, the coordination of green and attention to distinctiveness skews the vantage toward green and polarizes the focus. On level II, coordination of distinctiveness and blue includes blue but as a second-class member. On level III, lightness is added, which raises the brightness of the entire category via the foregoing linkage with green. On level IV, similarity weakly amplifies the recessive vantage sufficiently to cover at least many of the colors that the dominant vantage includes. Figure 3 shows that the dominant vantage is internally consistent: zooming in matches the process of concentrating progressively more on distinctiveness. But the recessive vantage is internally inconsistent: zooming in toward similarity matches dilation of concentration. Perhaps this contradiction is in part what motivates people to name, focus, and map the recessive vantage inconsistently as, for example, the Zulu speaker did when he placed its first mapping step in blue or assigned it a second focus (F20) at a centralized dominant position (Figure 2e). People may wish to convert this view to a dominant vantage to escape its internal disparity. Since only the recessive vantage is inconsistent, it is likely to be derived from the dominant vantage both psychologically and in historical time. This possibility matches the older age and wider usage of dominant hlaza in Zulu. (In the Mesoamerican data, among all coextensive relations between a native word and a loanword, the loan is recessive [MacLaury in press: general summary; 199la: Figure 8].) Conversely, the recessive vantage is likely to be the site of innovation, literary and poetic elaboration, and a point of departure for developing more abstract conceptualizations of the category. We will return to these hypotheses in section 4.8.

2. 5 Qualifications When people construct a color category as an analogous point of view, they do not imagine themselves standing in a field of graded hue. Rather, they organize this highly visual domain by nonvisual spatial reasoning of the kind that blind people use to construct a vantage point. A growing literature compares the cognition of people who are congenitally blind, blinded in childhood, and sighted but blindfolded (e. g., Zimler- Keenan 1983; Shimojo et al. 1989). First, it shows that all use

246

Robert E. MacLaury

the same strategies during experiments that involve mental rotation, scanning, mirror reversal, property verification, paired association, free recall, spatial reasoning, and viewpoint. Second, it shows that these tasks are largely independent of visual imagery and, thereby, suggests that much of cognition is not specific to any sensory modality. Some research shows that images are enhanced and altered by noncanonical sensory information, such as color, brightness, weight, and numerosity, even though the imagery does not parallel perceptual processes (Intons-Peterson-Roskos-Ewaldsen 1989). In this sense, sighted people construct a color category as a vantage in reference to their memory of perceptual unique hues or pure exemplars. However, the penetration of color vantages by modality-specific information does not indicate that the conceptualization of color categories is primarily visual, and it does not imply that cognitive vantages in other domains rely upon any visual imagery. Blind people accurately construct the color wheel (Marmor 1978).

2. 6 Implications Some implications of vantage theory are described in the following subsections.

2.6.1 Analogy on the level of coordinates The general idea that linguistic categorization is based on spatial analogy goes back to Byzantine scholarship, according to Anderson (1973; cf. Lyons 1977: 718-724). But categorization is not created by an analogy between space and a category as general entities. Rather, the analogy occurs specifically between spatial and mental coordinates.

2.6.2 Phylogeny It is unlikely that the human capacity to categorize evolved among proto-

human populations in vacuo, that is, without a precedent in other behavior (MacLaury 1986: 81, 207, 410; cf. Lakoff 1990: 73). Keeping one's bearings in time and space is a primordial aptitude that is shared by most biological species. Perhaps we differ from other animals by our ability to use analogy as a means to transfer the strategy of spatial reckoning to modes of abstract thinking, such as forming and using categories that are

Vantage theory

24 7

more than simple functions of physical reality. Vantage theory attributes to human categorization that particular prehistorical origin. Other theories treat categorization in the here and now without linking the capacity to a phylogenie speculation. 2. 6. 3 Human and animal categorization

Even insects have categories, and they keep their bearings by awareness of coordinates, for example, bees. But if only humans have combined the two capacities through analogy, then how do bees organize a category? Possibly their categories are fuzzy-set identities (Kay- McDaniel 1978). Perhaps apes and even neonate infants categorize as do bees rather than as do humans beyond infancy. 2.6.4 Innateness

Since the dominant-recessive pattern is produced by each individual independently of cultural prescription or physiological determinacy, one must inquire where the knowledge comes from. Where do people learn how to construct a category by analogy to a vantage in space? The only apparent answer is that the propensity is inborn, which implies that it resides in human genes and, thus, is instinctive. It would seem less radical to propose that categorization is an innate human propensity, irrespective of how categorization is achieved. But that seemingly safer idea artificially separates a category from its making. If categorization is innate, then the process of creating a category, maintaining it, and changing it must also be inborn. 2.6.5 Embodiment

On Johnson's (1987) view, people model abstractions after the design of bodily experience with space, force, sensation, and basic-level objects. They transfer such fundamental structure to abstract thought by analogy and metaphor. Vantage theory extends Johnson's "embodiment" to commonplace categorization to provide for the active involvement of the individual. 2.6.6 Vantages versus category

A vantage is necessarily embodied, because it is an arrangement of coordinates; only a person can compose the arrangement, and in the act this

248

Robert E. MacLaury

composer becomes part of the resulting point of view. In the case of coextensively named categories, the view is named over and above the rest of the category. The latter consists of the selected coordinates and the emphases on them apart from the way the viewer has arranged them. These too embody the person. Every category consists of coordinates, emphases, and at least one arrangement or vantage. In cases of singlevantage categories, such as English green, it is most difficult to tell whether its name applies to the collection of coordinates, to its vantage, or to both. In cases of "covert coextension" -coextensive vantages that share a single name (MacLaury in press: Chapter 8)-perhaps the set of coordinates is named instead of the viewpoints. But certainly the distinct arrangements are favored by anyone who applies two names from opposite slants or viewing angles. 2.6. 7 Function versus primary motives

A person's objective in forming a category is to establish a view on the world, at least one perspective from which to comprehend. There is also the functionalist notion that a category serves to reduce infinite variation to manageable groupings. This function is undeniable and, quite reasonably, it implies that the capacity to categorize would be favored by natural selection. The two purposes of categorizing are not at odds, because behavior can be selectively advantageous for reasons that are apart from whatever motivates it. However, the two notions can come into conflict when the functionalism is confused with motivation. People categorize and change categories in order to create and refine viewpoints; they do not feel themselves to be surviving as a species or to be reducing infinite variation. To explain data, a model must phrase theory in terms of the primary objectives of the people who categorize, rather than in terms of ultimate results and evolutionary benefits. 2.6.8 Relativity and Universality

A Zulu speaker seems to intermix his color terms more than does an English speaker. The intermixture is not the result of sloppiness or mental underdevelopment. Both speakers have accurately named different points of view. Vantage theory attempts to indicate how their views differ and the principles they share.

Vantage theory

249

3. Prototypes In Irian Jaya, 1969, Eleanor Rosch (under the surname Heider) began to develop prototype theory while visiting the Dugum Dani and conducting experiments with their color categorization on the basis of Munsell chips. Nine years later, she demoted her theory to a body of systematic observations that categories are graded (Rosch 1978), summarized as "prototype effects" or "typicality effects". But in the interim, the prototype notion had substantial influence throughout cognitive psychology, cognitive linguistics, and linguistic typology, as is detailed by books, anthologies, and synopses (Cuyckens 1984; Geeraerts 1989; Lehrer 1990; Medin-Smith 1984; Mervis-Rosch 1981; MacLaury 199lb; Neisser 1987; SmithMedin 1981; Taylor 1989; Tsohatzidis 1990; Van Brakel 1991). Rosch thought of a prototype as an exemplary member of a category that possessed the most attributes common to the category and least in common with prototypes of contrasting categories; other members pertained to the category to the extent that they shared attributes with the prototype. Different members could share different attributes and attributes could differ in importance, which gave the category a graduated structure rather than the cellular all-or-none property of so called logicians' categories hastily blamed on classical philosophers (cf. Lansing, this volume). Rosch made a series of interlocking decisions and assumptions that distinguish her prototype theory from vantage theory. Rosch began fieldwork with the intention of checking Berlin and Kay's prediction of universal foci: speakers of any language will focus basic color terms on a specific and delimited subset of Munsell chips among the 330 regardless of how many basic terms the language proffers. Dani seemed opportune for her investigations, because its speakers name color with only two basic terms, mala 'light-warm' and mili 'dark-cool'. After interviewing 40 Dani one at a time with a half-set of 165 chips (omitting odd-numbered columns), she reported foci as in Figure 4 (from Heider 1972: Table 1). Berlin and Berlin (1975: 84-85) call these "floating" foci, because only five of 80 choices match any of the unique hues (G2), marked by circles in the figure. But the patterns differ between categories: 15 of the 40 light-warm foci fall between unique red and unique yellow; 27 of the 40 dark-cool foci fall between unique green and unique blue, clustering in the category center (p < .01, chi-square; 15:25 vs. 27:13). Vantage theory would address this pattern as an example of greater polarization of foci in the category whose division is most advanced; Berlin

250

Robert E. MacLaury

and Berlin note that light-warm always divides before dark-cool in any language that uses and changes these. Although Rosch reported the Dani foci, she ignored them as she based her conclusions leading to the prototype concept on experiments that ceased to consider the whole domain of color but which proceeded by showing Dani sets of color chips isolated on cards of three types (Rosch 1973). One included a "focal" chip at center, another a focal chip on its edge, and the third only nonfocal chips. She called the cards "categories" and asked Dani to learn their names, which she borrowed from "sib" terminology. They learned to name the centered focal card fastest and the nonfocal card slowest, which led her to conclude that color-category foci are universal, that Berlin and Kay's prediction obtains in a nonwestern language of two color terms, and that advocates of linguistic relativity have overstated their case. With this conclusion, Rosch subscribed to the metaphysic that Berlin and Kay (1969: 4--5) express: ... the referents for the basic color terms of all languages appear to be drawn from a set of eleven universal perceptual categories, and these categories become encoded in the history of a given language in a partially fixed order.

On this view, there is one way to see and, thus, one way to name; languages differ in color naming because some people have not yet named what they see. This view seems convincing vis-a-vis the structural-

a Hota, the light-warm color category;

c D E

F G H

Hill; the dark-cool color category; 27 foci lletveen unique green and unique blue.

·~----------------~~ Figure 4. Dani foci of light-warm and dark-cool categories (from Heider 1972: Table I)

b

Vantage theory

251

ist metaphysic of arbitrary flux, which Bloomfield (1933: 140) had championed: Physicists view the color-spectrum as a continuous scale of light-waves of different lengths, ranging from 40 to 72 hundred-thousandths of a millimeter, but languages mark off different parts of this scale quite arbitrarily and without precise limits ... and the color-names of different languages do not embrace the same gradations.

But both the deterministic and the arbitrary notion of the relation between language and the world ignore a third way of viewing it: ... a selective use has to be made of the enormous range of sensitivity and of the correspondingly great number of possibilities for directing attention. Therefore, the observer is constantly making unconscious as well as conscious selections from his visual space. (Van Wijk 1959: 130)

Vantage theory, in espousing this third metaphysic, is at odds with both Rosch's determinist realism and Bloomfieldian arbitrary relativism. In brief, everyone sees the same world, but different people emphasize or suppress their awareness of different parts of it. It may be possible to predict conditions under which one set of selections will be favored over others, for example, members of isolated societies who are also free of harsh, externally imposed demands tend to favor brightness whereas members of complex, fast changing societies or societies under duress will favor hue (MacLaury 1992). Probably the difference derives from the degree to which an analytical view has been adopted. (Van Wijk followed this approach, thinking geographical latitude would predict categories of brightness or hue.) Rosch (1977: 197) bases her model on the functionalist assumption: "Categorization occurs ... to reduce the limitless variation ... to manageable proportions." Therefore, " ... categories would ... follow the lines of natural correlations of attributes, those that maximize the correlation and thus the predictability of attributes within categories". Her functionalist concept weds categorization with "the correlational structure of the environment". This concept leaves no latitude for a category to change, unless the environment changes. Thus, it is at a loss to explain colorcategory evolution as it advances in languages throughout the world in absence of physical alterations of wavelength or of the visual system. Her only possible account of change in color terms is that of trying to achieve the one predetermined world of eleven "focal colors" that everyone must see over and above brightness levels, desaturation, or the potential for different perspectives.

252

Robert E. MacLaury

When Rosch (1978) disavowed prototype theory, she expressed the need to consider the effect of cultural values, contrast within systems of categories, asymmetries, and other dynamics of categorization. It would seem that some formal concept of point of view would have emerged in Rosch's work if she had continued it. For example, her studies of "reference point reasoning" (Rosch 1975) and "canonical perspectives" (Palmer-Rosch-Chase 1981) verge on a notion of viewpoint. Rosch's work was finally dubbed "prototype effects" because it concentrated on measurement of behavior rather than underlying causes of gradation and best-example choices; her efforts to build a model were entwined with functionalism and a deterministic belief about a world-tolanguage relation. Vantage theory differs in positing that a focus of a category plus its content, shape, and size are entailed by a selection of fixed coordinates and a tension between mobile coordinates of differing strength. A category is an analogous point of view that people establish and alter in accord with the extent that they are motivated to routinely attend to detail. The stress is on process and method over function and form. 3

4. Point of view in linguistics and psychology The key question is whether vantage theory will furnish insights into unresolved issues of cognition beyond color categorization. Seven small studies explore the possibility. Color again is taken up in two final cases. Viewpoint is familiar to cognitive linguistics (Casad 1989), typology (DeLancey 1981), and cognitive psychology (Barsalou-Sewell 1984 [summarized in Barsalou 1987: 106-114]; Barsalou-Sewell, to appear [cited in Barsalou 1993: 46]; Palmer- Rosch-Chase 1981). In some semantic studies, viewpoint is taken for granted and thus left unmentioned, unless its location is altered in a marked case or otherwise crucial to an explanation. Therefore, Casad and Langacker (1985) bring viewpoint into their analysis of Cora locationals to explain usages of "inside" and "outside" of an otherwise undesignated line of sight that speakers canonically anchor at the base of a slope. Shepard and Metzler's (1971) seminal experiment with mental rotation demonstrates inadvertently that an image is a vantage, even when a standpoint is not implanted by convention. Although their purpose is to show that images are subject to quantifiable

w:r.•

,,.p

I Hllhllff -

.it

observation, the rotation requires a fixed coordinall' ()I " '"'''":"lltiJD1 1 mobile coordinate of a rotated part, and an imaginary pl~~t '' hi llltfttl while accomplishing the mental task. · 4.0.1 Reference points and asymmetry in a hierarchy

Asymmetry is fostered by the hierarchical relation of analogous spulhtl vantages. For example, in Rosch's (1975) "reference point reasoning", th1.1 cardinal number ten is the reference point of The trees are almost ten /i't'l tall. The reference point is a ground against which the figure of r~al height is assessed; the image requires an unacknowledged observer at an imaginary vantage point who matches figure with ground in the mind's eye. Rosch links reference points with category prototypes in experiments with asymmetry judgement; subjects equate degrees of similarity between prototypes and nonprototypical members with increments of spatial distance and, further, they judge nonprototypes to be more similar or "closer" to prototypes than are the prototypes to the same marginal members (cf. Tversky 1977). Rips (1975) replicates Rosch's asymmetries. He experiments with members of the bird category, such as prototypical robins and nonprototypical ducks; his subjects overwhelmingly assert that robins can more easily transmit a disease to ducks than ducks to robins. Although Rips does not entertain spatial notions, it is common knowledge that spatial proximity expedites spread of a disease. Rips's results suggest that the emphases on similarity and distinctiveness pertain to categories other than those of color, and construction of an analogous vantage is fundamental to categories of more than one domain. When category membership is assessed with its prototype as a fixed coordinate, attention to similarity is invoked as the mobile coordinate; the appraisal in reference to similarity is natural when a category is regarded from its innermost member outward, because attention to similarity is what unites the category in the first instance. Conversely, when membership is assessed with a marginal member as a fixed coordinate, from the outside inward, an appraisal in reference to distinctiveness is natural; from the external standpoint, attention to distinctiveness normally contrasts the prototype with everything else. The principle at work here is identical to that by which individuals assess different distances between color chips from dominant and recessive vantages on a coextensively named category. Rips introduces "given" and "target", other equivalents of ground-figure, landmark-trajector, type-token, and fixed and mobile coordinates.

~~

254

a

Robert E. MacLaury

Bird Schema: Gestalt of streamline shape. feathers. beak. wings. large chest. nesting. egg laying. flight ... etc . .SJ~ecies

Schema:

Robin Schema: Small. red breast. forages for worms ... .S.I!arrow Schema: Very small. drab. flocks. song

m

Duck Schema: Medium. aquatic. flat bill. webbed feet.quacks ... Kiwi Schemg: Medium. brown. long beak. flightless ...

b

(F) Fixed Coordinate; (M) Mobile Coordinate

c

Level I

(F) Bird Schema

Level II

(F) SimiloMty

E 0 0

... N

~

(M) Similarity (M) Species Schema

Level Ill (F) Species Schema (M) Distinctiveness

c Level I

/s

Level II

--..-_s

Robin Sparrow Duck Schema . . . Schema . . . Schema

Level Ill

D

d

D

D

Kiwi . . . Schema

D

Asymmetry Judgement Robin Schema D

+-

Duck Schema

./

~ ~Mobile FixedV

Robin Schema

Duck: Schema

~

Figure 5. The category of "bird" as a hierarchy of vantages

D

Mobile'-... " ~ixed

Vantage theory

255

The schema that composes the "bird" prototype must be a gestalt of streamline shape, feathers, a beak, wings, large chest, egg laying and nesting, flight, a certain "cocky" carriage in terrestrial locomotion, and so on, at least for Europeans and Anglo-Americans (Hunn 1975). Enough of these characteristics are possessed by even the flightless emu, kiwi, and cassowary to identify them as birds, and enough of the schema inheres in every avian species to give the category a clear margin in spite of its internal gradation (Wierzbicka 1990). The schema also enables both a robin and a sparrow to resemble it closely, explaining why both are prototypical (Figure Sa). The image schema of a prototypical bird is a figure-ground composition, because each part can be assessed against the total gestalt; a species is deemed less prototypical when a part, such as flight, is lacking. But the prototype is envisioned as an image schema at the level of the whole. The reference point, then, is a schematic image, against which the less complete gestalts of real birds are matched. In terms of vantage theory, the image schema is a fixed coordinate and real instantiations are mobile coordinates that are compared to the image. Moreover, the vantage point seems to be anchored. Rips's experiments suggest that the comparison is conducted as though the categorizer looked outward from a point near the image, almost as if assuming the robin's standpoint. As a vantage, the bird category consists of a hierarchical consideration of fixed and mobile coordinates at two levels of concentration: on level I, the "bird schema" is the fixed and "similarity" the mobile coordinate, then on level II, "similarity" becomes the fixed coordinate and the "species schema" the mobile one (Figure 5b). The image schema of a bird fits well with a folk taxonomy of levels (Figure 5c). On level II, separate species such as robins and ducks share schematic elements with the bird schema, but each species has a schema of its own. Contrast between species occurs exclusively at a third level of concentration, level III: species schema (fixed), distinctiveness (mobile). Thus, it is impossible to argue that a robin is to some degree a duck or vice versa, as Osherson and Smith (1981) argue for grizzly bears and squirrels in their parody of fuzzy gradation. Such arguments erroneously confine typicality to one level of contrast, neglecting the zooming between frames of reference. Three points are critical: First, level II and level III both constitute the taxonomic level of "species", but they represent different views of species: level II is the comparison of a species schema with the bird schema of

256

Robert E. MacLaury

level I, what Langacker (1988: 140) calls a "categorizing relationship"; the bird schema is the ground and the species schema is the figure; the comparison establishes that, say, a robin is a bird. Figure Sc represents degrees of similarity between the bird schema and the various species schemas with "S"s of distinct size, largest for strongest similarity; conversely, on level III, the reciprocal emphases on distinctiveness are represented with "D"s, smallest for weakest. The figure indicates directions of comparison with arrows. Whereas on level II the species schemas are compared with the bird schema, on level III species are compared only among themselves, as, say, a robin schema with a duck schema. This establishes that robins and ducks are distinct categories, even though both are birds. One of the species schemas is held to be the ground against which the other species schema is matched as a figure (Figure Sd). Second, when Rips sets up as a fixed coordinate a species schema that shares relatively less with the bird schema, the mobile coordinate of distinctiveness will be stronger than when he sets up as his "given" a species schema that shares more with the bird schema. Thus, in Figure Sd, robins are farther from ducks than ducks from robins, when they are compared from alternative vantages. Third, although both the bird schema and the attention to similarity are not concentrated upon at level III, they are retained "in the back of the mind" as presuppositions.

4.1 Double constructions Dray (1987) equates the prototype with noncategoricallinguistic marking of a reference point among two ambiguous options. She writes " ... 'shared category structures' ... are not qualitatively different from structures that we build in context ... " (1987: 33). She considers "double constructions", such as (1)

a. b. c.

d. e.

Salad salad, not fruit salad Its not cold cold, only cold The sauce is spicy hot, not hot hot. I mean whales whales, not Wales. Julie Julie, not Julie.

Some doubles pick out the prototype of a conventional category (a), and others pick out a canonical reference point, as an extreme on a scale (b) or a primary meaning (c). But other doubles distinguish private points

Vantage theory

257

of reference, such as a homonym choice (d) or one of two people with the same name (e). In one of Dray's examples, a bride, who has a throbbing headache, says to her groom on their honeymoon, "I have a headache, but I do not have a headache headache". Her double construction picks out a stereotypic "excuse" frame that is conventionally linked to such a context, as salad salad picks out a prototype. But Dray cites a foreigner with his visa in disorder who "had a headache at the airport, but not a headache headache". Here the expectation is real physical pain, not the bureaucratic snag reported metaphorically. Doubles treat in the same way both prototypes of categories and noncategorical reference points, and some of the points of reference are context-dependent expectations, stereotypic scenarios, or idiosyncratic projections. Doubles make a case for cognitive sameness among all. In the next example, the double names the fixed coordinate of what is plainly a vantage in space, suggesting that the sameness extends to a real point of view. Further, the double construction is applied to a fixed coordinate only after the coordinate has been transformed from an earlier mobile status by zooming in. Two principles-zooming and fixed reference-are common to both a spatial vantage and to categorization. The following is an article of personal experience: Mr. C and Mrs. P are driving on a boulevard to a particular store (Figure 6). There are three lanes in each direction and a divider between. C grew up in California where such thoroughfares are a way of life; he has "always known" that each direction has a "fast lane", a "center lane", and a "slow lane" (labeled F, C, and S in the figure). P is from Peru, where there are no three-lane roads; she does not share the three-lane schema or know its names: In (1 ), C is behind the wheel heading east and sticking to the slow lane on the extreme south side of the boulevard; P is navigating, looking north across the divider and across the westbound lanes for the store on the far opposite side. In (2), P tells C, "Take the middle lane". In (3), C changes over one lane to the center. Then in (4), P says, "No, the middle middle!", and in (5), C gets the point and takes the fast lane in time to make a left turn through an exit across the divider. C and P construct different vantages from the same point in space. In (1 ), C's fixed coordinate is the culturally prescribed frame of three lanes of one direction, each with a name; his mobile coordinate is his choice of lanes. In (2), P's fixed coordinate is the entire boulevard, unconstrained by cultural bias, and her mobile coordinate is also a choice of

258

Robert E. MacLaury

Mrs. '"P"

•

"Take the ml ddl e lane.·

East IJll I I

1. I

mI

I

i,f'::, II I I

I

•

~

2.

I I

Level I

, ,

I

~~II

II I

'!

'

I

I

West

3.

"No, the middle middle I"

~

~

I

I I I I I I I MI Ic I

Ill

j!

111

I I I I

!i I I

4.

1!!1 !!

II IMI

IF lc

r,

lei

~

F

c

i!

Level II

~

~~~~ i . I

I I

I

I

I

I

~

I I I

I

lli.!!!J (F) Whole Blvd (M) Choice of 11 Lone

~

IJlHLU (F) Choice of alene (M)"C"s" Choice

Figure 6. Applying the double construction to a vantage in space

lanes. In her wider frame, the "middle" is not in the center of the asphalt, but rather it is contiguous to the divider. But as C takes the center, P "zooms in" by transforming her chosen lane to a fixed coordinate and adopting C's choice as a mobile coordinate (3) and (4). P constructs her complex vantage like this: level I: "whole boulevard" (fixed), "her chosen lane" (mobile); then level II: "her chosen lane" (fixed), "the lane that C chose" (mobile). P's transformation of her chosen lane from mobile to fixed reference qualifies it for the double construction, the middle middle, which she can apply only at level II.

Vantage theory

259

4.2 Contraction of space The measurement of color categories shows that they are enlarged when regarded from a vantage of similarity. The attention to similarity decreases the perceived distance between the colors of a category so that more colors can be included within the vantage. Further, vantage theory stipulates that people attend strongly to similarity under conditions in which the impingement of novelty or the influx of new information is minimal or, at least, gradual. Probably blind people are exposed to less novelty than are sighted people, since vision is a rich information source. The difference would determine that blind people routinely attend more strongly to similarity than do sighted people, and that sighted people habitually attend more to distinctiveness. In turn, this difference in cognition might produce a systematic discrepancy between the ways that blind subjects and blindfolded sighted subjects construct a vantage of space during controlled experiments. Hollins and Kelley (1988) observe such a systematic difference in the accuracy with which the two groups of subjects keep track of locations in relation to their own standpoints. They explain a consistent lesser accuracy of blind subjects by allusion to "distortion of the stimulus field", which is a sensitive improvement over earlier discussion of similar observations. But their account does not constitute a model of what blind people think as they systematically misjudge location. How and why do they distort? Hollins and Kelley conducted their experiments on a small, circular plexiglass tabletop with two pointers affixed to its rim 90 degrees apart, called the "original" and "new" positions. Both blind and blindfolded sighted subjects familiarized themselves with locations of five common objects on the table from a station behind the original position. Then from both positions- half the subjects moved from the original to the new position and half worked in the reverse order-they located the objects one after the other in two tasks. In the first, they were allowed to touch the table while they replaced an object that had been removed, handed to them, and named. In the second, they only touched the pointer and directed it at their memory of each location as the experimenter named the objects one at a time. Degree of error was measured against a protractor and a graph. Hollins and Kelley find that blind and blindfolded sighted subjects perform the replacements with equal accuracy. During the task, blind

260

Robert E. MacLaury

subjects first touched the table edge in a few places, then moved directly to the targeted object. They might have measured distance against the reach of their arm. (In vantage theory, that sequence consists of establishing, first, a fixed coordinate and, second, a mobile coordinate.) But during the pointing task, the subjects could not touch the table or use their own reach as a guide. From the "new" position, both groups miscalculated to the right, which is least important here. Most important is that, from both positions, blind subjects consistently miscalculated at wider angles than the blindfolded sighted subjects. Thus, for example, from the original position, they pointed to the right of objects on their right side and to the left of objects on their left, whereas blindfolded sighted subjects-to a significant extent-did not aim as widely. Separate tests established that neither group lost track of the lateral distance between the objects themselves. The distortion occurred through frontal distance along the line that is established by point of view (Figure 7). The closer one moves toward a fixed lateral axis between objects, the wider will be the angle of pointing. The blind subjects pointed as if they underestimated the distance in front of them, believing that the vantage point and objects were closer to each other than they really were. In the words of Hollins and Kelley, " ... their memory for the locations of those

Figure 7. A blind subject's contraction of forward space with consequent misjudgment of

locations Note: Closed symbols represent real locations of objects, and they approximate a blind-

folded sighted subject's notion of the locations. Open symbols represent a blind subject's notion of the locations. Thin lines represent a blindfolded sighted subject's angles of pointing. Thick lines represent a blind subject's pointing angles (modified from Hollins- Kelley 1988: Figure 8).

Vantage theory

261

objects underwent a systematic transformation, with recalled fore-andaft distances shrinking relative to side-to-side distances" (1988: 387). Vantage theory offers that, in the experiments, blind subjects contracted distance to a greater extent than blindfolded sighted subjects because, throughout life, blind people attend more strongly to similarity than do sighted people. Attention to similarity contracts space, as it has been argued in the cases of the dominant vantage of color categories and asymmetry judgements.

4.3 Quintessential and representative prototypes Lakoff (1982: 27; 1987: 86-87) distinguishes best examples from typical members of a category; a best example of the category "soccer players" is an internationally famous player, whereas a typical player has less skill and negligible notoriety. A best example might be the sort of prototype that is maintained by connoisseurs, who pay close attention to a specific category and who discern details of which the layman is ignorant. Laymen tend to lump into a category the marginal members that the connoisseurs would reject posthaste, whereas the knowing experts favor examples at an extreme on a scale. The same cultivated difference pertains to a "true smoker" versus a "typical smoker". It also coincides with the Shoshoni choice of quintessential eagles and the Anglo-American choice of representative robins and sparrows as the bird prototype in a domain with which the Shoshoni are better acquainted (Hage- Miller 1976). Representative and quintessential categorization parallels dominant and recessive vantages, although each is offered by a different group of observers.

4.4 Coextensive vantages of a bird category Berlin, Boster and O'Neil (1981: Figure 3) report coextension on an intermediate level of Aguaruna folk ornithology, although they do not tout it as such. They worked with a whole domain. Their data concern two names that both cover six species of woodpecker, whose sizes vary notably (Figure 8). Applications of the names to the species suggest fixed and mobile coordinates of "large" (fixed), "similarity" (mobile) versus "small" (fixed), and "distinctiveness" (mobile). The former vantage has a slightly centralized focus and the latter a polarized focus, a dominantrecessive pattern. The number of speakers who name species 2 ( Celeus elegans) with the term sawake suggests that species 2, rather than species

262

Robert E. MacLaury

1, is the focus of the dominant vantage; but there is further evidence: three subjects designate species 1 with the name of an overlapping neighboring category, tatasham, which predominantly contains two markedly larger woodpecker species. The coordinate system proposed here corresponds only roughly to size, as Veliniornis affinis (species 6), the focus of dai, is a little longer than Veliniornis passerinus (species 5). Berlin, Boster and O'Neil (1981: 104) explain that focus choices are based as much on visual access: " ... V. affinis, which frequents the forest edge and clearing, is seen more often than V. passerinus, a bird of the deep forest. We believe V. affinis is selected as focus due to these facts", " ... the majority of the people have formed their category of sawake around the more common C. elegans and have merged the more striking but rare C. spectabilis into the already formed category". Since these data are aggregated, it is likely that not all speakers would have related the ranges coextensively when individually tested. However, each was interviewed by showing the feathered skin of one species after another and asking its name, as in the color-chip naming procedure. There were no procedures equivalent to focus selection or mapping. These might have consisted of displaying skins of many bird species at once, including the woodpeckers, and of asking each consultant, say, to "pick out the best sawake" or to "pick out all the sawake ... now more of the sawake . . . more yet . . . those still sawake even though they also have other names". Considering the singular method of data collection, resemblance of the aggregate to the dominant-recessive pattern may mean more than any individual's choice of names. 4

4.5 Time Lakoff and Johnson (1980: 41-43) discuss time as moving-object metaphors with front-back orientation. Time is plotted in reference to people, as in face the future, weeks ahead or the time will come; or time is plotted in reference to other time, as in the next week and the week following it. Figures 9a--c separate these expressions as three vantages of time. Figure 9a represents units of time on a fixed scale; the present is moving along the scale. Figure 9b reverses the relation of fixed and mobile coordinates such that the time line is moving and the present is fixed. Each relation warrants a different expression: the days ahead and the coming days. Figure 9c depicts a process of zooming in to a closer level of concentration on the mobile time line: first, a mobile unit of reference-called a "matching day" -is coordinated with the fixed present; then, this unit is

263

Vantage theory

25_

_25

--. __ --4--..

------

2 3 4 5 6

2

123456

-~----4 . . . . . _ 4 _ _

3

·-4·-·-4--.

4

5

~

-6

15 em

-~ :.E

.....

=>.,u .. a.

~

u ..

.. ..,."" c

u

...::> u ·e

.

E

.....

u""

. .. ·-"" ::>

:;:::., -~ a.=> 0

::.-1)

"- u a. .&:.

.. ·-o·-... ..

::> "E .._c

·;: >

.....

a.

.. E . > ·;:

Figure 8. Lexical identifications of six woodpecker species by 25 Aguaruna speakers Note: The vertical scale and bar graphs indicate the number of individuals who named each species sawake and/or dai during the indentification task (modified from Berlin-

Boster-O'Neill981: Figures 3 and 4).

264

Robert E. MacLaury

established as a stable point of reference with a mobile coordinate following it. The so-called "matching day" could be "farther off" on the time line, as is next week in Lakoff and Johnson's example: the next week and the week following it. Although the unit of reference has become a fixed coordinate on level2, it is still moving within the metaphor; "fixed" means stabilized as a point of reference at a particular level of concentration, not devoid of other motion.

a ttollile

Present

nxed

••• •• •I• • •I

The Dogs Ahead

b

• ••

tlollile Fixed

Present.

c ttelltle

c

1.

E

Fixed

The Comtog Dogs

•• •I• • •I

··+······ Present.

0 0 N

The Following Days

...

2.

• • +I• ' •, • • • •I Ftxedvtto•ne

1.

Present •

2.

Matching Day

Figure 9. Three vantages of time

/

Matching Day Days Following

+ I• •I

Vantage theory

265

4. 6 Dominant and recessive metaphors Probably there are no metaphors that pertain in two directions. Metaphors of a two-way appearance constitute separate levels of a hierarchy that share one coordinate. Metaphor is a special kind of vantage that is constructed by adopting domains as coordinates and by comparing one to the other. Metaphors usually are dominant, but they can be recessive. In dominant metaphors, the source domain is more concrete than the target domain whereas in recessive metaphors the target is more concrete. From either view, the source is a fixed coordinate and the target is a mobile coordinate, because the target is compared against the source and is the focus of attention; the relation pertains even though selected characteristics are attributed to the target from the source in a direction that opposes the course of comparison. Kovecses (1986: 117-120) classifies "primary" and "secondary" metaphors: "a storm is an angry person" and "anger is a storm". An angry person is more immediate to human experience than a storm, which, in turn, is more concrete than the abstraction of anger. Thus, there are two distinct metaphors: "angry person" (fixed), "storm" (mobile) versus "storm" (fixed), "anger" (mobile). They share "storm" as a coordinate, but in the different capacities of fixed and mobile. "Angry person" and "anger" are deceptively similar, but not identical. Feld (1982), in his ethnography of the Kaluli of Papua New Guinea, provides a case for comparison. The Kaluli believe that deceased relatives reside in tree tops as birds, and living Kaluli commune with their dead through an elaborate tradition of sorrowfully singing in imitation of bird calls. Thus, "birds are dead people" and "people are birds". People in life are probably more immediate to experience than wild birds, which, in turn, are more physical than bygone dead people, a formal relationship of A (fixed), B (mobile) versus B (fixed), C (mobile). Again, there are two separate but closely related metaphors, but not a bidirectional relationship of A (fixed), B (mobile) versus B (fixed), A (mobile). Figure 10 compares these metaphors. The "angry person-storm" metaphor is of the dominant type: greater abstraction is achieved by zooming in, which is equivalent to greater distinctiveness in a category. Kovecses notes that "anger is a storm" is poetic while "a storm is an angry person" is prosaic. "People-birds" is of the rarer recessive type. "People", the more accessible or "close" domain, is the target of less accessible "birds", whereas "birds" is the target of "dead people", the least accessible domain. The recessive metaphor zooms out toward the most remote concept

266

Robert E. MacLaury

Dead People

Domlnent

I

~

I

~

Birds

~·

-/-------., I Angry Person

Storm

T

Storm

Birds

People

I

Anger

1

~:

Recessive

Figure 10. Dominant and recessive metaphors

rather than zooming in toward the analytical outlook required of an abstraction. The recessive metaphor is imbued with art and poetry as the Kaluli express their identification of people with birds in elaborate dress and with the dead through the songs. Perhaps the recessive metaphor is composed in contradistinction to societal life on the terrestrial surface, whereas the dominant metaphor is not formed in contrast to another realm.

4. 7 Denotation and connotation Denotation is a clear concept, and customarily it warrants a section in major overviews of semantics (e. g., Lyons 1977: 206-215; Russell1974). But connotation is a catch-all term for many kinds of semantic nuance, which invites fewer and briefer treatments (e. g., Ullmann 1972: 355-357; Palmer 1981: 89-93). Connotative differences reside in distinct idealizations, as in baby versus infant; they concern positive and negative registers, as in statesman versus politician; they derive from associated scenarios, as pill versus tablet became contrastive after the advent of birthcontrol pills; they consist of frames that occasion relexification after becoming negative, as bane became poison, X-rated became NC 17, crippled became handicapped which, in turn, became disabled; they involve a welter of emotive overtones, and much more. Lyons notes that J. S. Mill, who coined the denotative versus connotative distinction, altered his characterization throughout his writings. The following discussion explores the capacity of vantage theory to model delicate shades of meaning. The example draws from Stanlaw's (1987) extensive investigation of connotation among Japanese color

Vantage theory

267

terms. Stanlaw used Berlin and Kay's original Munsell method to find that Japanese speakers name each of 36 color categories with both a native term and a loan term from English. Adopting "an extended feature theory of word meaning", he subjected 80 native judges to two variants of Osgood's "semantic differential test" to delineate the distinct senses of the paired color terms. Stanlaw's Munsell measurements produced observations that parallel those of the Mesoamerican survey: 1. Deriyed terms name areas between the unique hues, such as brown, purple, pink, and orange 2. Foci are pulled off center along the brightness dimension of light-dark 3. The foci of loan words are lighter and foci of native words darker 4. Although Stanlaw does not use the terminology of vantage theory, the word pairs name categories coextensively. However, Stanlaw found that 5. Japanese consultants associated the darker, broader native terms with things traditional, artistic, and cultured while they associated the lighter, narrower loan words with things modern and new; in practice, they chose each word of a dominant-recessive pair according to context, as murasaki would name the purple thread of a kimono while paapuru would name purple in a television commercial 6. On an emotive level, men thought that the loan words were "clever" while women thought they were "silly".

,.

~

Dominant Vantage

...

,

Paapuru Recessive Vantage

Purple

Simflortty----------Distlnctlveness Light

-----

Light

Modern

Mobile Coordinates

Fixed Coordinates

JLevel JLevel JLevel

Ill

JLevel

IV

II

Mobfle Coordinates

Figure 11. Japanese vantages on the purple category showing a continuum between denota-

tive and connotative meanings as they are established by coordinates at higher and lower levels of concentration (based on Stanlaw's 1987 extended features)

268

Robert E. MacLaury

In Figure 11, the hierarchy of meanings pertaining to murasaki and paapuru are diagrammed as zooming in through levels of increasing specificity within a pair of coextensive vantages. 1. The features "traditional/modern" and "clever/silly" are treated in the same way as are the coordinates of any vantage in the universalist model of color categorization; the universal coordinates are the reference points of the maximally purple hue and the emphases on similarity and distinctiveness; dark and light are widely used 2. The potential for adding lower-level coordinates, replacing them, or removing them is open-ended 3. Different features are part and parcel of different points of view and, therefore, the nonsynonymous coextensive terms are used to name the same color in different contexts 4. Nuances of meaning become increasingly subjective, subtle, and culture-specific at depths of the hierarchies. First and top-most is the perception of strong purple, a neural reflex. Second down are the universal cognitive emphases on similarity and distinctiveness, which further determine category width. Third are dark and light, which are open to a number of symbolic associations (Izutsu 1972). Fourth are the culture-bound values of traditional and modern. Fifth is the socially governed gender difference in emotional response. The hierarchic descent marks no divide between denotation and connotation, but places extremes of denotation and connotation at opposite ends of a cline.

4. 8 Prosaic and poetic A dominant coextensive term may be more appropriate for ordinary, functional, or even depreciative usage while the recessive term appears more in literature, poetry, aesthetic and delicate expression, or politeness. (Japanese native and foreign color-term pairs may seem to suggest a counterexample; however, the recessive loan words are linked with newness.) Forbes (1979: Figures 1 and 2) describes a coextensive relation in French of marron and brun (although she does not name their relation); marron is probably dominant because it covers more Munsell colors, whereas the focus of brun is polarized in consistency with a recessive view. French speakers most often apply dominant marron to eyes, clothes, trees, and food; they apply recessive brun to hair, paint, human skin (race?), and an animal's coat. But they switch to dominant marron to purvey the indelicate, as in un enorme rat marron.

Vantage theory

269

5. Summary Vantage theory originated as a model of color categorization. It may apply to cognition in other domains of language and culture. The theory holds that a category is an analogy between the fixed and mobile coordinates in physical space and other sensations and cognitions. Both in physical terrain and solely in thought, the intersection of coordinates composes a point of view. From such a mental vantage, people create, maintain, and change a category in an effort to comprehend the world amidst its predictable aspects as well as the novelty that it presents. They change their view by altering mobile coordinates and by rearranging or replacing fixed coordinates. They may also refine a vantage by zooming in, which they accomplish by converting a mobile coordinate to fixed status and adding a new mobile coordinate, or they widen it by zooming out in the reverse manner. This restricting and dilating of a category is also achieved by analogy to behavior in space. Vantage theory predicts that coextensive ranges will manifest the dominant-recessive pattern and that the relation between vantages will evolve through a continuum marked at points by near-synonymy, coextension, inclusion, and complementation. Although the evolution may begin at any point at which a second vantage of a category is constructed, subsequent developments will adhere to the order. The two vantages need not be separately named but, when they are, the semantic relation between terms will reflect the underlying phase of development. As the recessive vantage is at once internally inconsistent and analytical, it is given to greater indeterminacy and reflection or, at least, associated with the less usual. The worldwide statistical significance of the dominant-recessive pattern of coextension in color naming hints that people may construct categories, at least of color, by an innate method, although it would be odd if an inborn capacity were devoted to only one domain. The vantage model presents a specific hypothesis of this method and adds that categorization is as much an arrangement of coordinates as it is a selection of them and emphases among them. When people categorize, they are active agents inseparable from the viewpoints they so construct and name. 5 Notes 1. My fieldwork in Zulu was supported during 1991 by a grant for the invitation of an Overseas Research Fellow from the Institute for Research Development, Human Sciences Research Council, Pretoria, Republic of South Africa, solicited on my behalf by John R. Taylor.

270

Robert E. MacLaury

2. Shortly after arriving in Johannesburg, I conducted this interview with the cooperation of a Zulu watchman in the security compound at the University of the Witwatersrand through his limited English and my hastily acquired Zulu phrases. His terms are transcribed as I heard them. Some of his terms, e. g., kosazana, inamono, tsagatso, are not recorded in the standard Zulu dictionaries, and may be borrowings from non-Zulu languages and dialects. 3. Kay and McDaniel's (1978) fuzzy-set model of color categories expressed them as graded versions of theoretic sets; they sought to reconcile gradation with the notion that a category is a logical operation of either identity, union, intersection, or inclusion. The venture was prompted by prototype theory and by De Valois, Abramov and Jacob's (1966) finding in macaque monkeys of a modified neural analogue of Ewald Hering's nineteenth-century opponent-process model of the six elemental colors or perceptually irreducible primaries: unique red, yellow, green, and blue, and pure white and black. An identity contains one primary color at which its membership value is maximal while the values decrease in direct proportion to the extent that the color intergrades with neighboring primaries. A union contains two or three primaries with maximum membership at each, a half value between any two, and the category margins tailing down to zero membership at neighboring primaries; for example, the cool category is represented as a bimodal curve of fuzzy-set calculus, an intersection consists of the grading between primaries but not the primaries at maxima; an intersection, such as turquoise, cannot attain more than half the value of membership attained by the primaries. The fuzzy-set model shares all the problems of prototype theory, such as the inability to provide for change or to specify the role of the categorizer. It is based on the same functionalism and deterministic notion of the language-to-world relation. Its union operation predicts that foci of composite color categories will fall on ·unique hues, never between, but Rosch's foci of the Dani mili category in Figure 4 fall between green, blue, and black. It offers no account of coextension, such as that of the Zulu system in Figure 2. It supplies no insight into the dominant-recessive pattern of coextension or into the different behavior that typically accompanies each range of a coextensive pair, for example, the Zulu speaker's indeterminacy when naming and mapping recessive kosazana or people's tendency to name the recessive range with loanwords, rare words, local words, or words of a literary flavor. It cannot accommodate the continuum of nearsynonymy, coextension, inclusion, and complementation- including intermediate types in real behavior-or show how the types are linked to each other or explain how and why the continuum evolves. Space limitations prevent a full review, which, however, appears elsewhere (MacLaury in press: chapter 2). 4. Berlin (1992: Figure 5.1) adds (about five) choices to (I) Celeus spectabilis to bring its score in the histogram to an even level with the 23 choices originally assigned to (2) C. elegans by Berlin, Boster and O'Neil (1981: Figure 3). (Berlin [1992] scores Celeus spectabilis at 40 in his Table 5.5, which may be a typographical error.) The scores of the five other species retain their 1981 values. Berlin derives the 23 from a second score in Table 6 of the 1981 publication. He enters "modified and corrected" in the legend of his 1992 Figure 5.1. On the same page he writes The prototype of the generic taxon called sawake appears to be Celeus spectabilis, the rufous-headed woodpecker, although it is followed closely by C. elegans. All species of this genus are fairly similar in appearance. C. spectabilis is the most visually striking of the three species with a back covered profusely with small,

Vantage theory

271

black, heart-shaped spots. One of the synonymous names used to refer to it by splitters, apu sawake 'leader saw:ike', is indicative of its perceptual salience. This chief of the woodpeckers is probably a quintessential prototype while the drabber but more frequently viewed C. elegans remains the representative prototype, along the line originally argued by Berlin, Boster and O'Neil (1981). It is critical that C. spectabilis is, in the minds of at least three Aguaruna, a member of the abutting category of large woodpeckers, tatasham, whereas C. elegans is not. In the words of Rosch and Mervis (1975: 602), "the more prototypical a category member, the more attributes it has in common with other members of the category and the less attributes in common with contrasting categories ... [;] prototypes appear to be just those members of the category which most reflect the redundancy structure of the category as a whole", which renders C. spectabilis a poor candidate, at least, for the representative prototype. These woodpecker data never made an overwhelming case for the dominant-recessive pattern of coextension. Now with the various changes and reversals, they must be regarded as dubious support for any claim, be it Berlin's or mine. Throughout all, however, they have at least shown that some Aguaruna coextensively name the six woodpeckers. 5. Post Script. The foregoing, written in 1992, treats vantage theory as it was then, save minor editing in late Winter of 1994. Hill and MacLaury (this volume) incorporates a few refinements, such as formulae and notions of inherency and a preferred level of concentration, added to the theory in Spring. In Summer, I further mathematicized vantage theory, clarified its constructs, changed "space" to "space-time", and expanded its capabilities during the final revision of the book on color categorization (MacLaury in press). But enough developments have accrued since Summer to justify a second book of the same title and general application as the present paper, to begin in 1995.

References Anderson, John M. 1973 "Maximi Planudis in memoriam", in: Ferenc Kiefer- Nicolas Ruwet (eds.), Generative grammar in Europe, Dordrecht: Reidel, 20-47. Baines, John 1985 "Color terminology and color classification: Ancient Egyptian color terminology and polychromy", American Anthropologist 87: 282-297. Barsalou, Lawrence W. 1987 "The instability of graded structure: Implications for the nature of concepts", in: Ulric Neisser (ed.), 101-140. 1993 "Flexibility, structure, and linguistic vagary in concepts: Manifestations of a compositional system of perceptual symbols", in: Alan F. Collins, Susan E. Gathercole, Martin A. Conway, and Peter E. Morris (eds.), Theories of Memory. Hove, UK, and Hillsdale, USA: Lawrence Erlbaum Associates, 29-101. [Reprinted 1994.] Barsalou, Lawrence W.- Daniel R. Sewell Constructing representations of categories from different points of view. (Emory 1984 Cognition Project, Report 2.) [Manuscript, Emory University, Atlanta, GA.] to appear Constructing categories from different points of view. [Unpublished manuscript, Department of Psychology, University of Chicago.]

272

Robert E. MacLaury

Berlin, Brent 1992 Ethnobiological classification: Principles of categorization of plants and animals in traditional societies. Princeton, NJ: Princeton University Press. Berlin, Brent- Elois Ann Berlin 1975 "Aguaruna color categories", American Ethnologist 2: 61-87. Berlin, Brent-James S. Boster- John P. O'Neil 1981 "The perceptual basis of ethnobiological classification: Evidence from Aguaruna Jivaro ornithology", Journal of Ethnobiology 1: 95-108. Berlin, Brent-Paul Kay 1969 Basic color terms: Their universality and evolution. Berkeley- Los Angeles: University of California Press. [1991] [2nd edition, with colour plate, new preface, and updated bibliography.] Bloomfield, Leonard 1933 Language. New York: Holt, Rinehart and Winston. Casad, Eugene H. 1989 Speaker vantagepoint and canonical event structure in Cora. (Series A, Paper 270.) Duisburg: Linguistic Agency University Duisburg (L.A. U. D.). Casad, Eugene H.- Ronald W. Langacker 1985 "'Inside' and 'outside' in Cora grammar", International Journal of American Linguistics 51: 247-281. Cuyckens, Hubert 1984 "Prototypes in lexical semantics: An evaluation", in: Herwig Krenn- Jiirgen Niemeyer- Ulrich Eberhardt (eds.), Sprache und Text. Tiibingen: Max Niemeyer Verlag, 174-182. D'Andrade, Roy-M. Egan 1974 "The colors of emotion", American Ethnologist 1: 49-63. DeLancey, Scott 1981 "An interpretation of split ergativity and related patterns", Language 57: 626657. De Valois, Russell L.- Israel Abramov-Gerald H. Jacobs 1966 "Analysis of response patterns in LGN cells", Journal of the Optical Society of America 56: 966-977. Douglas, Mary 1966 Purity and danger: An analysis of the concepts of pollution and taboo. London- New York: Routledge. Dray, Nancy L. 1987 Doubles and modifiers in English. [Unpublished Master's thesis in linguistics, University of Chicago.] Feld, Steven 1982 Sound and sentiment: Birds, weeping, poetics and song in Kaluli expression, Philadelphia: University of Pennsylvania Press. Forbes, Isabel "The terms brun and marron in modern standard French", Journal of Lin1979 guistics 15: 295-305. Geeraerts, Dirk 1989 "Prospects and problems of prototype theory", Linguistics 27: 600-612. Hage, Per- Kristen Hawkes 1975 "Binumarien color categories", Ethnology 24: 287-300.

Vantage theory

273

Hage, Per- Wick Miller 1976 "'Eagle' = 'bird': A note on the structure and evolution of Shoshoni ethnoornithological nomenclature. American Ethnologist 3: 481-488. Heider, Eleanor Rosch 1972 "Probabilities, sampling, and ethnographic method: The case of Dani color names", Man 7: 448-466. Hollins, Mark- Elizabath K. Kelley 1988 "Spatial updating in blind and sighted people", Perception and Psychophysics 43: 380-388. Hunn, Eugene 1975 Cognitive processes in folk ornithology: The identification of gulls. (Working Papers of the Language Behavior Research Laboratory 42.) Berkeley: University of California at Berkeley. Intons-Peterson, Margaret- Beverly B. Roskos-Ewaldsen 1989 "Sensory-perceptual qualities of images", Journal of Experimental Psychology: Learning, Memory, and Cognition 15: 188-199. Izutsu, Toshihiko 1972 "The elimination of color in Far Eastern art and philosophy", Eranos 41: 429-464. Johnson, Mark 1987 The body in the mind: The bodily basis of meaning, imagination, and reason. Chicago- London: University of Chicago Press. Kay, Paul "Synchronic variability and diachronic change in basic color terms", Journal 1975 of Language in Society 4: 257-270. "Language evolution and speech style, in: Ben G. Blount-Mary Sanchez 1977 (eds.), Variability and change: Sociocultural dimensions of language change. New York: Academic Press, 21-33. Kay, Paul-Chad K. McDaniel 1978 "The linguistic significance of basic color terms", Language 54: 610-646. Kay, Paul- Bernt Berlin-William R. Merrifield "Biocultural implications of systems of color naming", Linguistic Anthropol199la ogy 1: 12-25. 199lb The world color survey. (Microfiche.) Dallas. International Bookstore. Kovecses, Zoltan 1986 Metaphors of anger, pride, and love. Amsterdam- New York: John Benjamins. Lakoff, George 1982 Categories and cognitive models. (Cognitive Science Report 2.) Berkeley: Institute for Cognitive Studies, University of California, Berkeley. Women, fire, and dangerous things: What categories reveal about the mind. 1987 Chicago: University of Chicago Press. "The in variance hypothesis: Is abstract reasoning based on image-schemas?", 1990 Cognitive Linguistics 1: 39-74. Lakoff, George- Mark Johnson 1980 Metaphors we live by. Chicago-London: University of Chicago Press. Langacker, Ronald W. 1987 Foundations of cognitive grammar. Vol 1. Stanford: Stanford University Press. 1988 "A usage-based model", in: Brygida Rudzka-Ostyn (ed.), Topics in cognitive linguistics. Amsterdam-Philadelphia: John Benjamins, 127-161.

274

Robert E. MacLaury

Lehrer, Adrienne 1990 "Prototype theory and its implications for lexical analysis", in: Savas L. Tsohatzidis (ed.), 368-381. Lyons, John 1977 Semantics. Vols 1-2. Cambridge: Cambridge University Press. MacLaury, Robert E. 1986 Color in Mesoamerica. Vol. I. (Ph.D. dissertation in anthropology, University of California at Berkeley.) University Microfilms International, Number 8718073. "Color-category evolution and Shuswap yellow-with-green", American 1987a Anthropologist 89: 107-124. "Coextensive semantic ranges: Different names for distinct vantages of one 1987b category", Papers from the twenty-third annual regional meeting of the Chicago Linguistic Society 1: 268-282. "Social and cognitive motivations of change: Measuring variability in color 1991a semantics", Language 67: 34-62. "Prototypes revisited", Annual Review of Anthropology 20: 55-74. 199lb "From brightness to hue: An explanatory model of color-category evolution", 1992 Current Anthropology 33: 137-186. Color and cognition in Mesoamerican languages: Constructing categories as in press vantages. Austin: University of Texas Press. to appear a. "The universal pattern of coextensive color naming: Cateogizing by analogy to points of view in space". [Submitted to Behavioral and Brain Sciences.] to appear b. Zulu color categorization. [Unpublished manuscript.] Marmor, Gloria Strause 1978 "Age at onset of blindness and the development of the semantics of color names", Journal of Experimental Child Psychology 25: 267-278. Medin, Douglas L.- Edward E. Smith 1984 "Concepts and concept formation", Annual Review of Psychology 35: 113-138. Mervis, Carolyn B.- Eleanor Rosch 1981 "Categorization of natural objects", Annual Review of Psychology32: 89-115. Neisser, Ulric (ed.) 1987 Concepts and conceptual development: Ecological and intellectual factors in categorization. Cambridge: Cambridge University Press. Osherson, Daniel- Edward E. Smith 1981 "On the adequacy of prototype theory as a theory of concepts", Cognition 9: 35-58. Palmer, Frank Robert 1981 Semantics. (2nd edition.) Cambridge: Cambridge University Press. Palmer, Stephen- Eleanor Rosch- Paul Chase 1981 "Canonical perspective and the perception of objects", in: John B. LongAlan D. Baddeley (eds.), Attention and performance. Vol. 9. Hillsdale, NJ: Lawrence Erlbaum Associates, 135-151. Rips, Lance J. 1975 "Inductive judgements about natural categories", Journal of Verbal Learning and Verbal Behavior 14: 665-681. Rivers, W. H. R. 1902 "The color vision of the Eskimo", Proceedings of the Cambridge Philosophical Society 11: 143-149.

Vantage theory

1905

275

"Observations on the senses of the Todas", British Journal of Psychology I: 321-396. Rosch, Eleanor Heider 1973 "On the internal structure of perceptual and semantic categories", in: Timothy E. Moore (ed.), Cognitive development and the acquisition of language. New York: Academic Press, 111-144. 1975 "Cognitive reference points", Cognitive Psychology 7: 532-547. 1977 "Human categorization", in: Neil Warren (ed.), Studies in cross-cultural psychology. Vol. I. London: Academic Press, 3-49. 1978 "Principles of categorization", in: Eleanor Rosch- Barbara B. Lloyd (eds.), Cognition and categorization. Hillsdale, NJ: Lawrence Erlbaum Associates, 27-48. Rosch, Eleanor-Carolyn B. Mervis 1975 "Family resemblances: Studies in the internal structure of categories", Cognitive Psychology 7: 573-605. Russell, Bertrand 1974 "On denoting", in: F. Zabeeh-E. D. Klemke-A. Jacobson (eds.), Readings in Semantics. Urbana -Chicago- London: University of Illinois Press, 141158. Shepard, Roger N.-Jacqueline Metzler 1971 "Mental rotation of three-dimensional objects", Science 171:701-703. Shimojo, Shinsu-ke-Masato Sasaki- Lawrence M. Parsons-Shuko Torii 1989 "Mirror reversal by blind subjects in cutaneous perception and motor production of letters and numbers", Perception and Psychophysics 45: 145-152. Smith, Edward E.- Douglas L. Medin 1981 Categories and concepts. Cambridge, MA: Harvard University Press. Stanlaw, James M. 1987 Color, culture, and contact: English loanwords and problems of color nomenclature in modern Japanese. [Unpublished Ph.D. dissertation in Anthropology, University of Illinois at Urbana-Champaign.] Taylor, John R. 1989 Linguistic categorization: Prototypes in linguistic theory. Oxford. Clarendon Press. Tsohatzidis, Savas L. (ed.) 1990 Meanings and prototypes: Studies on linguistic categorization. London: Routledge. Tversky, Amos 1977 "Features of similarity", Psychological Review 84: 327-352. Ullmann, Stephen 1972 "Semantics", in: T. A. Sebeok (ed.), Current Trends in Linguistics Vol. 9. The Hague- Paris: Mouton, 343-394. Van Brake!, Jaap 1991 "Meaning, prototypes, and the future of cognitive science", Minds and Machines I: 233-257. Van Wijk, H. A. C. W. 1959 "A cross-cultural theory of colour and brightness nomenclature", Bijdragen Land-, Taal-, en Volkenkunde 115: 113-137.

276

Robert E. MacLaury

Wierzbicka, Anna 1990 "'Prototypes save': On the uses and abuses of the concept 'prototype' m linguistics and related fields", in: S. L. Tsohatzidis (ed.), 347-367. Zadeh, Lofti A. 1965 "Fuzzy sets", Information and Control 8: 338-353. Zimler, Jerome-Janice M. Keenan 1983 "Imagery in the congenitally blind: How visual are visual images?", Journal of Experimental Psychology: Learning, Memory, and Cognition 9: 269-282.

The terror of Montezuma: Aztec history, vantage theory, and the category of "person" Jane H. Hill- Robert E. MacLaury

1. Introduction: The category of the person The proper anthropological understanding of 'persons'- human beings considered as "psycho-socio-biological individuals" (Spiro 1993: 117)-is today the site of intense debate. Universalists (e. g., Spiro 1993) argue that people in every society must understand personhood in essentially the same way. Relativists, suggests Spiro, miss evidence for cross-cultural regularities by essentializing local ethnopsychological stereotypes, failing to explore variation and change. Relativists (e. g., Shweder- Bourne 1984) argue that universalists neglect the complexity of local context and provide no place for the dialogic relationship between individual and culture. Vantage theory (MacLaury 1987, 199lb, 1992) opens a way out of this impasse. The model provides a formal account of categories of the person that uses a small universal set of dimensions, yet permits incorporation of the local understandings and choices of interested human actors. We present an account of person categories in terms of vantage theory through analysis of the textual representation of rulers in Aztec historical chronicles in the sixteenth century. This material does not, of course, reveal the full potential of Aztec ethnopsychological thought. The chroniclers aimed to sequence and detail major historical events, and to assert the place of the Aztecs themselves within universal history. They focus on the deeds of actors in these events, not on feelings and motives. Thus Aztec histories yield data on habitual and unreflecting (albeit elite) modes of understanding of persons. The problem that stimulates our analysis is an innovation in the understanding of persons manifest in the representation of Montezuma, ruler of Tenochtitlan, in Book XII of the Florentine Codex (Sahagun 1975; referred to hereafter as simply "Book XII"). Book XII is a history in Nahuatl of the Spanish conquest of Mexico. Compiled by the great Franciscan student of Aztec culture, Fray Bernardino de Sahagun, with the

278

Jane H. Hill- Robert E. MacLaury

aid of Aztec assistants, the history incorporates narratives by men who fought with the Aztec armies. Leon-Portilla emphasizes that the narrative structures and themes of Book XII "seem to stem from ancient forms of pre-Hispanic oral tradition" (Leon-Portilla 1974: 247). 1 While this is surely the case, Book XII is particularly revealing about the evolution of Aztec thought in the colonial period. It was created in the mid-sixteenth century by Aztec narrators who were in intensive contact with Spaniards and who sought, together with Sahagun and his assistants, to make meaning from chaos in the aftermath of conquest. One of the most dramatic and compelling themes in Book XII develops the terror and indecision of Moteuczomii Xocoyotzfn 'Montezuma the Younger', Tenochtitlan tlahtoiini, 'ruler of Tenochtitlan'. 2 In 1519 he was the most powerful lord in Mexico. Uniquely, the narrators of Book XII focused on the great ruler's inner feelings and motives. Montezuma is portrayed as distinctive from other rulers because of states located "inside" his body. What led to this unprecedented representation? We argue that post-conquest chaos in Aztec society encouraged elaboration of the person category, which was constituted in reference to a pair of coordinates. One point of reference was attention to similarity, and the other was attention to difference. As society broke down, the former emphasis weakened while the latter gained strength. An analytical "recessive" vantage on the category of person was innovated as an alternative to the original and "dominant" view of persons as stereotyped by role. The alternative regarded people individually. We link details of both form and content of the portrait of Montezuma in Book XII to the predictions of our vantage theory model. We first briefly introduce vantage theory, using the familiar domain of color categorization. We then sketch our approach to person categories. We exemplify the approach with a full analysis of an Aztec person category manifested in the representations of rulers in sixteenth-century Aztec chronicles written in the Nahuatl language, focusing on the innovative portrait of Montezuma in Book XII of the Florentine Codex. Finally, we summarize the implications of our approach.

2. Vantage theory: Color categories Vantage theory emerged in research on color terminology within the framework originated by Berlin and Kay (1969), and is easily understood in reference to examples from this domain. 3 In particular, vantage theory

Aztec history, vantage theory, and the category of "person"

279

solves certain problems in the fuzzy-set interpretation of color categorizations (Kay- McDaniell978). First, this interpretation lacks any explanation for change, even though the unidirectional evolution of color-category systems was a central claim of Berlin and Kay's ( 1969) work. 4 Second, the fuzzy-set interpretation provides an inadequate account of variation. Kay (1975) pointed out that color-terminology systems vary synchronically among speakers of single languages. Such variation is taken by fuzzy-set theorists to manifest the expected "noise" in probabilistic categorization. Subsequent findings show this account to be unsatisfactory. While the tendency of speakers cross-culturally to center color categories on universal "focal colors" was an important finding of Berlin and Kay (1969), some speakers "float" their preferred focal colors at a considerable distance from the expected unique hue points, and firmly insist upon this deviation. Further, such "polarized" foci are closely associated with other phenomena. The focus that is most distant from the expected unique hue is associated with the smaller of two extensional ranges when a speaker uses two terms with almost fully overlapping meanings. These coextensive patterns exhibit high statistical significance, such that understanding them as "random noise" is problematic (MacLaury 1987, to appear). Coextension will be discussed in more detail below. 5 Mac Laury (1986, 1987, 1991 b, in press, and in this volume) proposes that these results can be handled if the category named by a color term is considered as a collection of coordinates and the selective emphases on them. Human beings understand and organize their experiences of color by assuming a personal point of view or "vantage" which orders these coordinates and emphases into an arrangement. This vantage is constructed in the same way that people locate themselves in physical space, in terms of a limited selection from among the available coordinates. At least one will be a fixed coordinate and one will be mobile (these can be taken as "ground" and "figure" respectively, in the sense of cognitive grammar [Langacker 1987]). While speakers behave as if they were locating themselves in a space when they construct a vantage, the model does not entail that they literally envision themselves as standing in an imaginary field of graded hue, but that they keep track of categories in the same way that they keep track of locations in space, in reference to things that are fixed or backgrounded and things that are moving or, at least, profiled in high relief as if movable. Thus, like other processes of abstract thought, categorization is based on an analogy which equates mentally

280

Jane H. Hill- Robert E. MacLaury

manipulable percepts, images, and cognitions with the physical points of reference and the velocities that serve as coordinates for any creature that must keep track of its position in space and time. 6 Figure 1 models the color category "warm". Any vantage on "warm" requires the selection of two fixed perceptual coordinates, the hues red and yellow, accounted for in neurophysiological terms. The second pair of coordinates are two mutable cognitions, selected from a continuum of attention to similarity and attention to distinctiveness. Emphases along this axis are reciprocal: the more speakers emphasize distinctiveness, the less they note similarity, although neither can be entirely ignored. In the warm category, a large "composite" color category, the emphasis is on similarity. A vantage on the warm category determines the arrangement of the coordinates. First, coordinates assume fixed or mobile statuses. Second, two coordinates appear at the most preferred level of concentration. By "level of concentration", we refer to the fact that in imagining a category, a speaker can "zoom in" or "pan out" on this system of coordinates. "Zooming in" shifts a mobile coordinate to a "fixed" status as ground, while adding a new mobile coordinate as figure; "panning out" reverses this process. The most preferred level of attention is "panned out", in contrast to the kind of fine-grained and finicky attention required to "zoom in". The category, however, formally includes all levels as presuppositions and as potential sites of concentration. Indicated on the left side of Figure 1 are formulaic expressions for the cognitions at each of the three levels of concentration, in order of preference. Within the box, the process of "zooming in", whereby mobile coorLevels of Concentration

Fixed Coordinates

I

RS:S

Red

s

2

s v = s2

Similarity

s2

3

v o = s2-o

Yellow

D

Hobile Coordinates

Entailments

Similarity

Breedt h of Re nge

Yellow

Werm Cetegory

Distinctiveness

Limits of Renge

~ ~

Key: Red Unique or purest red Yellow Unique yellow

~Zoom-in

Figure 1. A model of the warm category as entailments of a point of view constructed in reference to coordinates

Aztec history, vantage theory, and the category of "person"

281

dinates shift to a new status as fixed, is modeled. At the right are brief statements of the entailments-the actual behaviors that a speaker will manifest-of each level of concentration (we use this shorthand notation in our discussion of person categories as well). At our first level unique red is the fixed coordinate; 7 the mobile coordinate is similarity. At this most preferred level of concentration, the choice of a best example is backgrounded and thereby default, and so is placed around unique red. The entailment "breadth of range" occurs because a speaker who is foregrounding attention to similarity will reach out to include as many examples as possible in the category centered on unique red. At the second level of concentration, the attention to similarity shifts to a new position as fixed coordinate or ground, replacing unique red, which assumes the status of a presupposition. A new hue, yellow, appears as mobile coordinate, becoming the center of attention against the backdrop of its similarity to unique red. Once yellow has been selected, the "warm" category is entailed. 8 At this point, we can clarify the shorthand expressions for the cognitions at the three levels of concentration. The expression "SY" is read as: "(at this level) attention to similarity is the fixed coordinate and the hue yellow is the mobile coordinate". At the right of the equal sign is the expression "S2 ". This is read as: "attention to similarity is doubled at this level". That is, attention to similarity has now appeared twice as a coordinate in the warm category, with a multiplicative effect. 9 On the third level, yellow appears as the fixed coordinate or ground, and attention to distinctiveness emerges as a new mobile coordinate, becoming the new center of concentration in this most-detailed level (YD). At this level, the attention to the distinctiveness guarantees that the warm category will be prevented from extending beyond yellow throughout the color domain. Note that the attention to similarity, required for construction of the composite warm category, must still be "in the back of the mind", even though attention to distinctiveness is foregrounded. The strength of this "background" against which distinctiveness is given attention is shown by the expression "S2 - D". Even though attention to distinctiveness is foregrounded at this finest level of the zoom in, it is not multiplied, since it appears only once as a coordinate in this arrangement. Figure 2 shows a coextensive organization of the warm category. We illustrate coextension in detail, because we will argue that coextensive person categories are attested in the representations in Book XII of the Florentine Codex. Coextension is a relational type (see note 5) that sometimes appears when speakers split a composite category into basic cate-

282

Jane H. Hill- Robert E. MacLaury Recessive

Domin11nt Entailments

fntaiJ ments

Broad Central Foe us Even Distribution Frequent Naming Stability

Narro'ol Marginal Focus Ske'oled Sparse Naming Indeterminacy

s

1.YS"'S

R Fixed

D

R

~

~ R

2

D

v

1.RD=D

~

~

2. s R =s2 3.RD=s2-o

s

v

D Mobile

3

v

s

Fixed

Mobile

2. o 3.

v = o2

v s = o2 -s

Figure 2. A model of the warm category as the entailments of two viewpoints constructed in reference of opposite arrangements of one set of coordinates

gories-in this case, "warm" is splitting into the basic categories red and yellow. A coextensive category has two vantages, dominant and recessive, each labeled with a distinct term. The dominant vantage is modeled in the left-hand box, focused for purposes of illustration in reference to yellow. Yellow is the fixed coordinate and attention to similarity is the mobile coordinate. Since attention to similarity contracts perceived distance between stimuli, this range will be broad (the speaker will name many colors with this term). Its focus will be centrally placed so as to evenly and equally represent colors of the category. Further, attention to similarity at the most preferred level of concentration entails the fact that this dominant vantage is relatively stable for the speaker; that is, the speaker can reproduce the same distribution and focus over many trials. As the level of concentration zooms in, red appears as the mobile coordinate. That is, this dominant vantage still has the properties of "warm". At the finest zoom-in, red shifts to the fixed coordinate and attention to distinctiveness, the mobile coordinate, limits the range. At the dominant range, similarity makes a double contribution on two levels of concentration, as expressed by the superscript in "S 2 " in the expressions for the second and third levels of concentration. The recessive vantage, shown in the right-hand box of Figure 3, is focused in red, which appears as the fixed coordinate at the most preferred level of concentration. Here, in contrast to the warm category in Figure 1, or to the dominant vantage in Figure 2, attention to distinctiveness appears as the mobile coordinate at the preferred level of concentration. The behavioral entailment of this attention is that the recessive viewpoint will be narrow, with few colors named with the special "red" term. At the second level, attention to distinctiveness shifts to a new position as

Aztec history, vantage theory, and the category of "person"

283

fixed coordinate, with the same multiplicative effect seen for attention to similarity in the dominant range. The double attention to distinctiveness in this viewpoint (expressed by D 2) may force its focus away from the category center near unique red into a polarized position, off the main axis that connects red and yellow in the Munsell array. The foregrounded attention to distinctiveness also restricts the size of this range. Should a speaker zoom in on this category, yellow appears as the mobile coordinate, but at less preferred levels of attention. Even though attention to similarity appears at the finest level of the zoom, it is outweighed by the doubled attention to distinctiveness centered on red, so that few examples of potential "yellow" may be admitted, and their admission may be rather unstable and indeterminate for the speaker, as expressed in the statement of entailments. Understanding of the formalism may be helped by imagining the behavior of a speaker with a coextensively named category like that shown in Figure 2 in a color-survey interview. In the first step of the interview, the speaker is presented with the 330 loose chips of the Munsell color array, one by one, in random order, and is asked to name each separately (Precedure I: "naming range"). In coextension, rerandomization of the naming responses to chips into a format of the rows and columns of the Munsell chart will yield a peculiarly chaotic distribution of two labels, overlapping across the same regions of the Munsell array and often looking like a checkerboard. 10 In the second procedure of the interview, "focus", the speaker is asked to choose a focus from a derandomized array of chips for each name she volunteered in Procedure I. Even where the naming range of two terms overlaps extensively, the speaker may locate their foci quite far apart. Finally, the speaker is asked to put a grain of rice on every color of the array that she is willing to include under each name. She maps each name separately and independently of other mappings (Procedure III: "mapping"). When the speaker stops mapping the range of a name, the results are recorded, and the request is repeated. Sometimes a consultant is willing to extend mapping through several such steps until she insists that no more colors can be included for that name. The consultant is likely to render the separate mappings of the two ranges step by step from opposite directions. She will map the dominant range in a few large steps, but will map the recessive range in many tiny steps, as she considers carefully what might be encompassed. The dominant range will be larger than the recessive range. All of these behaviors are diagnostic of coextension, and their significant association has been replicated in many languages around the world (MacLaury, to appear). A dominant range may be red-focused or yellow-focused; physiol-

284

Jane H. Hill- Robert E. MacLaury

ogy does not determine the dominant-recessive pattern of coextension. These behaviors are modeled in Figure 2 as the "entailments" of the cognitive activity of assuming a coextensive pair of vantages on a category.

2.1 Vantage theory: Person categories Vantage theory purports to be a general model of categorization. In regard to the category of persons, it predicts that human actors think of experiences of persons in their diversity, and the distinctions between persons and nonpersons, in terms of coordinates, selective emphases on them, and arrangements of these. Heelas (1981) and Lock (1981) have suggested that the diverse types of person categories attested in the ethnographic record can be arranged in terms of at least two universal "dimensions", location and controi.l 1 The location coordinates are "intrinsic' vs. "extrinsic". We relabel these coordinates as "inside" and "outside" respectively, to avoid the implication of necessity vs. accidence sometimes implied in those terms. 12 This pair of coordinates derives from a human necessity, the perception of self vs. other, conceiver vs. environment. Following Hallowell (1971), Lock argues that any conceptual system must make this distinction. However, in any specific behavioral environment the self-other boundary may be fuzzy. Precisely where it is drawn is quite variable crossculturally. The local boundary is the reference point for the locations, which is fixed within any particular culture. Action, motivation, career trajectory, identity markers, or other characteristics of persons considered to be salient and relevant to social life, or that separate persons from nonpersons, may be located either outside or inside the boundary. People may construct distinct arrangements of these location coordinates, muting attention to the outside in favor of the inside, or vice versa. However, both locations are always perceptually available. For Heelas and Lock "control" indicates the site at which the capacity for a person's action is held to originate. The control dimension is perceptually necessary, since human beings must be able to distinguish between their own activity and passivity. Lock observes that while control is "through neurological reality located within the individual" (Lock 1981: 30), the characteristic symbol orientations of a culture (or, we suggest, of an individual in a particular context) may locate control in other persons, natural entities, or forces, causing persons to be thought of as "under control". The alternative symbolic orientation is to identify person and control abilities, constructing persons as "in control". We find these terms for the control dimension confusing in some contexts, so we have relabeled them as "controlled" and "autonomous". In our analysis of an Aztec person category, we treat the identification of persons as controlled

Aztec history, vantage theory, and the category of "person"

285

or autonomous not as independent coordinates, but as entailments of the strengths of the attendances to similarity and distinctiveness respectively. 13 Further, we transform "autonomous" in Mesoamerican ethnographic terms as "unpredicted" or "unpredictable", as the periphery of animals is to the controlled center of human habitation (see Figure 3). Our other coordinates for person categories are taken from the continuum of attention to similarity versus distinctiveness. These pertain to any category (Medin-Goldstone-Gentner 1990), and research on color categorization shows that shifts between these coordinates, probably driven by environmental changes experienced by speakers, are a principal engine of variation and change. Attention to similarity yields the assimilation of persons to timeless stereotyped roles (for example, the stereotyped representations of human types in early Egyptian tomb paintings (Weeks 1979]), and can entail the assimilation of nonhuman entities and artifacts to the category of persons (for example, Zapotec naming of hills as persons [MacLaury 1989: Figure 3]). Attention to distinctiveness is strong when persons are considered primarily as "unique individuals" (as in Roman portrait sculpture). Attention to distinctiveness can function also to restrict the scope of person categories to some subgroup of human beings, as when infants, women, or slaves are considered to be nonpersons. A superb example of fine-grained attention to distinctiveness was the determination by the United States Constitutional Convention of 1787 that each slave would count as exactly three-fifths of a "free Dominant

Entailments

Social Stereotyped 0 s =s Controlled United Central Human s I = s2 Apart I o = s2 -o

0 Outside

Recessive

s

0

~

~

s

I

2

D

0

~

~ I

D

Fixed

Mobile

I

D

I

Inside

3

0

s

Fixed

Mobile

S Similarity

Entailments

Individual I D =D Unpredi cted Alienated Peripheral Animal o o = D2 Common o s = D2 -s

D Distinctiveness

Figure 3. A model of the "person" category (Florentine Codex Book XII) as the entailments of two viewpoints constructed in reference of opposite arrangements of one set of coordinates

286

Jane H. Hill- Robert E. MacLaury

person" for the purpose of enumerating the population of electoral districts. Once again, we emphasize that even though a particular speaker in a particular context may emphasize attention to one pole of the continuum, the other possibility is always available and can appear as the strongest coordinate should attention shift. Figure 3 models the Aztec person category that is required to account for the representations of rulers in Book XII of the Florentine Codex. We present it here in the abstract; the concrete realizations of the coordinates and their entailments in Aztec chronicles will be illustrated in section 3. The narrators of Book XII manifested their person category as a pair of coextensive vantages. The two vantages are opposite arrangements of the coordinates, which as a set constitute the single category of person. Yet the opposite arrangements do not constitute perfectly opposite points of view. They have different widths, degrees of acceptance, and association with the usual versus the exceptional or even the bizarre. The vantages are expressed in terms of abstract coordinates at three levels of concentration. Each level of concentration, expressed through zoom-in formulas, yields behavioral entailments. Categorizers accomplish these behaviors using locally pertinent symbolic material, which we sketch here only very briefly, pending more detailed exemplification below. The dominant vantage, shown on the left-hand side of Figure 3, yields all of the textual representations of rulers that we will consider below except the Book XII portrait of Montezuma. At the most preferred level of concentration, the fixed coordinate is outside location (0); the properties of persons to which categorizers attend unreflectingly are exterior to the local self-other boundary. 14 The mobile coordinate is attention to similarity. The cognition of this level of concentration is expressed as OS = S (0 stands for "outside" and S stands for "similarity"). The entailment is that candidate entities are recruited to personhood by reference to social stereotypes, triggered by external evidence (the Aztecs appealed to dress, speech, and other public acts considered to be characteristic of particular social roles). The second level of concentration zooms in, and attention to similarity shifts to ground. Inside location now appears as the figured or mobile coordinate. The cognition is expressed formulaicly as SI = S2 (I stands for "inside"). SI yields attention to those internal properties that link a person to human society. For the Aztecs these entailed principles of "control" and "orderliness" (as opposed to chaos). 15 The controlling entities are a set of cosmic animistic forces projected into the bodily organs. Hence they are "inside" each living human being, as opposed to "outside" and public, although knowledge of their nature is available to divin-

Aztec history, vantage theory, and the category of "person"

287

ers and other ritual adepts. These animistic forces were held to shape both moment-to-moment behavior and the trajectory of the life career. S2 expresses the multiplicative or doubled relation of S on level2 with S on levell. Hence similarity is particularly strong in this vantage. This entails the recruitment of as many examples as possible to this category and the specification of a highly stereotyped central exemplar of persons-this is, we will suggest below, the tlahtoiini, the ruler of each Aztec city-state. At the third, least preferred level of concentration, an extreme zoomin shifts the inside location to ground, with attention to distinctiveness as the mobile coordinate. The formula for the level is ID = S2 - D (D stands for "distinctiveness"). Attention to distinctiveness distinguishes kings from other people and ensures some outer boundary for the person category, distinguishing human persons from, for instance, wild beasts (a point to which we will return below). One entailment of ID is that at this level of concentration categorizers can attend to those internal aspects of a person which might set him apart from the rest of society. Thus Aztec rulers could be distinguished from other persons by the fact that the force of the deity Tezcatlipoca resided within the ruler and spoke through his voice. Concentration at this level permitted healers and diviners to attent to distinctive maladjustments of the cosmic forces that might cause deviance or illness, and propose appropriate curative disciplines. S2 - D expresses the fact that, while attention to distinctiveness (D) does occur in this vantage at the most fine-grained level of concentration, this is dominated by the twofold attention to similarity (S 2). Thus, even as the distinctive properties of the ruler are attended to, this occurs against the very strong background of his likeness to other persons, his assimilation to orderly human society. The recessive vantage, modeled on the right side of Figure 3, is evidenced in the Aztec chronicles only in the representation of the ruler Montezuma in Book XII. At the most preferred level of concentration, the fixed coordinate or ground is inside location; the mobile coordinate is attention to distinctiveness. The entailments of this level of concentration, ID = D, are that persons are seen in their uniqueness, according to properties located inside the self-other boundary. Zooming in to the second level of concentration, expressed as DO = D 2 , attention to distinctiveness shifts to a position as fixed coordinate or ground, with outside location appearing as the mobile coordinate. Fleshing out the entailments of this level with local symbolic material, we find that the Aztec evaluated distinctive behavior manifested externally as alien and unpredictable, perhaps even "out of control", outside human order in the chaos of the

288

Jane H. Hill- Robert E. MacLaury

periphery, more animal than human (especially, as we will see, in the case of the ruler). Difference was bad, and was understood as a disruption of the life trajectory, caused by loss or maladjustment of the controlling animistic entities. D 2 indicates that attention to distinctiveness is doubled in this vantage, with the effect of sharply restricting the number of entities that can be recruited to the category, and with the effect also of polarization, the location of the best examples at as great a distance as possible from the central exemplars of the dominant vantage. Finally, at the most careful and least preferred focus of the zoom, expressed as OS= D 2 -S, outside location appears as the fixed coordinate and attention to similarity is the mobile coordinate. Here, the entailment is that categorizers will concentrate on external properties (speech, dress, behaviors) that are general among humans and define the outer limits of the person category. In the Aztec view, these are behaviors that are common, vulgar, and unlordly. D 2 -S expresses the fact that even though attention to similarity appears at this level, it is still dominated by attention to distinctiveness. That is, an individual included in this category might be seen as "like the common people", but against a background of deviance, such that the commonness itself is seen as an aberration.

3. Representations of "persons" in the Florentine Codex We turn now to textual exemplification of the coextensive Aztec person category sketched above. The Florentine Codex is our most important source on Aztec ethnopsychology, although the vision of human nature found there is probably idealized and conservative. Underpinning this vision is the dominant vantage shown in Figure 3. Attention to similarity against a ground of outside location (OS) entails the deemphasis of individual differences and favors conformity to a series of idealized roles. Nonconformity was recognized only as "bad" (a-cualli 'not-good'). Book X of the Codex, The People, contains a famous catalogue of "good" and "bad" stereotypes. For carpenters, it prescribes: The good (cualli) carpenter is one who uses the plumb; who is resourceful; who uses the cord, marks with lines ... The bad (acualli) carpenter is one who breaks the work into pieces, who raises a clattering din; who is a nonchalant worker, a mocker; uncooperative, wasteful, squandering ... (Sahagun 1961: 27)

Characteristically, this prescription attends to action and behavior, "outside", public properties of persons, for both "good" and "bad" exemplars. Especially for the nobility, external appearance was paramount and

Aztec history, vantage theory, and the category of "person"

289

was highly stereotyped. Clothing and bodily ornament were regulated by sumptuary laws; the Florentine Codex devotes chapters to loving description of noble garments. That clothing was seen as "outside" in relation to the self-other boundary is evidenced by the fact that noble garments were clearly separate from the persons who might wear them. They were important tributary items, which might express the allegiance of a dominated province to the Aztec state. They could be exchanged between rulers as lavish expressions of alliance. Bodily movements, a public and external mode of expression, were stereotyped as well, and dancing was an indispensable art form, especially cultivated by high nobility. Rulers themselves often danced in public. In general, emotions are also considered as "outside" in this vantage: for the Aztec, these were in the main prescribed public manifestations. These manifestations were highly conventionalized, and were regulated by etiquette that specified when to laugh, shout, or weep. Speech as well was a public, conventional affair, a matter of the appropriate performance of a role. Book VI of the Codex, The Sayings of the Elders, (Sahagun 1969) lays down proper forms for public speech learned by young nobles during a rigid education in state-run schools. The Bancroft Dialogues (Karttunen- Lockhart 1987), a manual probably prepared to educate missionaries in polite Nahuatl speech, prescribes formulae for appropriate lordly courtesies between noble kin, stereotyped according to role (the mother, the younger brother, etc.), revealing that this prescriptive tradition persisted even in the late sixteenth century. What of the properties of persons that were located "inside" them, which could be attended to in the dominant vantage at the second, more exacting level of concentration (SI = S2 )? Lopez Austin (1980) shows that the Aztecs saw human persons as points of convergence of cosmic forces that shaped, not only the behavior and life trajectory of human beings, but the organization of the entire universe. While these existed outside and beyond each human being, during life they permeated certain interior organs. The tonal/i, a force derived from light and heat, determined fate and endowed human beings with varying types of temperament and degrees of vigor. This force derived from the day of birth (tonalli means "heat, day") and permeated the head and hair. A second animistic entity, ihfyotl, gave life-sustaining breath derived from the night wind. Located in life in the liver, this force shaped human passions. Most important for us here is the teyo/ia, an animating force that in life permeated the heart (yo/lot[), governing thought and feeling and requiring regulation by moderation and penance. The teyolia of each human being lived on after

290

Jane H. Hill- Robert E. MacLaury

death, residing in the appropriate afterworld (determined by the mode of death) after a brief period of earthly incorporeality. At the most fine-grained level of concentration, the extreme zoom-in to ID = S2 - D, the pair of coordinates ID are modified by the presuppositional contexts of the more preferred levels (OS and SI). These cognitions entail the definition of persons in terms of stereotyped rules, and the assimilation of all persons to an orderly universe due to the permeation within their bodily organs of animistic entities deriving from cosmic sources. Thus, ID entails primarily vertical distinction within the social system: in the case at hand, those qualities which distinguish the ruler from other human beings. This contrasts sharply with ID in the recessive vantage, where, at the most-preferred level of concentration, the entailment is horizontal and yields attention to individuality, contextualized at leve12 as "peripheral, animal" in contrast to humanness. This contrast should make clear the effects of presuppositions (the more-preferred levels of concentration) as contexts which affect the entailments of coordinates.

3.1 The Aztec tlahtoani 'ruler' During the last years before the Conquest the stratification of the Aztec society was intensifed, and nobles were strongly set apart from commoners (we can express this as an intensification of attention to distinctiveness resulting from zooming in within the dominant vantage: ID = S2 - D). The highest rulers, called tlahtoiini, in turn were distinctive among nobles; in many public contexts, no one was permitted to look at the ruler's face. Only one tlahtoiini ruled in each city state. The lists of the rulers of the greatest cities were memorized by every noble child, and pages of the Florentine Codex and other documents enumerate their qualities. Most important, the rulers were war leaders. 16 Says Book VIII of the Codex: 17 (1)

In tlahtoiini tliicateuct/i, motociiyotia: ftequiuh catca in yiioyotl. (Sahagun 1954: 51)

'As for the ruler,

lord of people was how he was named: As for his task, it was war.'

By this stereotype the ruler is exemplary of Aztec persons, since the role of warrior was shared by all. 18

Aztec history, vantage theory, and the category of "person"

291

The contrast between the "good ruler" and the "bad ruler" is set up in a special way. The good ruler is said to be like a bearer, carrying the state as a burden: (2)

tlamiima, tecuexiinoa temacochioa ... , (Sahagun 1961: 15)

'He assumes burdens, he carries people in his cape; he bears people in his arms.'

The ruler prays that he should not let this burden fall: (3)

Quen mach nenti in oniciitoyiihuih in onictepehxihuih in miicehual/i? (Sahagun 1969: 43)

'What will happen if I should cast into the torrent, if I should cast from the crag the common people?'

The "crag, ... the torrent" is a metaphoric couplet standing for the wilderness, the earthly manifestation of the zone of chaos that lay everywhere on the periphery of the orderly human world. The Florentine Codex texts locate the ruler who turns against his subjects precisely in this inhuman and disordered wilderness (DO = D 2). He is no longer a person at all. Instead he is the archetypical inhabitant of chaos, a wild beast, "with fangs, with claws". He is named tecuiini, literally "devourer of human beings". A passage in Book X details this vision: (4)

In tlahuelf/oc tlahtoiini, tecuani tzitzimitl, co/elect/i ocelot/, cuitliichtli. (Sahagun 1961: 15)

'The bad ruler is a maneater, a flying demon, a crawling demon 19 an ocelot, a wolf.'

Lordly speech is an especially significant public and exterior sign of the ruler's nature. Rulers (with high war leaders and divinities) take the principal speaking parts in Aztec chronicles. Indeed, the tlahtoiini was the quintessential "speaker": the word is the agentive form of the Nahuatl verb tlahtoii "to speak". By being designated as "speakers', manifesting that property which distinguishes humans from animals, the rulers again appear as exemplary Aztec persons. 20 But their speeches are stereotyped set pieces repeated in similar wording again and again throughout the surviving texts. Rarely are representations of the speech of rulers accompanied by explicit attention to feelings (other than lordly valor or piety expressed in the words of the speech themselves), and then only cursorily. The

292

Jane H. Hill- Robert E. MacLaury

reason for this is that attention to any inner states requires, in the dominant vantage, a very fine zoom-in to ID. This least-preferred level of attention is unlikely to be sustained over a long sequence of textual material in Aztec histories, which are not primarily biographical, but attend to actors only insofar as they play a major role in some chronicled event. An excellent example is found in a pivotal episode of Alvarado Tezoz6moc's Cr6nica mexicayotl [Chronicle of the Mexica] (1975 [c. 1609]). The Aztecs, crude barbarian newcomers, have secured the patronage of Achitometl, tlahtoiini of the Colhua Mexica, the most powerful state in the Valley of Mexico. As a token of his favor Achitometl gives the Aztecs his daughter, so that they may make her a goddess. But Huitzilopochtli, the War God, tells them that they must sacrifice the Colhua princess and dress a priest in her skin. The priests comply, and then invite Achitometl to sacrifice to the new divinity. Achitometl sacrifices quail and incenses the altar. He does not immediately see, through the smoke, the priest dressed in his daughter's flayed skin. Suddenly he recognizes what is before him: (5)

'Thus he saw the skin !non mah quittac in ehuatl in ce tliicatl of a person, in ichpoch of his daughter, did Achitometl. 21 in Achitometl. Cencah omomauhtih. He was very much startled. Then therefore he cries out, Niman ye fc tzahtzi, he cries out to them, quintzahtzilia to his warriors. in ftlahtohcahuiin. Ihuiin in fmiicehualhuan and to his commoners he says: quimilhuia: Who are you, you Aquihqueh in amehhuan Colhuaquehe? Colhuacans? Do you not see that they Cuix ahmo anquittah ca flayed oquixipeuhqueh my daughter? in nochpochtzfn? None of the evildoers will Ahmo niciin yezqueh in remain here, tlahuelflocqueh, tiquinmictfzqueh, We will kill them, we will destroy them, tiquinpohpolazqueh, niciin tliimfzqueh in here the evildoers will tlahuelflocqueh. end. And for that reason there Auh niman ye fc moyaotlah. was war.' (Alvarado Tezoz6moc 1949: 57-58)

Aztec history, vantage theory, and the category of "person"

293

For us, the poignancy lies in Achitometl's horror. But this moment is passed over in a brief zoom-in to the finest level of concentration of the dominant vantage, ID = S2 - D, in the expression, cencah omomauhtih 'he was very much startled'. The reflexive prefix, mo-, suggests that this state is conceptualized as located inside the self-other boundary. Attention to distinctiveness is almost certainly the mobile coordinate; Achitometl's horrible realization cannot have been regarded, even among the Aztecs, as an experience broadly shared among human beings. The chronicler, however, does not sustain this attention, but immediately pans out to the preferred level of concentration, OS, giving sustained attention to Achitometl's speech in a magnificent set piece in which the ruler addresses his warriors in stereotyped high language, opening with the rhetorical questions that occur repeatedly in lordly speeches, and closing with a typical call for the destruction of the enemy. 22 A second example of a failure to sustain attention to feelings is found in the Anales de Cuauhtitlan [Annals of Cuauhtitlan], attributed to the chronicler Chimalpahin. Here, another Colhua tlahtoiini, Cocox, has offered to make the prince of Cuauhtitlan, Iztactototl, his heir. The prince scorns his offer, saying that the Colhua state will fall, and the Cuauhtitlan state is greater. The Colhua lord's reaction is represented as follows; we give only the first few lines of his long speech: (6)

In oquicac cencah ic cualiin ic moyollitlacoh quihtoh: "Tlein quihtoa piltontli conetontli? Tlii hue! xictlahtlaniciin tie afz in taltepeuh? In aquin techpehuaznequiz? Cuix amo niciin ca miquiztli? Quenin tfxco tocpac ehuaz? ... " (Lehmann 1938: 158)

'When he heard it, intensely he was thereby angered he was thereby insulted. He said, "What does he say, the little boy, the little child? Should you all perhaps ask him what our city will become? Who is he to vanquish us? Can it be that here there is no death? How shall he before us, upon us rise up? ... "'

294

Jane H. Hill- Robert E. MacLaury

Briefly concentrating at the least preferred level of the dominant vantage, ID, the chronicler notes the fury of the Colhua lord with two verbs. The first is cualan 'he was angry'. There is some evidence that this was a public, "outside" emotion, appropriate to rulers in many contexts. However the second verb, rnoyollitlacoh 'he was insulted', displays the reflexive rno-, suggesting that this is conceptualized as an "inside" state. This state, further, is located in the heart, as evidenced by incorporated noun stem yo!- 'heart'. In contrast to this brief zoom-in to ID, the ruler's stereotyped speech, with its characteristic sequence of rhetorical questions and threats, is elaborate and sustained over many lines of text, strong evidence of panning out to the preferred level of concentration OS. 3.1.1 The portrait of Cuauhternoc in Florentine Codex Book XII: A "traditional" representation of a tlahtoani

Other than Montezuma, Cuauhtemoc is the most important ruler represented in Book XII. 23 Urging his people into battle against the Spaniards, Cuauhtemoc fulfills the warrior ideal. As with the portrayals of Achitometl and Cocox, the chroniclers' rhetorical strategies are grounded in attention to external properties of role performance. Appropriate to the representation of a ruler, Cuauhtemoc is given three speeches, in chapters 33, 38, and 41. The longest and most elegant (and eloquent testimonial to Aztec attention to costume as a major component of the externalized location of the socially-relevant qualities of persons) is Cuauhtemoc's call to battle in chapter 38. Hoping to terrify the Spaniards, he dresses his bravest warrior in a quetzal-owl battle dress that had belonged to the tlahtoani Ahuitzotl: (7)

Quitoh in Cuauhternoctzfn: "Jnfn Tlahuiztli, ftlahuiz catca in notechfuhcauh in notahtzfn Ahuitzotzfn. Ma yehhuatl conitqui, rna fpan omrniqui, rna conternahuizo/ti, rna fpan tetlattiti, ma quittacan in toyaohuan, rna quirnahuizocan. " (Sahagun 1975: 117)

'Honored Cuauhtemoc said: "As for this garb, it was his garb, of my late ruler, of my father Revered Ahuitzotl. Let him wear it, Let him die with it on, May he inspire awe, May he be seen in it, May our enemies see him, May they marvel at it."'

Aztec history, vantage theory, and the category of "person"

295

Cuauhtemoc gains most distinctiveness from other rulers through one poignant sentence in chapter 40. The passage describes the moment when the Aztec war leaders surrender: (8)

Auh in Cuiiuhtemoctzfn ftlan cah in Capitan. In quimolpflfya quetzalichpetztli, tlatlahcohuitectli, huftzitzilin ihhuiyo infc ocuiltecayo. Omach catziihuac, zan qufxcahuitica. (Sahagun 1975: 123)

'And there was Honored Cuauhtemoc near the Captain. He was wrapped in a cape, of shining maguey fiber, each half of a color, of hummingbird feathers in the style of Ocuillan. It was very dirty, it was all that remained to him.'

Today, we long to know what Cuauhtemoc must have "felt", what sensations raged inside him at this dreadful moment. But the Nahuatl-speaking narrators choose a characteristic trope of exteriority: his costume. As we noted above, attention to the ruler's garb is common in Aztec historical chronicles. Painted and embroidered and feathered capes of fine fabrics and exotic weaves were key items of sumptuary display among the Aztec nobility, and, since they were received in tribute, signalled the devotion of a lord's followers and the loyalty of his allies. Here, however, the stereotyped rhetoric terminates in a new twist: his filthy cape, "all that remained to him", becomes a stunning symbol ofCuauhtemoc's defeat. 24 This manifests attention to his distinctiveness, but through a reversal, pairing this attention with the ground of outside location (OD), rather than by a shift to the recessive vantage with its level-one cognition ID, or the second-level cognition DO. 3.1.2 The portrait of Montezuma: A distinctive representation of a tlahtoiini The representation of Montezuma in Book XII is organized largely according to the principles of the recessive vantage in Figure 3. Montezuma is differentiated from all other rulers through attention to his inner states. Montezuma's distinctness from other Aztec lords is shown in five ways: (1) Although as ruler he is above all a warrior, he refuses to attack the Spaniards. (2) The ruler is a "speaker", yet Montezuma is at a loss for words. (3) His inner states are elaborately represented. (4) His "inner" or "strategist" voice (GotTman 1974) is depicted, while in other texts only pub-

296

Jane H. Hill- Robert E. Mac Laury

lie speech is recounted. (5) The strategist voice uses colloquial language and rare conditional irrealis verbs. All except (1) are found in no other account of a Mexican ruler's deeds in any of the chronicles we have examined. 25 Montezuma's refusal to make war on the Spaniards is reported by all sources. He is said to have tried to destroy them by mobilizing his allies and by invoking sorcery, but he forbade his own armies to attack. He is like a witch, not a warrior king (DO = D 2). Book XII reports that Montezuma often wept. Many historians have taken this as evidence for his terror and confusion. However, it is unlikely that this weeping would have been understood by his people as evidence for any differentiated state of the ruler's mind. For the Aztecs, weeping was a public performance. Dignitaries often wept to show sincerity and seriousness. Cortez reported, in a letter to the King of Spain, how Montezuma and other assembled lords wept profusely during discussions of a peace treaty between the Spanish and the city of Texcoco (Cortes 1986 [1519-1520]: 99).

The "sayings of the elders" in Florentine Codex Book VI make clear that the ruler, like other highly-placed Aztecs, should spend considerable time in pious and public attention to the misery and transitoriness of the human condition. Through his conventional expressions of humility in prayer, the ruler exemplified mortal personhood before the gods. For instance, Book VI, chapter 12, prescribes that the new ruler, upon his ascension, should recite before assembled dignitaries the following prayer to Tezcatlipoca: 0 master, 0 our lord, 0 lord of the near, of the nigh, 0 night, 0 wind, thou hast inclined thy heart. Perhaps thou hast mistaken me for another, I who am a commoner; I who am a laborer. In excrement, in filth hath my lifetime been- I who am unreliable; I who am of filth, of vice .... (Sahagun 1969: 41)

Several passages in Book VI instruct the supplicant to weep (choca) when addressing the gods with such prayers, so that they might see his despair and pity him. Thus Montezuma's frequent public weeping cannot be interpreted as evidence of his inner states, and would not have been so interpreted by his contemporaries. This weeping is stereotyped exterior behavior (OS = S), evidence for the pitiable condition of the ruler as an exemplary human person vis-a-vis divinities, and mention of it can be motivated by the dominant vantage. Truly differentiating attention to Montezuma's inner states is absent in the early chapters of Book XII. When Spanish ships anchor on the Gulf coast for the second time in three years, Montezuma sends messengers with lavish gifts appropriate to divinities. In this section, the representa-

Aztec history, vantage theory, and the category of "person"

297

tion of Montezuma is shaped by the dominant vantage: the ruler advises his emissaries decisively, in appropriate high language, and rewards them with lordly generosity. But as the traps he laid with spies and sorcerers failed to stop the Spanish advance, Montezuma is said to have become afraid. By chapter six, Montezuma is reported to neither sleep nor eat, nor does anyone speak to him. While Montezuma's public displays of despair are probably conventional, the passages in Book XII that show him at a loss for words are at striking variance with other representations of rulers, who are usually portrayed through their oratory. These representations of the ruler's failure to speak are heavily evaluated. "Evaluation" (Labov-Waletzky 1967; Labov 1972) is that set of elements in a narrative discourse which departs from the minimum requirements of the narrative event sequence. Heavy evaluation is taken by narrative analysts to assert "reportability", departure from default elements of events (Linde 1978). This heavy evaluation can be analyzed in vantage theory as a textual icon of Montezuma's distinctiveness from the lordly ideal. These texts are entailments of the second-level cognition of the recessive vantage, DO= D 2 , representing Montezuma as exaggeratedly different, as evidenced by his public behavior, from what was expected of the ruler. Analogous to the polarization of the focal color in the recessive vantage of coextensively named categories, this textual elaboration locates Montezuma as far as possible from other rulers in the categorial space assigned to persons. For instance, in chapter 13, we find the following passage: (9)

Auh in oahcico yehhuiintin tftlantin iuhqui pohuilihqueh in Moteuczoma in iuh mochiuh in iuh quittahqueh. In Moteuczoma, in o iuh quicac, zii ohualtoloh zii ohualtolohtimotlalih ohuiilquechpiloh ohuiilquechpilotimotliilih. Aocmo ohualniihuat Zan ontlanauhtimotliilih Huehciiuhtica in iuhqui ont/apo/oh. (Sahagun 1975: 34)

'And when the messengers arrived Thus they recounted to Montezuma What had happened What they had seen. As for Montezuma, when he heard it, he just bowed his head, just sat, head bowed, He hung his head, sat, head hanging. No longer did he speak out, He sat as if deadly sick, He was delaying, as if thus destroyed.'

298

Jane H. Hill- Robert E MacLaury

By chapter nine, Montezuma's affective states are the specific focus of considerable elaboration, as shown in example (1 0). (10)

!nrc chiucnahui caprtulo: oncan mihtoa in quenin chocac Moteuczomatzln, ihuan in chocaqueh mexihcah, in ihcuac oquimatqueh, ca cencah chicahuacqueh in espafioles: Auh in Moteuczoma cencah tlatenmah, motenmah, momauhtih, mizahuih, quitlatenmachilih in altepetl. (Sahagun 1975: 25)

'As for the ninth chapter: There it is told how Montezuma wept, and the Mexicans wept, when they knew that the Spanish were so powerful, And Montezuma intensely prayed weeping in supplication, in awe, in astonishment, he prayed in supplication for the city.'

Bierhorst (1985b: 306) provides a translation which suggests a new understanding of this passage. While Dibble and Anderson translate cenca tlatenmah, motenmah as "loudly expressed distress, he felt distressed", Bierhorst gives tenmati 'to wail or complain, to worship with prayers' (he gives the te- 'human object' prefix with this verb (te-tenmati); here we see the stem with prefixes tla- 'inanimate object' and mo- 'reflexive object). Bierhorst's translation suggests that this passage may represent Montezuma as performing one of the duties of a ruler, weeping supplication of the gods, rather than as experiencing an exclusively personal "distress" suggestive of emotional breakdown. Unusual, however, is the embellished description of affect, realized through three different verbs-motenmah, momauhtih, mizahuih. In stead of tlatenmah, where the inanimate-object prefix tla-, surely locates "prayed weeping" as a public display oriented externally to the supplicant, we find the reflexive mo- in all three of these verbs, suggesting a shift to inside location, to private "feelings" and "motives", not to public display of the pitiable human condition. In comparison to the brief zoom-in to the finest level of concentration in the dominant vantage seen in examples (5) and (6), where only one verb refers to feelings located by the mo- prefix as probably "inside", this description is sustained. Within vantage theory, we can explain this prolonged atten-

Aztec history, vantage theory, and the category of "person"

299

tion: it is possible because the text is organized by entailments of the most preferred level of attention, ID, of the recessive vantage, instead of the least-preferred level, ID, of the dominant vantage (the source of the mentions of feelings in [5] and[6]). Repetition and parallelism like that seen in (9) marks Nahuatl high language, usually deployed in stereotyped locutions of heroism and piety. The rhetorical technique is familiar, but here it is turned to an unusually elaborate representation of inner states. Proliferation of these expressions sharply distinguishes the representation of Montezuma from that of Cuauhtemoc, or, indeed, of any other Aztec ruler who appears in the texts we have examined. 26 We turn now to the presentation in Book XII of the relationship between Montezuma and the cosmic forces that determine, in Aztec view, the ruler's destiny. In chapter seven, Montezuma is twice reported to have fainted: when he hears of the Spanish guns, and, in example (11), upon learning about the great Spanish dogs of war. As far as we know, no ruler or other high personage faints in any other Aztec chronicle. When enemies are terrified, conventional high language holds that they "marvel" (mahuizoa), as in (7) above. (11)

Auh in oiuh quicac in Moteuczoma, Cencah momauhtih, iuhquin yolmic, moyoltequipachoh, mo yollohzoman. (Sahagun 1975: 20)

'And when thus Montezuma heard it, he was very frightened, thereupon he fainted, his heart was afflicted, his heart was angered.'

Here the verb "to faint" and the two mo- ("reflexive") prefixed verbs that follow all incorporate the root yol: they evoke the teyolia, the cosmic force located in the heart. The verb translated "to faint", yol-mic, literally means "he died in regard to the teyolia". (The reference of the verb miqui need not be to literal death, as is clear from the other compound verbs like a-miqui 'to be thirsty' [literally, 'water-die'].) This affliction might result from sorcery or from excess. Fainting evidenced the weakening of Montezuma's teyolia, and, consequently, of his ties to cosmic order. In the dominant vantage, human beings resemble one another by virtue of sharing of the internal animistic forces (the entailment of the formula SI = S2, at the second level of concentration). A perception that this link is weakened can precipitate a shift to the opposite arrangement, found in the recessive vantage at the second level of concentration, DO = D 2). This cognition entails that a person so characterized will exhibit unpre-

300

Jane H. Hill- Robert E. MacLaury

dictable, alien, animal-like behavior. Thus this fainting is highly significant, and will be discussed further below. The representation of Montezuma's feelings as indubitably "inside", distinguishing him from other rulers, is particularly clear in a magnificent passage from chapter nine, seen in (12). Unlike the passage in (10), which might refer to a stereotyped public display of distress appropriate to a ruler, this section is difficult to interpret in any way other than as a representation of extreme distress "felt" by Montezuma in form of unique interior states. Since the rhetorical strategies here are shaped by the preferred level of concentration of the recessive vantage, sustained attention to Montezuma's feelings is possible, realizing the entailments of the cognition ID =D. The passage includes not only remarkable elaboration of verbs representing these feelings, but also unique rhetorical and grammatical devices found nowhere else in Aztec chronicles. These include a shift in register from high language to colloquial language, an unusually rich repertoire of verbal inflections, including a high frequency of a rare, complex, and highly marked verbal aspect, the conditional irrealis, and a shift from direct discourse reported speech to inner speech. Vantage theory provides an interpretation of this dimension of the text: these highly- marked devices are textual icons of polarization of Montezuma as an exemplar of the recessive vantage. (12) 2 3

4 5 6 7 8 9 10

11 12 13

14 15

Auh in iuh quicaquiya in Moteuczoma in cencah temoloh, in cencah matataco, cencah fxco tlachiyiiznequih in teteoh, iuhquin piitzmiquiya fyollo, yolpiitzmiquiya, cholozquiya, choloznequiya, mocholtfznequiya, mocholtfzquiya, motliitfzquiya, motliitfznequiya, quinni!tliitfzquiya, quinneiniiyilfznequiya in teteoh. Auh quimoyollotica,

'And thus when Montezuma heard of the great searching, of the great pursuit, because the "gods" wanted so to look at his face, thereby his heart was tormented, he was tormented in the heart, he'd flee, he would flee, he would get away, he'd get away, he'd hide himself, he would hide himself, he'd try to hide himself, he would try to hide from the "gods". And he'd been thinking,

Aztec history, vantage theory, and the category of "person"

quimoyollotiaya, 16 quimopfctica, 17 quimopfctiaya, 18 19 quiyocoxca, 20 quiyocoyaya; 21 fc moyolnohnotzca, fc moyolnohnotzfya, 22 fhtic quimolhufca, 23 fhtic quimolhuiaya, 24 25 canah oztoc calaqufz. 26 Auh cencah fntech moyollaliaya, fntech hue/ catca fyollo, 27 fntech tlacuauhtlamatia: 28 29 Cequfntfn quimo-

30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51

machfztiaya in quihtoaya: Ca ommati in Mictlan ihuan Tonatiuh fchan ihuan Tlalocan ihuan Cincalco infc ompahtfz in campa ye hue/ motlanequilfz. Auh ye hue/ ompa motlanequiliaya motlanequilih in Cincalco. Hue/ iuh machoc, hue/ iuh tempan motecac. Auh infn ahmo hue/it, ahmo hue/ motliitih, ahmo hue/ mfnax. Aoc yehuat, aoc tie tic, aoc yehuatfz. Aoc ye onneltic, aoc tle hue/ mochfuh in fntlahtol tlahciuhqueh,

301

he was thinking, he'd been imagining, he was imagining, he'd been inventing, he was inventing, thus he'd counseled himself, thus he was counseling himself, inside he'd said to himself, inside he was telling himself to go into some cave. And he felt much confidence in those that he trusted, in those that he was sure of: Some of them would know those who were saying: That they are known the Land of the Dead and Sun's House and Rain God's Paradise and Corn Goddess's House where there will be curing, where there will be what one most wants. And indeed for that place he was longing for it, he longed for Corn Goddess's House. So it was known, so it was being rumored. But this was not possible, he could not hide, he could not take refuge. No longer had he strength, no longer any use, no longer strength. No longer there they true, no longer did the words of his wizards serve,

302

Jane H. Hill- Robert E. MacLaury

52 infc quiyo/cuepca, 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67

with which they had turned his heart, with which they had broken his infc quiyol/apanca, heart infc quiyo/ma/acachoca, with which they had set his heart spinning, infc quitlacuepilfca, with which they had turned him backwards, those who had claimed to know in quimomachitocaca what was known, in ommatih, where things were named. in ompa omoteneuh. He only waited, Zan quimochiyeltih, he only resolved, zan moyol/ohtechfuh, he struggled in his heart, moyol/ohchichilih, he turned toward the peril, quihualcentlamih, quihualcentlancuiih in fyol/o, he bent his heart to the task, he wholly submitted quimocenmacac to whatever in zahzo tlein he might see, quittiiz, he might marvel at.' quimahuizoz. (Sahagun 1975: 26)

Montezuma attempts to retreat to a sacred cave for fasting, prayer, and autosacrifice. Priests and kings often practiced this discipline; but the Book XII narrators see Montezuma's purpose as "hiding" from the Spanish (DO = D 2). The retreat attempted is not a merely physical journey. It is a magical passage to a spirit kingdom, Cinca/co, Corn Goddess's House, sought through techniques of the vision quest in which Montezuma was particularly adept. But he does not have enough strength to do this; the requisite connection to the cosmos through his teyo/ia is weak. He retains life-force sufficient only to await passively the Spanish arrival (lines 60-68). 27 If Montezuma were understood by the Book XII narrators to be fully connected with the cosmos through his teyo/ia, then the states of his heart, yo/lot/, would have been seen to derive from cosmic balance. It would then be inappropriate to analyze narrative attention to such states as functioning to differentiate Montezuma from other rulers. But Montezuma's fainting, described in the passage in (11 ), and his failure to achieve a vision, both suggest weakening of teyolia. This implies that Montezuma's will is dominated, not by forces that lie outside him, but by internal

Aztec history, vantage theory, and the category of "person"

303

loci of control, his literal bodily organs with their weak residue of power. In the case of emotion verbs which incorporate the yol-element, this element will thus refer less to the cosmic teyolia than to an individualized dimension of it, located in Montezuma's interior within the physical heart, yollotl. The text is, in fact, quite explicit that Montezuma's feelings are located in fhtic 'inside him, his inside', in lines 23 and 24. This analysis is supported by other evidence about post-conquest Aztec ideology. By 1555, when Book XII was narrated, many Aztecs believed that their gods were dead. 28 Although idolatry persisted, the place of Aztec people in a chaotic new universe was uncertain. Montezuma is the exemplary first victim of this detachment from ancient cosmic order. We can trace in succeeding chapters of Book XII the appearance of representations shaped by the increasingly finer-grained zooming in within the recessive vantage. We have already seen realized the entailments of DO= D 2 , the cognition of the second level of concentration: Montezuma does not make war, nor does he speak, nor do people speak to him. Following the passage in (12), Montezuma essentially has no public existence as far as Book XII is concerned (except in the account of his meeting with Cortez, discussed below), and the narrators locate him at increasing distances from the orderly center of the human world. In chapter ten he leaves the imperial palace and moves to a smaller seat. He sends an impersonator as delegate to meet the Spanish. In the thirteenth chapter, Montezuma's soothsayers and priests come upon a drunk, who cries out that Montezuma now satisfies the stereotype of the bad ruler (OS = D 2 - S): He hath committed a fault, he hath abandoned the common folk; he hath destroyed the people. Because of him, they have been struck on the head; because of him they have been wrapped [in wrappings for the dead]. They have been laughed at; they have been mocked! (Sahagun 1975: 33)

When the Spanish arrive in Tenochtitlan, Montezuma goes to meet them on the Iztapalapa causeway, at the southeastern edge of his city. He greets Cortez with a last lordly speech, long and elegant. But the Spanish do not treat him as a ruler. Instead, they "look upon his face", they stare at him, even dismounting from their horses to examine him more closely. They touch him and caress him, and lead him by the hand. Brief attention to this moment is a representation appropriate to the entailment of the extreme zoom-in of the recessive vantage, OS= D 2 -S: Montezuma is treated like a common man. To the watching Aztecs, this shocking lese majeste must have made what came next- Montezuma is seized and made captive-almost anticlimactic. 29 Chapter 18 gives extended attention to

304

Jane H. Hill- Robert E. MacLaury

the sacking of Montezuma's palace. This attention is entailed by the cognition DO = D 2 at the second level of the recessive vantage: the sumptuary property that constitutes an especially important external dimension of the person of the ruler is stripped from him, leaving him startlingly distinct from the lordly ideal, as impoverised as a commoner (OS= D 2 -S): Thereupon was brought forth [Moctezuma's] own property, that which was indeed his personally, his very own lot, precious things all; the necklaces with pendants, the arm bands with tufts of quetzal feathers, the golden arm bands, and the bracelets, the golden bands with shells, to fasten at the ankle, and the turquoise diadem, the attribute of the ruler, and the turquoise nose rods, and the rest of his goods without number. They took it all. They possessed themselves of all, they appropriated all to themselves, they took all to themselves as their lot. (Sahagun 1975: 49)

Even after the massacre of the Aztecs by Pedro de Alvarado, Montezuma does not break his silence. Instead the courtier Itzcoatzin, not the ruler himself, tries to quiet the fury of the people. On the "Night of Sorrows", as the Spanish are driven from the city, Montezuma's body is found, lying alongside a canal along with that of his loyal follower Itzcoatzin. The ruler is treated like a commoner, cremated without ceremony in a remote location, while passersby shout insults. The account of Montezuma's cremation (in chapter 23) suggests that he has been assigned to the periphery, to the nonhuman world. To the degree that he is like a person at all, he is common. No honor is done him. His corpse burns "with a foul smell". The audience accuse him of the faults associated with the bad ruler, behavior as a wild beast, behavior that is "out of control". This blockhead! He terrorized the world; there was dread in the world, there was astonishment. This man! If anyone offended him only little, he at once disposed of him. Many he punished for imagined [faults] which were not real, which were only a fabrication of words. (Sahagun 1975: 66)

These are precisely what we would expect as textual realizations of the entailments of DO= D 2 , the second level of concentration of the recessive vantage. In contrast, proper honor is given to Itzcoatzin. His body is found with the ruler's, but the spokesman, not the lord, is cremated in the courtyard of a great temple in Tlatelolco, orderly center of the human world, "arrayed in the palace flag and in other paper goods" (Sahagun 1975: 66). While those who watched Montezuma's cremation shouted insults,

Aztec history, vantage theory, and the category of "person"

305

as Itzcoatzin's corpse burns the people weep, in that most characteristically Aztec performance of profound human solidarity. We turn now to the textual icons of "polarization" in the passage in (12). These are summarized in Table 1. Evidence for polarization is important to the vantage theory analysis, since our proposal that the representation of Montezuma is shaped by the recessive vantage requires that attention to his distinctiveness will be extremely strong. This is forced by the claim that in a recessive vantage attention to distinctiveness is the mobile coordinate at the most preferred level of concentration, and, furthermore, appears again as fixed coordinate at the second level (D 2). This prediction is borne out: the rhetorical system in ( 12) manifests an extraordinary elaboration of differentiating thematic material. Table 1. Rhetorical systems in the portrait of Montezuma in Book XII

Marking in chronicles in general Less marked, high frequency - - - - - More marked, low frequency 1. Verbal inflection

a. Pefective; b. Imperfective; c. Pluperfect; d. Conditional-irrealis 2. Reported speech voice a. Direct discourse; b. Indirect discourse; c. Quasilocution; d. Inner voice 3. Register a. High language; b. Colloquial language Marking Reversal in portrait of Montezuma Lower frequency Higher frequency

First, the passage in (12) exhibits an unusually diverse array of verbal inflections, including the conditional irrealis, the most highly marked Nahuatl verb construction. In classical Nahuatl, as in most languages, the main line of narrative is usually constructed through preterite verbs. These can be seen in the passage in example (ll ), where all the verbs are of this type. Example (12) includes preterites (for instance, a series of preterites appear in the coda of the passage in lines 59 to 67, but also has imperfect, pluperfect, and irrealis verbs, in addition to the rare conditional irrealis. These inflections are illustrated here: 1. Preterite verbs: The preterite is the "zero" form, marked either by truncating the verb stem (as in mfnax, 'he hid himself' in line 46, from mo-fnaya-perfective), by a suffix -h (seen in the verbs in lines 59 to 62,

306

2.

3.

4. 5.

Jane H. Hill- Robert E. MacLaury

or -c on stems that do not truncate, as in quimocenmacac 'he submitted himself entirely' in line 64.3° Pluperfect verbs ending in -ca, with the preterite stem. Examples are seen at lines 15, quimoyollotica 'he'd been thinking' and 17, quimopfctica, 'he'd been imagining'. These are paired frequently in this passage with imperfect verbs, giving a sense of duration. Imperfect verbs ending in -ya, seen in lines 16, quimoyollotiaya 'he was thinking', and 18, quimopfctiaya 'he was imagining', paired with the pluperfect. Irrealis verbs ending in -z, e. g., ompahtiz in line 36. Conditional irrealis verbs ending in -z-(ne)qui-ya '-irrealis-conditional-imperfect', e. g., cholozquia (line 7) and motlatfzquiya (line 11).

The conditional irrealis construction is peripheral not only in its rarity, but in its morphological complexity. As can be seen above, the construction requires three suffixes: the irrealis, the conditional (Andrews [1975] believes that this affix, which appears either as -nequi or as -qui, is derived from the verb nequi 'to want'), and the imperfect. Thus the form must be considered highly marked, and the presence of eight such verbs, in lines 8 to 14, is a remarkable feature of the passage. Second, the rhetorical system in (12) exhibits a development in the representation of lordly reported speech that is unique among the sixteenth-century chronicles: a movement away from direct discourse through indirect discourse and quasilocutional discourse, culminating in a representation of the "inner voice" of the ruler. This movement constitutes an icon for a movement in the locus of action away from public performance, typified in lordly speech, to an interior realm of feelings. Book XII shares with other sixteenth-century Nahuatl chronicles techniques for the elaboration of incident through the use of reported speech in direct discourse, which Longacre (1976) and Larson (1978) have shown appear universally in oral literature. Students of reported speech have distinguished a variety of types. Universal in the languages of the world is "direct discourse", where speech is represented from the point of view of the represented speaker, not the narrator. Propositional content of the speech is represented "faithfully", within the local cultural canons for such fidelity (Leech 1978). This is the preferred representation in Aztec literature; the speeches of Achitometl in (4), Cocox in (5), or Cuauhtemoc in (6) above exemplify it. Less common, both in Aztec literature and in the world's languages, is "indirect discourse". Indeed, there are languages which do not exhibit

Aztec history, vantage theory, and the category of "person"

307

this type (Li 1986), and it is quite rare in Aztec sources. Indirect discourse reflects the point of view of the narrator, not of the represented speaker. An example is seen at lines 23 to 25, repeated here to show its details: (12)

[lines 23-25] lhtic quimolhulca, lhtic quimolhuitiya, canah oztoc ca!aqulz. 'Inside himself he'd been telling himself, inside himself he was telling himself to go into some cave.'

Here, the verb calaqulz in the represented speech is in the third person, showing the point of view of the narrator. A passage in direct discourse, from the point of view of the represented speaker, would exhibit firstperson nicalaquiz, 'I will enter ... '. A third type of representation of reported speech, not usually grouped with this category by scholars, is also found in (12). We call this "quasilocutional". Quasilocutionals include representations of states which must be presumed to have locutionary or at least "propositional" content, even though this is not represented. Examples in (12) include moyolnohnotzca ... moyolnohnotzlya 'he'd been counselling himself, he was counselling himself', lines 21 and 22. We include in this category expressions of propositional attitude like that in lines 38 to 41: Auh ye hue! ompa motlanequiliaya, motlanequillh in Cincalco 'And indeed for that place he was longing, he longed for Corn Goddess's House'. The abundance of quasilocutional expressions in (12) is unusual both in Aztec literature and in the languages of the world, since many languages have only a few verbs of this type. Quasilocutional expression further reduces propositional content, which is often represented only very sketchily or is available only through inference by the hearer. The passage in (12) exhibits a fourth type of reported speech, an "inner voice" of Montezuma, of the type called "strategist" by Goffman (1974), found in lines 44 to 51. We argue that the voice in these lines is an inner voice, not the voice of the narrator, since in form an content it departs sharply from the narrative strategies found elsewhere in the passage. The passage contains brief expressions of despair: ahmo hue/it, ahmo hue! 'no way, impossible' (lines 44 to 46); aoc 'never' (lines 46 to 51). The lines are very short, and the verbs themselves are strikingly brief compared to the remainder of the passage: contrast motlatih 'hide himself', mlnax 'take refuge', and other verbs in lines 44 to 51 with those in line 52 and beyond, when high language, signified by long verbs with incorporated objects and compounded morphology and chains of parallelisms with subtle changes in meaning, begins anew the central rhetorical tendency of the passage.

308

Jane H. Hill- Robert E. MacLaury

An additional reason that we have concluded that lines 44 to 51 represent an inner voice is that they are dominated by expressions which we take to be almost entirely "subjective" (Banfield 1982), expressions equivalent to English forms like Damn! and Alas! Hill has heard ahmo huelit(i) 'no way, it's impossible' and aoc 'no longer' used by contemporary Nahu- ·. atl speakers as "expressives", indicating affective states ranging from casual dismissal to hopelessness. Among speakers today such expressive forms are merely quotidian. But in a sixteenth-century literary work like Book XII such "subjective" elements are deeply colloquial, in striking contrast to the high-language. public discourse of Aztec ruler seen again and again in the chronicles. This colloquialism is a shocking departure from the locutionary ideal of lordly grandeut, valor, and optimism in the fact of adversity. 31 It is evidence that this brief climactic passage is organized by entailments of the least preferred level of concentration in a recessive vantage, with its arrangement OS= D 2 -S, materially realized with locutionary evidence for unlordly vulgarity. Why are there no locutionary verbs introducing lines 44 to 51? We believe that this absence occurs because the narrators are exploiting a rhetorical technique that Longacre (1976) calls "drama". '~Drama", the representation of human speech without accompanying locutionary verbs, is a preferred rhetorical strategy for the construction of narrative climax or "peak"; its presence, among the many other evidences of "disturbance" of the poetic and rhetorical structure of the narrative in lines 44 to 51 (not the least of which is the extraordinary sequence of eight negatives, seven of them initial in their clauses) suggests that these lines are the climax ofthis story of Montezuma's role in the history of the conquest. His fate is decided here. All that follows is "coda", the tying up of loose ends. Students of iconicity in language (cf. Haiman 1985) recognized that marked forms and forms that are longer or more complex than usual often stand for distance of what is represented from a prototype. Our theoretical position requires an extension of this useful generalization; such an icon can stand not only for distance from a prototype within a single category (as, for instance, in the English expression light green), but for the differentiation of a recessive vantage from the dominant vantage at the discourse level of expression of a category. 3.1.3 Summary of the paired vantages in Book XII

We have presented detailed textual evidence that the narrators of Book XII saw rulers from a pair of coextensive vantages on the category of

Aztec history, vantage theory, and the category of "person"

309

person. They represent Cuauhtemoc from the point of view of a dominant vantage, through stereotyped techiques that can be found in many other sixteenth-century Nahuatl language chronicles. The fixed coordinate or ground of this vantage is "outside" location; thus, Cuauhtemoc is characterized primarily through speech and dress. The narrators presupposed that Cuauhtemoc was "controlled": this is an entailment of the second level of concentration of this vantage, and the presence of the usual array of cosmic forces as fully realized in Cuauhtemoc's nature is evidenced by his vigor and courage in battle. In contrast, the portrait of Montezuma is unique in both form and content. Shaped by the recessive vantage, it grounds his distinctiveness not only in reversals (his failure to fight or to speak), but in "feelings" explicitly located inside the self-other boundary. Montezuma, in contrast to Cuauhtemoc, is "uncontrolled" and autonomous, his connection to cosmic and human order both weakened. Thus he waits helplessly for the Spanish arrival, scorned by his people as an inhuman being who has mistreated them and abandoned them to disorder, who permitted himself to be treated like a common man, thereby placing all Aztec lives in jeopardy. Once we have determined that the portrait of Montezuma was constructed as a recessive vantage, we can account for the details of the highly marked narrative strategies summarized in Table 1. We take these details to be icons of "polarization" of the focus of the recessive vantage. Polarization occurs because in a recessive vantage attention to distinctiveness is extremely strong, functioning both as the mobile coordinate at the most preferred level of attention, and as the fixed coordinate at the second level (D 2).

3.2 Representations of Cuauhtemoc and Montezuma in other chronicles Other sixteenth-century texts that portray Montezuma and Cuauhtemoc assimilate both figures to the stereotyped array of rhetorical techniques for representing rulers, and hence to the dominant vantage. Thus, the recessive vantage is not elicited by any intrinsic uniqueness of Montezuma: we can draw no conclusions about what the great lord was "really" like. Instead, the recessive vantage provides a structure that is precisely what makes attention to his uniqueness possible. Texts constructed from the perspective of the dominant vantage represent him as like other rul- . ers. The earliest known Nahuatl conquest chronicle, the Anales de T/atelolco [Annals of Tlatelolco ], is thought to date from 1528. We infer from '

310

Jane H. Hill-Robert E. MacLaury

excerpts published by Leon-Portilla (we have not been able to examine a Nahuatl version) that this chronicle treats Montezuma as a conventional ruler, neither "bad" nor deviant. He orders that the Spanish be properly greeted and commands that the ill-fated festival of Toxcatl (that ended in the massacre of the unarmed Aztec celebrants by the troops of the infamous Pedro de Alvarado) go forward. Most historians consider this to be the event that gave the Aztec "war party", opposed to the faction around Montezuma who desired peace with the Spaniards, the upper hand. The 1528 chronicler gives Montezuma a speech protesting the massacre: The king Moteuczoma, who was accompanied by Itzcohuatzin and by those who had brought food for the Spaniards, protested:: "Our lords, that is enough! What are you doing? These people are not carrying shields or macanas. Our lords, they are completely unarumed!" (Leon-Portilla 1962: 131)

Leon-Portilla's translation suggests that this speech adheres to the model we have seen elsewhere. Montezuma encourages peace, but by defending his people, not by urging them to concede. The speech is courteous and includes one of the rhetorical questions favored in lordly style. The third portrayal of Montezuma contrasts sharply with the treatment of the same incident in Book XII, where, as we have seen, -Montezuma delivers no speech during or after the massacre, but sends his spokesman, Itzcoatzin, to confront his outraged subjects. The Anales de Ia conquista de Tlatelolco en 1473 yen 1521 [Annals of the conquest ofTlatelolco in 1473 and 1521] gives speeches to both rulers. This text represents Cuauhtemoc as a stereotypical warrior-ruler, vigorous in defending his city. Cuauhtemoc's speech in (13) is in traditional style: (13)

Auh in tlahtoani Cuauhtemoctzfn in oquilhualmihtolhufa: Xiquincaquican nepapan tlaxcaltecah Ale polihufz in altepetl in Tlatilolco in ihtauhca. Ayac zan quimonfxtaquilfz. In ahzo tlanehuflo? In amechonpahpaquilitihqueh in mexihcah tepilhuan in tlatilolcah? (McAfee and Barlow 1945: 331)

'And as for the ruler Cuauhtemoc he came saying: Hear them, the Tlaxcaltecans: Never will the city be lost, Tlatelolco its fame. No one will even touch it. Could it diminish? Does it not belong to you, nobles of Mexico, nobles of Tlateloloco?'

Aztec history, vantage theory, and the category of "person"

311

Cuauhtemoc's warlike stance is contrasted with Montezuma's temerity. But in this text, Montezuma is represented not through his feelings, as in Book XII, but through deviant speech. (14)

In Cuauhtemi5ctzfn in ye fpatiuh in a/tepetl in iuhqui coztic tei5cuitlat/ atlacuezonan In xochitlalpan mocuep in a/tepetl. Auh in Moteuczi5matzfn quihualihtoaya, Ma fxquich cua/ani/iztli mexihcah, rna ximocehuican. McAfee and Barlow 1945: 330)

'As for Cuauhtemoc, it was his treasure, the city was like yellow gold, a water lily a flowery paradise the city bloomed. 3 2 But as for Montezuma he came saymg, "Let there not be anger among the Mexicans, 33 Let them be at peace"'.

In summary, the evidence suggests that treatment of a tlahtoani through stereotyped speeches is not tied to particular personages, but is applicable to all. It counters the argument that the great difference between the representations of the two rulers in Book XII is simply a matter of sketching contrasting personalities.

4. Why does the recessive vantage appear in Book XII? Why does the Aztec category of the person develop a recessive vantage? Why did not all Aztecs continue throughout the colonial period to understand persons through the dominant vantage constructed on the coordinates of outside location and similarity, a type of vantage very widely attested in the ethnographic record? MacLaury (1991a) has shown that the reciprocal attentions to similarity and distinctiveness are vulnerable to environmental factors. 34 In particular, rapid change can trigger a shift toward attendance to distinctiveness. The preference for sweeping categorization motivated by attention to similarity is challenged by rapid change: people begin to prefer fine-grained differentiation of phenomena. In the case of Aztec person categories, the extraordinary chaos of the conquest itself certainly provides a context appropriate for attention to distinctiveness. 35 When this is turned to persons, universal constraints specify which coordinates may be selected while person categories split

312

Jane H. Hill~ Robert E. MacLaury

and change. If attention to similarity is grounded "outside" the self-other boundary, then attention to distinctiveness will access the "inside" ground, even if this has hitherto been ignored. If the control locus "controlled" is entailed by the dominant vantage, then attention to distinctiveness against the ground of inside location can summon into consciousness the entailment of "unpredictable, out of control". But it is the shift in the cognitive attentions to similarity and distinctiveness that drives the change. This specification of the origin of category change contributes an essential increase in explanatory power to the Heelas- Lock theory of person categories. The vantage theory characterization of cognitive innovation thus permits a formal account of a phenomenon often noted by anthropologists: catastrophic change can yield novel ideologies. Vantage theory suggests why attention to distinctiveness and the consequent recessive vantage emerge in the relatively unreflective context of the historical chronicle in association with the figure of the great lord Montezuma. This is important, since Klor de Alva (1987) proposes that, as evidenced by the complaints of the missionaries that their parishioners were unable to accomplish sincere Christian confessions, inner states were simply not salient for Aztec sinners. Why, then were Sahagun's consultants able to attend to Montezuma's inner states? We have pointed out above that while rulers were in many ways archetypal Aztec "persons", they were also highly differentiated from other people. Vantage theory predicts precisely that a recessive vantage is likely to be emerge first at such a perceptually differentiated locus, and is unlikely to be focused in parts of categories which exhibit inherently low differentiation, such as some subcategory of "ordinary people", or, especially, on the presupposed self of the confessing sinner. While descendants of the Aztec royal families retained many privileges under Spanish rule, they were certainly less different from ordinary people by far than was the last Tenochtitlan tlahtoani with his well-known autocratic claims and his pivotal historical role. Thus a representation of Montezuma is an obvious locus for the development of the recessive vantage, which, by the very attention to distinctiveness that defined it, was quite resistant to the incorporation of additional members. Thus our results do not contradict those of Klor de Alva. Vantage theory permits principled speculation on why the Book XII narrators, alone among the chroniclers, display the recessive point of view. The other sixteenth-century chroniclers shared with the Book XII narrators both the context of catastrophic change and the dilemma of accounting for the Aztec defeat. The usual answer is that Sahagun's con-

Aztec history, vantage theory, and the category (if "person"

313

sultants fought for Tlatelolco, where Cuauhtemoc was the highest lord, rather than for Tenochtitlan, Montezuma's seat. Thus they had nothing to lose from emphasizing the failure of the lord of Tenochtitlan. Yet other chroniclers blamed Montezuma for the Aztec defeat also, but still constructed their accusations from the dominant vantage. We believe the answer lies in the "conservatism" of the other chroniclers of the period, who are self-consciously traditional. Their writings emulate closely the textual structure of the "day-books" and the "yearbooks" that were recited in pre-conquest times, with pictographic codices as mnemonic aids, by prestigious specialists who learned their skills in the schools for nobles. These documents, written in the Roman alphabet by authors who were baptized Christians and often also officers of the Spanish government, even include in their conservative tendency the use of the Aztec calendrical system in their enumeration of the passing years. Hanks (1986) has pointed out the problems faced by Mesoamerican nobles, who derived their status, valuable under Spanish rule, from the rank system of the pre-conquest indigenous societies. Where the biographies of other chroniclers are known, we find that they claimed lordly rank. For instance Alvarado Tezoz6moc, author "of the Cr6nica mexicayotl [Chronicles of the Mexica] cited in (5), claimed to be a member of the Tenochtitlan royal family. Chimalpahin Cuauhtlehuanitzin, probable author of the passage cited in (6), claimed descent from the Colhua royal line and from the rulers of Cuauhtitlan (although he was probably a member of the lower nobility at best [Schroeder 1991]). If an important diacritic of high noble status in the post-conquest period was mastery of traditional genres (along with literacy in the Spanish style), then we would expect most chroniclers to use the conservative dominant vantage, in spite of environmental pressures towards attention to the distinctiveness. Sahagun gives no attention to the genealogies of the consultants for Book XII. He describes them as personas principales, translated by Dibble and Anderson as "principal persons". The Spanish term principal in fact meant "princely, of the nobility". We do not know what these narrators expected that Sahagun and his assistants would do with their dictations, but certainly their expectations were not the same as those of a chronicler like Tezozomoc or Chimalpahin, who wrote as individuals and signed their work, for the very reason that Book XII was obviously a collective project. While their narration exhibits sufficient properties of Aztec high language to dignify the topic, they might have felt little need to be "conservative" in their choice of representational strategies in order to further

314

Jane H Hill- Robert E. MacLaury

a claim of nobility, and so could succumb freely to any inclination to attend to distinctiveness. 36 Whatever their reasons, the important datum is that the Florentine Codex narrators found the words to talk about what happened fhtic 'inside' Montezuma. That they could do so at such length is evidence of a cognitive innovation, their recessive vantage of the category "person", constructed with attention to the distinctiveness of the fallen lord on the new ground of inside location.

5. Some implications of a vantage theory formalization of the category of the person Vantage theory addresses requirements for a general theory of person categories. In our view, one of the most powerful cases that we can make for the theory is to show how it accounts for intricate details of the rhetorical and grammatical systems of a specific text. We are aware of no other discussion of person categories that attempts such a detailed reconciliation of category theory with the results of discourse analysis. At the more general level, vantage theory provides a fonnal mechanism-the shift of cognitive attention from similarity to distinctivenessthat accounts for variation and change. Since this attention occurs at reciprocal strength on a sliding scale, different vantages of the "person" can occur at different points along its length, both cross-culturally and within local and individual categorial repertoires. Thus the theory allows for vantages to be split between contexts, a phenomenon repeatedly noted in the literature. The vantage theory account, like the Heelas- Lock theory, is universalist: it permits all coordinates to be cognitively available at all times to all categorizers. Unlike the Heelas- Lock theory, vantage theory provides a formal motivation for this availability: categories must include the several levels of concentration. "Zooming in" and "panning out" by shifting coordinates from the status of figure to the status of ground requires four coordinates in the case of a composite category (as in Figure 1). Coordinates can be present as ground, as figure, or as presupposition, thus permitting any coordinate to be "in the back of the mind" in any such act of categorization. This formal result accounts for an important ethnographic finding: that a dominant pattern of sociocentric person categorization can coexist with recognized zones of egocentricity, and vice versa. Shweder and Bourne (1984: 191) point out that Oriya people, who nor-

Aztec history, vantage theory, and the category of "person"

315

mally focus on exteriority, "do peer into one another's hearts and minds". Lutz (1987, 1988) observes that even though Ifaluk discourse characteristically locates emotions "outside" in the performance of role in social interaction, Ifaluk will say, when specifically asked, that emotions are located in people's "insides". We claim that this occurs because, even in cultures where the preferred discourse about persons manifests the coordinates of outside location and similarity as ground and figure respectively (the classic "sociocentric" category), the least preferred level of the zoom-in, with inside location as ground and attendance to distinctiveness as figure, will always be available should such close attention be forced. Where the dominant vantage locates the fixed coordinate at the most preferred level of attention internally to the self-other boundary, the extreme zoom-in is also available and will yield, by formal requirement, the opposite result. Thus "egocentric" Americans can understand deviant behavior as occurring when an "individual" succumbs to "peer pressure". This formal result addresses an important problem in ethnological theory. Universalists, including both Mauss (1938) and Hallowell (1960, 1971 ), and as recently argued in detail by Spiro (1993), assert that all human beings must, at some level, have self-awareness, with the corollary that they will believe that others have self-awareness as well. Universalists maintain that this is required both by theoretical necessity and the ethnographic record. Even relativist Richard Shweder has observed that one purpose of the study of exotic ethnopsychologies is to reveal to us our own "less conscious selves" (Shweder 1991: 108). This insight is constructively addressed by the formalization of multiple levels of awareness in the complex structure of categories. Any particular vantage may be the one favored in a wide variety of discourses and expressed through the most common forms of behavior, but the complex structure of a vantage makes all coordinates available should attention focus in and concern itself with distinctiveness. A shift in attention toward distinctiveness at the preferred level of concentration can bring new coordinates into full attention in a new recessive vantage, which may come to have coequal status as a complementary view, or even to outrank the original favored vantage. Like the Heelas- Lock theory, the vantage theory model predicts a small number of major types of person categories. To fill out the inventory of predicted types and seek ethnographic attestation for them would require a monograph. The theory obviously does not account for all diversity. However, it throws interesting diversity into strong relief. The

316

Jane H. Hill- Robert E. MacLaury

Berlin- Kay account of color categorization is often criticized for neglecting attested diversity. However, precisely its strong universalist claims invite ethnographic examination of salient exceptions, while permitting us to recognize "ordinary" cases. That the Hanun6o recognize a dimension of succulence that may supersede attention to hue (Conklin 1955) becomes a matter for scrutiny in new perspective, and not simply one more piece of ethnographic trivia in an undifferentiated relativist landscape. However, when yet another four-term color system without extraordinary complications is identified, we need not puzzle over the exotic local properties of mind from which it emerged, as ethnographers did for years before the Berlin- Kay breakthrough. Our proposal for person categories should similarly enable the distinction between truly exotic exceptional cases, and those forms of thought that are unmarked and ordinary in human societies. Like Heelas and Lock, we recognize that many details of person categorization must exhibit cultural specificity. By requiring the entailments of each level of concentration to be filled out with locally-appropriate symbolic material, we permit the required specificity. For instance, Montezuma's "inside" locus of will happens to have been located by the Book XII narrators, following local cultural logic, in his heart, but the ethnographic record of course attests to other organic and inorganic loci for such properties of persons-the brain, the skin, or the breath. An interesting question is whether these loci may also exhibit some systematicity and constraint, a question entertained by Lock (1981 ). Similarly, the Aztecs believed in the cosmic animating forces, while Babylonian astrologiers found controllers in the stars and modern psychiatrists look to the chemistry of the neurotransmitters. For the Aztecs (as for other Mesoamerican peoples), an uncontrolled human was a being of the chaotic periphery, the wilderness of unpredictable wild animals. In the medieval European tradition, deviants were consigned to the zone controlled by the forces of evil, who were all too orderly, deployed as demon armies led by their great general, the fallen angel Satan. Those interested in the coherence of metaphoric systems might do well to focus their search for diversity on these less constrained entailments, rather than continuing to belabor such dimensions as sociocentricity and egocentricity. For instance, Shweder and Bourne (1984) argued that person categories are best explained as epiphenomena of more general metaphorical systems in world view. They suggest that the Oriya person category, which is probably analyzable in our terms as a vantage constructed on the location ground "outside", with attendance to similarity as figure or mobile coordinate at the most preferred level, is produced as a logical

Aztec history, vantage

theory~

and the category of "person"

.J 1 '

corollary of a "holistic", "organic" view of the world. Clearly, a similar explanation is available for the Nahuatl perspective, one rooted in the theory of cosmic forces with a nexus in the living body. The dominant Aztec view looks very much like a case of what Shweder and Bourne call "organic holism". However, there are many problems with Shweder and Bourne's account. First, is not a person category type like "sociocentric organic" so broadly defined as to wash out many of the details that relativists hold to be fundamental? In vantage theory, the specification of a very few major coordinates is motivated, but this is not the case within relativist perspective. Second, given the many attested cases of "sociocentric organic" person categories, are these always, or usually, associated with "holistic-organic" world view? If this is the case, does not this coincidence suggest a need for general explanation? We certainly concur with Shweder and Bourne that relative coherence in systems of cultural knowledge is not surprising. However, a requirement of coherence creates serious problems for an account of change-as Shweder and Bourne point out (1984). Indeed, it creates a serious explanatory challenge for Shweder and Bourne, since almost 50 percent of their Oriya informants exhibit an "egocentric contractualist", not a "sociocentric organic", view of persons. To what world view might this minority subscribe? Attention to variability is central for us. Further, we require a satisfactory account of the details of representational strategies, in this case the structure of our text. We do not see how attention to the broader Nahuatl world view would help us account for fine details of discourse practice. This was only made possible by recognizing the formal properties of vantages. Many apparent complications and exceptions to the small list of coordinates that we have proposed may be accounted for by extension of elements of the basic category types through a short inventory of cognitive processes like those proposed by Lakoff (1987), such as image schema transformation, metonym, or metaphor. Other variants may be understood as involving the penetration of other category systems into person categorization. Careful exploration of these possibilities, using as evidence fine details of discourse and behavior, seems more likely to enlarge ethnographic understanding than is the labeling of global dichotomization as relativist thought. While permitting an account of change in person categories, vantage theory is in no way an "evolutionary" model. The shift in strength from similarity to distinctiveness propels movement through the continuum. But there is no inherent directionality between the outside and inside location coordinates, nor between attention to similarity and attention to distinctiveness. Thus the model is not "presentist" (Shweder and

318

Jane H. Hill- Robert E. Mac Laury

Bourne 1984: 163); it does not imply that a person category formed by attendance to similarity and outside location with an entailed external controller is somehow an imperfect approximation to one emphasizing distinctness and inside location, and entailing autonomy. Indeed, the emergence of outside location or the entailment of "control", often associated with "concrete" or "primitive" logic, is very familiar in contemporary society. A vantage composed from attention to similarity ("sociocentricity") as the figured coordinate against a ground of location "outside" tatemae, is well-known as the dominant person categorization of modern Japanese. 37 Obviously this category cannot be archaic. Instead, any group of human beings might construct this vantage, but the detailed symbolic realization of outside/similarity is distinctive depending on whether it is conducted by a Japanese or an Oriya, an Ifaluk or a Pintupi. Ethnography and history are charged with elucidating these details, but there is no contradiction between these local developments and the general principles of vantage theory. Vantage theory addresses three important concerns of relativist thinkers: the requirement of "domain specificity" (Shweder 1991), the requirement that category theory attend to the subjectivity and agency of cognizers, and the requirement that theory account for dialogical interaction between individual and culture. As to domain specificity, vantage theory does aim to be a very general account of human categorization. The analogy to the construction of a location in space-time is fundamental, and the cognitive attentions to distinctiveness and similarity may prove to be coordinates of all categorial vantages. However, the vantage theory account of any specific category must address those dimensions of human perception and cognition appropriate to it. Thus, perceptions of hue or judgements about brightness provide coordinates for color categorization. Outside or inside location relative to the self-other boundary, and the control entailments that account for activity and passivity, are specific to the categorizations of persons and selves. Other types of categorizing acts will presumably exploit their own domain-specific coordinates. As to the requirement of subjectivity, the categorizer in vantage theory is active, assuming a position in reference to selected coordinates, placing them in more than one arrangement, and variously attending to similarity and distinctiveness. Since only a thinking human being can arrange, select, and emphasize, the conceptualizer is actively tied to any category. The arrangements and emphases are cognitions, and are not bound by universals of perception. As to the requirement of dialogicality, choices of attention in particular interact closely with the external environment, its change or stability. Categorizers draw from their local semiotic envi-

Aztec history, vantage theory, and the category of "person"

319

ronment the specific stuff of meaning through which persons are understood: the willful heart, the sumptuous robes, the standards of eloquence. But in some particular context, thinkers may find the obvious options inadequate to their purposes. In that case, thinkers can recruit a new coordinate, perhaps from some remote domain.

6. Conclusion We interpret variation in the Aztec representations of rulers in historical chronicles in terms of vantage theory, a model originally developed to address change and variation in color categories. We thus treat differences in theme and content in discourse as reflections of different points of view. We find that the Aztec narrators of The History of the Conquest could represent rulers from both a dominant and a recessive vantage. They constructed their dominant vantage at the most general level of concentration by arranging a fixed coordinate or ground of "outside" location and a mutable coordinate or figure of attention to similarity. The recessive vantage is modeled by a fixed coordinate of "inside" location and a mutable coordinate of attention to distinctiveness. The two vantages yield contrasting Aztec views of "person" as manifested in the most salient kind of human being, the ruler or tlahtoani. From the dominant vantage, the ruler performs an externally visible social role whose prescription is invariable from one exemplar to the next; from the recessive vantage, the ruler traverses inscrutable interior feeling states whose sheer elaboration mark him as an individual without counterpart, and which entail his separation from cosmic power. We interpret the former vantage as dominant because it twice incorporates the coordinate of similarity; it is attested in portrayals of persons in a number of Sahaguntine texts, for instance, Florentine Codex Book VII Kings and Lords and Book X The People, as well as in the portrait of Cuauhtemoc in Book XII. It is found as well in the historical chronicles noted in note 25. The latter vantage is recessive for twice incorporating the coordinate of distinctiveness. It represents a rare minority strategy, which we have found only in Book XII among the historical chronicles. The textual representation of this vantage uses a variety of highlymarked lexical, rhetorical and grammatical devices which can be interpreted as icons for the polarization of this recessive vantage, used to represent the tlahtoani Montezuma. In overview, we hope to demonstrate the utility of vantage theory for the exploration of categorization in domains besides color. We argue that

320

Jane H. Hill- Robert E. MacLaury

the Heelas- Lock (1981) theory of indigenous psychologies can be augmented by an additional dimension of attendance to similarity and distinctiveness, and by the basic insight that a category is always at least one vantage or point of view, to provide a theory of universal constraints on person categories with greater explanatory power. Notes I. The exact contribution Sahagun himself made to the language of Book XII is not known. We concur with most recent authorities in judging the text to be in all significant respects the words of the Aztec narrators; our argument for this position is beyond the scope of the present paper. Since Sahagun states that he intended the history to "record the language of warfare and the weapons the natives use in it, in order that the terms and proper modes of expression for speaking on this subject in the Mexican language can be derived therefrom" (Dibble-Anderson, in Sahagun 1982: 101), the elaboration of vocabulary in the narrative may reflect this goal. Leon-Portilla (1974) considers some debates on authorship. For a recent review of the historical status of the work, see Cline (1988). 2. Moteuczoma means "He shows anger, frowns in a lordly manner". This Nahuatl word is quite difficult to pronounce in both Spanish (where the accepted equivalent is Moctezuma) and English (where is it usually Montezuma). Since neither is like [mote:kwso:ma:], we see no harm in using the ordinary English name for this historical figure. 3. Lucy (1992) criticizes color-research methods as imposing the English-language category of "color" on subjects of diverse linguistic and cultural backgrounds. While this may be an interesting point for some theoretical purposes, it is in fact the case that thousands of people speaking hundreds of languages from every region in the world have willingly, indeed, often enthusiastically, participated in color surveys using the methods developed by Berlin and Kay and their colleagues. We believe that it is fair to conclude that every human being has easy access to the distinctions of hue that English speakers use to categorize colors, even though the hue distinctions may not be exploited for local cultural purposes. Some languages name colors as bands of brightness or as combinations of brightness and hue (Mac Laury 1992; Casson- Gardner 1992). 4. The same criticisms apply to Rosch's (1973) accounts of color categories. 5. "Coextension" is one of a series of relational types distinguished within vantage theory. The relational types are phases on a continuum that is seen as speakers split composite categories, like "warm" or "cool", into basic categories like "red" and "yellow" or "green" and "blue". The sequence is as follows: near-synonymy ("two terms name one category"); coextension ("the range of one term is slightly larger and evenly distributed, whereas the other is smaller and skewed, yet the two ranges encompass each other's foci"), inclusion ("the skewed range has retracted to the colors surrounding only its focus, whereas the larger range retains its size and continues to cover both foci; either focus may be placed on the extreme outer edge of its range, or both foci may be 'polarized' in that way"); and complementation, where "both ranges have retracted to separate foci and neither range encompasses the other's focus" (MacLaury 1992: 141142).

Aztec history, vantage theory, and the category of"person"

321

6. Evidence is accumulating that many processes in human cognition may be fruitfully understood in terms of spatial metaphors. Lyons (1977, 2: 718) refers to this as the "localist" hypothesis. 7. "Warm" categories can be focused in either red or yellow; we illustrate a category focused in red. 8. For instance, we can imagine a speaker selecting "blue" as the mobile coordinate at this level. Such a selection is unattested in color research and is only possible in theory; however, some Salish languages extend a green focused term to cover yellow (Galloway 1993: 663, Figures g and i) while others extend the green focused category throughout blue (Mac Laury 1991 b: Figure 3). 9. Unique hues, such as RandY, cannot multiply, (as *R 2 or *Y 2), as they are physiologically fixed sensations and unenhanceable beyond threshold. Emphases on S and D can multiply as S2 and D 2 , as they are cognitive points on a sliding scale and, thus, are inherently mutable or "mobile". However, in the zooming process, inherently fixed coordinates, such as R or Y, can be treated as though they are mobile by shifting them to a figure position, while inherently mobile coordinate, S and D, can be treated as fixed by converting them to a ground. Yet in either position, the latter remain multiplicative due to their inherent scalarity; R and Y remain pegged to their thresholds. 10. See, for instance, Figure 3a in MacLaury (1992). MacLaury (1986, 1987, 199la, 1992, in press) has published a number of diagrams illustrating coextension by speakers of Mesoamerican languages. II. Technically, Lock (1981) states that these are the universal dimensions of "self conception", not of person categories. The two types of categorization are obviously closely related, presumably by projection of self-understanding onto the conceptualization of others, and the projection of private sensations into the public, social arena. We lack evidence on the Aztec concept of the "self', and are considering here only unreflecting and habitual representations of others. However, Lock's dimensions are appropriate to this task. 12. We are indebted to Joel Robbins for this suggestion. 13. Lock (1981) proposes an additional dimension of time, which he suggests is the source of self-continuity and hence of moral responsibility, wherever this is located. We do not exploit this dimension here. In the full account of vantage theory (MacLaury in press: chapter 6) emphases on similarity and distinctiveness are constructed an analogies to time as a function of relative motion; both are scales marked by absolute poles affording ample gradation of coordinate points between. 14. We cannot develop here a complete argument for the nature of this "local boundary". There is some evidence that it is designated by the root fx, the precise meaning of which is controversial (cf. Lopez Austin 1992). The root appears in terms for parts of the face, including the forehead and the eyes, and can apparently be extended to mean the front of the human body, and even the body surface in general (Lopez Austin 1980). It appears also in many expressions having to do with intelligence and perception. To present oneself before another person was to be rxco 'at his fx'. The fx element apparently designates a more abstract conception than does ehuatl, the literal, physical "skin". This latter term can also be used for the bark of trees, the peel of fruit, or for the hides of animals. 15. Handling "control" as an entailment of this level obviates the need for an independent control coordinate, as suggested by Heelas (1981) and Lock (1981). It may be that

322

16.

17.

18.

19.

20.

21.

22.

Jane H Hill- Robert E. MacLaury person categories in some other societies are best modeled using an independent control coordinate, however. Joyce Marcus (1992) points out that even a very unsuccessful tlahtoiini, the ruler Tizoc, had a commemorative stone carved to celebrate his military victories. Tizoc's battles resulted in the loss of more men than his warriors were able to capture, but these defeats produced captives whose very existence attested to Tizoc's qualities as a warrior, and who could be sacrificed to sacralize his assumption to the office of speaker. All Nahuatl passages given here are normalized to the transcription given in the analytical dictionary of Kartttunen (1983). Since published paleography is available for all passages cited, we preferred a normalized transcription that would ease the task of nonspecialist readers. Texts are formatted in lines according to the theory of Nahuatl ethnopoetics of Bright (1990). Parallel lines are aligned. Elements indended beyond the parallelism are simply continuations of lines. The English translations have been modified to reflect this ethnopoetic format. The original paleography and translations can be seen by consulting the sources cited for each text. Where only a translation is given, this is in the wording given in the cited source. Every adult man was expected to take up arms on order of the ruler and fo follow war leaders into battle. Even the reproductive power of women was assimilated to the warrior role; at the moment of birth, midwives gave war cries celebrating that the woman had been a brave warrior and taken a captive, the baby. Women dying in their first childbirth, like warriors dead in battle, went to the Sun's heaven. Further evidence that the ruler is an exemplary and central member of the category of persons comes from the metaphor of society as a great eagle. In this metaphor, the noble vassals and the common people are "the wings, the tail", while the ruler resides in "the hair place, the head place" (Maxwell-Hanson 1992). The "head" is very widely available in human languages as a metonym for the body, and can stand for it as its exemplary element. Dibble-Anderson translate, "a demon of the air, a demon". We propose "crawling demon" for the second mention, on the theory that the first element of coli!lectli is from co lot! "scorpion". Clendinnen (1991: 80), following an analysis by Thelma Sullivan, suggests that the voice of the ruler was sometimes thought to be the voice of the god Tezcatlipoca, "Lord of the Near, the Nigh". In Nahuatl, an order Verb Object Subject (saw, skin, Achitometl) is not highly marked, as in English (although it is somewhat unusual to spell out both subject and object with full nouns). The marked order in Nahuatl can be seen in example (2), where the subject, in tlahtoani 'the ruler', is in initial position. In order for the reader who does not know Nahuatl to follow the translation, we have attempted to keep the English translation in a line-by-line relationship to the Nahuatl original. However, a translation like 'Then Achitometl saw the skin of a person of his daughter' would more closely reflect the rhetorical tone of the original. The reader should keep this in mind. In fact, this particular text is ambiguous as to whether in Achitometl is the subject of the verb, or whether it is the possesor in a genitive phrase: in ichpoch in Achitometl means "the daughter of Achitometl". A Spanish chronicler, Fray Diego Duran (1965 [1581]), recounts the same episode. Duran gives a more elaborate representation of Achitometl's reaction: " ... vico al ques-

Aztec history, vantage theory, and the category of "person"

23. 24.

25.

26.

27.

28.

323

taba junto a! idolo sentado, vestido con el cuero de su hija, una co sa tan fea y orrenda, que cobrando grandisimo temor y espanto, solt6 el encensario que en las manos tenia, sali6 grandes voces y diciendo ... " [... he saw who was seated next to the idol, dressed in the skin of his daughter, a thing so hideous and terrible, that being seized by the greatest fear and horror, he dropped the incense burner that he had in his hands, and went out giving great shouts and saying ... ] (the speech that follows is almost identical to that given in example [4]). Duran's sense of the moment, like our own, required a somewhat greater elaboration of the ruler's inner state. His name, Cuauhtem6c, means "Descended in the manner of an eagle (cuauhtli)". Presumably the narrators meant that the cloak was all that remained of Cuauhtemoc's lordly battle garb: his feathered and jeweled headdress, his earrings, nose-piece, bracelets and anklets, the banners attached to his back, and his weapons would all have been stolen when he was captured. These include Alvarado Tezoz6moc (1975), Lehmann (1938), Lehmann-Kutscher (1958), Zimmermann (1963), and Zimmermann (1965). We have also reviewed Bierhorst's (1992a, 1992b) edition of the texts seen also in Lehmann (1938). It is possible that such representations may have occurred in dramatic performance or in poetry, but a review of Horcasitas (1974) for the former and Bierhorst (1985a) for the latter revealed no examples. It is possible that this elaboration is one of Sahagun's language lessons, in which he is emulating the repetitious style usually associated with accounts of heroism and piety; in a new domain, he is trying to fit in as much vocabulary as he can. However the elaboration and repetition is entirely consistent with the theoretical treatment that we propose here. Cincalco 'Corn Goddess's House', is both a literal cave in the rocky hill of Chapultepec and a mythical paradise, the western meeting point of earthly world and divine realm. Gillespie (1989) suggests that Montezuma's attempt to go to Cincalco resonates strongly in Aztec history: Cincalco is where the last lord of the Toltecs, Huemac, went to kill himself upon the fall of his kingdom. In some versions of the history, Huemac dies after losing a struggle with Quetzalcoatl (the figure with whom many Aztecs identified Cortez). Duran's Historia elaborates the story as follows: Montezuma repeatedly sends messengers to the cave to offer gifts to Huemac and request permission for the ruler to enter Cincalco and live there as Huemac's servant. When Montezuma has finally met the rigorous prescriptions for fasting and abstinence imposed by Huemac, he goes to the cave at night, drawn by a mystical light that illuminates the entire city. But a priest of the temple learns of the ruler's plans in a dream and speeds to intercept him. As the great light fades, the priest shames Montezuma into returning to his city. Note that in this version of the story, even though it was redacted by a Spanish chronicler, Montezuma is "controlled", his fate determined by forces outside him. All motivation emanates from immortal beings: Huemac and the mystic voice that speaks to the priest. Thus Montezuma in this version of the story is treated from the perspective of the dominant vantage, as a human being in an orderly world. The most famous expression of this point of view appears in Sahagun's Coloquios y doctrina cristiana [Colloquies and Christian doctrine], in which an Aztec priest tells the Franciscan missionaries: (i)

Ye rna ca timiquiclin, Ye mii ca tipolihufciin, Tel ca teteoh in omicqueh. (Sahagun 1986 [1524]: 149)

'May we not die, May we not disappear, Even though the gods have died.'

324

Jane H. Hill- Robert E. MacLaury

29. It is interesting that the passage in chapter 16 (Sahagun 1975: 45) that describes how the Spanish stared at Montezuma and carressed him is immediately followed by a list of the lords who accompanied Montezuma to the climactic meeting. And, states the text, "When Montezuma was made captive, they not only hid themselves, took refuge, [but] abandoned him in anger". The Nahuatl verb is qui-tlahuel-cauhqueh, "him-angrilythey left". This is an additional realization of the entailments of the recessive vantage: other great lords disassociate themselves from the deviant king. 30. Andrews (1975) considers the-ca number suffix; on his account the "zero marking" of the Nahuatl preterite is thus even more apparent. 31. One possibility is that these lines, since they are very colloquial, represent a "popular" voice, the mob's evaluation of Montezuma's situation. We prefer to understand them as an inner voice which is colloquial in accordance with the entailment "common" of the arrangement OS = D 2-S, but the popular voice would also be unique in Aztec literature. 32. These lines use a very common metaphor of the battlefield as a flowery garden. Their literal sense is that Cuauhtemoc was willing to drown his city in the blood of his fallen warriors. 33. McAfee- Barlow (1945) translate this phrase beginning with ma with a positive meaning, "All is anger toward the Mexicans". We translate it in the negative "admonitive" sense (Andrews 1975: 56), which seems to make better sense. 34. MacLaury (1991 a) examines the impact of the construction of the Panamerican Highway through a Tzotzil-speaking community, comparing color categorization there to that in another town lying off the road. 35. Very rapid change in representation of persons in the context of catastrophic acculturation can be found elsewhere in Native American texts. Ellen Basso (1987) has studied Kalapalo stories in which the heroes are great warriors, "bowmasters", trained from youth to defend their people in the dangerous time when the advance of the whites disrupted traditional relations among Amazonian peoples. Reflecting from a vantage point of deep pacifism and a horror of violence which dominates present-day Kala palo, story-tellers devote autobiographical intensity to the inner states of the bowmasters as they make their murderous decisions, and concentrate on the bowmasters' development through complex sequences of human relations. These representations contrast with the stereotypical behavior of other mythical figures in Kalapalo stories. 36. One line of speculation about why the recessive vantage a pears in Book XII addresses the interaction with Sahagun himself. As a Christian missionary, he might have been interested in an "inner" account of Montezuma's failure and pressed the narrators for such. Or, in his editing of their materials, he might have unconsciously imposed a vantage based on inside location and distinctiveness. This speculation receives little support, for in Sahagun's own 1585 Spanish version of the history of the conquest, based more loosely on the original narration than the Florentine Codex history, the portrait of Montezuma is considerably attenuated, with little attention to the ruler's inner states (Sahagun 1989 [1585]). Cline (1988, 1989) observes that Cortez, a shadowy figure in Book XII, emerges in sharp focus in the 1585 manuscript. 37. But note that tatemae, the carefully managed exteriority relevant for sociocentricity, is complemented by h6nne, the internal private realm relevant to an egocentric perspective (Kondo 1990; cf. Matsuki, in this volume). The two perspectives may be complementary vantages, not a dominant and recessive vantage as in the Aztec case. We have illustrated here how discourse analysis provides a way to address this question.

Aztec history, vantage theory, and the category of "person"

325

References Alvarado Tezoz6moc, Fernando 1975 Cr6nica mexicayotl. Mexico: Instituto Nacional de Antropologia e Historia/ [c. 1609] Universidad Nacional Aut6noma de Mexico. Andrews, J. Richard 1975 Introduction to classical Nahuatl. Austin: University of Texas Press. Banfield, Ann 1982 Unspeakable sentences. London: Routledge and Kegan Paul. Basso, Ellen Changing ideology in Kalapalo warrior biographies. [Paper presented to the 1987 Annual Meeting of the American Anthropological Association, Chicago, IL.] Berlin,Brent-Paul Kay 1969 Basic color terms. Berkeley. University of California Press. Bierhorst, John 1985a Cantares Mexicanos: Songs of the Aztecs. Stanford: Stanford University Press. 1985b A Nahuatl-English dictionary and concordance to the Cantares Mexicanos. Stanford: Stanford University Press. History and mythology of the Aztecs, the Codex Chimalpopoca. Tucson: Uni1992a versity of Arizona Press. Codex Chimalpopoca, the text in Nahuatl with glossary and grammatical 1992b notes. Tucson: University of Arizona Press. Bright, William 1990 "'With one lip, with two lips': Parallelism in Nahuatl", Language 66: 437452. Casson, Ronald- Peter Gardner 1992 "On brightness and color categories: Additional data", Current Anthropology 33: 395-399. Clendinnen, Inga 1991 Aztecs. Cambridge: Cambridge University Press. Cline, Susan L. 1988 "Revisionist conquest history. Sahagun's revised Book XII", in: J. J. Klor de Alva- H. B. Nicholson- E. Quinones Keber (eds. ), The Work of Bernardino de Sahagun. (Studies on Culture and Society, Volume 2.) Institute for Mesoamerican Studies of SUNY Albany. Austin: University of Texas Press, 93106. 1989 "Introduction", in: Fray Bernardino de Sahagun, Conquest of New Spain, (1585 revision.) Salt Lake City: University of Utah Press, 1-21. Conklin, Harold 1955 "I:Ianun6o color categories", Southwestern Journal of Anthropology 11: 339354. Cortes, Hernan 1986 Letters from Mexico. Translated and edited by Anthony Pagden. New Haven: [1519-1520]Yale University Press. Duran, Fray Diego 1965 [1581] Historia de las Indias de Nueva-Espafia e Islas de Tierre Firme, Torno I. Mexico: Editora Nacional.

326

Jane H. Hill- Robert E. MacLaury

Galloway, Brent D. 1993 A Grammar of Upriver Halkomelem. (University of California Publications in Linguistics 96.) Berkeley: University of California Press. Gillespie, Susan D. 1989 The Aztec Kings. Tucson: University of Arizona Press. Goffman, Erving 1974 Frame Analysis. New York: Harper and Row. Haiman, John (ed.) 1985 !conicity in syntax. Philadelphia: John Benjamins. Hallowell, A. Irving 1960 "Self, society, and culture in phylogenetic perspective", in: S. Tax (ed.), Evolution after Darwin. Vol2. Chicago: University of Chicago Press, 309-372. 1971 Culture and experience. 2nd edition.) Philadelphia: University of Pennsylvania Press. Hanks, William 1986 "Authenticity and ambivalence in the text: A colonial Maya case", American Ethnologist 13: 721-744. Heelas, Paul 1981 "The model applied: Anthropology and indigenous psychologies", in: Paul Heelas-Andrew Lock (eds.), Indigenous Psychologies, London: Academic Press, 41-63. Hill, Jane H. 1990 "Weeping as a metasignal in a Mexicano woman's narrative', in: E. B. Basso (ed.), Native Latin American cultures through their discourse, (Special Publication of the Folklore Institute, Indiana University.) Bloomington, In: Folklore Institute, Indiana University, 29-50. Horcasitas, Fernando 1974 El teatro nahuatl. Mexico: Instituto Nacional de Antropologia e Historia/ Universidad Nacional Aut6noma de Mexico. Karttunen, Frances 1983 An analytical dictionary of Nahuatl. Austin: University of Texas Press. Karttunen, Frances-James Lockhart 1987 The art of Nahuatl speech: The Bancroft dialogues. (Nahuatl Studies Series No. 2.) Los Angeles: UCLA Latin American Studies Center. Kay, Paul 1975 "Synchronic variability and diachronic change in basic color terms", Language in Society 4: 257-270. Kay, Paul-Chad McDaniel 1978 "The linguistic significance of basic color terms", Language 54: 610-646. Klor de Alva, Jorge 1987 Telling lives: Confessional autobiography and the reconstruction of the Nahua self. [Paper presented to the Annual Meeting of the American Anthropological Association, Chicago, IL.] Kondo, Dorinne 1990 Crafting selves; Power, gender, and discourses of identity in a Japanese workplace. Chicago: University of Chicago Press. Labov, William 1972 "The transformation of experience in narrative syntax", in Language in the inner city, Philadelphia: University of Pennsylvania Press, 354--396.

Aztec history, vantage theory, and the category of "person"

327

Labov, William-Joshua Waletzky 1967 "Narrative analysis, in: June Helm (ed.), Essays on the verbal and visual arts, Seattle: University of Washington Press, 12--44. Lakoff, George 1987 Women, fire, and dangerous things. Chicago: University of Chicago Press. Langacker, Ronald 1987 Cognitive grammar: An introduction. Stanford Stanford University Press. Larson, Mildred 1978 The function of reported speech in discourse. Dallas: Summer Institute of Linguistics. Leech, Geoffrey 1978 "Natural language as metalanguage: An approach to some problems in the semantic description of English", Transactions of the Philological Society of London 1976-1977: 1-31. Lehmann, Walter 1938 Die Geschichte der Konigreiche von Colhuacan und Mexico. Stuttgart: W. Kohlhammer. Lehmann, Walter-Gerdt Kutscher 1958 Das Memorial Breve acerca de Ia Fundacion de Ia Ciudad de Culhuacan (Chimalpahin). Stuttgart. W. Kohlhammer. Leon-Portilla, Miguel 1962 The broken spears. Boston: Beacon Press. 1974 "The problematics of Sahagun: Certain topics needing investigation", in: Munro S. Edmonson (ed.), Sixteenth century Mexico. Albuquerque: University of New Mexico Press, 235-255. Li, Charles N. 1986 "Direct and indirect speech: A functional study", in: F. Coulmas (ed.), Direct and indirect speech. Berlin: Mouton de Gruyter, 29--46. Linde, Charlotte 1978 "The organization of discourse", in: T. Shopen-J. M. Williams (eds.), Style and variables in English, Cambridge, MA: Winthrop Publishers, 84--114. Lock, Andrew 1981 "Universals in human conception", in: Paul Heelas-A. Lock (eds.), Indigenous psychologies, London: Academic Press, 19-36. Longacre, Robert 1976 An anatomy of speech notions. Lisse. Peter de Ridder. Lopez Austin, Alfredo 1980 Cuerpo humano e ideologia. Mexico: Universidad Nacional Autonoma de Mexico. 1992 "Cuerpos y rostros", Historicas 34: 35--46. Lucy, John A. 1992 Linguistic diversity and thought. Cambridge: Cambridge University Press. Lutz, Catherine A. 1987 "Goals, events, and understandings in Ifaluk emotion theory", in: Dorothy Holland- Naomi Quinn (eds.), Language and cultural knowledge. Cambridge: Cambridge University Press, 290-312. 1988 Unnatural emotions. Chicago: University of Chicago Press. Lyons, John 1977 Semantics. Vol. 2. Cambridge: Cambridge University Press.

328

Jane H. Hill- Robert E. MacLaury

MacLaury, Robert E. 1986 Color in Mesoamerica. Vol. I. A theory of composite categorization [Ph.D. dissertation in anthropology, University of California at Berkeley.] 1987 "Coextensive semantic ranges: Different names for distinct vantages of one category, in: Barbara Need-Eve Schiller-Anna Bosch (eds.), Papers from the twenty-third annual meeting of the Chicago Linguistics Society. Chicago: Chicago Linguistics Society, 268-282. 1989 "Zapotec body part locatives: Prototypes and metaphoric extensions", International Journal of American Linguistics 55: 19-54. "Social and cognitive motivations of change: Measuring variability in color 199la semantics", Language 67: 34--62. 199lb "Exotic color categories: Linguistic relativity to what extent?", Journal of Linguistic Anthropology 1: 26-51. "From brightness to hue: An explanatory model of color category evolution", 1992 Current Anthropology 33: 137-186. to appear The universal pattern of coextensive color naming: Categorizing by analogy to points of view in space. [Manuscsript submitted to Behavioral and Brain in press

Sciences.] Color and cognition in Mesoamerican languages: Constructing categories as vantages. Austin: University of Texas Press.

McAfee, Byron- Robert Barlow 1945 "Anales de Ia conquista de Tlatelolco en 1473 yen 1521", Academia mexicana de Ia Historia, Memorias 4(3): 326--339. Marcus, Joyce 1992 Mesoamerican writing systems. Princeton: Princeton University Press. Mauss, Marcel 1985 [1938) "A category of the human mind: The notion of person; the notion of self', in: Michael Carrithers-Steven Collins-Steven Lukes (eds.), The category of the person, Cambridge: Cambridge University Press, 1-25. Maxwell, Judith-Craig Hanson 1992 On the manners of speaking that the old ones had: The metaphors of Andres de Olmos in the TULAL manuscript. Salt Lake city: University of Utah Press. Medin, Douglas L.- Robert L. Goldstone- Deirdre Gentner "Similarity involving attributes and relations: Judgments of similarity and 1990 difference are not inverses", Psychological Science 1: 64-69. Rosch, Eleanor 1973 "Natural categories", Cognitive Psychology 4: 328-350. Sahagun, Fray Bernardino de 1954 Florentine Codex Book VIII: Kings and Lords. Charles E. Dibble-Arthur J. 0. Anderson (translators). Salt Lake City: University of Utah Press. 1961 Florentine Codex Book X: The People. Charles E. Dibble-Arthur J. 0. Anderson (translators). Salt Lake City: University of Utah Press. 1969 Florentine Codex Book VI: The Sayings of the Elders. Charles E. DibbleArthor J. 0. Anderson (translators). Salt Lake City: University of Utah Press. 1975 Florentine Codex Book XII: The History of the Conquest. Charles E. DibbleArthur J. 0. Anderson (translators). Salt Lake City: University of Utah Press. 1982 Florentine Codex: Introductions and Indices. Charles E. Dibble-Arthur J. 0. Anderson (translators). Salt Lake City: University of Utah Press.

Aztec history, vantage theory, and the category of "person"

329

1986 [1524) Coloquios y doctrina cristiana. M. Leon-Portilla (editor and translator). Mexico: Fundaci6n de Investigaciones Sociales/Universidad Nacional Aut6noma de Mexico. 1989 [1585) Conquest of New Spain. ( 1585 revision; translated by Howard F. Cline.) Salt Lake City: University of Utah Press. Schroeder, Susan 1991 Chimalpahin and the Kingdoms of Chalco. Tucson: University of Arizona Press. Shweder, Richard A. 1991 "Cultural psychology: What is it?", in Thinking through cultures, Cambridge: Harvard University Press, 73-!10. Shweder, Richard A.- Edmund 1. Bourne "Does the concept of the person vary cross-culturally?" in: Richard A. 1984 Shweder- Robert A. Levine (eds.), Culture theory, Cambridge: Cambridge University Press, 158-199. Spiro, Melford E. 1993 "Is the western conception of the self 'peculiar' within the context of world cultures?", Ethos 21: 107-153. Weeks, Kent "Art, word, and the Egyptian world view", in: Kent R. Weeks (ed.), Egyptol1979 ogy and the social sciences: Five studies, Cairo: American University in Cairo Press, 59-81. Zimmermann, Gunter 1963 Die Relationen Chimalpahins zur Geschichte Mexicos. Part I. Hamburg: Cram, de Gruyter. 1965 Die Relationen Chimalpahins zur Geschichte Mexicos. Part 2. Hamburg: Cram, de Gruyter.

Selection of Japanese categories during social interaction 1 Munekazu H. Aoyagi

But schemata ... cannot simply refer to knowledge of the physical world. In fact I would argue that cognitive approach to discourse must build on interaction J. J. Gumperz

1. Introduction As part of the study of language as "the construal of the world", the

present paper discusses the applied significance of meaningful organization in understanding social interaction. Here, I intend to show that categorization processes are themselves social strategies by which a social individual contributes to the incessant constitution and maintenance of social reality. Using vantage theory as an organizing framework, and Japanese language and culture as the data source, I will discuss the cognitive organizations that give rise to nihon-go no kotobazukai 'styles of Japanese speech'. Defined as "the relationship between a speaker (or hearer) and a situation that s/he conceptualizes and portrays", the concept of "construal" offers a promising approach in understanding the role of language as constitutive practice (Langacker 1987; Taylor, this volume). This concept suggests the contribution that subjects' knowledge and judgments in specific contexts of interaction make to the creation of meanings and categories. This knowledge and these judgments are further influenced by positions, backgrounds, interests, and assumptions that the individuals have as members of a society. Thus, meanings and categories invoke diverse perspectives, or what postmodernists would call "multivocality'. Variable reapplication of these perspectives to social practices may also lead to gradual social change. In this respect, the construal perspective efficiently explains social dynamics and complexities. Based on this construal approach, the present study attempts to bridge the gap between an individual's subjective view and what can be considered as objective knowledge, or the gap between cognitive and sociologi-

332

Munekazu H. Aoyagi

cal dimensions of meaning. As Shore (1991: 21) contends, a study of such comprehensive mental representations should consider double movements where cultural forms are personalized at the very moment that personalized experience is conventionalized. These movements "suggest an impulse toward reproducing the subject's physical experience of the world in the mental representation of conventional sign-forms" (Shore 1991: 15). Shore's framework is substantiated by vantage theory and its mechanisms as well as processes of meaningful construction: spatial analogy; vantages and coordinates; zooming processes; and the shift of coordinates (MacLaury 1987, 1989; Aoyagi 1989). The following case that I will discuss exemplifies the efficiency of the model. In brief, the subject's experience of his position in the world is reproduced in the mental space as personalized convention with respect to vantages and coordinates. By zooming as well as shifting processes, one can edit one's mental representation, allowing the representation to change flexibly according to different situations. At the same time, this mental representation serves as the source of communicative practice with other members of the society that conventionalizes one's action. Seen in this light, the cognitive organizations that produce nihon-go no kotobazukai consist of a dynamic, two-way process which is of interest because it shows the impact of the personal viewpoint that the Japanese have upon their linguistic choices. As I will discuss below, the viewpoint is constituted by the way a Japanese individual perceives personal and situational information to which he is exposed. In the past, scholars of cognitive science have proposed a variety of semantic models (Brown-Gilman 1972; Minami 1974; Uno 1985; Brown- Levinson 1987). But these models are less favored because they neglect the contribution that viewpoint makes to the creation of meaning.

2. Basic terms, definitions, and data source Nihon-go no kotobazukai consists of word categories that range from polite to vulgar expressions. They are based on social criteria of power, distance and formality. By power, I mean the ability and authority an interacting agent exercises over a given situation. Distance refers to the extent of familiarity between interacting agents. Formality is any demeanor from tense to relaxed that interactive settings allow. The continuum of formality corresponds loosely to the poles of public-formal versus private-casual milieux. As I will show, a Japanese speaker bases language choice on his evaluation of a hearer in reference to power and distance. The

Japanese categories during social interaction

333

speaker will, then, compare power and distance with formality. The details concerning the contents, the systematic relationships, and the dynamic processes of these categories are developed on the basis of vantage theory. The models presented below are constructed from interviews with five Japanese participants. Examples drawn from daily conversations, drama, and documents support the data collected from these interviews. Some words and metaphoric expressions provide further clues to the structure and processes of categorization. It should be noted that the conversations in this study refer to the so-called hyoojun-go, or "standard Japanese". Dialect differences will not be treated here even though they prevail in Japan.

3. Honorifics and cognitive models Japanese, like Arabic, Javanese, and other languages, is known for delicate levels of honorifics that range from polite to vulgar. While many studies regarding styles of Japanese speech are available (e. g., Minami 1974; Uno 1985), only a handful of them mention the mental processes by which these distinctive expressions are achieved. Most of these sources simply categorize honorifics and describe occasions in which they are used, or are expected to be used. Brown and Levinson (1987) go beyond such simple descriptions by providing a semantic model that incorporates language of politeness. The model represents a system that is composed of a series of binary choices of ideal expressive behavior. This system operates on the assumption that any expression that the speaker produces during interaction with the hearer can be disturbing, or "face threatening", to the hearer. The speaker, therefore, chooses from an array of behavioral categories that may or may not minimize the threat. "Politeness" is the end product of such selection processes. Brown and Levinson contend that the first choice the speaker makes from an array of behavioral categories is either "on record", clarifying the reason for taking an action, or "off record", having more than one unambiguously attributable intention for taking an action (1987: 68, 69). Within the category of "on record", the options of "acting baldly without redress" and taking a "redressive action" coincide (1987: 69). Regarding "redressive action", both "positive politeness", by which the speaker treats the hearer as a member of his ingroup or as a friend, and "negative politeness", where the speaker does not interfere with the hearer's freedom to act, exist (1987: 70). The choices go on.

334

Munekazu H. Aoyagi

This framework is problematic because it confines the mental process to a prefixed set of codes. It does not determine whether people would actually incorporate this rigid, mechanical framework into the production of their expressions. I contend that humans can associate linguistic, social, environmental, and personal information with their values and perspectives more flexibly. This is where observation regarding the impact of personal viewpoint on meaning has its significance. Here, vantage theory provides an efficient explanation.

4. Vantages and coordinates Some of the significant aspects of vantage theory include spatial analogy, zooming processes, shifts of figure and ground, and behavioral entailments. First, the theory holds that categories are vantages that are constituted by analogy with personal experience in a physical space (MacLaury 1987). In such experience, a figure is constructed against a ground, or a fixed spatial coordinate. Various perspectives of the figure are projected by moving or rotating the figure on the ground along an axis. Vantages are edited images that result from the projection. Power and distance are two referential coordinates for personal evaluation. Power is given by an asymmetrical relationship between speaker and hearer. Once established, it cannot be altered either according to different situations or by individual preference. This makes power a fixed coordinate or "ground". Distance is, on the other hand, preferred by a symmetrical relationship between speaker and hearer which can be altered. Distance functions, therefore, as a mobile coordinate or "figure". The speaker constitutes a vantage of personal evaluation regarding the hearer in reference to both power and distance in any interaction. Second, the theory postulates the processes of "zooming in" and "zooming out" (MacLaury 1989). By zooming in, the initial figure becomes a ground and a new figure is adopted. Zooming out is the reverse process. The ground becomes a figure which then stands against a new ground of wider scale. Imagine a flower vase on a table. Initially, the vase is a figure and the table is a ground. By zooming in, one can look into the vase, where the vase becomes a ground in which the figure of a flower is located. By zooming out, the view may move from the table on which the vase is located to the room in which the table is located. A new ground of room is introduced, and the table becomes a figure. Through

Japanese categories during social interaction

335

a series of zooms, vantages are constructed and linked with each other in narrow and broad scopes. These processes play their roles in analyzing the contents of power, distance, and formality (e. g., the factors of age in power, acquaintance in distance, and, say, extreme formality). Third, the theory considers shifts of figure and ground (Aoyagi 1989; MacLaury 1989). A figure can be transposed from one ground to another, creating a new vantage; or one ground can have different figures, producing different vantages. In the "flower in the vase" example, one may shift the ground from the vase to the garden, composing different images with the same flower. The vase can also contain a cactus instead of the flower, creating different mental sketches. Finally, the theory argues that the particular expression is selected as an entailment of the mental processes (MacLaury 1987, 1989). In this sense, the particular hearer's expression that is selected from styles of Japanese speech is considered an entailment of the figure-to-ground manipulations in which personal and situational evaluations are made.

5. The criteria for the selection of styles of spoken Japanese Styles of Japanese speech are characterized by changes in the form of words and word endings, copula and affixes. For example, the set of verbs "to eat" in its second or third person form ranges from otabe ni naru or omeshiagari ni naru (very polite), taberareru (polite), taberu (plain), to kuu (vulgar). The stem of these verbs tabe can constitute different compounds with respect to the different levels of politeness. Similarly, the first person verb "to eat" can alter between itadakimasu (very polite), tabemasu (polite), taberu (plain), and kuu (vulgar). Another example is the auxiliary verb masu that, attached to verbs, functions as a distance marker. Thus, if one uses masu in the phrase "I go", it becomes a distant expression. If one does not use masu, it is rather direct: (1)

a.

b.

Polite: Watakushi I Direct: Ore I 'I go.'

wa (particle)

ikimasu. go

wa (particle)

iku. go

The copula de gozaimasu (polite-distal), desu (distal), and da (direct) that mean "to be" also mark styles of Japanese speech. These are attached to

336

Munekazu H. Aoyagi

the end of a sentence, to imply different levels of formality. Consider the following forms of the sentence "I am twenty-six": 2

a.

b.

c.

Polite distal: Watakushi wa (particle) I Distal: Watashi wa I (particle) Direct: Ore wa I (particle) 'I am twenty-six.'

nijuuroku twenty-six

de gozaimasu. (copula)

nijuuroku twenty-six

desu. (copula)

nijuuroku twenty-six

da. (copula)

These examples also indicate that one can alter the first person pronoun "I" from watakushi (formal), watashi (plain), to ore. Ore is very casual and is normally used by males only. I contend that power, distance, and formality are semantic criteria that, fashioned after actual human relationships, affect these word choices in an interaction: power is determined by the asymmetrical relationship between speaker and hearer; distance by the symmetrical relationship; and formality by the degree of formality in which the interaction takes place. Brown and Gilman's (1972) [1960] study of pronouns of address in some European languages provide support for this hypothesis. They observe that two forms of such pronouns, formal and casual, exist in such languages as French, Spanish, and Italian. Power and distance determine these variable usages. Selection among the formal-casual variables depends on the speaker's level of power relative to the hearer, and his familiarity (distance) with the hearer. So the speaker will use a casual pronoun, such as tu in French, for a familiar colleague, and a polite pronoun, vous, for an unfamiliar senior. For Japanese, the power criterion takes a form of the so-called "ranking consciousness". As Nakane (1970: 26) contends, this consciousness is deeply rooted in Japanese social behavior: A Japanese finds his world clearly divided into three categories, senpai [seniors], koohai [juniors], and dooryo. Dooryo, meaning "one's colleagues", refers only to those with the same rank, not to all who do the same type of work in the same office or in the same shop floor; even among dooryo, differences in age, year of entry or of graduation from school or college contribute to a sense of senpai and koohai.

Once the rank is established on the basis of seniority, it is applied to all circumstances, controlling social life and individual activity to a great

Japanese categories during social interaction

337

extent (1970: 29). In terms of vantage theory, the rigidity of power contributes to the speaker's recognition of power as a fixed coordinate. Senpai, dooryoo, and koohai, as explained by Nakane, are themselves metaphors of the asymmetrical relationship. The literal translation of senpai is "one who is ahead", dooryoo is "one who is at the same level", and koohai is "one who is behind". Words that are related to these suggest that Japanese perceive power on the basis of imaginary height. Senpai is synonymous with meue (literally) 'above one's eye level', jooshi 'person at the upper level', and uwayaku 'who one plays the higher role'. On the other hand, koohai is associated with meshita 'below one's eye level', karyoo 'person of lower rank', and shitayaku 'one who plays the lower role'. Doohai is a synonym of dooryoo meaning "person of equal rank". The metaphor of height is further supported by the term jooge-kankei 'upand-down relationships'. Nakane discusses further the behavioral outcomes of this asymmetrical relationship: In everyday affairs a man who has no awareness of relative rank is not able to speak or even sit and eat. When speaking, he is expected always to be ready with differentiated, delicate degrees of honorific expressions appropriate to the rank order between himself and the person he addresses. The expressions and the manner appropriate to the superior are never to be used to an inferior. Even among colleagues, it is only possible to dispense with honorifics when both parties are very intimate friends. (Nakane 1970: 30)

Nakane (1970: 27) also gives attention to pronouns of address as a linguistic example. A suffix san (Mr./Mrs.) accompanies an individual name, as in Tanaka-san, if that individual is a senpai. Kun replaces san when the individual is a koohai. In case of a dooryoo, no suffix is attached. Emphasis on ranking consciousness does not, however, mean that it alone influences Japanese behavior. Yoneyama (1976: 37) introduces another equally influential criterion, distance. Distance is a symmetrical relationship that consists of three personal categories: miuchi, families and relatives; nakama, companions; and tanin, strangers. Miuchi are "insiders" with whom the speaker closely associates, while tanin are "outsiders" with whom the speaker does not establish a close relationship. Nakama refers to an ambiguous group of people who stand between miuchi and tanin. During an interaction, the speaker applies one of these categories to the hearer depending on the degree of familiarity he recognizes between himself and the hearer. Thus, distance signifies imaginary distance. This perception is supported by the metaphorical term shinso-kankei 'closeand-far relationships'.

338

Munekazu H. Aoyagi

The speaker's selection of a personal category based on distance also has behavioral consequences. For example, the speaker may act unreservedly with miuchi and nakama, exercising hadaka no tsukiai, literally "naked relationships". However, he is likely to act reservedly with tanin. Linguistically, a possessive pronoun watashi no, 'my', may precede miuchi, as in watashi no miuchi 'my relative', and nakama in watashi no nakama 'my companions', but not tanin. Thus, miuchi and nakama imply those who are close enough for the speaker to treat them as his possessions. The speaker is also likely to speak causally with his miuchi and nakama, while he may speak cautiously with tanin. In terms of vantage theory, distance is a mobile coordinate. It can change with respect to time, contexts of interaction, and personal preference. The speaker who originally regards the hearer as a tanin may eventually consider her as a nakama. Conversely, the speaker who associates with the hearer as a miuchi may eventually treat her as a tanin. Formality is another criterion that affects the selection of styles of Japanese speech. Formality is represented in public contexts such as official interactions and the political arena. Here, the speaker is expected to behave considerately toward others, while the others have the initiative of action. Lebra (1976) calls such a behavioral expectation "social preoccupation" and "empathy". The speaker, suppressing his desire to commit willful acts, must perform in accordance with prescribed customs and norms. This stylized behavior is translated in Japanese as keeshikibatta, literally "form-stressing". Overwhelmed by prescribed modes of action, the public-formal contexts of interaction often produce tension between speaker and hearer. They can thus kataku naru, 'become hard', in such a context. Informality is, on the other hand, represented by private-personal contexts. Here, the speaker may release herself from the rules of behavior, although it is incorrect to say that he will not exercise any consideration toward others. While formality emphasizes empathy, informality focuses on autonomy. The formal-informal distinction is associated with the further Japanese notions of tatemae and honne. Honne has to do with one's inner wishes, while tatemae with cultural standards. In a formal context, expressing honne is inappropriate. In an informal context, it does not matter (Lebra 1976: 136; Matsuki 1989: 10, 11). Linguistically, formality leads the speaker to choose between speaking politely and impolitely. An example of this is shown in Conversation A (A.l-A.8) in the Appendix with the verb taberu 'to eat'. Its imperative form changes across different situations of formality, although it is applied to the same hearer.

Japanese categories during social interaction

339

6. The process of the selection of styles in Japanese speech Based on vantage theory, the process of style choice can be perceived as a network of coordinate configurations that are arranged in several phases. The speaker constructs a vantage, or a personal viewpoint, in each phase. Figures 1 to 5 will guide through one of the possible dynamics of this process. The total overview of this process is summarized as follows: Figure 1: The speaker determines the content for power. Figure 2: The speaker determines the content for distance. Figure 3: The speaker evaluates the hearer on the basis of the selected contents for power and distance. Figure 4: The speaker evaluates formality and incorporates this situational evaluation with his personal evaluation of the hearer. Figure 5: Associating the personal-situational evaluation with linguistic knowledge, the speaker selects an expression. Figure 6: Overview of Figures 1-5. I will elaborate upon this process on the basis of my data. The numbers in parentheses in the following discussion refer to interviews, conversations, and figures in the Appendix.

6.1 Establishing power Phase I (Figure 1a): In interacting with the hearer, the speaker must determine the contents for power and distance (4.7-4.9). Age, position, status, achievement, skills, and so on are subcriteria of power (1.2, 2.1, 4.4). These criteria range from inherent to achieved (2.1). To characterized power, the speaker makes power a ground in which the range of characteristics of subcriteria stands as a figure. Phase II (Figures 1b to 1d): Zooming in to the figure of the initial configuration of Phase I, the speaker classifies the subcriteria according to which portion in the range of characteristics, between inherent and achieved, they belong. Here, each subcriterion is a figure that stands against the ground, the range of characteristics. For example, age is inherent, and skills are achieved (2.1 ). Phase III (Figure le): In order for the speaker to determine which subcriterion is the most important for power in a given context, he must rank the criteria (1. 7, 1.8, 2.2, 2.3). The speaker accomplishes this by

340

Munekazu H. Aoyagi

zooming in to the figure in the configuration of Phase II. Ranking by priority becomes a figure for the ground subcriterion. Phase IV (Figures lf to lh): Zooming in to the figure in the configuration of Phase III, the speaker locates each subcriterion as a figure in the ground, the selected ranking of priority. Here, the speaker determines the ranks of the criteria he considered: e. g., position is prior, age is secondary, skills are tertiary. These evaluations are subject to situational change, and more than one subcriterion may become equally important as another (1.6-1.8, 2.1-2.3).

Characteristics

c

b

~~

zo

Ranking

d

+

Prior

e

Subcriteria -Age -Position ·Skill

Final

411111

Figure

. Aoom out zoomm/t

Rank

Figure I. Determining the power

Ground

::::::::7 ~ . _..

...

t

Rank

· ,.

Rank

Japanese categories during social interaction

341

6.2 Establishing distance Phase I (Figure 2a): The content of distance is determined in a way similar to that used for determining the content of power. Subcriteria of distance include kinship, membership, acquaintance, years of relationship, the depth of the speaker's knowledge about the hearer, and personality (1.4, 2.5, 3.2, 3.3, 4.4). These criteria range, like power, from inherent to achieved (2.5). The range of characteristics becomes a figure that stands in the ground, distance. Characteristics

PHASE I

PHASE II

~00, zo

PHASE IV

d

c

b

Rank

+

Subcriteria

Rank

-Kinship - Membership - Acqua!ntance

PHASE III

Rank

Ground I~ to PHASE V

+Ground~ Figure 2. Determining the distance

342

Munekazu H. Aoyagi

Phase II (Figures 2b to 2d): The speaker classifies each of the subcriteria with respect to characteristics that range from inherent to achieved. He will zoom in to the configuration of Phase I (2.5), making each criterion a figure that stands in the ground, the selected portion in the range. For example, kinship is inherent, acquaintance is achieved, and membership is intermediate. Phase III (Figure 2e): By zooming in, the speaker applies ranking by priority to the subcriteria. Ranking is a figure, and the set of subcriteria is the ground (1.7, 1.8, 3.5-3.7). Phase IV (Figures 2f to 2h): By zooming in further, the speaker ranks each subcriterion with respect to its relevance for the interaction with the hearer (1.6-1.8, 3.6-3.8). As in the case of power, this ranking is also sensitive to situational change (1.8, 3.8).

6.3 Integrating power and distance Phase V (Figures 3a, 3b): Once the speaker determines the subcriteria of power and distance, he will associate power and distanceto evaluate the hearer. The speaker begins by shifting the ground in each of the configurations of Phase IV (Figures lf, 2f) from ranking to the hearer (4.84.10). Phase VI (Figures 3c, 3d): By zooming in to the figures of Phase V configurations, the speaker introduces new figures, grading, against their grounds, the selected subcriteria. Grading is based on a metaphoric height for power (up high, up low, down high, and down low), and metaphoric distance for distance (very close, fairly close, far, and very far) (4.8). The speaker perceives the hearer's personal position in each of these metaphoric coordinates in accordance with the subcriteria selected. Phase VII (Figure 3e ): The speaker integrates the grading of the hearer based on power with the grading based on distance (4.8, 4.13, 4.14). To do so, the speaker zooms in to the figure of power configuration in Phase VI (Figure 3c). He also shifts the ground of distance configuration in Phase VI (Figure 3d) from the selected subcriterion for evaluating distance to the grading of the hearer based on power. Consequently the hearer's personal position in the imaginary distance becomes the figure that stands against his position in the imaginary height. This produces a final personal evaluation of the hearer: e. g., "up far", "down far", "up close", and "down close".

Japanese categories during social interaction

343

PHASEV

zoom~ut. zoomm

t

Subcriterion Acquaintance

c

.;zoomout

Ground!~

zoomm

+around)l

PHASE VII

zoom~ut. zoomm

T

(to PHASE VIII) Figure 3. Evaluating the hearer on the basis of selected categories

6.4 Completing the evaluation Phase VIII (Figure 4a): Degrees of formality of an interaction range from formal to informal: the formal is associated with the public milieu, and informal with the private, casual, or personal milieu (5.8-5.10). The speaker zooms in to the Phase VII configuration, constructing the degree of formality as the figure that stands against the speaker's personal evaluation of the hearer, with respect to imaginary height and distance. Even if the speaker grades the hearer as respectful, she may downgrade the rating within a casual milieu (5.7-5.10, A.2-A.5). However, this

344

Munekazu H. Aoyagi

PHASE IX

Figure 4. Determining the level of formality

downgrading does not occur if the speaker regards the hearer very respectfully (5.6, 5.11, 5.12). This indicates that situational formality evaluation is subordinated to the personal evaluation (power and distance) in terms of cognitive priority. This observation is consistent with Lebra's (1976) argument that Japanese are more concerned with personal relationships than with situations, and that situational evaluation constitutes a microscopic, short-range view of the Japanese behavior: While evincing a trans-situational persistence in behavior patterns, the Japanese also show sensitivity to situational change and readiness for situational adjustment. ... It would be dangerous, however, to infer a lack of moral integrity in this situational adaptability. Situational fluctuation constitutes only a part of Japanese behavior, and fluctuation often takes place at a surface level or for a reason that does not bear on ethics. (Lebra 1976: Ill)

However, Lebra contends that the Japanese use dichotomy to distinguish one situation from another. One the one hand, there is a dichotomy of uchi versus solo. Uchi refers to "in, inside, internal, private", whereas solo means its opposite-"out, outside, external, public". "Where the demarcation line is drawn varies widely: it may be inside vs. outside, an individual person, a family, a group of playmates, a school, a company, a village, or a nation." On the other hand, there is a dichotomy of omole versus ura. Omole indicates "front", or what is exposed to public attention, and ura 'back' implies what is hidden from the public eye (Lebra 1976: 112).

Japanese categories during social interaction

345

Lebra proposes further that uchi and ura, solo and omote, and solo and ura combine to form intimate, ritual, and anomie situations, respectively. These three combinations are situational domains in which behavior is likely to change: in the intimate situation, ego perceives alter as an insider and feels sure that his behavior toward alter is protected from public exposure; in the ritual situation, ego perceives alter as an outsider and is aware that he is performing his role on a stage with alter or a third person as audience; and in the anomie situation where ego defines alter as an outsider, intimacy between ego and alter is ruled out, and ego is also freed from the concern that an audience is watching his behavior (Lebra 1976: 112). In my observation, there are no clear boundaries between intimate, anomie, and ritual categories. Uchi, solo, ura, and omole constitute four relative poles of one and the same mental space, rather than two independent sets of dichotomies; and there are many in-between categories such as "more omole than ura", or "more uchi than solo". Thus, what Lebra describes as three situational domains of Japanese behavior are the three behavioral entailments of a combined personal-situational evaluation, or the evaluation of the speaker's action toward the hearer based on power, distance, and formality. Phase IX (Figure 4b): When power, distance, and formality are evaluated, they are compared with the corresponding degree of respect. The speaker zooms out from Phase VIII configuration. The degree of respect, which ranges from respectful to disrespectful, becomes the ground in which the figure, the total grading of power, distance, and formality stands (4.16).

6.5 Selecting an expression Phase I (Figure Sa): To establish the linguistic expression, the speaker must select a proper linguistic item from his knowledge of Japanese. The speaker emphasizes the specific linguistic item by making it a figure which stands against the ground, Japanese (4.20, 4.21). Phase II (Figure 5b): Zooming in to the figure of Phase I configuration, the speaker finds that the item he has chosen ranges from respectful to disrespectful in terms of the selection of styles of Japanese speech. Here, the degree of respect stands as a figure against the ground, the linguistic item (4.16-4.18). In this phase, the speaker also observes the levels of linguistic expression (4.22). For instance, he will consider the request form of the verb

346

Munekazu H. Aoyagi zoom in~ worn out

Linguistic Item

Degree

PHASE X

...

Respect

b

Phase III

·lzoomout ~"·:::::7 ~

womm

Expressions a

Phase I

1. Respectful 2.Polite 3. Casual 4. Disrespectful

Omeshiagarikudasai Tabetekudasai Tabete Kue

Figure 5. Selecting an expression (examples from the Japanese verb)

"to eat" that changes from omeshiagari kudasai or otabe kudasai (respectful), tabete kudasai (less respectful), tabete (casual), to tabero or kue (vulgar) (Uno 1985: 131-133). Phase X: The association of personal and situational evaluation with a specific linguistic expression is possible by zooming out from the configuration in Phase IX. The configuration is Figure 5b, where the degree of formality becomes the figure against the ground, the selected linguistic item such as the request form of a verb "to eat". The vantage of this configuration results in the actual expression that the speaker produces in interaction with the hearer (4.16-4.22).

7. Dynamics of the selection of Japanese styles in action Based on the way Japanese participants talk about cognitive categories, it has been shown that delicate levels of Japanese honorifics are achieved through the cognitive organization that incorporates a series of personal and situational vantages. These vantages are, in turn, subjective images that are created on the basis of spatial analogy, zooming processes, and shifts of coordinates in analytical space. The following discussion concentrates on the social applicability of the model. Directing attention to the dymanics of the selection process, I will observe different articulations of the process in terms of conversational patterns.

Japanese categories during social interaction

347

Figure I

Figure 2 PHASE!

PHASE II

~~"'

-~

PHASE III

Ronk

/

Ronk Position om 11 nd f

Filtlln!J

1'~HASEV _,,,?'•-""

Figure 3

.~----G~·~-=-4--~ ·~~

*

Subcriterion

Suhcriterion

.;·. ····

PHASE VI

PHASE IX

Figure 5

-···

tnwwno'IUI

LinguiNiiciiCIII

PHASE X

wom1n

Grade

Phase II PHASE VII

Expressions

(to PHASE VIII)

Figure 6. An overview

I. Respectful 2. Polite 3.Casual

Omeshiagurilllda.wli Tubetebdtmli Tubele

4.Disre~>pectful

K"

348

Munekazu H. Aoyagi

The order of the specific scenario that has just been followed represents one of many possible orders that could occur. In reality, the process of style selection is a dynamic feedback that is given by multilevel loops. Being reversible, it may begin at any phase, and may take any course. The speaker may start, for instance, from personal and situational evaluations and select a corresponding expression (5.7, 5.8); or he may think of a certain expression and judge it in terms of personal and situational evaluation (5.9, 5.10). He may also proceed from any intermediate phase and eventually fulfill the entire process. In Figures 1 to 5, reversibility of the process is indicated by two-way arrows. If zooming in takes place from one configuration to another, zooming out takes place in the reversed direction. However, because the styles of Japanese speech constitute a system, no configuration of figure and ground by itself achieves the process. The speaker is normally aware of cardinal configurations: (1) the power-distance configurations of Phase VII (Figure 3e); and (2) the power-distance-formality configuration of Phase VIII (Figure 4a). These configurations are the so-called "prototypes" (Sweetser 1987), or simplified representations of all other configurations. They are foremost in mind, while the rest are entailed (4.24). The dynamics of the selection of Japanese styles also allow the speaker to go through the process repeatedly. The speaker may evaluate and revalue power, distance, and formality, enabling expression to shift from one context of interaction to another. This process or "repeated evaluation" is observed in the conversations B, C, and D in the Appendix. The following two cases provide details of the process.

7.1 Shift of power; constant formality (conversation B) Conversation B (B.1-B.ll) refers to the first encounter between KT and AH. Their primary task is to introduce themselves to each other in an attempt to determine power and distance. On the scale of formality, they select 'formal' style in this context, as any first encounter can generally be ritualistic and formal. Under this constant formality, the information regarding power is emphasized: it becomes clear that KT is much more senior than AH. Before any information about power is known, both KT and AH incorporate the copula desu (see section 5) in their utterances to be polite with each other (B.1-B.4). This form changes after the information be-

Japanese categories during social interaction

349

comes known: for KT, AH is "low" in power; for AH, KT is "high". Since KT has no need to show deference to AH, he "casualizes" his expression. Thus, the polite copula desu drops out of KT'S utterances (B.5, B.7, B.9, B.ll). On the other hand, AH maintains desu since he must keep his deference to KT (B.6, B.8, B.lO).

7.2 Shift of distance and formality; constant power (conversation C) In conversation C (C.l-C.l6), the situation begins with a formal encounter between TS and OF (Set 1). But since TS and OF are old colleagues, their intimate relationship in the past is recovered and eventually takes over the formal interaction (Set 2). This is the case where power is constant, and the shift is focused on distance. Both TS and OF's grading of distance alters from "distant" in Set 1 (C.l-C.3), where TS and OF meet for the first time after long separation, to "close" in Set 2 (C.l4-C.l6), in which the old friendship reasserts itself. Thus, the copula desu and the auxiliary verb masu (section 5) appear in Set 1, whereas they disappear in Set 2. In one utterance, desu is replaced with the casual form da (C.l4).

8. Sociolinguistic implications The dynamic network of associated cognitive and linguistic categories does not force a speaker to act passively according to mechanical processes of personal and situational evaluations. Instead, the speaker may actively use the network as a means to complement discourse strategies. To borrow from Gumperz (1982: 166), the network becomes a set of schemata that enacts speech activity to some communicative goals. Under such circumstances, the chosen linguistic expressions are inferential signals. The speaker uses them to propose certain attitudes that he wishes the hearer to infer on the basis of power, distance, and formality. For instance, he may use the abundant honorific expressions to avoid conflicts, smooth the communication, or maintain esteem in interactions with the hearer (Uno 1985: 35-39, 44-46). But the speaker cannot attain these goals unless the hearer comprehends his message. In discourse strategy, the indexed personal and situational evaluations need not correspond to the speaker's actual evaluations. The speaker may, for example, behave respectfully toward the hearer to avoid conflict

350

Munekazu H. Aoyagi

although he does not respect the hearer. One such case is found in conversation D. The store conversation between CS (a customer) and LK (a clerk) indicates what the Japanese call ingin-burei 'nasty nice' or 'impolitely polite'. Here, the polite copula desu as well as the polite auxiliary verb masu mark LK's utterances, showing LK's respectful performance toward CS. The principal rule among the Japanese is that clerks must act deferentially to customers. Yet, the content of LK's utterance indicates that he is refusing to offer his service to CS. The following interview was conducted after this interaction in order to clarify the ambivalence of polite and unkind (the interviewer is "Q"): Q:

Naze ano okyakusan no rikuesto ni oojinakatta n desu ka? [Why didn't you reply to the customer's request?] LK: Saabisu shitaku nakatta kara desu yo. [I didn't want to serve him.] Q: Sore wa naze desu ka? [Why is that?] LK: Anna shitsurei na wakamono ja saabisu shitaku naku mo narimasu yo. [I wouldn't want to serve such an arrogant youngster.] Q: Dewa naze teinei na kotoba de hanashita no desu ka? [Then why were you talking politely to him?] LK: Sore wa ichiou shigoto desu kara ne ... [It's my job, you know ... ]

This indicates that the actual degree of respect that LK has toward CS is low, although LK behaves politely toward CS. LK's vantages that are based on age ("CS is a youngster") and personality ("CS is arrogant") are recessive, whereas his dominant vantage is based on position ("CS is a customer"). The dynamics of style selection in Japanese allow LK to evaluate CS on the basis of these three criteria at the same time and consider one of them as the priority selection. This shows that selection of a vantage from the series of vantages constitutes interaction beyond personal feelings.

9. Conclusions Using vantage theory as an organizing framework, this study has shown the processes by which distinctive Japanese expressions come into effect. The selection of styles reflects the way in which the Japanese, through

Japanese categories during social interaction

351

dynamic mental processes, negotiate with propriety in interactions. When interlocutors meet, they reveal to each other their places in the predetermined power hierarchy. When they establish this as a fixed relation, other decisions follow. These decisions are flexible, and they vary with personal distance and degrees of formality. Different theories and models about the relationship between cognitive criteria and honorifics, in languages of Japan and elsewhere, have been offered (Brown and Gilman 1972 [1960]; Uno 1985; Brown and Levinson 1987). But they confine cognition to pre-fixed, mechanical processes by ignoring the contribution variable personal viewpoints make to the constitution of meaning. By virtue of vantage theory, the present study has gone beyond these static frameworks to propose a dynamic model. More generally, the present study has shown how the selection of social interaction among the Japanese is "construed" by cognition. I have demonstrated the mechanism of association between linguistic criteria and their users' perception of the world. Such mechanism functions dynamically, allowing change from one context to another, and "multivocally", allowing several perspectives to coexist at the same time. The process of linguistic selection through this dynamic, multivocal scheme indexes communicative strategy. In this sense, the selection of Japanese styles functions as a creative construction of idiosyncratic analogies that bridge the gap between personal experience and the conventional world.

Appendix Abbreviations in the following interviews include "Q" for questioner and "Sj" for subject or participant. Each interview is prefaced by a brief explanation of the participant's background (sex, age, occupation, and education). In conversations B and C, the names of the participants are indicated by the initials of their family and given names. Transcripts of these interviews and conversations are given in Japanese, followed by their English translations. Interview 1. Female, 29, office lady, college educated (1.1) Q: (1.2) Sj:

(1.3) Q:

Aite to no chikara-kankei ni wa tatoeba donna mono ga arimasu ka? Tatoeba nenrei taka seibetsu taka ... shakai-teki na chii taka so no hito no kooseki taka ... nooryoku nante mon janai kashira. Jaa aite to no kyori wa nan de kimemasu ka?

352

Munekazu H. Aoyagi

(1.4) Sj:

(1.5) Q: (1.6) Sj:

(1.7) Q:

(1.8) Sj:

(1.1) Q: (1.2) Sj: (1.3) Q: (1.4) Sj:

(1.5) Q: (1.6) Sj:

(1.7) Q:

(1.8) Sj:

Sore wa tatoeba sono aite ni kansuru chishiki taka tsukiatteta nensuu taka ... nani ka no guruupu de onaji menbaa ka doo ka taka ... Aite no chikara arui wa kyori o handan- -suru toki ima yutta yoo na kijun o zenbu ippen ni kooryo-suru no desu ka? Ichiou zenbu kooryo-suru ka mo shirenai keredo ... yapparijookyoo ni oojite dare ka hitotsu ka mata wa ikutsu ka ni shooten o ateru to omou wa. Tatoeba kono hito to kono jookyoo ni wa toshi taka ... Ichio zenbu o kooryo-shite sore karajookyoo ni oojite hitotsuka ikutsu ka o erabu to iu koto wa, kooryo-sareru arayuru yooso no naka kara dare ga ichiban taisetsu ka, dare ga niban-me ka, soshite dare ga sanban-me ka tte iu hairetsu o tsukuru to iu koto desu ka? Soo ne. Kaisha nan ka ja yapperi haitta toshi da taka chii da taka o saisho ni miru-shi, sono hoka no basho de wa toshi ga saisho kama sirenai-shi ... What examples do you have for a power relationship between you and your hearer? For example, age, sex, ... social status, achievement, and ability. How would you determine, then, distance between you and the hearer? That's by, for example, how much I know about the person, number of years in which I have been relating to the person, ... whether or not I am a member of the same group as the person ... Do you consider these criteria that you have just mentioned as a whole when you judge power or distance? I will consider all of them in a way ... but I'll emphasize one or a few of them in accordance with the situation. For example, I'll apply age for a particular person in a particular situation. Does the fact that you consider all in a way and yet emphasize one or a few with respect to the situation mean that you organize a priority of what is the most important, what is the second, and what is the third out of all the factors that you can think of? I think so. In a company I'll consider seniority and position as primary; and in other places age might have a priority.

Japanese categories during social interaction

353

Interview 2. Male, forties, company director, college educated (2.1) Sj:

(2.2) Q: (2.3) Sj: (2.4) Q: (2.5) Sj:

(2.1) Sj:

(2.2) Q: (2.3) Sj: (2.4) Q: (2.5) Sj:

Chikara no naka ni mo nenrei toka nenkoo toka nooryoku toka samazama na mono ga aru kedo, ikutsu ka wa yori senten-teki dashi, hoka no wa yori kooten-teki nan jaa nai ka na. Tatoeba nenrei nan ka wa jibun no ishi ni wa kankeinaku sadamatchaushi nooryoku demo chishiki no yoo na mono wa jibun no doryoku shidai de yori ooku erareru-shi ... Dewa sorera, o ichioo bunrui-shita ue de sono naka kara dore ga ima koko de kooryo-subeki ka o erabu no desu ne? Soo iu koto desu ne. Shinso-kankei ni tsuite wa doo desu ka? Kekkyoku onaji desyoo. Shinso ni wa kazoku ka yososama ka toka, a to wa menbaashippu ya tsukiatta nensuu ya ga aru kedo, kazoku ka doo ka nante iu no wa senten-teki ni kimaru n jaa nai kana. Even among the criteria of power, there are various things like age, seniority, and ability; but I think some are more inherent and some are more achieved. For example, age is determined regardless of your will, and knowledge among the abilities can increase based upon your effort. You will categorize these criteria first and then select the one which is the most important here and now. Right? That's right. How about distance? I think it's the same. In distance there are family or outsider, and others like membership and years; but a thing like family is determined, I think, inherently.

Interview 3. Female, 51, housewife, high-school graduate (3.1) Q: (3.2) Sj:

(3.3) Q: (3.4) Sj: (3.5) Q: (3.6) Sj: (3.7) Q:

Kyori no kijun to shite wa donna mono ga kangerareru? Sono hito ni kansuru chishiki ga fukakunakya dame desyoo. Sore wa son dake fukaku tsukiatte iru wake dakara. Motto hoka ni wa? Soo ne, uchi ka so to ka no kankei ya jinkaku . . . Tsumari konomeru ka konomenai ka . . . nan ka ja nai? faa sono naka de dore ga ichiban juuyoo da to omou? Sore wa hito ni yotte chigau daroo kedo ... Soo ne, yappari tsukiai no fukasa kana ... Sore wa jookyoo ni yotte mo kawaru mono nano?

354

Munekazu H. Aoyagi

(3.8) Sj:

Soryaa sao da wane. Ikura shiriai demo kenka o sureba yappari aka no tanin no yoo ni toriatsukau kama shirenai-shi ... Nakanaori o sureba mata miuchi mitai ni mo nariuru-shi. Sore wa jookyoo no chigai tte iu n ja nai no?

(3.1) Q: (3.2) Sj:

What criteria can be considered for distance? You'll certainly need a good knowledge of the person, since you will have such an in-depth interaction with that person. Any others? Well, insiders or outsiders, and personality ... that is, whether one can like or dislike ... something like these. What do you think is the most important among these? That may vary according to different persons ... All in all, I think it's the intensity of the relationship. Do they change according to the situation? Indeed, I may treat someone I know as a stranger if I have a fight with her ... , we may become like family again if we make up again. Isn't that a situational difference?

(3.3) Q: (3.4) Sj: (3.5) Q: (3.6) Sj: (3. 7) Q: (3.8) Sj:

Interview 4. Male, 27, college student (4.1) Q: (4.2) Sj: (4.3) Q: (4.4) Sj: (4.5) Q: (4.6) Sj: (4.7) Q: (4.8) Sj:

(4.9) Q: (4.10) Sj: (4.11) Q: (4.12) Sj: (4.13) Q:

Naze aite ni yotte kotoba-zukai o kaeru no? Sore wa hito ni yotte tachiba ga chigau kara da to omou. Sono tachiba tte iu no wa nan de kimaru no? Sore wa yappari jooge-kankei taka shiriai ka doo ka taka ... Jaa chikara-kankei to jibun to no kyori? Sao dane. Jaa sono aile no tachiba no handan kara kotoba no seisan made o doo iu fuu ni saigen-suru? . . . . Mazu aite ga, tatoeba, toshi-ue ka doo ka tte iu no o miru deshoo. . .. Sore to itsho ni sono aite ga jibun ni chikai kankei ni aru ka tooi kankei ni aru ka o miru deshoo. . .. Sore de sono futatsu o kumi-awasete dono hen ni iru ka tte iu no o kimete kara kotoba o erabu to ... Matte. Mazu wa aite to iu mono ni chikara aruiwa kyori to iu mono o atehameru to. Sao. Sore kara sono chikara ka kyori ga dare gurai no reveru ni aru ka o miru to. Soo. Tsugi ni sono chikara no reveru to kyori no reveru o kumiawaseru to.

Japanese categories during social interaction

355

(4.14) Sj: Un soo. (4.15) Q: Sore de ... (4.16) Sj: Sore de sono hito ni dare gurai sonkei taka teinei ni dekiru ka kimete, sono sonkei-do ni yotte kotoba o kimeru. (4.17) Q: Kotoba o kimeru tte iu no wa so no hit a e no sonkei-do ni hittekisuru sonkei-do no mono o kotoba no taikei kara erabu tte iu koto? (4.18) Sj: Sao. (4.19) Q: So no kotoba de nani ka rei o agete mite. (4.20) Sj: Rei tte itte mo bunpoo-teki ni iron na no ga aru kedo ... (4.21) Q: Jaa erabu kotoba-sonkei-go taka teinei-go taka yori zatsu na hanashikata taka-no naka ni mo iroiro na bunpoo no mono ga aru wake? (4.22) Sj: Sao sao, sore de sono naka kara tatoeba dooshi nan ka o erande, tatoeba meshiagaru taka omeshiagari ni naru taka iu wake. ( 4.23) Q: Sana katei no naka de ichiban juuyoo tte iu ka, toku ni fukaku chuumoku-subeki dankai tte aru? (4.24) Sj: Dakara ichi-ichi hitotsu-hitotsu no dankai o ishiki-teki ni au nante iu koto wa shinai ja nai. Kekkyoku patto aite o mita toki omou no wa sana hito no chikara to kyori no kumiawase de, sore ga dare gurai ka de kotoba ga detekuru tte kanji. Dakara sono tachiba no handan no dankai ga ichiban jyuuyoo mitai. (4.1) Q:

Why do you switch your style of speech with respect to different persons? (4.2) Sj: I think that's because different persons have different positions. (4.3) Q: What determines those positions? (4.4) Sj: That's, of course, something like hierarchy or whether or not you know the person ... (4.5) Q: Is it power and distance? (4.6) Sj: I guess that's it. (4.7) Q: Then how would you reconstruct the process between your judgment of that person and your production of a speech? (4.8) Sj: ... Well, first I'll look, for example, whether he is older or younger than me .... Then I'll see whether he stands close to or far from me .... Then I'll combine them together and select my speech style accordingly. (4.9) Q: Wait, so first you attribute power or distance to a person. (4.10) Sj: Yes.

356

(4.11) (4.12) (4.13) (4.14) (4.15) (4.16)

Munekazu H. Aoyagi

Q: Sj: Q: Sj: Q: Sj:

(4.17) Q:

(4.18) (4.19) (4.20) (4.21)

Sj: Q: Sj: Q:

(4.22) Sj: (4.23) Q: (4.24) Sj:

And you decide what the levels of power and distance are. Yes. And then you combine the two levels of power and distance. Yes, that's right. And ... ? And I will determine the degree in which I can be respectful or polite to the person, and according to that degree of respect, I determine my expression. To determine the expression means to select language with a degree of respect that matches with the degree to which the person is respected? Yes. Give me an example of that type of expression. There is a variety of grammatical examples ... So there is variable grammar within the language that you use-like respectful, polite, or much rougher ways of talking? Yes indeed. I'll select a verb out of those and say something like meshiagaru (polite) and omeshiagari ni naru (very polite). Is there a phase that is most important, or that one should put a deeper emphasis upon? You don't consciously trace the phases one by one every time, you know. After all, you think about the combination of power and distance at the moment you see the person; and your expression is produced with respect to the degrees of that combination. So, I guess judgment of position is the most important.

Interview 5. Female, 21, college student (5.1) (5.2) (5.3) (5.4) (5.5) (5.6)

Q: Sj: Q: Sj: Q: Sj:

(5.7) Q: (5.8) Sj:

Meue no hito ni wa dono yoo ni hanasu? Mochiron, sonkei-go ya teinei-go o tsukatte teinei ni hanasu. Onaji meue demo sensei toka senpai toka de wa chigai ga aru? Yappari aru ne. Doo iu fuu ni chigau? Tatoeba sensei da to kei-do no takai sonkei-go o hanasu kedo, senpai nan ka da to tashoo yawarageru. Sore ni sensi ni wa zutto keigo o tsukau kedo, senpai ni wa zutto tsukau koto nai. Donna toki senpai ni keigo o tsukawanai no? Rirakkusu-shiteru toki toka jookyoo ga keishikibatte nakute puraibeeto na toki.

Japanese categories during social interaction

357

(5.9) Q: Jaa jookyoo ni yotte mo kotoba ga kawaru wake? (5.10) Sj: Un, kawaru to omou. Keishikibatta toki yori keishikibatte nai

toki no hoo ga motto rafu na hanashikata ni naru to omou. Sensei ni taishite zutto keigo o tsukau tte iu koto wa sensei no bawai keishikibatte yooga rirakkusu shite yooga kankeinai tte koto? (5.12) Sj: Soo, sensei ni wa donna toki demo keigo ni naru. (5.11) Q:

How would you talk to your seniors? I'll of course use respectful or polite expressions. Among the seniors, is there any difference between people like your teachers and your senior friends? (5.4) Sj: There is. (5.5) Q: How does it differ? (5.6) Sj: For instance, I'll use honorifics that convey a high degree of respect for teachers, but I'll ease a little bit for senior friends. Also, I'll use honorifics continuously for teachers, but not for the senior friends. (5.7) Q: When do you not use honorifics for your senior friends? (5.8) Sj: In a relaxed setting, or when the situation is informal and private. (5.9) Q: Then the speech changes with situations as well? (5.10) Sj: Yes, I think so. My speech will become rougher in informal contexts than in formal ones. (5.11) Q: Does using honorifics continuously for your teachers mean that you use them regardless of whether it is in a formal context or in a relaxed one? (5.12) Sj: Yes. I use honorifics for my teachers all of the time.

(5.1) Q: (5.2) Sj: (5.3) Q:

Conversation A An imaginary conversation. The speaker (male, 25) is a college student. He was asked to create the utterance "Please eat this" with respect to various situations. Note the request forms of the verb "to eat": omeshiagari kudasai (very polite), otabe kudasai (relatively polite), tabete kudasai (polite), tabete (direct), and tabero or kue (vulgar). (A.l)

The speaker faces his professor (high power, far distance) in public (formal context):

Zehi

kore o

omeshiagari kudasai.

by all means this (particle) eat

please

358

Munekazu H. Aoyagi

(A.2)

The speaker faces his father (high power, close distance) at home (informal context): Toosan, kore tabete. father this eat

(A.3)

The speaker faces his father in public (formal context): Toosan, kore o tabete kudasai. father this (particle) eat please

(A.4)

The speaker faces his senior friend (high power, close distance) in public (formal context): Senpai, zehi kore o otabe kudasai. senior by all means this (particle) eat please

(A.5)

The speaker faces his senior friend at home: Senpai, kore o tabete kudasai. senior this (particle) eat please

(A.6)

The speaker faces his colleague (equal power, close distance) at home: Kore tabete ya. this eat (particle)

(A. 7)

The speaker faces his younger brother (low power, close distance) at home: Kore tabero yo. this eat (particle)

(A.8)

The speaker is angry at his younger brother (low power, far distance): ii daroo! Katte ni kue ba as you like (particle) eat (conjunction) good (copula)

Conversation B An actual conversation. KT and AH are students at the university. KT

is much older and higher in seniority than AH. Set 1 (B.l, B.2) takes place when they meet for the first time, Set 2 (B.3B.ll) when they are getting to know about each other, and Set 3 (B.l2B.15) when they have learned about each other. Note the copula desu (or da) and auxiliary verb masu.

Japanese categories during social interaction

Set 1. (B.l) KT: (B.2) AH: (B.l) KT: (B.2) AH: Set 2. (B.3) KT: (B.4) (B.S) (B.6) (B.7) (B.8) (B.9) (B.lO) (B.ll)

AH: KT: AH: KT: AH: KT: AH: KT:

Aa, kondo kita AH-san desu ka? KT desu. Aa, KT-san desu ne. AH desu. Zehi yoroshiku onegai-shimasu. Oh, you are Mr. H who just came. I am KT. Oh, you are Mr. KT. I am AH. How do you do?

AH-san wa moo dare gurai daigaku de yatte rareru no desu ka? Karekore sannen desu. Are, jaa ima mada gakushi-katei na no? Aa, sao desu. KT-san wa? Baku wa ima hakase-katei desu yo. Aa ... sao desu ka. Jaa ima ikutsu? Nijuuroku desu. KT-san wa? Baku wa ima choodo sanjuuichi.

(B.3) KT: (B.4) AH: (B.S) KT: (B.6) AH: (B.7) KT: (B.8) AH: (B.9) KT: (B.lO) AH: (B.ll) KT:

How long have you been at the university? Around three years. Well, you are in the Bachelor's program, then. No? Yes, that's right. How about you? I'm in the Ph. D. program now. Oh ... Is that right? How old are you then? I'm twenty-six. How about you? I'm just thirty-one.

Set 3. (B.l2) (B.l3) (B.l4) (B.l5)

Yoo, AH-kun. Konki wa donna kurasu o totteru no? Ee, seijigaku aa futatsu ni, gaikoku-go to tookei desu. Hoo, omoshiroi kumiawase da ne. Sao desu ka?

KT: AH: KT: AH:

(B.l2) KT: (B.l2) AH: (B.14) KT: (B.l5) AH:

359

Hey, AH. What classes are you taking this semester? Two political-science classes, one foreign language, and one statistics class. Huh, what an interesting combination. Is that so?

360

Munekazu H. Aoyagi

Conversation C A drama. TS (male) and OF (female) are both in their thirties, old classmates who once had an affair with each other. They meet again at an alumni party, long after they have separated. Their romance grows once more. Set 1 (C.l-C.l3) takes place when they first meet after many years. Set 2 (C.l4-C.l6) takes place after their relationship has been reestablished. Note the copula desu (or da) and auxiliary verb masu (and its past tense mashita). Set 1. (C.l) OF: (C.2) TS:

(C.3) OF: (C.4) TS: (C.5) OF: (C.6) TS: (C.7) [OF (C.8) TS: (C.9) OF: (C.lO) TS: (C.ll) OF: (C.l2) TS: (C.13) OF: (C.l) OF: (C.2) TS:

(C.3) OF: (C.4) TS: (C.5) OF: (C.6) TS: (C.7) OF (C.8) TS: (C.9) OF: (C.IO) TS:

TS-san deshoo? Ee, TS desu. Oboete ite kudasatta n desu ne? ... Hajimete desu yo ne, doosookai de oai suru no wa. Boku mo sonna ni nesshin ni sanka suru hoo ja nakatta kedo. Nagoya ni otsutome nan desu ka? Ee. Maa tsumannai koto yattemasu. Watashi TS-san to hanashiteru! lya, boku no hoo koso ... smiles longingly, but she looks away from TS.] Nani ka? lie, ii n desu. Nani ka arimashita? Tokidoki omoidashite ita n desu, Nagoya no rekoodo-yasan no koto . ... /ya ii n desu. ... Aa! Ano ame ga futte ta ... Soo, ame ga futte ta wa ...

Isn't it Mr. TS? Yes, I'm TS. You remembered me. This is the first time I met you at an alumni meeting; although I haven't been too serious about participating either. Are you working in Nagoya? Yes. But I'm doing something worthless. I'm talking with you! No. It's me who is talking with you ... smiles longingly, but she looks away from TS. Something? No, nothing. Something mattered?

Japanese categories during social interaction

(C.ll) OF: (C.l2) TS: (C.l3) OF:

361

Sometimes, I remember the record shop in Nagoya. . .. That's all right. Oh, yes! When it was raining ... Yes, it was raining ...

Set 2. Mter their reunion, TS and OF took a walk into the woods together. They encounter the scene of a suicide.

(C.15) OF: (C.16) TS:

Sanna ni ki ni suru koto wa nai yo. Igai ni shokku ukeyasui n dana. Betsuni ... Tada urayamashii tte kanjita dake. Demo shinu ki ni nareba nan demo dekiru n da.

(C.14) TS: (C.15) OF: (C.16) TS:

Don't worry so much. You get shocked unexpectedly easily. Not really ... I only felt jealous. But you can do anything once you have the will to die.

(C.l4) TS:

Conversation D

An actual conversation. CS is a customer who comes to the book store where LK works. CS is looking for the book called Y and written by X. CS is a student, much younger in age than LK. CS behaves arrogantly in the store, and LK does not like his attitude. Thus, LK is not willing to offer his services to CS. Note the copula desu and auxiliary verb masu (and its denial form masen). (D.l) CS: (D.2) LK: (D.3) CS: (D.4) LK: (D.5) (D.6) (D.7) (D.8)

CS: LK: CS: LK:

(D.9) CS: (D.lO) LK:

Anoo, chotto kikitai n da kedo, sotchi ni X no kaita Y tte hon oite nai? Aa, taihen mooshiwake nai n desu ga, ima chotto oite masen. Ee, nai no? Jaa chuumon dekiru? Aa, tabun watakushi-domo no hoo de wa chotto sagashi-kanemasu ga ... Doo ni ka sagase-nai? Doko no shuppan-sha desu ka? Dokko tte itte mo ... Sore mo shiraberare nai no? Chotto muzukashii desu ne. Shuppan-sha ga wakaranai to doo ni mo narimasen ... Moo ii wa. Sore kara sotchi ni hon no risuto ga oite-atta to omou kedo, sore nai? Ima chotto kirashite run desu yo ne. ... Gakusei-san desu ka.

362

Munekazu H. Aoyagi

(D.ll) CS: (D.12) LK: (D.13) CS: (D.14) LK:

Soo da kedo. Ano risuto wa kyooju ya sensei-gata dake ni owatashi shiteru n desu yo ne ... Moo iii Hoka de sagasu. Taihen mooshiwake gozaimasen.

(D.l) CS: (D.2) LK: (D.3) CS: (D.4) LK: (D.S) CS: (D.6) LK: (D.7) CS: (D.8) LK: (D.9) CS: (D.lO) LK: (D.ll) CS: (D.12) LK: (D.13) CS: (D.14) LK:

I want to ask you if you have a book by X, called Y. I'm very sorry, but we are out of it now. Oh, you don't? Can I order it, then? Well, probably it would be impossible for us to look for it. Can't you look for it somehow? Which publisher is it? I am not sure which ... Can't you check that up, as well? It's difficult unless we know which publisher it is ... Never mind. I think you had a book list. Can I get it? We are out of it now.... Are you a student? Yes, I am. We only give the list to professors and instructors ... Never mind! I'll look for it somewhere else. I'm terribly sorry.

Note 1. I am most grateful for the support of Robert E. MacLaury and John R. Taylor. I am also thankful for critical comments and suggestions by Jane H. Hill, Susan U. Philips, Keiko Matsuki, and Scott Hurley. I dedicate this paper to the memory of my colleague, Shawn D. Wiser, for whom the dynamics of mind always was the source of human existence and practice.

References Aoyagi, M. Hiroshi 1989 Vantage theory and decision making: Japanese choice of language level in social interaction. [Paper presented at 6lst South West Anthropological Association Annual Conference, Long Beach, California, April 12-14.] Brown, Penelope-Stephen C. Levinson 1987 Politeness: Some universals in language usage. Cambridge: Cambridge University Press. [1978] [Originally published as "Universals in language usage: Politeness phenomena", in: Goody, Esther N. (ed. ), Questions and politeness: Strategies in social interaction. Cambridge: Cambridge University Press, 56--289.] Brown, Roger-Albert Gilman 1972 "The pronouns of power and solidarity", in: Pier P. Giglioli (ed.), Language and social context. Harmondsworth: Penguin, 252-282. [1960] [Originally published in: Thomas Sebeok (ed.), Style in Language. Cambridge, MA: Massachusetts Institute of Technology, 253-276.]

Japanese categories during social interaction

363

Gumperz, John J. 1982 Discourse strategies. Cambridge: Cambridge University Press. Langacker, Ronald W. 1987 Foundations of cognitive grammar. Vol. 1. Stanford: Stanford University Press. Lebra, Takie S. 1976 Japanese patterns of behavior. Honolulu: University of Hawaii Press. MacLaury, Robert E. 1987 "Coextensive semantics ranges: Different names for distinct vantages of one category", in: Papers from the twenty-third annual meeting of the Chicago Linguistic Society 1: 262-282. 1989 Vantages and coordinates: Spatial analogy in categorization. [Paper presented at the Symposium on Cognitive Grammar and American Indian Languages, 60th South West Anthropological Association Annual Conference, Riverside, California, April 28.] Matsuki, Keiko 1989 Japanese double cognition and communication. [Paper presented at 88th American Anthropological Association Annual Conference, Washington DC, November 14-19.] Minami, Fujio 1974 Gendai nihon-go no koozoo [The structure of Modern Japanese]. Tokyo: Taishuukan Shoten. Nakane, Chie 1970 Japanese society. Berkeley: University of California Press. Shore, Bradd 1991 "Twice-born, once conceived: Meaning construction and cultural cognition", American Anthropologist 93(1 ): 9-27. Sweetser, Eve E. 1987 "The definition of a lie: An examination of the folk models underlying a semantic prototype", in: Dorothy Holland- Naomi Quinn (eds.), Cultural models in language and thought. Cambridge: Cambridge University Press, 4366. Taylor, John R. [this volume] "On construing the world". Uno, Yoshikata 1985 Keigo o dono yoo ni kangaeru ka [How to examine the honorifics]. Tokyo: Nan'undoo. Yoneyama, Toshinao 1976 Nihon-jin no nakama ishiki [Japanese group consciousness]. Tokyo: Koodansha.

Genus, species, and vantages Jeff Lansing

1. Introduction In this paper I will briefly examine several theories of categorization in reverse historical order. My point will be to show that the structuralist theory of categories based on bundles of necessary properties can be understood as an historical corruption of an original vantage theory. By vantage theory here I mean the application of a certain cognitive process; the core of this process involves the separation of a mobile (moving or movable) "coordinate", which is in the foreground as seen from a certain vantage (point of view), with respect to a fixed background coordinate. This core can be extended by "zooming", whereby a former mobile coordinate becomes the fixed coordinate, and a new mobile coordinate is separated out. 1 Let me begin with a modern (straw man) version of the "classical" theory of categorization. 2 This version, in which categories have "essences" (which we are to understand as "all parts [moria] ... whose destruction causes the destruction of the whole" [Taylor 1989: 22], goes as follows: To say that an X is a Y is to assign an entity X to the category Y. We do this by checking off the properties of X against. the features which define the essence of the category Y; our knowledge of this set of features characterizes our knowledge of the meaning of the word Y. In [Metaphysics f.4], Aristotle singled out two defining features of the category MAN (and hence two features in the definition of the word man), namely [TWO-FOOTED] and [ANIMAL]. These two features are, individually, necessary for the definition of the category (the destruction of either causes "the destruction of the whole"); if any of the defining features is not exhibited by the entity, then the entity is not a member of the category. Jointly, the two features are sufficient; any entity which exhibits each of the defining features is ipso facto a member of the category. (Taylor 1989: 23) 3

So in this theory there are three basic kinds of objects: words (such as man), categories (such as MAN), and entities (such as the son of Sophroni-

scus, who just happens to be approaching now). These are connected via meanings, essences, and features. Categories have essences which are

366

Jeff Lansing

defined by features, which in turn are exhibited by entities. Words have meanings, which are also defined by features. Knowledge of meanings is characterized by knowledge of essences; for example, we characterize our knowledge of what the word man means by knowing that [TWO-FOOTED] and [ANIMAL] together make up the essence of the category MAN. It is important to notice here the equal status of the two features: both are equally necessary (since both are part of the essence), and each contributes equally to their joint sufficiency. For we can easily imagine a different version of the theory in which ANIMAL is a category (rather than a feature), and MAN is the subcategory of these entities which are members of that category that exhibit the feature [TWO-FOOTED]. 4 It is also important to notice what gives these features their status: in this theory the features are necessary because they are (somehow) connected with an important consequence, which is the destruction of the whole (the man). But we can easily imagine an alternate version of the theory in which the status of the features comes from something else, such as being connected with the greatest number of consequences. 5 In summary, the "classical" theory ties three levels of objects together (words, categories, and entities) using a democratic system of features, which are chosen for their consequences. As we shall see, vantage theory is hidden in this theory in the following way: essences are mobile coordinates which are seen from the vantage of their consequences, against a fixed background of entities.

2. The fixity of species In this section I discuss a slight complication which was introduced into the classical theory, certainly by the time of Linnaeus. 6 Recall that the level of entities is related to the level of categories because on the one hand entities exhibit features, and on the other hand features define categories (through essences). The complication consists in understanding a sort of inverse of these two relations (exhibiting, and defining, together) as also being a relation -call is "aboutness" -whereby categories are about entities. Linnaeus's thinking concerning this aboutness relation changed as he made new discoveries: at first he used this relation to develop a taxonomy of plants, but as his thinking changed, he began to develop a systematics. Linnaeus's taxonomy or classification of plants by features (attributes, characters) was based on the axiom that "we count as many species as

Genus, species, and vantages

367

there were different forms created in the beginning". 7 This is usually taken to mean that Linnaeus believed (at that time) that God had created at least some of the entities in each of the now-existing categories (i.e., species) of plants at the time of the creation, or in other words, that there is a fixed link between the level of categories and the level of entities. Because new categories could not come into existence, seeds from an entity (a plant) in one category (one species) could only develop into an entity in the same category. Thus the category that an entity would be in was fixed even before that entity came into existence, in fact, even from the beginning. Later, research with hybrids convinced Linnaeus that the actual situation in nature must be more complex. Knowing that entities in one category could exhibit affinities (relationships of similarity) with entities in a different category, Linnaeus proposed a systematics or processual theory, based on hybridization, which could explain such affinities. The Creator, Linnaeus now [writing in 1762] believed, created in the beginning only one species in each natural order; in other words, a very small number of species. These original species had the ability to fertilize each other, and this they immediately did. In this way one species arose for every genus; all the genera were thus virtually primordial, even if they were genetically derived by means of a first hybridization. Over the ages the primordial species within the genus has crossed with species from other genera; the manifold species which have come down to us, and which are often difficult to distinguish, have emerged gradually. (Eriksson 1983: 97-98)

This systematics which Linnaeus proposed represents a considerable deviation from the classical theory, no matter how it is interpreted. The interpretive issue is whether we are to understand there to be something other than entities-such as species-at the level of entities or not. In other words, is there something which actually participates in such processes as crossing and emerging? If we assume that there is something over and above entities which actually exists-we can think of this something as a part of life to make it seem more real (for surely life is real)-then the complication to the classical theory reduces to a complication of the fixed background component. But there is a problem. As Olivier Rieppel ( 1991: 97) argues, "ancestry is under determined by observation of similarity". So we can never know what these real somethings, these parts of life, actually are. So they must be theoretical constructions. So they don't exist at the level of entities. If, on the other hand, we assume that only entities are actual, so that species must be at the level of categories, then we have to introduce a

368

Jeff Lansing

new relation between entities and categories. Since we can't observe categories, this new relation will be what tells us about the ancestry of categories. So this is an "aboutness" relation between entities and categories. But we already had an "aboutness" relation between categories and entities: categories are about nature, which is composed of entities. So entities and categories are about each other. However we decide to understand this circularity, it is clear that we have deviated considerably from the classical theory if circularity is part of what we now believe in. Vantage theory tells us that this deviation is exactly what we should expect. Why? The way vantage theory looks for additional structure is by "zooming in", that is, by fixing a previous foreground as a new background and by then separating out a new foreground, or set of one or more mobile coordinates. But Linnaeus's systematics looks for additional structure by trying to "zoom in" on the background. Vantage theory tells us that we can't focus on the background and still keep it as the background (the level of entities), because we can't be at two vantage points at the same time. So we should expect that either the background (the level of entities) should come to the foreground (the level of categories), or else an oscillation between vantages should occur. And these are exactly the two possibilities which we just found occurring. In summary, the first (and most recent) corruption of the original vantage theory, which we just examined, Linnaeus's attempt to fix the relation between categories and nature, has so far proved to be unworkable. 8 Vantage theory tells us that this is exactly what we should expect. We can take this agreement as evidence that categories are mobile coordinates in the foreground, against a fixed background of entities. In the next section we proceed to uncover more of the original theory which underlies the "classical" theory of categories.

3. A Neoplatonist version Porphyry, the most famous disciple of Plotinus, was also known because of his Eisagoge [Introduction (to Aristotle's Categories)], which was the beginning of any course of study of logic for more than a thousand years. This work explains the relation between genus (genos) and species (eidos) as follows: The individual is contained by the species and the species by the genus, for the genus is a kind of whole, the individual a part. The species is both a

Genus, species, and vantages

369

whole and a part, a part of another and a whole, not of another but in others. The whole is in the parts. 9

Evidently the relation between the genus and the species is the same, for Porphyry, as the relation between the species and the individual. An individual is a part of a species, but a species is also a part of a genus. This is rather different from the theories which we considered earlier, for there genera and species were related at the same level (the level of categories), whereas the relation between species and individuals was a relation between levels (the level of categories and the level of entities). But there is an even more important difference implicit here, as we shall see. Porphyry gives us a concrete example of how we are to understand the relation between genera and species. Let us make the meaning clear with reference to one category. Substance is itself a genus; under this is body; and under body animate body, under which is animal; under animal is rational animal, under which is man; under man are Socrates, Plato, and particular men. Of these substance is the highest genus, and it is genus only, while man is the lowest species, and it is a species only. Body is a species of substance but a genus of animate body. Animate body is a species of body but a genus of animal. Animal is a species of animate body, but a genus of rational animal. Rational animal is a species of animal, but a genus of man. Man is a species of rational animal, but it is not also a genus of particular men. It is a species only. Every species which is predicated immediately prior to individuals will be a species only, never a genus. 10

This is the so-called "tree of Porphyry" that was the model for the urge to classify everything which dominated seventeenth-century science. 11 What is interesting about it here is the way the relation between genus and species can combine with itself to form a larger "tree" structure. We begin with something very general (in this case, with "substance"), and make this the fixed background. We then separate out a mobile foreground (in this case "body", which is movable) from a certain vantage (which is not clear in this case). And then we make that mobile coordinate be the new fixed background, and separate out a new mobile foreground (in this case "animate body", which moves), and so on. In other words, we "zoom in" on the individual. Another way to see the difference between Porphyry's theory and Linnaeus's taxonomy is to count species. For Linnaeus there were as many species as there were primordial forms. But for Porphyry every genus (except for the ten at the top) could also be a species. So there were very many (exponentially many) more species than there were primordial forms. The species were not "fixed" in the same way. Rather, just as in

370

Jeff Lansing

vantage theory, the relation between genus and species was more like a process which could be applied to almost anything, even itself. In the next section we will see how this (almost) unrestricted process was itself a corruption of an earlier, more restrained theory.

4. Aristotle's partonomy Barbara Tversky and Kathaleen Hemenway, in an attempt to find out what it is that makes some categories more "basic" than others, distin~ guished partonomies from taxonomies: Knowledge organization by parts (partonomy) [can be] contrasted to organization by kinds (taxonomy). Taxonomies serve to organize numerous classes of entities and to allow inference from larger sets to sets included in them. Partonomies serve to separate entities into their structural components and to organize knowledge of function by components of structure. (Tversky-Hemenway 1984: 169)

This sort of functional definition, while it is useful for testing whether a given organization is a partonomy or a taxonomy, does not tell us what sort of organization to look for, beforehand. However, we already have an example of taxonomy, in Linnaeus's early work, so all we are missing is an example of a partonomy. Pierrre Pellegrin has argued (persuasively, evidently, since his argument has been widely accepted) that Aristotle intended his version of the "clas~ sical" theory to be, not a taxonomy, but a partonomy. 12 We can see this by looking at how a definition of an animal (or a plant) species would go, for Aristotle. Such a definition would require two sorts of procedures, division and coordination. Division develops within a homogeneous do~ main by "fasten[ing] on a function or on an organ proper to that func~ tion. Thus one may validly divide footed animals into 'solid~footed' and 'split-footed'" (Pellegrin 1985: 100). In this way division gives us species of parts, but not species of animals. Coordination, the second stage in the definition of an animal (or a plant) species, combines several species of parts simultaneously, instead of following a single line of division, as we saw happening in Porphyry's theory. Aristotle explains coordination in Politics 4: Therefore just as, in case we intended to obtain the kinds (efde) of animal, we should first define the properties belonging to every animal (for instance some of the sense-organs, and the machinery for masticating and for receiving food, such as a mouth and a stomach, and in addition to these the

Genus, species, and vantages

371

locomotive organs of each of them), and if there were only so many necessary parts, but there were different varieties of these (I mean for instance certain various kinds (gene) of mouth and stomach and sensory organs, also of the locomotive parts as well), the number of possible combinations of these variations will necessarily produce a variety of kinds of the animal (eide zdou) (for it is not possible for the same animal to have several different sorts of mouth, nor similarly of ears either), so that when all the possible combinations of these are taken they will produce as many species (eide) as there are combinations of the necessary parts. 13

Pellegrin takes care to point out the hypothetical character of this passage (this is how a species would be defined if we were to attempt doing so). For our purposes, we can note than by the genera (gene) of mouth, stomach, etc. Aristotle means different varieties of these parts. One interpretation of this is that the genus mouth is the undifferentiated collection of (varieties of) mouths, and that each of the certain various kinds is a species or form (efdos) of mouth: the narrow mouth, for instance, or the broad mouth. Pellegrin goes on to point out that in fact Aristotle never actually uses coordination to define an animal species, but that there are interesting uses of this procedure, nonetheless. In Parts of Animals, 1.3, Aristotle, arguing against the Platonic tradition, shows that the method of dichotomy (such as we saw used in the construction of the tree of Porphyry) can never arrive at an actual animal species; actual species are too complex to be defined by a single differentia. For example, web-footed birds (which are bipeds) have a broad beak which enables them to dig for roots easily, just as pigs are rooteating quadrupeds because their broad snout enables them to dig. 14 So "broad-mouthed", which is a sort of "mouthedness" (which in turn is a way of having head parts), cuts across the categories produced by "biped" and "quadruped", which are sorts of "footedness". Pellegrin (still contending that Aristotle never uses coordination to define animal species) looks at the following section of this argument, which shows that the method of dichotomy has to arrive at a single differentia. And clearly there cannot be more than one such differentia [sc. distinguishing feature, such as biped or toed]; for by proceeding continuously one reaches the last differentia (though not the final differentia which is species). And this is either the toed alone or the whole compound, if one divides off man, for example, putting footed with biped or toed. If man were merely a thing with toes, this method would have shown it to be his one differentia. But since in fact he is not, he must necessarily have many differentiae not under one division. IS

Aristotle's (rather weak) point here seems to be that, since man is not the only animal with toes, one differentia is not enough to define an animal

372

Jeff Lansing

species. But Pellegrin brings out a much more subtle point. Pellegrin notices that in this passage Aristotle is actually talking about two different axes of division: number of feet, and shape of feet. (For "two-footed" is not a sort of "toedness", nor is "toed" a sort of "two-footedness".) So the real point of this passage is that since man is both two-footed and toed Gust as ducks are both two-footed and broad-mouthed), then the method of dichotomy, which can only arrive at "two-footed" or at "toed", can never define man. The confusion here was due to the fact that both number and shape were applied to the same organ -the foot. Thus "for Aristotle the homogeneity of the domain in which division is carried out must be very rigorous: to delimit an axis of division it is not enough to limit it to one function or even to one organ, one must also apply division to that organ according to the same point of view" (Pellegrin 1985: 103, original emphasis). Restated in terms of vantage theory, Pellegrin's point is that for Aristotle three things were required, in order to delimit an axis of division. First, there must be a fixed coordinate, which he calls the homogeneous domain of division. Next, the mobile coordinate, usually an organ or part of the animal, is required. And finally there has to be a vantage point, which Pellegrin calls the point of view. 16 These three things together will give us the different species of parts, such as "two-footed", or "toed". In turn, taking combinations of these species of parts together simultaneously will give us species of animals. In other words, Aristotle's partonomy consists of vantage theory, together with a coordination procedure. In summary, "knowledge" can be organized either by a taxonomy or by a partonomy (or by a systematics). Organization by partonomy requires two procedures in order to define naturally occurring species: division from a certain point of view (which is vantage theory), 17 and coordination. Division by function is not enough. In their study of "basic" categories, Tversky and Hemenway claim that "for object and biological categories, primary or basic category cuts seem to follow natural breaks in the correlational structure of attributes in the world". They argue that such breaks "are determined by part configuration", where "separate parts will have separate functions, ... similar parts will have similar functions, ... [and] together, parts form an organized, integrated, functional whole" (Tversky- Hemenway 1984: 189). This is not so, however, for the (presumably) basic category MAN. And this is because, as we have just seen, it is not enough to limit an axis of division to one function, or even to one part. Tversky and Hemenway neglected to include vantages in their account.

Genus, species, and vantages

373

5. Conclusion Aristotle analyzed substance as a combination of: (i) the matter which was at work being that substance, and (ii) the form which the substance has under its description as that particular substance. 18 The matter was the undifferentiated .background, which vantage theory calls the fixed coordinate. The form-a term that in both Greek (efdos) and Latin (species) refers to what is seen -was the foreground, which vantage theory calls the mobile coordinate. Arid the description or identification of the substance from a certain point of view, is what vantage theory would call the vantage. Aristotle suggested that this theory could be applied to biology, and that when combined with a coordination procedure, could delimit the varieties of animals which everyday experience reveals to us. So in this original application, vantage theory was (part of) the way to delimit species. Porphyry, in carrying out the Neoplatonist program of reconciling Aristotle's work with Plato's, made a subtle change, whereupon the significance of categorization theory was transferred from the process of delimiting categories to the product of that delimiting process. As A. C. Lloyd explains, Porphyry's was a pure logician's program, wherein categories, as "semantic representations", were explicated solely in terms of words and entities (Lloyd 1990: 53). 19 Linnaeus took Porphyry's semantic representations (which- as Lloyd [1990: 49-53] argues-were concepts for Porphyry) and made them actual. In his taxonomy they were primordial creations, deriving their actuality directly from God. Later, in his systematics, some were primordial, while others were actualized through hybridization. The structuralist version of categorization theory tried to maintain the notion, from Linnaeus's systematics, that there were higher ("primordial") and lower ("hybrid") categories, with the lower ones somehow deriving part of their reality from the higher ones, while divorcing the resulting structure from actuality. It is still possible, though, to recognize in this version the original vantage theory, hidden beneath the layers of historical change. Notes 1. See MacLaury's (this volume) introduction to vantages. I should point out that I am not using the complete theory here. 2. According to its author, this theory is classical in two senses: first, "in that it goes back ultimately to Greek antiquity", and second, "in that it has dominated psychology,

374

JeffLansing

philosophy, and linguistics (especially autonomous linguistics, both structural and generative) throughout much of the twentieth century" (Taylor 1989: 22). 3. Note that whether or not the two features mentioned actually do define the category in question is not at issue here. 4. As Joan Kung pointed out, essential features, for Aristotle, were not the same thing as necessary features (even though there could be overlap). For example, [WINGED] seems to be an essential feature for bats, since bats without wings would not be bats, but it is not an essential feature for ants, since the same species of ants can have members with and members without wings. This opens up the possibility that for Aristotle it was the way that an entity had features that made the difference (cf. Kung 1977: 367). 5. One could object that having features with the greatest number of consequences is not a property of an entity because it is a statistic, but that is irrelevant for the classical theory, since the status of features is only an issue at the level of essences, and so is independent of the level of entities. 6. J. Ramsbottom (1938: 200) writes that "so far as I have been able to discover, Linnaeus's statements about the fixity of species are the most definite made up to his time". 7. Species tot numeramus, quot diversae formae in principia sunt creatae. (Quoted in Ramsbottom 1938: 196). 8. Whether this will eventually work or not is still being debated in the pages of Cladistics and Systematic Zoology. 9. Translation by Warren (1975: 41). 10. Translation by Warren (1975: 35-36). 11. See Slaughter 1982. 12. Pellegrin calls Aristotle's version a "moriology" (after Gr. moria 'part'). 13. Translated in Pellegrin (1985: 101). 14. Cf. Part of Animals, 3.1. 15. Translation by David Balme, quoted in Pellegrin (1985: I 02). 16. This suggests an explanation for Joan Kung's conclusion that it is the way an entity has a feature that matters. (See note 4 above.) Having feet, from the point of view of the number of feet, is a way of being footed. 17. Meaning, of course, just the fragment of the full theory with which we have been working here. 18. Cf. Kosman (1987: 372). 19. More precisely, Porphyry's program was a pure model theorist's program, as opposed to Aristotle's more proof-theoretic methods.

References Eriksson, Gunnar 1983 "Linnaeus the botanist", in: T. Frangsmyr (ed.), Linnaeus: The man and his work. Berkeley: University of California Press, 63-109. Kosman, L. Aryeh 1987 "Animals and other beings in Aristotle", in: A. Gotthelf-J. G. Lennox (eds.), Philosophical issues in Aristotle:~ Biology. Cambridge: Cambridge University Press, 313-338. Kung, Joan 1977 "Aristotle on essence and explanation", Philosophical Studies 31: 361-383. Lloyd, Antony C. 1990 The anatomy of Neoplatonism. Oxford: Clarendon Press.

Genus, species, and vanta);eS

375

Pellegrin, Pierre 1985 Aristotle: A zoology without species", translated by A. Preus, in: A. Gotthelf (ed.), Aristotle on nature and living things. Pittsburgh, PA: Mathesis Publications, 95-115. Ramsbottom, J. 1938 "Linnaeus and the species concept", Proceedings of the Linnean Society of London 150: 192-219. Rieppel, Olivier 1991 "Things, taxa, and relationships", Cladistics 7: 93-100. Slaughter, Mary M. 1982 Universal languages and scientific taxonomy in the seventeenth century. Cambridge: Cambridge University Press. Taylor, John R. 1989 Linguistic categorization: Prototypes in linguistic theory. Oxford: Clarendon Press. Tversky, Barbara- Kathleen Hemenway 1984 "Objects, parts and categories", Journal of Experimental Psychology: General 113(2): 169-191. Warren, Edward W. 1975 Porphyry the Phoenician: Isagoge (translation, introduction and notes). Toronto, Ontario: The Pontifical Institute of Mediaeval Studies.

On construing the world of language Nigel Love

Discussion of the connections between language, thought and culture tends to focus on how the language we speak influences the way we think about the world and the cultural practices we erect in the light of our language-influenced Weltanschauung. Less often explored is the influence of culture on thought about language itself. Unlike questions such as "What is a laser?" or "What is a Barna?", the question "What is a language?" is not one that admits of the kind of value-free answer that might be supplied by the physicist or the zoologist in response to the former. On the contrary, the answer to "What is a language?" is bound to be a product of culture-specific conditioning of various kinds. This state of affairs is somewhat obscured by the fact that, since the nineteenth century at least, mainstream Western linguistics has prided itself on being the "science of language", and therefore projects an image of itself as culture-neutral, in the sense that physics, chemistry or biology are culture-neutral. That is to say, while nobody denies that Western science is the product of a particular culture, its findings, if true, are held to be universally true. The laws of physics do not cease to hold in those parts of the world where the inhabitants' intellectual history is such that they could not have formulated them for themselves. The facts vouchsafed by the science of human physiology are what they are, even for those who have no inkling of them. And so on. If, for instance, Chomsky's theory of language acquisition is valid at all, then the claim that it is a scientific theory implies that it is and always has been valid for all human beings, irrespective of the particular ideas about language that may be entertained at different times and places in different human societies. That the Western "science of language" in fact lacks the cultureneutrality necessary for scientific status may be suggested by considering, in general terms, the answer to the question "What is a language?" offered by modern linguistics, and where it comes from. Mainstream linguistic theory since Saussure rests, first, on asserting the primacy or priority of spoken language. Speaking-hearing is the only mode of language use with which the linguist seriously concerns him- or herself. Other modes are treated as secondary, or ancillary. Secondly,

378

Nigel Love

mainstream linguistics sets out to explain spoken language by assuming the existence of cotemporal totalities of context-free verbal possibilities. That is to say, "language", as a form of behavior manifest in the course of specific social interactions, is explained with reference to abstract systems called "languages", each consisting essentially of a fixed and specifiable set of correlations between "forms" identified at various levels, on the one hand, and their "meanings", on the other. Any given piece of linguistic vocal behavior-that is, any utterance-is treated as a concrete instance of some part of the abstract system in question, which is held to exist in advance of any actual utterance in which it is brought into play, and which remains invariant across differences of various kinds between the contextual circumstances of actual speech events. To say this is to generalize over a broad range of possible attitudes to abstract linguistic systems. One linguist may be involved simply in describing them. Another may be mainly interested in the light shed by their structural properties on the nature and organization of the human mind. Or this latter concern may be taken to the point where "languages" themselves are seen as merely epiphenomenal, and hence no longer the real object of study (see, e. g., Chomsky 1986). For present purposes these disagreements do not matter: all that needs to be established here is that behind spoken language, as an observable behavioral phenomenon, the linguist sees languages, in one or another version of the sense just outlined. Why should this be so? It is not as though spoken languages are simply there for the finding, by anyone who inspects the empirically observable phenomena of speech and verbal interaction. Nor do linguistic theorists waste much time in trying to justify their postulation of "languages" with reference to some realm or dimension of observable fact. Saussure was the first to advocate a science of spoken language based on the concept of a cotemporal totality of context-free verbal possibilities. Indeed, that rather cumbersome phrase can more or less stand as a definition of the Saussurean langue. But Saussure does not show-in fact makes no attempt to show-that the langue is a reality, in any sense like that in which the chemist, say, shows and is obliged to show that molecules are a reality. When he speaks of the idiosynchronic hat de langue, Saussure is not referring to some fact about the organization of linguistic phenomena that he has empirically discovered, but to a postulated theoretical entity to which he proposes to attach the name "a language" (langue). To take another example, Chomsky's version of synchronic structuralism is of course rather different from Saussure's (and Chomsky's more

On construing the world of language

379

recent theorizing may be said to have taken him out of the realm of structuralist linguistics altogether). But in respect of the concept of a language in play, at least in his earlier work, the similarity is striking. Chomsky starts, or started, by setting up as his object of study a so-called "ideal speaker-listener" located in a "completely homogeneous speech community" and endowed with the mysterious attribute of knowing his language "perfectly" (Chomsky 1965: 3). This idealization clearly serves as a device for ushering in the concept of a language under discussion. The idea of a language as something that might in principle be known "perfectly" is the idea of something fixed and, at least in one sense, finite. And since the set of speech events in which a language is held to be manifest is not in any sense finite, the "language" here must be understood as a delimited abstraction from speech events. In other words, a context-free totality. While "complete homogeneity" is the stipulation that ensures that this totality is analyzable as a consistent and uniform linguistic system. So the twentieth-century Western science of spoken language is concerned with the analysis of objects that are not in any obvious sense "given" to experience. Where does the idea come from that there are such objects? Or, to put the question somewhat differently, what is it that lends plausibility to the general idea of the cotemporal totality of context-free possibilities as a postulate on which to found a science of spoken language? What makes the Saussurean idea of the synchronic language state at least seem like a realistic basis for the analysis of verbal behavior? Why, when we first encounter the Chomskian "ideal speaker-listener", is there a tendency to think that we are dealing with an unobjectionable statement of an appropriate and necessary idealization? The Cours de linguistique generate (Saussure 1916) is founded on an explicit appeal to the reader's sense of what languages are like for the individual language user. Saussure's objection to the nineteenth-century historical linguistics that he hoped to supersede was that it dealt in diachronic changes operating over such long time spans that they do not impinge on the individual's linguistic experience. Although the individual unconsciously participates in the process of bringing about such changes, they are not real to him. So what then is real to him, where language is concerned? What does he experience? Saussure's answer is that the individual experiences a spoken language as a cotemporal totality of contextfree verbal possibilities-and, what is more, Saussure's reader is at least initially inclined to agree with him. Given that the world does not on inspection appear to contain any such things, why should that be? Be-

380

Nigel Love

cause of a culture-specific background of ideas about language shared by Saussure and his readers. There can be few who have encountered the Cours de linguistique generate without prior exposure to what some linguists disparagingly refer to as "traditional grammar". In Saussure's time, the embryonic discipline called "general linguistics" was pursued by students who would have had a thorough instruction according to traditional methods in one or more foreign languages. And it is traditional grammar that has bequeathed to modern linguistic science its commitment to the idea that behind speech there are objects called "languages". For the traditional grammarian a language consists of a fixed inventory of microunits (words) displayed in a dictionary as a set of correspondences between forms and meanings, which are combinable into macrounits (sentences) according to the rules laid out in a grammar book. The network of ideas underlying traditional grammar involves imposing on the continuum of linguistic differences between people at different times and places an analysis in terms of discrete linguistic systems (languages). Traditional grammar abstracts from the interactive behavior deemed to involve a given such system what it sets up as the strictly linguistic aspects of that behavior. It then projects this abstraction as a body of knowledge which, if acquired by a learner, might be put to use in interactive episodes with existing speakers of the language in question. This conception of languages is indeed primarily a pedagogical tool, in use in Western educational institutions since classical antiquity for teaching them. The traditional pedagogue's idea of what a language is has a number of features worth stressing in the present context. First, it emphasizes the idea of a language as an object; language use as a form of behavior is treated as secondary to the object-indeed, as being founded on, and only possible because of, knowledge of the object. This feature is the basis for the well-known distinctions, in Saussurean and Chomskian versions of structuralist linguistic science, between langue and parole on the one hand, and "competence" and "performance" on the other. Second, a language as presented by the grammarian is a synchronic (or cotemporal) object. Learning Latin, French or German is not the same thing as learning the history of Latin, French or German. And this is where the idea of the synchronic language state gets its plausibility. We are inclined to think of languages in this way, not because that accords with our day-to-day experience as language users, but because we have undergone an overt training in languages that presents them thus.

On construing the world of language

381

Third, traditional linguistic pedagogy treats a language as a closed system-as a totality. Take the most detailed grammar book and the largest dictionary, and there, for practical purposes, you have it: the whole language. Chomsky's idea of a language as something that might be known "perfectly" is a projection or transposition of the idea that one might, in principle, arrive at an exhaustive knowledge of a language, conceived as the totality of the information contained in the grammar book in combination with the dictionary. Fourth, the fixed synchronic system upheld by the language teacher does not allow for dialectal or regional variation. If the language in question is, say, French, then it is a homogeneous, monolithic standard Parisian literary French that is taught. This idea, too, is simply taken over by the modern structuralist science of language; both Saussure and Chomsky assume, for all theoretical purposes, a homogeneous speech community. In all these ways, what purports to be a culture-neutral science of language embodies a conceptualization of languages that was already in daily use for purposes of formal linguistic education in the culture whose product that science is. To say the least of it, that is a remarkable coincidence. What, in turn, are the cultural prerequisites for traditional grammar itself? First, there is a particular concept of linguistic correctness. The idea of a completely homogeneous speech community is a linguist's fiction. Not only is every individual different, his experience of language is unique, and it is probably beyond his vocal control to replicate his utterances exactly. These features of the human condition ensure that language will always display phonetic, phonological, morphological, syntactic and lexical variation, in both time and place. Like any other kind of behavioral differentiae, linguistic variants are liable to social evaluation. That is the general reason why there are such things as linguistic norms and concepts of linguistic correctness; and we may take it as universally the case that societies will entertain some such notions. But their precise nature may be expected to vary from culture to culture. One of the key metalinguistic ideas of Western culture is that it makes sense to assess the appropriateness or acceptability of linguistic expressions judged in a dimension of "pure linguisticity". That is, as well as recognising that the unique utterance may, in its context, be esthetically or pragmatically good or bad, clever or foolish, right or wrong, efficacious or communicationally inadequate, etc., we also recognise a property (sometimes called "grammaticality") that potential utterances either

382

Nigel Love

have or do not have in the abstract. This dimension of correctness is perhaps the most significant and pervasive characteristic feature of the Western attitude to language. 1 In the Western tradition the idea of correctness is bound up with the history of grammatical inquiry, whose origins may be traced to the preSocratic philosophers of the fifth century B. C. 2 Grammar emerged as one facet of what the Greeks called logos, a concept that embraced not only language but also the capacity for articulate discourse and the power of reason itself. Rationality came to be the subject matter of the modern discipline whose name is an etymological continuation of logos; it is the lack of a clear distinction between the concepts "language" and "reason" that explains why early Greek grammatical speculation is so tightly connected with early Greek consideration of logic. Nevertheless, with the Stoics grammar began to emerge from about 300 B. C. as an autonomous branch of inquiry. The Stoics were the first to be interested as much in linguistic form as in semantics. The Alexandrian grammarians continued the Platonic-Aristotelian-Stoic tradition, which culminated in the grammars of Dionysius Thrax (late second century B. C.) and Apollonius Dyscolus (second century A. D.). It has been said that subsequent grammarians up to the present day have done little more than copy or reword these two works. Dionysius attempted to dissociate his work from philosophy more rigorously than any of his predecessors. This stance is evident in the definition of techne grammatike 'the art of writing' that begins his treatise. According to Dionysius the domain of grammar is the usage of poets and prose writers. It has six parts: (i) accurate reading, with due regard for prosody; (ii) exegesis of poetical figures; (iii) annotation of phraseology and allusions; (iv) discovery of etymologies; (v) an account of linguistic patterns; (vi) criticism of poetical works ("the noblest part of grammar"). All that "grammar" is now understood to comprise is given under (v). In the subsequent development of grammatical studies the linguistic patterns in question came to be, essentially, those that could be dealt with by the application of sets of distinctions pertaining to word classes (the "parts of speech"), word composition (prefix, suffix, etc.), grammatical categories (number, mood, tense, etc.), syntactic relations among words (agreement, government, etc.), sentence types (negative, interrogative, etc.) and sentence composition (subject, predicate, clause, etc.). The grammars of both Dionysius and Apollonius are clearly focused on literary norms rather than spoken usage. The implication that literary writing is the only form of language

On construing the world of language

383

worth analyzing in this way was to be profoundly influential in the subsequent history of Western linguistic thought. The Romans learned the grammatical concepts and categories designed for and from Greek and applied them to Latin. Faced with discrepancies between the structures of Greek and Latin, Roman grammarians often chose to misrepresent their native tongue rather than compromise the authority of (Greek) grammar. This testifies to the cultural reverence that the Romans had for Greece. In any case, for the Roman the study of Greek was a prerequisite for all other learning. So the Romans came to the examination of their native tongue with certain linguistic ideas firmly entrenched: for instance, that there are three grammatical numbers (singular, dual and plural); that there are five verbal moods (indicative, subjunctive, optative, imperative and infinitive); and that the noun phrase may include a definite article. The fact that Latin has neither a dual nor an optative nor a definite article did not deter Roman grammarians from using these categories in their analyses: the forms of the dual, it was said, happened to be identical with those of the plural-except where some zealous hellenizers went so far as to attach Greek dual inflections to Latin words. The optative, which Greek itself lost in the Alexandrian period, continued to be listed in grammars of modern language until quite recent times. Even Varro (116-27 B. C.), the Roman grammarian credited more than any other with originality of thought, sensitivity to the Latin language, and an inclination to reason about his observations and classifications rather than merely follow the Greek model, included articuli as a category of Latin grammar, listing under it the demonstratives is and hie. In this adherence to a mode of description designed for a different language we see an important dimension of grammatical prescriptivism: the implication is that where the structure of Latin differs from that of Greek we are dealing with an inferiority that ought in principle to be remedied. This attitude is a recurrent topos in European linguistic studies: in its turn the grammatical structure of Latin came to be a norm such that, to the extent that they deviated from it, the modern European vernaculars were seen as deficient or degraded. Our evidence for what a Roman education in grammar was like is chiefly provided by Quintilian (b. 35 A. D.). 3 In his Institutio Oratoria Quintilian, unlike Dionysius Thrax, regards grammar as divisible into just two parts: (i) interpretation of the poets (poet arum enarratio); and (ii) the study of correct expression (recte loquendi scientia). As far as the latter is concerned, the most important factor to be taken into account in establishing the grammatical correctness of an expression is usage (con-

384

Nigel Love

suetudo), which Quintilian defines as "the agreed [linguistic] practice of educated men". The significance of this depends on how we identify the "educated". Quintilian's answer is the educated are those who have followed the approved educational curriculum, as laid down by educators such as Quintilian. The authority for linguistic correctness is thus ultimately vested in the educational system itself. Quintilian's argument sets the seal on a linguistic prescriptivism that was to dominate European culture thereafter. The prescriptivism was self-sustaining. Grammar defines correct usage because it is based on the consensus of the educated, while the educated speak correctly because they have studied grammar. Unlike so many other aspects of classical culture, the Graeco-Roman grammatical tradition was maintained throughout the Middle Ages, though essentially on the basis of just two works: the grammars of Donatus (d. 335) and Priscian (d. ?526). Neither deviated greatly from the format established by the Alexandrian grammarians. Donatus and Priscian remained in continuous use all over Europe throughout the Middle Ages, and in the Renaissance they rose to great prominence. Donatus was to be the model for all pedagogical grammars for centuries; his name became synonymous with grammar itself. Priscian's importance is attested to by the preservation of his grammar, despite its length, in over a thousand manuscripts in various European libraries. The grammatical tradition inaugurated in classical antiquity combined with the linguistic self-consciousness engendered by post-Renaissance European nationalism to usher in a great project of "standardization" of the languages of European nation-states. Politics is the most important factor in the process by which a language or dialect is chosen for standardization. It is unlikely that all the communities that see themselves, however vaguely, as constituting a political unit will be equal in political power. They will vary in population, natural resources, acquired wealth, the establishment of industries and governmental bodies, and even political will. Only one community will be recognised as the region's capital, where political and cultural-including linguistic-institutions will be concentrated. The actual mechanism whereby a politically favored linguistic variety becomes standardized consists essentially in the suppression of alternative modes of expression- of free variation in the use of communicationally equivalent and therefore competing inflectional forms, grammatical constructions, lexical items, spellings, etc. Suppression may take the form of either simply condemning and discouraging the use of the disfavored alternative, or else of fostering a potential or incipient functional differen-

On construing the world of language

385

tiation of the variants. This weeding of the linguistic garden has as its objective the "fixing" of the language, and it is the chief task of the prescriptive grammarian. Fixing the language in this way is important because: (i) the great languages of classical antiquity were so fixed; (ii) a fixed language is in itself (not to mention a course of instruction in the prescriptions arrived at via the fixing process) held to be conducive to clarity and precision in the use of words; (iii) some of the communicational purposes for which a standard language is required (certain juridical processes, for instance) demand it. For a standard language, as compared with a nonstandard variety, displays "elaboration of function". It has to be usable across the whole range of legislative and administrative operations carried out by the government of a state, and for educational, scientific and literary purposes of various kinds; all of which require a written language with a high degree of uniformity and, in many cases, a large and extensible technical vocabulary available only to the literate. A standard language is, therefore, primarily a written language. It may, of course, have a standard oral interpretation-that is, a particular way of speaking a standard language may come to be a prestige norm-but this not need happen and, if it does, it is essentially independent of standardization as such. (Thus one may talk of a standard language being spoken with different accents.) Indeed, it is only when a language has associated with it the consistent use of a writing system that the question of standardizing it can arise. Acquaintance with the practice of writing is perhaps the most fundamental prerequisite for the Western concept of a language. The notion of correctness in the dimension of "pure linguisticity" is crucially dependent on it. The articulation of norms governing classes or types of utterance, considered in abstraction from contextualized speech events, requires that the verbal aspect of contextualized speech events be analyzed as manifesting recurrently instantiated abstract invariants. A distinction between classes and member, or types and tokens, is needed. So long as language is purely spoken language there is no possibility of drawing such a distinction in a concerted or systematic way, partly because purely spoken language is purely context-bound language, and partly because purely spoken language cannot yield the consistently reliable metalinguistic discourse required for articulating it. Where attempting to generalize over unique utterances is concerned, a pre- or nonliterate culture faces the difficulty that there is nothing to do it in except unique utterances themselves. One important part of the problem is how to make it clear that some utterances are to be taken as utterances, whereas others are to be

386

Nigel Love

taken as the names of something more abstract than utterances. It may be suspected that primeval attempts to talk about language were considerably hampered by the logically imposed fact that the "names" of utterance types are homophonous with the utterances themselves. ("Utterance type" here means the sense in which, for example, dogs bark is the type of which a particular utterance "dogs bark" is a token. The typographical differentiation has no oral counterpart.) Talk about utterance types sufficiently stable and thoroughgoing to permit the concept of a language as a complete and coherent system of types requires a radically new development. That new development was the use of writing as a linguistic medium. Writing meets the need for a consistently reliable concrete correlate of the idea of apprehending a unique utterance as an instance of an underlying invariant by providing a means of citing the invariant in question in a medium other than oral utterance itself. It was Saussure who inaugurated the tradition whereby the modern science of language claims to offer descriptive (rather than prescriptive) accounts of the structure of spoken (rather than written) languages. In modern linguistic theory it is common ground: (a) that there is a sharp distinction to be drawn between prescribing and describing usage, and that the descriptive linguist, unlike the traditional grammarian, is engaged in the latter task; and (b) that the oro-aural medium has a claim on the linguist's attention that precludes consideration of what is seen as no more than a convenient ancillary notation. But this declaration of intent is undermined (a) by the essential normativity of grammar, and (b) by the dependence of any conception of grammar on writing. In the first place, vindicating the idea that spoken language has any consistent structure at all demands of the analyst a cutting and 'smoothing of the raw data of speech, at all levels of description from phonetics to semantics. Every spoken utterance is a unique physical act, and at the finest level of detail no two utterances are phonetically identical. However narrowly the subject matter of the description is defined -whether it be, e. g., "Modern English", "twentieth-century East Anglian English", "post-war Suffolk English", "post-1980 Ipswich fishmongers' English"phonetic norms have to be established for it, abstracting away from all sorts of idiosyncratic variations in the acoustic signal, both sporadic and systematic, ranging from differences in voice quality to differences in the precise articulatory realization of certain phones. As far as morphology and syntax are concerned, no descriptive linguist would be content merely to set down faithfully what he hears said. Spontaneous unscripted

On construing the world of language

387

speech is full of phenomena that have to be ignored if any analysis of its grammar is to be forthcoming at all. As for semantics, not only is it likely that different speakers, and the same speakers on different occasions, mean and understand different things by the words that they use and hear spoken, there could be no way that merely describing utterances could possibly yield their meanings anyway. So whatever "describing" spoken language means, it cannot in any obvious sense be solely a matter of attending to the properties of empirically observable phenomena. Hence the idea that utterances are superficially diverse manifestations of some portion of a delimitable, and therefore in principle exhaustively describable, linguistic system- a "language", envisaged as a set of invariant abstractions underlying the vagaries of speech. Furthermore, "describing" spoken language apparently requires the prima facie contradictory procedure of stating linguistic "rules". A modern grammatical description of a language often takes the form of a system of rules defining the sequences of words that constitute well-formed sentences of the language. Further sets of rules then specify how these sequences are pronounced and what they mean. Descriptive linguists have explicitly denied that their use of the term "rule" has anything to do with prescribing standards of correctness; and there has been prolonged debate among philosophers as to what in that case they might, or in principle could, mean by it (see, e. g. Itkonen 1978; Baker- Hacker 1984). What is clear enough, however, is that rules must in one sense or another imply norms. How are these notions compatible with the claim that what is being described is a phenomenal reality rather than a culturally determined theoretical construct? The answer is that the theoretical construct-the abstract, delimitable, determinable linguistic system -is (at least in the more basic, broadly defined cases, such as Modern English, if not always in the more precisely defined ones, such as post-1980 Ipswich fishmongers' English) held to be something real for speakers themselves. This is the justification for the conceptual distinction crucial to all versions of modern descriptive linguistics between langue and parole, or "competence" and "performance", such that the langue, or the speaker's linguistic competence, and not parole, or the speaker's actual linguistic performance, is the subject matter of the description. The idea, entertained no doubt by speakers themselves, that their speech instantiates (parts of) an abstract linguistic system, is founded on the invention of writing, without which the notion of an abstract, purely oro-aural linguistic system of any degree of complexity or comprehensiveness could not be grasped. And the further idea, also entertained by

388

Nigel Love

speakers themselves, that their speech is an imperfect, perhaps idiosyncratically distorted, concrete realization of (parts of) an abstract linguistic system is a product of a tradition of normative grammatical inquiry that itself depends on the availability of writing, and on the way that tradition has latterly been exploited in the elaboration of standard written languages for the nation-states of post-Renaissance Europe. In Saussurean theory these ideas manifest themselves in the postulation of a self-consistent, fixed and invariant language (a langue) that exists in the "collective mind" of a linguistic community. In Chomskian theory they underpin the postulation of an ideal speaker-listener. The relevance of prescriptive grammar lies in the implication, or outright assertion, that speech falls short of a norm enshrined in the langue of the collectivity, or the unconscious linguistic knowledge of the individual. The relevance of language standardization lies in the fact that such idealizations would be self-evidently absurd, if not inconceivable, were there not in fact a large measure of linguistic homogeneity prevailing in the communities whose linguistic theories these are-a state of affairs that does not antedate the introduction of compulsory universal education in Europe in the nineteenth century, and the inclusion in the curriculum of that education formal instruction in the standardized national language. Thus the most fundamental assumption of the modern science of language projects as a culture-neutral theoretical ideal a linguistic uniformity whose existence in fact-to the extent that it exists at all-is the culminating achievement of two millennia of socially and culturally conditioned normative thinking about language. Whether a value-free "science of language" is a realizable possibility at all must be doubtful. What is not at all doubtful, on the other hand, is that a would-be linguistic science must start by emancipating itself from what is no more than a profoundly important but nonetheless culturally parochial way of construing linguistic phenomena. Notes I. There is no reason to think that "grammaticality" in this sense is a universal or necessary metalinguistic concept. For instance, Miihlhiiusler (1987: 10) reports that "intuitions about ... grammaticality are notoriously unreliable or non-existent among speakers of many Pacific languages, in particular those with no training in a European standard language, as deciding on contextless grammaticality is not one of the metalinguistic games they tend to engage in". 2. For more detailed discussion of the historical background see, e. g., Robins (1967), Joseph (1987). 3. Cf. Harris-Taylor (1989: 59-74).

On construing the world of language

389

References Baker, G. P.-P. M.S. Hacker 1984 Language, sense and nonsense: A critical investigation into modern theories of language. Blackwell, Oxford. Chomsky, Noam 1965 Aspects of the theory of syntax. Cambridge, MA: M.I.T. Press. 1986 Knowledge of language: Its nature, origin, and use. New York: Praeger. Harris, Roy-Talbot J. Taylor 1989 Landmarks in linguistic thought: The Western tradition from Socrates to Saussure. London: Routledge. Itkonen, E. Grammatical theory and metascience: A critical investigation into the method1978 ological and philosophical foundations of "autonomous" linguistics. Amsterdam. Benjamins. Joseph, J. E. 1987 Eloquence and power: The rise of language standards and standard languages. London: Pinter. Miihlhausler, Peter 1987 "The politics of small languages in Australia and the Pacific", Language and Communication 7: 1-24. Robins, R. H. 1967 A short history of linguistics. London: Longman. Saussure, Ferdinand de 1916 Cours de linguistique glmerale. Paris: Payot.

Index of names

Abelson, Robert I 0 Abramov, Israel 270 n3 Achard, Michel 23, 45 Achitometl 292-294, 306, 322 n21 Alvarado, Pedro de 304, 310 Alvarado Tezoz6moc, Fernando 292, 313, 323 n25 Anales de Cuauhtitlan (Annals of Cuauhtitlan) 293 Anderson, Arthur J. 0. 298, 313, 320 nl, 322 nl9 Anderson, John M. 246 Andrews, J. Richard 306, 324 n30 n33 Anttila, Raimo 132 Aoyagi, Munekazu H. 18, 332, 335 Aristotle 18, 365, 368, 370-374 nl2 nl9 Atlas Jay D. 91 Austin, John L. 1-4, 6, 8-9, 11-12,20 Avicenna 160, 164-165 Baines, John 239 Baker, G. P. 387 Balme, David 374 nl5 Banfield, Ann 308 Barlow, Robert 310, 324 n33 Barsalou, Lawrence W. 91, 252 Bartholomew, Doris 31 Basso, Ellen 324 n35 Beek, H. H. 156 Bendix, Edward H. 77 n9 Bennett, David C. 97 Berlin, Brent 17, 232, 237, 249-250, 261-263, 267, 270-271 n4, 278-279, 316, 320 n3 Berlin, Elois Ann 237, 249 Besnier, Niko 137 Bierhorst, John 298, 323 n25 Bierwisch, Manfred 6 Bloomfield, Leonard 251 Boorde, Andrew 162 Boster, James S. 261-263,270-271 n4-5 Bourne, Edmund J. 277, 314, 316-318 Bowden, John 120-124, 127-128, 133

Bright, William 322 n 17 Brown, Penelope 332-333, 351 Brown, Roger 332, 336, 351 Brugman, Claudia 115, 120-121, 133,227 Bybee, Joan L. 130-131 Campbell, Lily B. 161 Carey, Kathleen 75 Casad, Eugene H. 6, 19, 23, 29-30,46 n6, 252 Casson, Ronald 320 n3 Cervantes 165 Cervantes Seevedra, Miguel de. See Cervantes Chapman, Allan 162 Chase, Paul 252 Chimalpahin 293, 313 Chomsky, Noam 377-379, 381 Chronicle of the Mexica 292, 313 Cincalco 30 I- 302, 323 n27 Clark, Herbert 120 Claudi, Ulrike 119, 121, 125 Clendinnen, lnga 322 n20 Cline, Susan L. 324 n36 Cronica mexicayotl 292, 313. See also Chronicle of the Mexica Cruttwell, Patrick 161 Cocox 293-294, 306 Colhua Mexica 292-294, 313 Conklin, Harold 316 Cook, Kennth W. 23, 45, Cortes, Heman 296. See also Cortez Cortez 296, 303, 324 n36 Cuauhtemoc 294-295, 299, 309-311, 313, 323 n23-24 Cuyckens, Herbert 249 D' Andrade, Roy 227 n6, 234 Davitz, Joel R. 150 n3 Dean, Paul 77 nl3 DeLancey, Scott 252 De Valois, Russell L. 270 n3

392

Index of names

Dibble, Charles E. 298, 313, 320 nl, 322 nl9 Diepgen, Paul 156 Dirven, Rene 6, 8-9, 19, 96-97, 99 Doctor Tissot 153-154 Douglas, Mary 238 Don Quixote 165 Donatus 384 Draper, John W. 161 Dray, Nancy L. 91, 256-257 Dr. Johnson. See Johnson, Samuel Duran, Fray Diego 322-323 n22 n27 Di.irer, Albrecht 161 Dyscolus, Apollonius 382 Egan, M. 234 Ekman, Paul 13, 168, 192 Elyot, Thomas 162 Eriksson, Gunnar 367 Esperson, Jane 227 n5 nlO Feld, Steven 265 Fillmore, Charles J. 77 n3, 101 Florentine Codex, The 277-278, 281, 286, 288-291,294-303, 306-309, 311-314, 319-320 nl, 324 n36 Fodor, Jerry A. 11-12 Forbes, Isabel 268 Friesen, Wallace V. 13, 168, 192 Galen, Claudius 157, 160 Galloway, Brent D. 321 Gardner, Peter 320 n3 Geeraerts, Dirk 10, 13-15, 30, 174, 177, 181-182, 184, 249 Gentner, Deirdre 285 Gibbs, Raymond W. 174 Gillespie, Susan D. 323 n27 Gilman, Albert 332, 336, 351 Giv6n, Talmy 23-24, 44 Godderis, Jan 156 Goffman, Erving 295, 307 Goldberg, Adele 227 n5 n 10 Goldstone, Robert L. 285 Grondelaers, Stefan 10, 13-15, 181 182, 184 Gumperz, John J. 331, 347

Hacker, P. M. S. 387 Haiman, John 308 Hage, Per 238, 261 Hallowell, A. Irving 284, 315 Hanks, William 313 Hanson, Craig 322 n18 Harris, Roy 388 n3 Harvey, William 158 Hawkes, Kristen 238 Hawkins, Bruce W. 23, 29, 45, 72 Heelas, Paul 284, 314, 316, 321 n15 Heider, Eleanor Rosch 249-250 see also Rosch, Eleanor Heider Heine, Bernd 10, 19, 119-125, 127, 131, 133 Hemenway, Kathleen 370, 372 Hering, Ewald 270 n3 Hill, Jane H. 18, 271 n5, 308 Hippocrates of Kos 153-155 Hollins, Mark 259-261 Hopper, Paul J. 46 Horcasitas, Fernando 323 n25 Huitzilopochtli 292 Hunn, Eugene 255 Hi.innemeyer, Friederike 119, 121, 125 ibnSina. See Avicenna Intons-Peterson, Margaret 246 Irwin, James R. 156 Itkonen, E. 387 Itzcoatzin 304-305, 310 Izutsu, Toshihiko 268 Jackendoff, Ray 4, 28-31 Jackson, Stanley W 165 Jacobs, Gerald H. 270 n3 Jakel, Olaf 15, 19, 197, 226-227 n5 Johnson, Mark 13, 15, 23, 29, 52, 95, 137, 174, 183, 197, 199. 203, 205, 219-220, 225-227 n2, 247, 262, 264 Johnson, Samuel 162 Johnson-Laird, Philip 8, 29 Joseph, J. E. 388 n2 Kail, Aubrey C. 161 Karttunen, Frances 289, 322 n17 Kay, Paul 4, 17, 232, 237-238, 247, 249-250, 267, 270 n3, 278-279, 316

Index of names Keenan, Janice M. 245 Kelley, Elizabeth K. 259-261 Klibansky, Raymond 156, 161, 164 Klor de Alva, Jorge 312, 326 King, Brian 185, 190, 193 Kondo, Dorinne K. 324 n37 Kosman, L. Aryeh 374, nl8 n4 Kovecses, Zolt:in 10-15, 137-141, 145, 147, 149-50 n3 n9, 155. 166-177, 181, 184, 190, 265 Kung, Joan 374 n4 nl6 Kutscher, Gerdt 323 n25 Labov, William 297 Lakoff, George 4, 13, 15, 23, 32, 52, 77 nl n3, 81-85, 87-90, 92, 95, 137, 150 n2 n6, 166, 174, 181, 183-184, 190, 197-199, 203, 206, 219-221, 225-227 n2 n4-7 nlO, 246, 261-262, 264, 317 Langacker, Ronold W. 4, 6-7, 19, 23, 25-27, 31-32, 35-37,41-42,44,46 nl n7, 52, 54, 56, 63, 67, 75-77 n2 n4 n5 nl4, 78 nl6, 252, 256, 279, 331 Lansing, Jeff 18, 117nl,249 Larson, Mildred 306 Lass, Roger 133 n9 Lebra, Takie S. 144, 338, 344-345 Leech, Geoffrey 306 Lehmann, Walter 293, 323 n25 Lehrer, Adrienne 249 Lekens, Benjamin 133 n5 Leon-Portilla, Miguel 278, 310 Levenson, Robert W. 13, 168, 192 Levinson, Stephen C. 332-333, 351 Lewis, Clive S. 3 Li, Charles, N. 307 Linde, Charlotte 297 Lindeboom, Gerrit A. 156 Lindner, Susan 23, 29-30, 45 Linnaeus, Carolus 18, 366-369, 373-374 n6 Lloyd, Antony C. 373 Lock, Andrew 284, 316, 321 nll nl5 Lockhart, James 289 Lodge, David 102, 105, 107, 109-113, 115, 118n8 Longacre, Robert 306, 308

393

Lopez Austin, Alfredo 289, 321 nl4 Love, Nigel 19 Lucy, John A. 320 n3 Lutz, Catherine A. 137, 194-195, 315 Lyons, John 78 n 15, 246, 266, 321 n6 McAfee, Byron 310, 324 n33 Macaulay, Monica 120 McDaniel, Chad K. 247, 270 n3, 281 MacLaury, Robert E. 16-19, 120, 133, 232, 234, 237-238, 241, 244-246, 248-249, 251, 270-271 n3 n5, 277, 279, 283, 285, 311, 320 n3 n5, 321 n8 nlO, 324 n34, 332, 334-335, 373 nl McNeill, David 93 Major, Ralph H. 156 Manney, Linda 23 Marcus, Joyce 322 nl6 Martin, Willy 99, 104, 107, 115-116 Marlowe, Christopher 162 Marmor, Gloria S. 246 Matsuki, Keiko 6, 10, 13, 19, 184, 324 n37, 338 Matthei, Edward 9 Mauss, Marcel 315 Maxwell, Judith 322 Medin, Douglas L. 91, 249, 285 Merrifield, William R. 232 Mervis, Carolyn D. 249, 271 n4 Metzler, Jacqueline 252 Miller, George A. 8, 29 Miller, Wick 261 Mill, John Stuart 266 Minami, Fujio 332-333 Montezuma the Younger 277-278, 294300, 302-305, 307-314, 319-320 n2, 323 n27, 324 n29 n31 n36 Muhlhiiusler, Peter 388 nl Munro, Pamela 186, 190, 194 Nakamura, Akira 137 Nakane, Chie 336-337 Needham, Rondney 133 n4 Neisser, Ulrich 249 O'Neil, John P. 261-263, 270-271 n4-5 Onions, C. T. 228 nl3 Origuchi, Shinobu 150 n7

394

Index of names

Ortony, Andrew 175 Osgood, Charles 267 Osherson, Daniel 255 Ovid 160 Ovidius Naso, Plurius. See Ovid Pagliuca, William 130-131 Palmer, Frank R. 266 Palmer, Stephen 252 Panofsky, Erwin 156, 161, 164 Parsons, Lawrence M. 245 Pellegrin, Pierre 370-372, 374 nl5 Perkins, Revere D. 130-131 Piaget, Jean 221 Pike, Kenneth L. 45 Plato 223, 369, 373 Plotinus 368 Priscian 384 Pope, Maurice 161- 162, 171 Porphyry 18- 19, 368-3 70, 373 Putnam, Hilary 2 Quetzalcoatl 323 Quintilian 383-384 Quintilianus, Marcus Fabius. See Quintilian Radden, Gunter 106, 227 nl Ramsbottom, J. 374 n6-7 Reddy, Michael, J. 226, 228 n20 Rice, Sally Ann 23, 45 Rieppel, Oliver 367 Ripa, Cesare 159-161, 169 Rips, Lance J. 253, 255-256 Rivers, William H. R. 238-239 Robins, R. H. 388 n2 Roediger, Henry L. 227 n6 Roeper, Thomas 9 Rosaldo, Michelle 150 n9 Rosch, Eleanor Heider 249-253, 270-271 n3-4, 320 n4. See also Heider, Eleanor Rosch Roskos-Ewaldsen, Beverly B. 246 Rubba, Johanna 46 Rudzka-Ostyn, Brygida 77 nl Rundell, Michael 227 n5 Russell, Lord Bertrand A. W. 266 Ryle, Gilbert 219

Sahagun, Fray Bernardino de 276, 289291, 294-298, 302-304, 312-313, 320 nl, 323-324 n26 n28-29 n36 Salmond, Anne 227 n6 Sasaki, Masato 245 Saussure, Ferdinand de 377-381, 386 Saxl, Fritz 156, 161, 164 Schafer, Jiirgen 156, 158, 161-162 Schank, Roger 10, Schroeder, Susan 313 Searle, John R. 15-16 Seneca, Lucius Annaeus 160 Sewell, Daniel R. 252 Shakespeare, William 162 Shepard, Roger N. 252 Shimojo, Shinsu-ke 245 Shinmura, Izuru 142 Shore, Bradd 332 Shweder, Richard A. 277, 314-318 Siegel, Rudolph E. 155-156 Sinclair, John M. 227 n5 Slaughter, Mary M. 374 nll Smith, Edward, E. 249, 255 Smith, Michael B. 23, 32, 42, 45-46 n2 Socrates 369 Solomon, Robert C. 186, 192-193, 195 Soufas, Teresa S. 165 Spiro, Melford E. 277, 315 Stanlaw, James M. 266-267 Starobinsky, Jean 165 Sternberg, Robert J. 227 n6 Stolz, Thomas 120 Sullivan, Thelma 322 Summers, Della 227 n5 Svorou, Soteria 120, 133 Sweetser, Eve E. 225, 348 Talmy, Leonard 25 Tanz, Christine 120 Taylor, John R. 5, 23, 30, 77 n2 nl3, 249, 331, 365 Taylor, Talbot J. 388 n3 Tenochithin 277-278, 303, 312-313 Tezcatlipoca 296, 322 n20 Tissot. See Doctor Tissot Tizoc 322 nl6 Thrax, Dionysius 382-383 Thompson, Sandra A. 46

'

I

Index of names

Tlatelolco 310 Tlatelolco, Anales de, 309-311 Torii, Shuko 245 Toxcatl, festival of 310 Traugott, Elizabeth 23, 119, 132, 174, 225 Tsohatzidis, Savas L. 7, 19, 77 n2, 249 Tuggy, David 23, 32 Turner, Mark 150 n2, 198 Tversky, Amos 253 Tversky, Barbara 370, 372 Ullmann, Stephen 266 Uno, Yoshikata 332-333, 346, 349, 351 Vandeloise, Claude 23, 45 van Brake!, Jaap 249 van Hoek, Karen Ann 23, 76 Van Wijk, H. A. C. W. 251 Verhagen, Arie 46

Varro, Marcus Terentius 383 Vicary, Thomas 162 Virchow, Rudolf 159 Vollmer, Gerhard 221 Warren, Edward W. 374 n9-10 Waletzky, Joshua 297 Webster, John 162 Weekley, Ernest 228 nl9 Weeks, Kent 285 Wierzbicka, Anna 9, 23, 95, 255 Williams, Edwin 6 Wittgenstein, Ludwig 11 Yamanaka, Joota 150 n7 Yoneyama, Toshinao 337 Zimler, Jerome 245 Zimmermann, Gunter 323 n25

395

Subject index

Abstraction 199, 211, 219, 385 degrees of 40, 53, 61, 72 abstract domain 197 Adverb: topographic adverbials 33 of degree 118 African languages 120-33. See also Cushitic, Niger-Congo, Nilotic, Swahili, Wolof, Zulu Agent 105, 117 volitional 27 unspecified 31 role of 77 human cause 106 Aguaruna 261-262, 270-271 Alignment 57-58, 64-65 Ambiguity 112-113, 158 American languages. See Aguaruna, Chickasaw, Cora, Kalapalo, Mixtec, Nahuatl, Pintupi, Shoshoni, Tzotzil, Zapotec Analogy 231-232, 240, 242, 245-247, 332 Analytical view, analyticity 239, 243 differentiation 238 Anger Ifaluk 136 Japanese 13, 136-50 models of 13-14, 181-95 theory of the humors 153-77. See also Metaphor Animistic forces 286, 289-290, 299, 302, 316 Arabic 333 Arbitrariness 32 Bloomfieldian notion of 251 Saussurean doctrine of the linguistic sign 16 Archetypal experience 63 semantic role 53. See Conceptual archetype Arrangement of coordinates 240, 247-248, 269,280, 284, 318-319, 339 Asymmetry 18, 133

between center and periphery of a category 18, 255-261 conceptualizer and object 46 n7 objective 66 possessor-possessed alignment 57-59 of power between speakers 334, 336337 versus a symmetrical relation 18 Atemporal character 67 relation 54, 64, 72 Aztec 18, 277-324 Bachelor, the ICM 4. See also Spinster Background 53, 209, 281, 365, 367-369, 373 backgrounding 25, 28-29 backgrounded 77, 279 Base and profile. See Figure/ground Basic level of categorization 238 Binumarien 237-238 Blind, cognition of the 245-246, 259-261 Bodily experience 61. See also Embodiment, Experience Body parts 58, 121-33 competing models 128 human/ anthropomorphic model versus animal/zoomorphic model 121-33 Body-part terms 10, 122-132, 285 etymology of 125 Border/boundary 280, 345 fuzzy 5, 284 self-other 284, 287-289, 293, 309, 318, 321 nl4 Case marking 45 accusative versus dative 45 Categorization 7, 38, 42, 56, 90, 108, 110, 331, 333, 337-338 commonalty with noncategorical cognition 256-261 function of versus motivation for 248 fuzzy-set model of 270 n3

Subject index

history of theory on 365-368, 373 human versus animal 247 implicit 86-88 measurement of 232-235 ofbirds 18,253-256,261-262 of color 16-18, 231-246, 266-268 of the person 18, 277-278, 284-324, esp. 284-285 and 314-319 preexisting 11 0 patterns 41 prototype model of 249-252 vantage theory of 237-46, 269 structure 39 psychological distance 240 versus vantages 247-248. See also Asymmetry, Prototype, Vantage theory Categorizing relationship 27, 40, 256 in syntax 40 Category. See Categorization Causation 8-9,95-117 causal relation 8 direct versus indirect 95 Dutch 96 German 96 English 96-97 noncausal relation 9 Change of state 28-29 verb 131, Chickasaw 194 / Chinese language 185, 188-190, 193 Mandarin 74 Classical theory 365-366, 374 n4-5: words, categories, and entities via meanings, essences, and features. See also Linguistic Theory Coextension 17, 237, 241, 248, 261, 269271, 279, 281, 320 n5; coextend 231 coextensive 234, 240-41, 268, 283-284 coextensively 248, 262, 283, 297 coextensive ranges 232, 237 coextensive vantages 243-245, 248, 261, 286 Cognitive base 128 domain 52, 57, 60, 65, 77 events inside the individual 19 structure 52

397

mediation 6 model 197-198 transfer 120, 130. See also Idealized cognitive model Cognitive Grammar 23, 45, 51, 56, 58, 63, 76 contrastive 28 exegesis of 32-34, 51-56 universal grammatical categories 61 Color basic colors 17, 249-250 category boundaries 244 color terms 16, 233-235, 242-243, 249, 267-269 display 232-236 domain of 18, 232-236 ethnographic surveys 241 focus/foci 17, 231-232,234-237,240244, 249-50 interviewing methods 231 - 236 neural processing 16 spectrum 17, 232-233 unique hue 16-17 Zulu 232-236, 242-243. See also Categorization Commonalty 53. See also Abstraction Competing models. See Body parts Complementation 237-238, 270, 320 n5, Complex expression compositional motivation I 0 component structures 54-56 composite expression 54-55, 63, 70-71 correspondences between subparts 5455, 70 higher level of composition 54 semantic and phonological integration 54 Componential 116 componential analysis !57 component 84-85 Composite structure 32, 40, 55, 63, 70 compositional patterns 54 constructional schemas 54 Compositionality 33 Compositional path 34 Compound lexical 43 literal versus metaphorical meaning 43;

398

Subject index

Concept: causal 8 complex 8 nuances of 8 source and target concepts of grammaticalization 119-32 relational 121 Conceptual archetype 52-53, 61, 63, 67, 77 Conceptual content 53, 55, 65 contentful conception 60 contentful 61 · Conceptual reification , 70 Conceptualization 6, 13, 108 complex 44 detailed 52 of emotions 15, 181-184, 186, 190-195 of thought 198,206,218, 220 Conceptualizer 60 Conceptual structure 28-29, 31-32, 36 unipolar 32 Connotation. See Denotation Constraint 183, 192 Construal 4-6, 19, 24, 29, 36, 45, 53, 103, 116, 331 abstract 68 alternate 32, 45 scene 31 extent 29 relation(ship). 4, 6, 27-28, 37 of scenes. 32 of the situation 5, 46 of mental processes 15 patterns of 19 temporal 33 theory of 16 traversal 29 universals of 19 Construction: grammatical 5, 23 higher-order 77 of vantages 231, 240-241, 332, 335 Construe 4, 7-8, 26-27, 351, 377 conceptualizations 44 Content requirement 27, 32 Context dependent on 7, 90-92 contextual availability 89 figurative 106

sociocultural context 136-50, 181-95, 338, 358 variation in accord with sociocultural context 149-50, 186 Control 31, 284-286, 312 Conventionalization 36, 190 conventionalized meaning 40. See also Imagery Coordinate 231, 245, 278-280, 284-286, 317, 319, 332, 334, 342, 346 fixed and mobile 18, 231, 240, 244, 252258, 260, 262-267, 280-283, 287, 315-316, 319, 337-338, 365, 368369, 372 Coordination in Aristotle's partonomy 370-373 Cora 28, 32-34, 36, 38, 44-47, 252 Creativity linguistic 42 sociolinguistic 332 Crosscultural semantics 137-50, 181-95 Cross-language differences 6, 119-133 Cross-linguistic 130 Crossover of vantages 242-243 Cuauhtemoc in narrative 294-295, 306, 309-311, 313, 319, 323 Cushitic 121 Dani 249 Danish 226 Deixis 5 Definiteness 58, 63 Demonstrative 131, 133 Denotation and connotation 266-268 Diachronic 132-133 prediction 132 versus synchronic 133, 155 Discourse analysis. See Rhetorical systems Distinctiveness. See Similarity and distinctiveness Distributive suffix 39 Dominant and recessive 242-243 ranges 241-242, 283 metaphors 265-266 terms 17-18, 243, 268 vantages 244-245, 253, 278, 282, 288, 290, 292, 296, 299-300, 303-304, 308-309,311-312,314,319

Subject index

word-pairs 267 Dominant-recessive pattern 241-244, 247, 261-262, 269 of coextension 241, 269-271 Dominion 60, 70, 73-76, 182 function of 74 of reference point 66 Double-construction 256-258 Dummy 75 Egypt, ancient 237, 285 Elaboration 40, 56 site 55 structure 63 Embodiment 183, 186, 191-192, 194-195 Emotion 10, 13, 101, 114, 137-150, 164165, 170-71, 183-195. See also Anger, Metaphor, Metonymy English 28, 32, 34, 38, 46, 76, 386-387 color naming 232-233 metaphors of mental activity 197-228 metaphors and metonomies of anger 136-42, 184, 187, 189-190 speakers 31 syntax 41 Entailment 244-245, 281-282, 286-287, 290,303-304,312,316,318,321 nl5, 324 n29, 335 Eskimo 239 Etymology 223-225 Event 101 classes of 28 coordination of events 27 Evolution linguistic 120 of categories 237-240, 279, 285, 317 of differentiation 238-240 Existence in relation to possession 51, 75, 78 n15 Expert theories 181, 192, 227. See also Folk theories Experience, physiological 181-184. See also Embodiment Explanation 182. See also Motivation Expression 52 impersonal 75 linguistic element of 52

399

Feature theory of meaning: extended 267 See also Classical theory, Semantics, Structuralism Figure primary 54, 65 secondary 65 Figure/ground 231, 243-244, 253-256, 340-342 active/passive alternation 77 alignment 27, 65 base/profile 44, 55-56, 61, 70-71, 73 landmark/trajector 44, 53-56, 60-61, 64-66, 69-70, 73-76, 97, 105, 253 reversal of 77 schematic landmark 55, 62 schematic trajector 71, 77 organization 53, 67 relationship 30 reversal 30 Finite clause, ICM of 45 Florentine Codex 277-278, 286-291, 294-299, 302-303, 310-313, 316, 319 Frame 76, 205 framing 40, 75 French 45, 380-381 Focusing: on events 26, 34 focus of attention 37. See also Color Folk theory 82, 181, 197, 212 folk model 220 folk model versus scientific knowledge 174-176. See also Idealized cognitive model Foreground 5. See also figure/ground From-Meaning: pairings 51 continuum 51 Functor 28-29 as schematic class 31 Grammar 54 cognitive 27, 32, 44 sanctioning role of 44 Grammatical categories 119 constructions 27 functions 119 structure 32, 45 meaning of 45 head 56

400

Subject index

Grammaticalization 76, 119-33 primary and secondary sources of 132 Genus and species 367-373 German 226, 228, 380 case 45 verbs of cognition 46 change of state 131 Germanic 228 Gestalt 255 experiential 30 Greek 228 Alexandrian-period 382-383 ancient 15 Grounding predications 46 n7 Hanun6o 316 Head noun 41 possessed 61, 67 Hebrew 226 Highlighting 25, 27 Historical analysis 170-173. See also Diachronic Holistic and nonholistic reading 101-102 Humors, the theory of 13-14 versus metaphors of anger 153- 177, 182, 184, 194 Hungarian 185-190 Iconicity 308-309 Idealized Cognitive Model (ICM) 77, 197198, 203, 209, 216, 219-220, 222-223, 226 Idiomaticity 108- 109, 111 Ifaluk 137, 194, 315.318 Inflection 300, 305 causative-applicative 45 conditional-irrealis 306 indicative and subjunctive 45 imperfective 306 perfective 305 pluperfect 306 Imagery 33 contrastive 44 conventional 23, 32, 36, 40, 45 perceptual 23 complementary 169 Image schema 29, 52- 53, 223, 225, 228, 317

basic ability 59 Image-schematic relation 60, 207 Impersonal, third singular 34 Inanimate 102 Inclusion 237, 270, 320 n5 Indeterminacy 245 Indexicality 91. See also Deixis Indo-European 228 Indonesian 226 Innateness cognitive abilities 52 image-schematic abilities 61, 65 method of categorizing 24 7 Instantiation 52, 136 Intrinsic relationship 67-70 Invariance hypothesis 197 Italian 226 Item, assignment of 5 discrete 1-2 lexical 125 Javanese 333 Japanese 137-150, 185, 187-191, 206, 318,331-362 Kalapalo 324 n35 Kaluli 265-266 Knowledge versus belief 162 Landmark, of a relation 30 environmental 121, 123. See also Figure/ ground Language. See Linguistic theory Languages African 10 Oceanic 10. See also African languages, American languages, others by language name, e.g., "Latin." Latin 228, 380, 383 Level of concentration: in a categorical vantage 243-245, 255-256, 267-268, 280-281, 286, 303-304, 314 in a spatial vantage 258 exacting 289 preferred 282, 290, 293-294, 299-300, 305, 315 least preferred 294, 308 Level of structural organization 56, 65

Subject index

level of composition 62 level of constituency 70-71 clausal level 77 Linguistic structures directly attested 27 schematic 27 Linguistic theory, history of Greco-Roman 382-384 Middle Ages and Renaissance 384 Nationalistic 19, 384-385 Saussurean 377-381, 388 Chomskian 379-380, 388 Location in relation to possession 51, 75, 78 nl5 of participants 59 Locational entity 74 propositions 131 Locative complement 73 nouns, adverbs, and adpositions 120 relationship 75 search domain of 73 path to endpoint 77 Luisefio 35 Malayalam 226 Meaning implicit 27 lexical 81 literal 43 literal versus specialized 41 literal versus metaphorical 43 spatial 18 spatial versus nonspatial 96, 115 Memory 212-215, 219, 259-260 Mental contact 58-60, 70, 76 mental access 60-61 Mental rotation 252 Mesoamerican languages 232. See Cora, Mixtec, Nahuatl, Zapotec Metaphor 41, 53, 56, 63, 108, 136-50, 316-317, 322 n18, 337 alternative 218, 221-223, 225 conceptual 198, 204, 216 dead 225 in a cluster 15 fading 67

401

from computer science 15-16 dominant and recessive 265-266 mind as workshop 15-16 nonliteral expression 5 of anger 13,137-50,181-95 of conduit 226 of journey 222 of mind 196-228 of vision 15, 130 ontological 199, 205, 207 source domain 15, 121, 130, 221, 223 spatial 66 structural 199, 206, 213 target domain 15, 121, 219, 221 primary source domain 127, 132 secondary source domain 132 Metaphoricall2, 56, 97, 171-73 image 169 space and distance 200 versus literal 5 scenario 206 structuring 67 Metaphorically extended meaning 57, 77, 97 Metonymy 163, 317, 322 nl8 nonliteral expression 5 ofemotion 13, 138, 144, 190-191 Mixtec 121 Model-integrated components (MIC) 198-200, 203-204, 206-207, 209, 213-214, 216, 218,222-223, 226 Montezuma in narrative 276-278, 295314, 319-320, 323-324 Motivation 44, 219 bodily versus conceptual 183 multiple 1 of expressions 181-184 of semantic extensions 10, 41 etymological or semantic 164 local versus global 164 Name, vacuous 85. See also Dummy Nahuatl Classical 278, 289, 290-295, 297-311, 320 n2, 322 n 17 n21, 323 n28 Tetelcingo 45 Near Synonymy 237, 270, 320 n5 near synonyms 234

402

Subject index

Network foci of 34 of knowledge 34, 136 Niger-Congo 131, 133 Nilotic 121 Nominal expression 53-55 derived 26 possessor 61 versus relational 70 Nominalization 5, 42, 69-70, 72 atemporal thing 5 in relation to possession 51 temporal process 5 nominalized 26, 33 Noun: category 77 phrase, complex 44 Objectivity, degrees of 37 property versus name 16 Oceanic languages 121-33 Orientation 53 expressions of 10. See also Vantage point Oriya 316, 318 Part/whole relation 68 Participant processual 69 relation to-event 7 Particle adverbial 75 verb particle 30, 41 Passive passivization 46 sentence 26 subject 26 verb phrase 27 Path 29, 97 actual versus possible 60 alternate 32 bounded 28 compositional 35 directed 30- 31 endpoint 29 evolutionary 67 pathway 29 Perfect aspect, in relation to possession 51, 75-76

Perspective 5, 53, 96 global 26 mental route 5 multiple perspectives 24 of the speaker 5. See also Figure/ground, Deixis, Vantage Phonological form paired with meaning 32, 51 pole 72 Phylogeny of categorization, speculation on 246-247 Physiological explanation 167-168 Pintupi 318 Poetic and prosaic 265-268 Polarization versus centralization 238, 241, 243, 249-250, 268, 279, 283, 288, 297, 300, 305 Polysemy 72, 77, 142 polysemous 51 Possessor 30, 57, 61, 64, 70, 78 Possession 6, 56, 58, 61, 63, 75, 77-78, 83-84 cognitive linguistic affinities 51 constructions 6, 51 linguistic category 51 obligatory 57 relation 6, 30, 51 Possessive determiners 77 locative 66, 77 locutions 58 constructions 58, 63 English 61 elements 61, 71 inflection 70 main verb 63 morpheme 62, 67 prepositional 66-67 Possessives 56-57, 59 locative 67 have-constructions 73-76 of-constructions 69-73 Preposition 8, 54-55, 63, 94-117 by, 26, 97 contact 99 Dutch 8, 96-117 passim English 8, 45, 70, 96-117 passim French 45, 106

Subject index

German 8, 96-117 passim object of 70 of-constructions 69-73 possessive 67 selection of 9 spatial versus nonspatial 97-99 verb final 40 Prepositional phrase 25, 77 object of a preposition within 26 Presupposition 5, 53, 243, 256, 280-281 Profile 54 nominal 70-71 tempora126 relational 53, 72 regional 72 unprofiled 67. See also figure/ground Profile determinant 55-56, 63, 70 profile determinance 56 shift 78 Profiled relationship 54, 64-65, 75 Prominence 27, 53 degree of 25-27 focal 65 relations 65 Propositional phrase 27 verbs 45 Prototype 32, 45, 51, 53, 56, 308, 343 asymmetry in relation to 253, 256 category 61, 99 model 231, 249-252 effects 249 quintessential versus representative 261, 271 Prototypica136-37, 39, 53, 58-59,61,64, 70, 77, 86-88, 98, 114 scenario of anger 138, 144-146 nonprototypical anger 147-49 versus atypical member of a category 88 temperaments 156 Prototypicality 31 of subject 77. See also Typicality Psychological reality 31 Purview of a vantage 240 Reality subjective 3 objective 3-4. See also World

403

Recessive. See Dominant Register 305-309 Reference point 5, 7, 59-60, 73, 76 alternative 130 as coordinate and image schema 255 basic 120, 133 cognitive 17 fixed and mobile 231 fixed reference body 240 implicit 25 implied 71 inherent 59 in a frame of reference 255 in space or locus of relevance 74 low-profile 25 of a dominion 71 of a search domain 72 participants 59, 64 private 256-257 role of 25 temporal 76 ubiquity of 59 versus dominion 74-75. See also Coordinate, Figure/ground Reference-point characterization 75 function 64, 73-74 model 58, 61, 65-67, 70, 73, 75-76 reasoning 252-253 relationship 60-61, 63-67, 69-70 value 73, 75 Relational complement 70 expression 53, 56 noun 69, 133 predicate nominative 70 value 70 versus nominal 70 Relativity versus universality 238-239, 248, 266-268, 277-278, 314-319 Rhetorical systems 295-297, 300-302, 305-309,314-315,317,319. See also Inflection, Voice, Register Salience alternate 32 cognitive 59 primary and secondary 40

404

Subject index

Samoan 45-46 Scanning sequential 27, 54, 64 mental 52 direction of 53 Scenario 11, 24, 44, 348 memory 213 metaphorical 206 problem-solving 209 viewed 24 Scene 45, 60 components of 4 from location of the observer 5 objective 35 Schema 13, 136 categorization schema 42 constructional schema 55 substantive types 220 schemata 349 Schematic 25 associative relationship 30 characterization 57, 60 conception 53 degrees of schematicity 27 landmark 55, 61 level 64 networks 27 pattern 40 target 63 value 52 verb do 27 Schematicize 127 Script 10-12 as a pattern of events 10-11 Scope 53 of predication 27 of visual field 36-37 substructure within 53 Search domain 72-73, 77 Selection 215 for comment 24-25 Selfhood Japanese honne versus tatemae 144, 318, 324 n37, 338. See also Categorization of the person, Border/boundary Semantic description 91 extension 38, 41-43, 44, 72, 77

endpoint of semantic extension 44 function 2 interpretation 92 pole 6 role 116 specifications 62 Semantics cognitive 3-4, 81, 92 experiential 23 feature 116 formal 23-24, 28 model-theoretic 2 objectivist 19, 81 -83 of natural language 28 structure of 23, 32 subjective 23 truth-conditional 2, 20, 5, 23 versus conceptualization 6 Semantic relations continuum of 270, 320 n5 typology of 237-238, 270, 281, 320 n5. See Near Synonymy, Coextension, Inclusion, Complementation Semiotic 159, 161. See Symbolic Shoshoni 261 Similarity and distinctiveness attention to 17, 238, 240, 243, 256, 259, 261, 278, 280-283, 285, 287-288, 295, 305,309,311-316,318-320 coordinates of 244-245, 261, 317, 324 n36 emphases on 240, 245, 253, 256, 268, 279 in reference to 243-244 perceived 43 Situation 36, 63, 95, 101 conceptualized 4, 19, 23, 25 final state 26 objective 8 precision of characterization 4 separation versus contact 106 specificity of portrayal 4 state of affairs 6 successive states 26 viewed 24 Sociolinguistic deference, familiarity, formality, honorifics, politeness, power, styles 331-358 passim South Africa, Republic of 232, 269

Subject indl'x

Spanish 32, 35, 38, 41, 44, 47, Spanish Conquest of Mexico 18, 277, 294298,303-305,311-314 Spatial concepts 120 configurations 117 orientation 127-132 reasoning 231, 245-246 relationship 54 term 33 verb 29. See also Temporal, Profile Speaker role 23-24, involvement 25 Specialization 40 of participants 59 Species. See Genus and species Specification, semantic 55, 62 Specificity, levels of 32, 52-53, 63 Spinster, the ICM 4. See also Bachelor Standard language 19, 384-385 Stereotype 277, 285-289, 294, 300, 309, 324 n35 Structuralism 365. See Classical theory Subject clausal 72, 76 role of 31 of a clause in relation to possession 51 Subjectivity 25 degrees of 37 subjectification 76 versus objectivity 46 Sublexical 35 Submetaphor 197 subcomponent 213 submodel 198 Swahili 122, 226, Symbolic 6 bipolar structures 32 simple and complex 44 target structure 44 local semiotics 316-319 Symbolization 32 conventional 34 lexicon-grammar continuum 51 nonarbitrary 32 of conceptualization 51 relationship 38 Symmetrical relationship 58

405

Synonymy versus coextension 17 nonsynonymy of terms with one objective referent 35, 81-83 nonsynonymous 82 Syntax 9 arbitrary 9 formal24 semantically motivated 9 Tahitian language 186-187, 189-190 Tamil language 226 Target 60-61, 63, 76 versus dominion 70 Taxonomic tree 19 taxonomy 366 taxonomy versus partonomy 370-371 Temporal 103 domain 33, 76 evolution 64, 77, 132 extension 77 succession 33 verb 29. See also Spatial, Profile Tense past 27 future 131 Tense-aspect 27, 131 past perfective 27, 131 past participle 114 progressive 131 Theories, Aristotle's theory of forms 18 about emotions 12 as semantic models 14 linguistic 19, 44 of anger 12 Time time-arrow 54, 64 models of time 262, 264 Toda of Southern India 239 Trajector. See Figure/ground Transformation 101, 258 Transitivity 45-46 sliding scale of 46 Turkish 226 Type, choice of 5 of a finite set 1-2 type/token 253, 386 Typicality 88, 249, 255

406

Subject index

versus atypicality 89-90. See also Prototypicality Tzotzil 324 n34 Universality 53, 131, 248, 268 Universals 61 of categorization 18, 237-238, 241-243 of cognition 226 of color-category foci 249-250 of construa119 diachronic 132 of language 51, 56 of anger metaphor 136 of structure 34 Vantage 240, 244-246, 248, 252-253, 255-256, 259, 262, 268, 334-335, 346, 350, 366, 373. See also Dominant andrecessive, Coextensive Vantage point 231, 234, 245, 253, 255, 260 perspective 248, 251-252, 331 speaker's 24, 36-37, 45, 53 alternate 32 point of view 231, 242 slant 248 standpoint 231, 252-253, 259 viewing angle 248 viewpoint 232, 240, 246, 252, 332, 334 Vantage theory 16-18,247-249,252,255, 260-261, 269, 277-288, 296-298, 303304, 311-320, 332-334, 337-339, 365, 370, 372 Verb, intransitive 85 transitive 105 Vestige of a physical meaning 64, 73 of metaphorical origin 65 of an active sense 73

relic 176 vestigial 76 of humoral belief 159 Viewpoint 5, 36 mental route 5 Vision as source domain 223, 225-226 Voice 300, 305 direct discourse 300, 306, 310-311 indirect discourse 306 quasilocutional 306 inner voice or strategist 295-296, 300302, 307, 324 n31 Whole-to-part 7 Whorfian 6 Wolof 186-187 Words versus domain 177 World 81-82, 95, 197, 336 arbitrary 251 ephemeral 238-239 etymology of "world" 3 mind-independent 81 language as 377-388 possible worlds 3 objectivist picture of the 6 projected worlds 4, 231, 251, 269 the real 1-2, 4, 231, 250 Weltanschauung, 377 word-to-world "loosely speaking" 4 Yaqui 35 Zapotec 285 Zooming in and zooming out 243-245, 255, 257-258, 262, 280, 282-283, 294, 303, 314-315, 332, 334, 339-340, 342343, 345-348, 365, 368 Zulu 234-237, 244-243, 269-270

Contributors

Munekazu H. Aoyagi University of British Columbia, Vancouver, Canada Eugene H. Casad Summer Institute of Linguistics, Tucson, Arizona, USA Rene Dirven University of Duisburg, Germany Dirk Geeraerts University of Leuven, Belgium Stefan Grondelaers University of Leuven, Belgium Bernd Heine University of Cologne, Germany Jane H. Hill University of Arizona, USA Olaf Jakel University of Hamburg, Germany Jeff Lansing University of California, San Diego, USA Mgel Love University of Cape Town, South Africa Zoltim Kovecses Eotvos Lorand University, Budapest, Hungary Ronald W. Langacker University of California, San Diego, USA Robert E. MacLaury George Washington University, Washington DC, USA Keiko Matsuki University of Arizona, USA John R. Taylor University of Otago, New Zealand Savas L. Tsohatzidis Aristotle University, Thessaloniki, Greece