Cognitive and communicative approaches to linguistic analysis

Citation preview

Cognitive and Communicative Approaches to Linguistic Analysis

Studies in Functional and Structural Linguistics (SFSL) Taking the broadest and most general definitions of the terms functional and structural, this series aims to present linguistic and interdisciplinary research that relates language structure — at any level of analysis from phonology to discourse — to broader functional considerations, whether cognitive, communicative, pragmatic or sociocultural. Preference will be given to studies that focus on data from actual discourse, whether speech, writing or other nonvocal medium. The series was formerly known as Linguistic & Literary Studies in Eastern Europe (LLSEE).

Founding Editor John Odmark Honorary Editors Eva Hajicˇová

Petr Sgall

Charles University

Charles University

General Editors Yishai Tobin

Ellen Contini-Morava

Ben-Gurion University of the Negev

University of Virginia

Editorial Board Alexandra Y. Aikhenvald

Jim Miller

La Trobe University

University of Auckland

Joan Bybee

Marianne Mithun

University of New Mexico

University of California, at Santa Barbara

Nicholas Evans

Lawrence J. Raphael

University of Melbourne

Emeritus CUNY

Victor A. Friedman

Olga Mišeska Tomic´

University of Chicago

Leiden University

Anatoly Liberman

Olga T. Yokoyama

University of Minnesota

UCLA

James A. Matisoff University of California, Berkeley

Volume 51 Cognitive and Communicative Approaches to Linguistic Analysis Edited by Ellen Contini-Morava, Robert S. Kirsner and Betsy Rodríguez-Bachiller

Cognitive and Communicative Approaches to Linguistic Analysis Edited by

Ellen Contini-Morava University of Virginia

Robert S. Kirsner University of California, Los Angeles

Betsy Rodríguez-Bachiller Kean University

John Benjamins Publishing Company Amsterdam
/
Philadelphia

8

TM

The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.

Library of Congress Cataloging-in-Publication Data International Columbia School Conference on Linguistics (6th : 1999 : Rutgers University) Cognitive and communicative approaches to linguistic analysis / edited by Ellen Contini-Morava, Robert S. Kirsner and Betsy Rodríguez-Bachiller. p. cm. (Studies in functional and structural linguistics, issn 0165–7712 ; v. 51) Chiefly revisions of papers presented at the 6th International Columbia School Conference on Linguistics held at Rutgers University in Oct. 1999. Includes bibliographical references and index. 1. Linguistic analysis (Linguistics)--Congresses. 2. Cognitive grammar--Congresses. 3. Semantics--Congresses. 4. Semiotics-Congresses. I. Contini-Morava, Ellen, 1948- II. Kirsner, Robert S., 1941III. Rodríguez-Bachiller, Betsy. IV. Title. V. Series. P126.I58 2004 410--dc22 isbn 90 272 1560 X (Eur.) / 1 58811 566 6 (US) (Hb; alk. paper)

2004057066

© 2004 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

Contents List of contributors Introduction Robert S. Kirsner

vii 

Part I: Cognitive Grammar 1. Form, meaning, and behavior Ronald W. Langacker

2

2. Cataphoric pronouns as mental space designators Michael B. Smith

6

Part II: Theoretical issues in classical sign-based linguistics 3. Monosemy, homonymy and polysemy Wallis Reid 4. On the relationship between form and grammatical meaning in the linguistic sign Mark J. Elson 5. Revisiting the gap between meaning and message Joseph Davis

93

3 55

Part III: Analyses on the level of the classic linguistic sign 6. The givenness of background Zhuo Jing

77

7. The relevance of relevance in linguistic analysis Bob de Jonge

205

8. A sign-based analysis of English pronouns in conjoined expressions Nancy Stern

29

9. Semantic oppositions in the Hebrew verb system Noah Oron and Yishai Tobin

235

10. Grammaticization of ‘to’ and ‘away’ Kumiko Ichihashi-Nakayama

26

vi

Contents

Part IV: Below and above the level of the sign 11. Interaction of physiology and communication in the make-up and distribution of stops in Lucknow Urdu Shabana Hameed

277

12. Between phonology and lexicon Yishai Tobin

289

13. Length of the extra-information phrase as a predictor of word order Ricardo Otheguy, Betsy Rodríguez-Bachiller, and Eulalia Canals

325

14. Word-order variation in spoken Spanish in constructions with a verb, a direct object, and an adverb Francisco Ocampo

34

15. Estrategias discursivas como parámetros para el análisis lingüístico Angelita Martínez

36

Index of names Index of subjects

38 384

List of contributors Eulalia Canals 1517 8th Avenue Brooklyn, NY 11215 USA [email protected]

Zhuo Jing Heinrich Heine Strasse 64 D-45657 Recklinghausen Germany [email protected]

Ellen Contini-Morava Department of Anthropology University of Virginia Charlottesville, VA 22906 USA [email protected]

Bob de Jonge Dept. of Romance Studies Arts Faculty, Groningen University P.O. Box 716 9700 AS Groningen The Netherlands [email protected]

Joseph Davis School of Education, R6207-A City College New York, NY 10031 USA [email protected] Mark J. Elson Slavic Languages and Literatures PO Box 400783, 109 Cabell Hall University of Virginia Charlottesville, VA 22904-4783 USA [email protected] Shabana Hameed “Baitul Hameed” 4/1407 Medical College Road Aligarh – 202002 U.P. India shabhameed@rediffmail.com Kumiko Ichihashi-Nakayama 1-29-78 Tama-cho Fuchu-shi, Tokyo 183-0002 Japan [email protected]

Robert S. Kirsner Department of Germanic Languages 212 Royce Hall – UCLA Los Angeles, CA 90095-1539 USA [email protected] Ronald Langacker Linguistics, 0108 University of California, San Diego La Jolla, CA 92093-0108 USA [email protected] Angelita Martínez Juncal 655, Primer piso – A 1062 Buenos Aires Argentina angema@filo.uba.ar Francisco Ocampo University of Minnesota Department of Spanish and Portuguese 37 Folwell Hall, 9 Pleasant St. S.E. Minneapolis, MN 55455 USA [email protected]

viii

List of contributors

Noah Oron Department of Foreign Literatures & Department of Behavioral Sciences Ben-Gurion University of the Negev P.O. Box 653 84105 Be’er Sheva Israel [email protected] Ricardo Otheguy Graduate Center, CUNY 365 Fifth Avenue New York, NY 10016 USA [email protected] Wallis Reid Graduate School of Education Rutgers University New Brunswick, NJ 08903 USA [email protected] Betsy Rodríguez-Bachiller Department of English Kean University Union, NJ 07083 USA [email protected]

Michael B. Smith Department of Linguistics 320 O’Dowd Hall Oakland University Rochester, MI 48309 USA [email protected] Nancy Stern School of Education, R6/207B City College New York, NY 10031 USA [email protected] Yishai Tobin Department of Foreign Literatures and Department of Behavioral Sciences Ben-Gurion University of the Negev P.O. Box 653 84105 Be’er Sheva Israel [email protected]

Introduction On paradigms, analyses, and dialogue Robert S. Kirsner University of California, Los Angeles

He who knows only his own side of the case knows little of that. His reasons may be good, and no one may have been able to refute them. But if he is equally unable to refute the reasons on the opposite side, if he does not so much as know what they are, he has no ground for preferring either opinion. John Stuart Mill, On Liberty

The majority of the fifteen chapters included in this volume are revised and expanded versions of talks originally presented at the Sixth International Columbia School Conference held at Rutgers University in October 1999. This conference was notable for attempting to bring about not only an exchange of ideas between various sign-based linguists — Columbia School, Guillaumean, and others — but also, and perhaps more importantly, a dialogue between sign-based linguists and representatives of Cognitive linguistics.¹ Significantly, one of the plenary speakers was Ronald W. Langacker who, together with George Lakoff, founded the movement in linguistics which has come to be known as Cognitive Grammar. The title of this volume — opposing Cognitive and Communicative — refers to the fact that while Cognitive linguistics views language as an example of cognition and invokes such notions as concepts, thoughts, and imagery in all their richness (Langacker 1991: Chapter 1), sign-based linguistics, in particular the Columbia School, draws a rather strict line between the linguistic system itself as a purely communicative instrument (one which may offer only the sparsest and roughest hints towards the message the speaker wishes to convey) and that message itself and how it is conceptualized by and dealt with by language users. As Contini-Morava puts it (1995: 8): Thus the main distinguishing characteristic of sign based grammatical theory is the principle of one form — one meaning. This principle, like the linguistic sign itself, follows from the function of language as a communicative instrument. The most efficient communicative instrument is one in which there is a one-to-one correspondence between the signaling units and the things signaled. It therefore makes sense, if one is beginning with a functional orientation, to use this as an initial methodological assumption when beginning a linguistic analysis. Of course, the extent to which the one-form — one-meaning principle holds in a given case is an empirical question since . . . neither forms nor meanings are given in advance of an analysis.

2

Robert S. Kirsner

Consider in this vein also the characterization of the Columbia School as a “minimalist” linguistics versus Cognitive Grammar as a “maximalist” linguistics in Kirsner (1993: 81–82) and the suggestion that the two approaches are at least in part complementary. The chapters contained in this volume testify that dialogue between sign-based linguistics and Cognitive Grammar is indeed possible. To be sure, the individual members of the various approaches represented are not necessarily ready to abandon their own frameworks. Nevertheless, they demonstrated a willingness to consider the perspectives on language and linguistic data which other frameworks can provide. This book constitutes an invitation to the reader to undertake the same sort of critical comparison. What is learned from one approach which is not learned from the other, and vice-versa? What are the strengths and weaknesses of each approach? While it seems simply silly to compare and contrast entities as dissimilar as apples, paperclips, and kangaroos, it is possible to compare apples, grapes, and oranges. Similarly, while a comparison of Columbia School Linguistics with Generative Grammar might be of little use, because of the tremendous differences in the two schools’ orienting premises,² comparison and contrast between Cognitive Grammar and the Columbia School is potentially fruitful; cf. Langacker’s assertion (1987: 4) that there is a “natural affinity” between Cognitive Grammar and the Columbia School. In spite of the fact that some Columbia School linguists have felt otherwise (e.g. Contini-Morava 1989: 35–44; Huffman 1997: 21–23, 329–333), one can suggest that each school ignores the criticism of the other only at its own peril: cf. Kirsner (1993: 103–107; 2003). There are perhaps two ways of introducing these chapters. The first might be in terms of style of argument and presentation. The chapters by Langacker and Reid may both be called examples of “interparadigmatic dialogue” in that, in addition to presenting an illustrative analysis, each discusses and evaluates the other’s theoretical framework. (Because of the importance of this dialogue, I shall discuss these two chapters in more detail than the others.) Davis’s chapter, in contrast, may be considered “intraparadigmatic dialogue” in that it suggests a refinement within Columbia School theory, taking issue with earlier Columbia School analyses, but without comparing that theory with others. The remaining chapters are all “paradigmatic monologues.” Each presents an analysis within a single theoretical paradigm, arguing against previous traditional or generative analyses of the relevant phenomena, but without directly taking issue with either the Columbia School or Cognitive Grammar. Each analysis itself is to be taken as a demonstration of how its theoretical framework can be applied and as a recommendation to the reader to consider seriously the specific theoretical principles which underlie the analysis. The only chapter which might not to seem to fit here is Ocampo’s, but his might perhaps be viewed as a useful example of “the agnostic paradigm of paradigmlessness”; i.e., as illustrating a largely a-theoretical search for patterns in data. The second and perhaps more direct and transparent way of introducing the chapters is to survey them simply as listed in the Table of Contents, where — purely

Paradigms, analyses, and dialogue

for the sake of convenience — they are sorted into four categories, depending on whether they (a) focus on Cognitive linguistics, (b) present theoretical and methodological discussions within sign-based linguistics, (c) present analyses of units the “size” of or at the level of the classical linguistic sign, or (d) discuss units or combinations of units at other levels above or below the analytical level of the classical linguistic sign (in traditional terms: phonology, morphology, “syntax”, and discourse). A second principle for ordering the chapters, used within (b) and (c), was to proceed from the central or most paradigmatic examples of Columbia School work to approaches taking issue with it or mixing it with other approaches. To begin at the beginning, then: Part I, containing two chapters, focuses on Cognitive linguistics. Ronald W. Langacker’s “Form, meaning, and behavior: The Cognitive Grammar analysis of double subject constructions” first succinctly reviews the basic conceptual apparatus of Cognitive Grammar, from which it is immediately apparent that this approach embraces many notions such as construction and constituency, subject and topic which have no clear counterpart in the Columbia School. Langacker’s subsequent discussion of double subject constructions in a number of languages raises a crucial issue for all sign-based linguists and certainly for Columbia School linguists. How legitimate is it to apply the very same term — double subject — to widely different languages with different morphologies? From a Columbia School perspective, very different signal-meaning units are utilized in the different languages, with substantively very different meanings, and different oppositional structures among those meanings, and different inferential routines made possible by the different meanings. Langacker argues that the striking similarity in semantic-pragmatic mechanisms underlying the use of such constructions across many languages forces the linguist to recognize double subjects as a genuine linguistic phenomenon requiring a unified analysis. In other words, the functional similarities are judged as outweighing, are more important than, the structural differences between the languages. In the final section of his chapter, Langacker presents a critique of the Columbia School. With considerable wit, he discusses various Columbia School concepts (e.g., strategies for the communicative exploitation of a meaning as something separate from that meaning) and forms of Columbia School argumentation, such as the “slippery-slope” rhetoric invoked to refute the notion of polysemous signs.³ Confronted with reasoning like “If X is allowed, there will be a natural progression to Y”, where Y is a totally arbitrary stopping point, to be rejected because of its very arbitrariness,⁴ Langacker offers the observation that “by refusing to descend a slippery slope one does not succeed in making the landscape flat.” He asserts that the “where does one stop?” question does in fact have an answer if one makes use of the Cognitive Grammar concepts of entrenchment, conventionalization and the notion of structured polysemy (cf. Nikiforidou 1991). An additional point which Langacker raises repeatedly is the relatively limited scope of phenomena for which the Columbia School takes analytical responsibility: “The more narrowly linguists define their own work, the less they will contribute to the global enterprise of figuring out how

3

4

Robert S. Kirsner

language works.” All in all, Langacker has done Columbia School linguists a great service by familiarizing himself with their approach and crafting a reasoned and provocative rebuttal to some of its concepts and argumentation. Nevertheless, areas of disagreement will remain. One point in Langacker’s chapter that Columbia School linguists would vigorously dispute is his claim that Columbia School theory involves “a tacit assumption that the meaning of a whole is merely the sum of the abstract meanings of its parts (i.e. the component lexical items), and that anything beyond this is attributable to contextual interpretation based on communicative strategies and other ‘human factors’”. In particular, Langacker asserts that a Columbia School analysis would have no way to explain the difference in interpretation between The big dog frightened a tiny cat and A big cat frightened the tiny dog because they consist of the same lexical items. This is not the case. A Columbia School account of these English sentences would invoke not only the meanings of the lexical items, but also two explicit word-order signals: the order AdjectiveNoun, a signal within the grammatical system of DIFFERENTIATION, and the order Participant1-Event-Participant2, a signal within the grammatical system DEGREE OF CONTROL.⁵ Of course, the Columbia School does not treat all orderings of words as explicit grammatical signals. The question of which orderings of which elements are to be taken as explicit signals in a given language as well as the question of how to explain the effects of word order when it cannot be considered a signal are analytical problems which must be handled on a case by case basis; see Otheguy et al, and Ocampo, this volume, for discussion. However, it must also be acknowledged that word order phenomena have not been a major focus of Columbia School research. Whether this is a weakness returns us to the question, explicit or implicit in all the chapters in this volume, of how the linguist defines the object of study. The second chapter exemplifying the Cognitive Grammar approach, Michael B. Smith’s “Cataphoric pronouns as mental-space designators,” utilizes Giles Fauconnier’s theory of mental spaces and concepts of iconicity developed by John Haiman to analyze sentences in English, German, and Russian such as We abhor it that the mayor absconded with the money versus We know that the mayor absconded with the money versus * We know it that the mayor absconded with the money. Smith argues that English it, German es, and Russian tot refer to the mental space (or “knowledge structure”) created by certain “structure-building” verbs: in these examples, the referent of the subordinate clause. On the one hand, such “prefiguring” of the subordinate clause with an it, es, or tot can increase the conceptual distance between the verb and the clause. Its presence or absence seems to correlate with whether the event or state of affairs referred to by the clause is viewed as unfavorable or not (so that the speaker would want to “distance” himself from it); cf. abhor versus know above. In other instances, the cataphoric pronoun, providing an additional reference to the content of the subordinate clause, communicates a foregrounding of, or emphasis on, that content. For example, with verbs which can communicate either cerebral knowledge or sensory experience, presence of a cataphoric pronoun favors the “sensory experience” reading. Compare Henry loves to read timetables. He even

Paradigms, analyses, and dialogue

knows when the Pennsylvania train arrives versus If you live near the Santa Monica Airport, you (really) know it when the planes land. Why the whole house vibrates! Theoretical issues which arise in “classical” sign-based analysis are the focus of Part II of the volume. Presenting an analysis of the English prepositions in, on, and at, Wallis Reid’s “Monosemy, homonymy, and polysemy” provides an articulate and vigorous defense of what John Taylor, in a number of publications, has called the “two-level” model or the “semiotic model” (2003), one which — as the Columbia School does — rigorously distinguishes between “a linguistic-semantic level of meaning, and an essentially non-linguistic, conceptual level of interpretation” (Taylor 1995b: 3–13). This two-level model, in contrast to Langacker’s network model of meaning (1988), rejects the concept of polysemy as a pre-theoretical folk notion. The first argument Reid offers for this rejection is that the concept of polysemy does not fit with the Columbia School’s view of linguistic meaning as instrumental rather than necessarily representational: cf. Huffman (1997). Whether language is in fact representational is an empirical issue, to be decided by actual analyses, and one should not adopt a richer theory before first seeing if a sparser theory works. If the message communicated with an utterance is in fact not built up in a water-tight manner from meanings acting like little semantic building blocks or Lego pieces, but is instead inferred from meanings envisaged as rough hints (which taken together comprise the barest “semantic skeleton” consonant with the message), then various contrasts in extra-linguistic reality need not be viewed as being built into the language itself. In other words, an analysis of in as a purely linguistic signal — a hint rather than a picture of a scene — need not differentiate between the different sorts of physical locations of the referents of bird, bullet, coffee, and crack in There’s a bird in the tree ( ’ in the tree’s branches), There’s a bullet in the tree ( ’ in the tree’s trunk), There is coffee in the cup ( ’ in the hollow cavity formed by the glass or plastic or bronze making up the cup), and There’s a crack in the cup ( ’ in the material out of which the cup is formed, creating the above mentioned cavity, but probably not the material making up the handle).⁶ More crucial for the consideration of English as a communicative instrument (rather than a representational device) are the contrasts, the oppositions, between the prepositions in Hank met Nancy at/on/in the street or The lumberjack is at the tree (’ standing at the base of the vertical tree, waiting to begin work), The lumberjack is in the tree (’ standing among and enveloped by the branches, sawing off the smaller branches before the larger ones and before attacking the trunk), and The lumberjack is on the tree (’ standing on top of the upper surface of the now sawed-down tree, which is lying horizontally on the ground). A second reason Reid offers for rejecting polysemy, first articulated by Ruhl (1989), is that many signs appear to be polysemous only because they are analyzed by linguists relying on their own intuitions and contemplating artificial data separated from any context. That way, one simply cannot see the contribution of the context to the message communicated with the word. Once one examines large numbers of attested uses of the word in full context, as in Ruhl (2002), what at first appeared to be clearly distinct isolated senses, with “spaces in between,” merge into a continuum

5

6

Robert S. Kirsner

which cannot be chopped up on a principled basis into distinct, discrete units.⁷ Ockham’s razor then requires one to at least consider a monosemic analysis. Without recapitulating Reid’s analysis here, we draw attention to the fact that he uses several concepts analogous to those invoked by Cognitive Grammar but accords them a different theoretical status. Reid discusses conceptualization at length, but views different conceptualizations as different extralinguistic strategies for the use of a (monosemic) meaning rather than different (polysemous) meanings in themselves. For example, Reid proposes that there is a conventionalized trajectory strategy for the use of on, based on the use of on for linear location (The bird is on the telephone wire) and linear movement (The plain is on course), which accounts for the difference between get in the car and get on the bus. Buses, as part of a public transportation system, follow a known sequence — a fixed trajectory — of scheduled stops; private cars do not. (Of course one can say Pete got in the bus, but the message probably involves a private bus, perhaps a Volkswagen Minibus. And if you do say He got on the minibus, you are suggesting that the minibus in question is now following a known trajectory; perhaps it is a shuttle bus, picking up people at their homes but taking them all to a fixed destination, such as the airport.)⁸ A crucial point is that for Reid, conventionalized strategies for exploiting meanings, based on specific conceptualizations of reality, are no more a part of the communicative instrument called a language than Bach fugues are a part of the musical instrument called a harpsichord. Just as it takes years of practice for a harpsichordist to learn to meld individual notes into a product understood as music by an audience, so does it take years of practice for the language user to learn to manipulate the individual meanings signaled by the individual signals of the language to produce utterances which are understood as communicating something by his or her interlocutor. What Reid is suggesting is that an analysis of a language as a communicative instrument (illustrated here with in, on and at) should concern itself primarily if not exclusively with what is in the instrument itself, what its structure is, and nothing else, just as the analysis of a harpsichord should lay out the strings and the bridge and the nut and the registers and the keyboard, etc., but not discuss the musical styles of Johann Sebastian Bach or Theolonius Monk or the set of skills which a competent harpsichordist needs to have acquired. The problem which now presents itself is really nothing less than that of the very purpose of linguistics. How immediately “useful” should a linguistic theory be and in what ways? How much of the usage of a language is it the linguist’s obligation to explain and describe and how much is the responsibility of other disciplines? Should the linguistic analysis of a language be instantly and directly convertible into a textbook for learning that language, as the Cognitivist’s reluctance to separate linguistic from cultural and encyclopedic knowledge might suggest? We may perhaps advance the discussion by extending the musical instrument analogy introduced above just a bit more. If one is a harpsichord expert, then one can accomplish something with a barebones technical description of a given harpsichord — perhaps compare different

Paradigms, analyses, and dialogue

harpsichords made by different craftsmen. Furthermore, armed with an analysis, a blueprint, of a harpsichord, the expert can proceed to take the harpsichord apart, separate the individual pieces on the floor of a warehouse, and verify independently and intersubjectively, by inviting in other experts, that the description of each part is precisely what the blueprint says it is. And he or she can then reassemble the harpsichord in the way shown by the blueprint and verify that, by hitting this particular key, a different noise is produced than by hitting that particular key. And then the expert can pick up a musical score, note that the musical score is really not in the harpsichord itself, and learn how to play the harpsichord to mimic the score. None of this is possible with language. A Columbia School analysis without a detailed discussion of the data on which the analysis is based (e.g. Reid’s analysis of in, on, at presented without his consideration of strategies) is roughly like the harpsichord expert’s technical description of a harpsichord, which would be useless to someone who, never having even seen a harpsichord before, having perhaps only heard a Bach fugue once, now wants to learn how to play one. Furthermore, there is no way to lay out the parts of a language on the floor of a warehouse the way one can lay out the keyboard, strings, etc. of a harpsichord and hence no way to be sure that one can really separate the harpsichord of language (the semiotic instrument) from the music it produces or the talents of the harpsichordist. Cognitivists will undoubtedly suspect that, in the case of language, there is in fact no separate or intersubjectively separable “harpsichord,” no purely semiotic level. Given that such a situation is not inconceivable, it might be nothing more than a relatively fruitless cerebral and ultimately purely esthetic exercise to attempt to partition the phenomena of language into an individual “harpsichord”, with individual strings and bridges etc., producing individual notes, all of this entirely separate from, on a different level from the subject and answer and countersubject composing a fugue played on it. What Reid is really attempting to do is to take the music produced by the initially unanalyzed complex of “harpsichord + harpsichordist + musical score” and use performed fugues (perceived messages of texts and conversations) to deduce what the blueprint of the harpsichord is, how the strings and keys and other constituents are arranged. As I have suggested earlier (Kirsner 2002), this kind of approach might not appeal to folks who want to learn how fugues are played.⁹ But, Columbia School linguists will counter that any analysis which does not attempt to separate some sort of relatively parsimonious “harpsichord of language” from the complex of “harpsichord + harpsichordist + musical score” is doomed to remain eternally programmatic, making perhaps attractive majestic and ambitious claims about the mind, but never actually delivering on its promises. At this point it is up to the reader to judge how productive these two approaches to linguistics — the cognitive and the communicative — ultimately are. In any case, the merit of Langacker’s chapter and of Reid’s chapter is that they have demonstrated how each of their approaches is utilized in a linguistic analysis and have discussed where they agree and where they disagree. We may perhaps even perceive modest signs of convergence: First of all, there is Reid’s approval of analyses of the

7

8

Robert S. Kirsner

kind outlined in George Lakoff and Mark Johnson’s book Metaphors We Live By. Second, Reid’s own verbal discussion of how one strategy (linear location) can lead to another (the trajectory strategy) could possibly be viewed as a notational variant of nodes in a conceptual network if not an orthodox semantic network. At the same time, we may observe a recent desire among cognitivists to once again take up Ockham’s razor, recognize the open-endedness of all contexts and the ongoing creativity of language users, and thereby reduce the degree of crystallized, fixed, conventionalized polysemy postulated in their analyses; cf. Janssen (2003) for stimulating discussion. And, as was already indicated above in endnote 7, Tyler and Evans’s Cognitive analysis of over (2003), clearly concerned with reining in the “rampant polysemy” (Taylor 1993b: 13) found in earlier Cognitive linguistic works, makes use of some notions with clear counterparts in Columbia School linguistics. Mark Elson’s tightly-wrought chapter “On the relationship between form and grammatical meaning in the linguistic sign” challenges the widely held assumption that the inflectional ending of a word (i.e. the desinence) explicitly signals all of the grammatical meaning(s) associated with the word. Utilizing data from the history and dialects of Slavic and Romance languages, Elson proposes that oppositional relationships within paradigms — specifically between hierarchically organized subparadigms — permit endings to communicate far more than they may be shown to actually signal. The theoretical merit of this chapter is its demonstration that the kind of distinction the Columbia School makes between meaning and message (cf. Contini-Morava 1995: 5–7) applies not only to phenomena traditionally subsumed under the term “syntax” but also within the word, to morphology as well. Elson’s command of his data, together with his relentlessly semiotic argumentation, should induce many contemporary morphologists to rethink their positions. Joseph Davis’s “Revisiting the gap between meaning and message” argues against the theoretical construct of a strategy for the communicative exploitation of a meaning, a construct first introduced into Columbia School thought in García (1975), made use of in e.g. Kirsner (1979), Sussman Goldberg (1995), discussed at length in Reid (1995), and appealed to by Reid in the present volume. Davis suggests that the very notion of “strategy” is an illegitimate reification of an aspect of the inferential process whereby a message is “leapt to” from a meaning. In strict Columbia School thought (and in direct opposition to what is envisaged in Cognitive Grammar), this process — in reality involving multiple factors and cues — is held to be outside the data for which the analyst can reasonably claim analytical responsibility. Accordingly, Davis urges linguists to dispense with the concept of “strategy.” Text counts and other quantitative data (a basic pillar of Columbia School validation techniques going back to Diver 1969) should be viewed not as tests of a hypothesized “strategy” but as a test of the hypothesized meaning of the signal being studied. One predicts statistical skewings in the use of the signal on the basis of (i) the hypothesized meaning and (ii) the general expectation that, for communication to be possible at all, meanings will tend to be used coherently, synergistically, with most cues to the message “pointing in the same direction.”

Paradigms, analyses, and dialogue

Readers concerned with the status of linguistics as a science (and hence with the precise role of Popperian falsifiability in linguistic analyses; cf. Kirsner (1989) and the discussion of Karl Popper in Chalmers (1999) ) will find much to think about in Davis’s assertion that “A CS [Columbia School ] signal-meaning hypothesis can be neither proved nor disproved; it can be only accepted or rejected.”¹⁰ Another crucial issue is the possibility that, by jettisoning the concept of strategy, Davis may be further restricting the scope of phenomena for which a Columbia School linguistic analysis would claim responsibility, thereby making the Columbia School approach more vulnerable to Langacker’s criticism that it attempts to explain too little about Language to be interesting. The idea that there might be a finite number of conventionalized strategies made possible by — motivated by — the meaning of a signal, yet underdetermined by that meaning, allows one to explain crucial “negative data” — why a signal is not used in all the imaginable ways which might be compatible with its hypothesized meaning. As Langacker notes, strategies have been one way in which the Columbia School has dealt with data which Cognitive Grammar would treat by invoking polysemy. It would seem to be such data which at least have the potential of inducing Columbia School linguists to abandon their initial, plausible hypothesis that language is a sparse communicative tool and accept the view that it is (at least in part) a richer, representational system.¹¹ Note that Reid’s chapter explicitly raises this possibility. To the extent that Davis’s rejection of the concept of strategy narrows the empirical scope of Columbia School analyses (namely by focusing analyses on only data which actually occur and not on conceptually possible data which do not occur), it could be viewed as elevating the idea of sparse purely hint-like meanings from a hypothesis to a dogma. This issue is the subject of vigorous current debate among Columbia School linguists. In any case, if Davis’s proposal were to be accepted, many earlier Columbia School analyses making use of the concept of strategy, including those by the movement’s founder William Diver, would have to be rethought and redone. Part III presents five analyses: three of Indo-European languages, one of a Semitic language, and one of a Yuman language. The analyses by Jing, de Jonge, and Stern are primarily influenced by Columbia School theory. Oron and Tobin meld Columbia School theory with a Jakobsonian style of semantic analysis, especially as further extended by van Schooneveld (cf. Tobin 1990: 74–77). Finally, IchihashiNakayama invokes Grammaticalization Theory (cf. Hopper and Traugott 1993) and, implicitly, the possibility of a network analysis linking the different senses of the forms which developed at different times.¹² Zhuo Jing’s chapter “The givenness of background: A semantic-pragmatic study of two Modern German subordinating conjunctions” demonstrates that the standard reference and pedagogical grammars’ characterization of both wenn ‘if, when’ and als ‘if ’ as massively polysemous is fundamentally flawed. First, such a characterization suggests erroneously that the two forms can be synonymous, which is never the case. Neither does it explain why als is never used with reference to future time. Drawing on the Columbia School view that the meaning signaled by a

9

0

Robert S. Kirsner

form can be an instruction to the hearer rather than a description, Jing proposes that als signals BACKGROUND GIVEN while wenn signals BACKGROUND NOT GIVEN. She demonstrates in detail how these apparently sparse meanings can be used in rich discourse contexts to guide the hearer’s interpretation, buttressing her analysis with textual examples and frequency counts from both fiction and non-fiction (philosophic) texts. Another good example of the Columbia School approach to linguistic analysis is Bob de Jonge’s “The relevance of relevance in linguistic analysis: Spanish subjunctive mood”. This chapter presents a semantic analysis of the Spanish subjunctive endeavoring to explain both (a) its obligatory use in subordinate clauses with certain verbs and certain conjunctions and (b) its variable use in other environments, where it competes with the indicative. His fundamental claim is that the subjunctive, in all its uses, indicates CONTEXTUAL RELEVANCE OF AN ALTERNATIVE FOR THE OCCURRENCE EXPRESSED BY THE VERB and he offers both qualitative data (examination of examples in context) and text counts using different contextual variables in defense of this hypothesis.¹³ Nancy Stern’s “Sign-based analysis of English pronouns in conjoined expressions” investigates the use of simple pronouns and their corresponding self- forms in collocations such as the huge gap between herself and Stefan versus the intimate connection between her and Stefan. One of her empirical findings is that -self is used where there is conceptual distance between the conjuncts while the simple form is used where there is conceptual closeness. These and other observations support her analysis of -self as signaling INSISTENCE ON AN ENTITY. By urging the hearer to focus specifically on one member of the two entities referred to in the between phrase, by insisting that the hearer focus on that entity as opposed to others, the speaker hints to the hearer that the two entities are somehow unequal, that there is a distinction to be made between them and, hence, that they do not form a unified pair. The converse is shown by what Stern calls the “conspiratorial” tone of between you and me in such utterances as Between you and me, she’s an awful singer, in which the speaker invites the hearer to share his opinion that the singer stinks; cf. the unidiomatic ?Between you and myself, she’s an awful singer. An additional merit of Stern’s chapter is its underscoring of the fact that the use of the self-forms is multiply-determined. In other contexts, factors other than conceptual distance between conjuncts may influence the speaker’s choice of a bare pronoun or its corresponding self-form. Noah Oron and Yishai Tobin’s “Semantic oppositions in the Hebrew verb system: A sign-oriented approach” takes as its starting point the analysis of the Hebrew verb laid out in Tobin (1995) and presents a detailed discussion of two derivational infix patterns or binyanim: the (traditional simple) PAAL pattern exemplified by katav ‘(he) wrote,’ based on the triconsonantal root k-t-v ‘writing,’ and the corresponding (traditional reflexive) HITPAEL pattern exemplified by hitkatev ‘(he) corresponded.’ They argue convincingly that preceding traditional and modern sentence-based analyses of the Hebrew verb, which do not postulate single invariant

Paradigms, analyses, and dialogue

meanings for each of the patterns, all ultimately degenerate into relatively unhelpful lists of heterogeneous uses. They present evidence that the binyan system is not frozen, as is sometimes thought, but is exploited for neologisms. The heart of the chapter is a demonstration that the differences between such varied PAAL forms as ‘ran,’ ‘breathed,’ ‘cried,’ and ‘was furious’ and the corresponding HITPAEL forms ‘ran back and forth,’ ‘panted,’ ‘whined,’ and ‘became furious’ can best be understood as examples of the opposition between the meaning SINGLE PERCEPTION OF ACTIONS, STATES OR EVENTS and the meaning MULTIPLE PERCEPTION OF ACTIONS, STATES OR EVENTS. Of particular merit, I believe, is the authors’ statement that the meaning of a HITPAEL form cannot be mechanically predicted from the meaning of the corresponding PAAL form but is rather motivated by it. One can still detect the semantic ingredients contributed by the HITPAEL pattern (to the meaning of the root as seen in the PAAL pattern) even if the verb-form in question has gone on to develop new, specialized idiosyncratic or idiomatic usages.¹⁴ Kumiko Ichihashi-Nakayama’s contribution, “Grammaticalization of ‘to’ and ‘away’: A unified account of -k and -m in Hualapai,” grapples with a classic problem in American Indian linguistics. She argues that the morphemes -k and -m in the Hualapai language, each of which is associated with a number of different “functions,” each has a single basic meaning. The morpheme -k signals INSIDE/TOWARDS FOCAL POINT and -m signals OUTSIDE/AWAY FROM FOCAL POINT. In different linguistic contexts, such as with nouns referring to places, with motion verbs, and in sentence-final position, the morphemes are associated with different specific messages: allative versus ablative, towards speaker versus away from speaker, speaker does or does not vouch for the statement, same subject across two clauses versus different subject, discontinuous events in different times and places versus iteration of one activity at a single time and place.¹⁵ Particularly stimulating is the issue Ichihashi-Nakayama raises concerning the best explanation for the wide range of uses of the forms, synchronic or diachronic. From the vantage point of Grammaticalization Theory, the various functions can be seen as representing distinct diachronic stages. Turning now to the final section of the volume, Part IV, we may note that Shabana Hameed’s chapter, “Interaction of physiology and communication in the makeup and distribution of stops in Lucknow Urdu,” is the only one in the volume dealing with phonology and the only one illustrating William Diver’s uniquely quantitative approach to phonology: Phonology as Human Behavior.¹⁶ Rather than classifying consonants in the customary textbook way, in terms of points of articulation, Hameed analyzes them according to the articulators producing them and devotes considerable discussion to the ranking of articulators on a scale of adroitness. Physiological factors interact with communicative ones in such a way that, while there is a tendency toward full combination of the five separate Urdu articulators with both aspiration and voice, there are various asymmetries in the distribution of stops in word-initial and word-final position. Previously, most studies in Phonology as Human Behavior have dealt with Romance, West Germanic, Scandinavian, Slavic



2

Robert S. Kirsner

and Semitic languages: cf. Tobin (2002). The present chapter shows what can be learned about phonology from an Indic language with much more complex phonological units. The data were obtained during several years of field work. The second chapter in this final section concerns an intriguing problem in (sub) morphemic analysis. Yishai Tobin’s “Between phonology and lexicon: The Hebrew triconsonantal (CCC) root system revolving around /r/ (C-r-C)” constitutes the initial stage of an attempt to find semantic subsystems within tri-consonantal Hebrew roots. Here Tobin outlines the hypothesis that C-r-C roots, containing the trilled /r/ phoneme, communicate the general notion of ‘a change in structure,’ which may be specified further in individual cases. This chapter might remind readers both of Bolinger’s discussion (1950 a, b) of phonesthemes and also of Tobin’s earlier work on word systems in Modern Hebrew (Aphek and Tobin 1988). Clearly, such an intriguing hypothesis will have to be tested by comparison with C-C-C roots containing other medial consonants. Furthermore, the notion ‘change of structure’ will have to be operationalized so that different analysts can agree as to what does and what does not fall under this rubric. The observable word orders of a language may profitably be viewed as the “vector sum” of a number of different influences. Where word order is not used as a fixed sign, i.e. an invariant signal of an invariant meaning, it may be open to more general communicative influences, such as the strategy of placing given information before new. The chapters by Otheguy, Rodríguez-Bachiller, and Canals on the one hand and by Ocampo on the other illustrate, respectively, the Columbia School approach and an eclectic approach to word-order variation. Otheguy, Rodríguez-Bachiller and Canals’s chapter, “Length of the extra information phrase as a predictor of word order: A cross-language comparison” is best characterized as an experiment. The presence or absence of dedicated word-order signals in a language is treated as the independent variable and the observable word order of certain other constituents in that language is treated as the dependent variable. The chapter asks if there should be a difference in the placing of prepositional phrases and adverbs modifying verbs (thereby providing “extra information” about the event) in one language where the relative order of subject, object, and verb is a dedicated sign of participancy (Degree of Control) in the event (English) and two where it is not (Spanish, Catalan). In other words, how does the presence or absence of a word-order signal in a language interact with other factors or principles influencing word order? The authors propose a cognitive-iconic principle — Group together signals whose meanings form cognitive units: keep together what goes together — according to which conceptually related constituents (such as verbs and adverbs) should tend to be grouped together.¹⁷ In English this principle clearly conflicts with the placement of the lower controller (“direct object”) immediately following the verb. Meticulous use of translations (from English to Spanish and from Spanish and Catalan to English) provide an analytical control. Various quantitative predictions are made about the placement and relative “bulk” of the extra information. These are largely confirmed with frequency counts from texts.

Paradigms, analyses, and dialogue

Ocampo’s “Word order variation in spoken Spanish in constructions with a verb, a direct object, and an adverb” attempts to delineate the syntactic, cognitive (including iconic), pragmatic, and prosodic features which correlate with different aspects of word order in cases where the relative order of verb, direct object, and adverb cannot be (or at least has never been) considered a sign, i.e. an invariant signal of an invariant meaning. The merit of this chapter is that it teases apart the influence of a number of different factors on word order (including “higher order” ones, such as the general communicative technique of putting given information before new), and also attempts to come to grips with the simultaneous influence of more than one factor. Furthermore, as Columbia School papers have almost always dealt with written languages, it is therapeutic to confront the different kind of complexity which spoken language data provide.¹⁸ The last chapter in the volume, Angelita Martinez’s “Variación sintáctica y estrategias discursivas,” discusses the variable use of the Spanish clitic pronouns lo, la and le in regions in Argentina with a Quechua substrate. The main thesis of her chapter is that, under the influence of discourse strategies in Quechua (made possible by the particular signals and meanings of that language), speakers manipulate the Spanish dative versus accusative contrast (signaling greater versus lesser “potency” of participants in events; cf. García (1975)) in ways not seen in other dialect areas to create expectations and suspense in narratives. It is now time for the cartographer to put away his tools and for the headwaiter to take his menu (and even his tray of hors d’oeuvres) back to the kitchen. As it is never possible for the map to be the territory or the menu to be the meal, the reader is now invited to turn from these brief and hopefully enticing remarks to the chapters themselves.

Notes * I want to thank my fellow editors Ellen Contini-Morava and Betsy Rodríguez Bachiller for critical feedback on an earlier version of this Introduction. I also wish to acknowledge the support of Grant 2964 from the Faculty Grants Program of the Committee on Research, Academic Senate, University of California, Los Angeles. 1. For earlier volumes which developed out of such conferences, see Contini-Morava and Sussman Goldberg eds. (1995) and Reid, Otheguy, and Stern eds. (2002). 2. For earlier critiques of Generative Grammar from a sign-based perspective, see Garcia (1975), Kirsner (1979), and Huffman (1997). 3. See, for example, Huffman’s discussion of Smyth’s (1920) treatment of the Greek dative (1997: 16–17). 4. I have adopted this characterization from Williams (1995). Note that Shermer (1997: 58) explicitly classifies the slippery slope argument as a fallacy. 5. For more on the System of DIFFERENTIATION in English, see Diver (1984: 23–28). For a Columbia School discussion of adjective-noun order in Spanish, see Klein-Andreu (1983).

3

4

Robert S. Kirsner For a brief presentation of the System of DEGREE OF CONTROL in English (where CONTROL is not to be confused with the use of the word control in Generative Grammar), see Reid (1991: 174–178). Huffman (1997: 294–298) compares CONTROL systems in English and French. In the present volume, Otheguy, Rodríguez-Bachiller, and Canals contrast English , which has an explicit System of of DEGREE OF CONTROL, with Spanish, which does not. 6. I have taken the liberty of choosing different (and perhaps more “extreme”) lexical items than Reid’s to underscore his point that he is viewing language as a purely communicative rather than a fully representational system. All in says here is that the associated lexical item is to be taken as a location and that three dimensions are relevant rather than none (at) or a number of dimensions greater than zero and less than three (on). Conceptualizing the actual physical relationship involved in locating the item in question requires real-world knowledge of the properties and uses of the nouns’ referents. (Birds are capable of sitting on the branches of a tree, while kites can become entangled in those branches without sitting in them, and a bullet can do neither.) 7. See now also the last chapter of Sampson (2001), entitled “Meaning and the limits of science,” where it is argued that word meanings are essentially indeterminate. Here, of course, Cognitivists would argue, first, that the kind of general meanings proposed by Ruhl are so ethereal, so ineffable, as to be communicatively — and analytically — useless (cf. Taylor 1995a: 267). Second, they would suggest that there are indeed principled bases for segmenting the apparent message continuum without ending up with unconstrained, infinite polysemy. In their analysis of the English preposition over, Tyler and Evans (2003) suggest that previous treatments (by Brugman and by Dewell) have been “too fine-grained,” (postulate too many distinct nodes and paths) and that it is possible to distinguish intersubjectively between senses, (which are linguistic and stored in memory) and “constructed meanings” (which involve conceptual and online processes.) What is exciting for the Columbia School linguist here is that Tyler and Evans’s argument invokes (2003: 145–6) — without saying so explicitly — both the notion of semantic opposition and the idea that the speaker selects that particular preposition (from a closed system of prepositions) which “fits best” with the scene being described. Compare the Columbia School notion of “the least inappropriate meaning relative to a given message” in Kirsner (1979: 32–33). 8. Note the importance of lexical domains here and knowledge of the properties and social use of referents in the world. When physical enclosure by the transporting device is not possible, in is less likely to be used and on is used in the linear location sense without suggesting a fixed trajectory. One gets on a horse, on a bicycle but not usually on the 5:30 horse or the 7AM bicycle to the station even though one gets on a train. The possibility of salient enclosure seems to be the relevant variable in such cases as She was sitting on a stool/on a chair/ in an easy chair/ in a wheelchair. She was sitting on a wheelchair seems to suggest that the wheelchair is not functional as a wheelchair, perhaps because it has tipped over on its side. 9. For further discussion of how different the aims of a Columbia School analysis are from those of other theoretical frameworks and why they are different, see Chapter 10 of Reid (1991) as well as Reid (2002). There is a danger, of course, that if the theoretical constructs of the Columbia School cannot eventually be “triangulated” (cf. Wolpert 1992: 124–5) by other disciplines such as psychology and cannot prove themselves to be useful, outside of linguistics, to people attempting to understand language from other vantage points, it will suffer the fate of typewriters when the personal computer (with word-processing programs) was developed. As Verhagen (1992: 302–3) points out in his argument for the superiority of

Paradigms, analyses, and dialogue Cognitive Grammar over Generative Grammar, computers did not “scientifically disprove” typewriters. They just allowed people to do much more than they ever could with a typewriter while also performing the functions of one, so that eventually most people, except for a Luddite or two, “simply” stopped using typewriters. 10. For a thorough discussion of what has been called falsificationism, see Chapters 5 through 6 in Chalmers (1999). For further material on the scientific status of linguistics, see Chapter 9 of Reid (1991) and also Cognitive Linguist Laura Janda’s discussion of prediction in linguistics (2000: 5). She makes the interesting argument that the notion that linguistics is (or should be) an exact science is actually a folk theory of science, not unlike the conduit metaphor of language mentioned by Reid in his chapter in the present volume. 11. Columbia School linguists do, of course, accept the concept of iconicity — for example the idea that, in the absence of any additional information such as that provided by the explicit signals before and after, the very order of mention of events (and hence the processing of that order in interpretation) usually reflects (represents?) their extralinguistic order in time: He entered the room, put on his jacket, and straightened his tie refers to a different sequence of activities than He straightened his tie, put on his jacket, and entered the room. 12. For detailed discussion of the possibility of recognizing semiotic networks within a fundamentally Columbia School analysis (while preserving the theoretical differences between the Columbia School and Cognitive Grammar), see Contini-Morava (2002). 13. A historical note: De Jonge’s analysis is similar to an unpublished analysis of the Latin subjunctive once presented by William Diver in his “Structure of Latin” courses at Columbia University in the mid-1960s. At one point Diver argued that the Latin subjunctive indicated OCCURRENCE QUESTIONED while the indicative signaled only OCCURRENCE. To question an occurrence (but without asking someone whether it actually happened or not) is to raise the possibility of alternatives — of the event not happening or happening differently, which seems close to De Jonge’s postulated meaning. 14. Tobin and Oron’s contrasting “motivation” to “prediction” is quite compatible with Langacker’s use of these same terms; cf. Langacker (1991: 517) and elsewhere. 15. Readers familiar with Contini-Morava (1991) might wonder if the meaning signaled by -m, associated with event discontinuity, could not be analyzed as having a deictic component. 16. For a full explanation of the theory as well as its clinical application to the treatment of children with speech disorders, see Tobin (1997). 17. This principle appears to have been first enunciated by Otto Behagel in the 1920s: “Das oberste Gesetz ist dieses, daß das geistig eng Zusammengehörige auch eng zusammengestellt wird.” or “What belongs together mentally is placed together syntactically.” (This citation and its translation are from Slobin 1985: 228). 18. An exception is Klein-Andreu (1989).

References Aphek, Edna and Tobin, Yishai. 1988. Word Systems in Modern Hebrew: Implications and Applications. Leiden and New York: E. J. Brill. Bolinger, Dwight. 1950a. “Rhyme, assonance, and morpheme analysis.” Word 6: 117–136.

5

6

Robert S. Kirsner Bolinger, Dwight. 1950b. “Shivaree and the phonestheme.” American Speech 25: 134–5. Brugman, Claudia. 1981. The story of over. Berkeley: M. A. Thesis, Linguistics Department, University of California. Published [1988] as The Story of Over: Polysemy, Semantics, and the Structure of the Lexicon. New York: Garland Press. Chalmers, A. F. 1999. What is This Thing Called Science? Third edition. Indianapolis/Cambridge: Hackett Publishing Company. Contini-Morava, Ellen. 1989. Discourse Pragmatics and Semantic Categorization: The Case of Negation and Tense-Aspect with Special Reference to Swahili. Berlin: Mouton de Gruyter. Contini-Morava, Ellen. 1991. “Deictic explicitness and event continuity in Swahili discourse.” Lingua 83: 277–318. Contini-Morava, Ellen. 1995. “Introduction: On linguistic sign theory.” In Contini-Morava and Sussman Goldberg (eds.), 1–39. Contini-Morava, Ellen. 2002. “(What) do noun class markers mean?” In Reid, Otheguy, and Stern (eds.), 3–64. Contini-Morava, Ellen and Sussman Goldberg, Barbara (eds.) 1995. Meaning as Explanation: Advances in Linguistic Sign Theory. Berlin: Mouton de Gruyter. Dewell, Robert, 1994. “Over again: image-schema transformations in semantic analysis. Cognitive Linguistics 5 (4): 351–380. Diver, William. 1969. “The system of Relevance of the Homeric verb.” Acta Linguistica Hafniensia 12 (1):45–68. Diver, William. 1984. The Grammar of Modern English. Unpublished textbook for Linguistics G6801, Columbia University. Fauconnier, Gilles. 1985. Mental Spaces. Aspects of Meaning Construction in Natural Language. Cambridge: MIT Press. García, Erica. 1975. The Role of Theory in Linguistic Analysis: The Spanish Pronoun System. Amsterdam: North-Holland. Goldberg, Barbara Sussman. 1995.“The -ra and -se opposition in Spanish.” In Contini-Morava and Sussman Goldberg (eds.), 381–404. Hopper, Paul J. and Traugott, Elizabeth Closs. 1993. Grammaticalization. Cambridge: Cambridge University Press. Huffman, Alan. 1997. The Categories of Grammar: French lui and le. Amsterdam/Philadelphia: John Benjamins. Janda, Laura. 2002. “Cognitive linguistics.” SLING2K February 2000. Manuscript. Revised version forthcoming in Fowler, George (ed.) Slavic Linguistics 2000. Bloomington: Slavica. Janssen, Theo. A. J. M. 2003. “Monosemy versus polysemy.” In Cognitive Approaches to Lexical Semantics [Cognitive Linguistics Research 23], Cuykens, Hubert; Dirven, René; and Taylor, John R. (eds.), 93–122. Berlin: Mouton de Gruyter. Kirsner, Robert S. 1979. The Problem of Presentative Sentences in Modern Dutch. Amsterdam: North-Holland. Kirsner, Robert S. 1989. “Does sign-oriented linguistics have a future?: On sign, text, and the falsifiability of theoretical constructs.” In Tobin (ed.), 161–178. Kirsner, Robert S. 1993. “From meaning to message in two theories: Cognitive and Saussurean views of the Modern Dutch demonstratives.” In Conceptualizations and Mental Processing in Language [Cognitive Linguistics Research 3], Geiger, Richard A. and Rudzka-Ostyn, Brygida (eds.), 81–114. Berlin: Mouton de Gruyter.

Paradigms, analyses, and dialogue Kirsner, Robert S. 2002.“The future of a minimalist linguistics in a maximalist world.” In Reid, Otheguy, and Stern (eds.), 339–371. Klein-Andreu, Flora. 1983. “Grammar in style: Spanish adjective placement.” In Discourse Perspectives on Syntax, Flora Klein-Andreu (ed.), 143–179. New York: Academic Press. Klein-Andreu, Flora. 1989. “Why speech seems ungrammatical.” In Tobin (ed.), 25–44. Lakoff, George and Johnson, Mark. 1980. Metaphors We Live By. Chicago and London: University of Chicago Press. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar. Volume I. Theoretical Prerequisites. Stanford: Stanford. University Press. Langacker, Ronald W. 1988.“A usage-based model.” In Topics in Cognitive Linguistics [Current Issues in Linguistics 50], Rudzka-Ostyn, Brygida (ed.), 127–161. Amsterdam/Philadelphia: John Benjamins. Langacker, Ronald W. 1991. Foundations of Cognitive Grammar. Volume II. Descriptive Application. Stanford, California: Stanford University Press. Nerlich, Brigitte; Tood, Zazie; Herman, Vimala; and Clarke, David D. (eds.) 2003. Polysemy: Flexible Patterns of Meaning in Mind and Language. Berlin and New York: Mouton de Gruyter. Nikiforidou, Kiki. 1991 “The meanings of the genitive: A case study in semantic structure and semantic change.” Cognitive Linguistics 2 (2):149–205. Reid, Wallis. 1991. Verb & Noun Number in English: A Functional Explanation. London and New York: Longman. Reid, Wallis. 1995. “Quantitative analysis in Columbia School theory.” In Contini-Morava and Goldberg (eds.), 115–152. Reid, Wallis. 2002. “Introduction: Sign-Based Linguistics.” In Reid, Otheguy, and Stern (eds.), ix-xxi. Reid, Wallis; Otheguy, Ricardo; and Stern, Nancy. (eds). 2002. Signal, Meaning, and Message: Perspectives on Sign-Based Linguistics. [Studies in Functional and Structural Linguistics 48]. Amsterdam/Philadelphia: John Benjamins. Ruhl, Charles. 1989. On Monosemy: A Study in Linguistic Semantics. Albany: State University of New York Press. Ruhl, Charles. 2002. “Data, comprehensiveness, monosemy.” In Reid, Otheguy, and Stern (eds.), 171–189. Sampson, Geoffrey. 2001. Empirical Linguistics. London and New York: Continuum. Shermer, Michael. 1997. Why People Believe Weird Things. Pseudoscience, Superstition, and Other Confusions of Our Time. New York: W. H. Freeman and Company. Slobin, Dan. I. 1985. “The child as linguistic icon-maker.” In Iconicity in Syntax. [Typological Studies in Language 6], Haiman, John (ed.), 221–248. Amsterdam/Philadelphia: John Benjamins. Smyth, Herbert Weir. 1920. Greek Grammar. Cambridge: Harvard University Press. Taylor, John R. 1995a. Linguistic Categorization: Prototypes in Linguistic Theory. Second edition. Oxford: Clarendon Press. Taylor, John R. 1995b. “Models of word meaning in comparison: The two-level model (Manfred Bierwisch) and the network model (Ronald Langacker).” In Current Approaches to the Lexicon. [Duisburger Arbeiten zur Sprach- und Kulturwissenschaft 24)], Dirven, René and Vanparys, Johan (eds.), 3–26. Frankfurt am Main: Peter Lang. Taylor, John R. 1999. “Cognitive semantics and structural semantics.” In Historical Semantics and Cognition [Cognitive Linguistics Research 13], Blank, Andreas and Koch, Peter (eds.), 17–48. Berlin and New York: Mouton de Gruyter.

7

8

Robert S. Kirsner Taylor, John R. 2003. “Cognitive models of polysemy.” In Nerlich, Tood, Herman, and Clarke (eds.), 31–48. Tobin, Yishai. (ed.) 1989. From Sign To Text: A Semiotic View Of Communication. Amsterdam/ Philadelphia: John Benjamins. Tobin, Yishai. 1990. Semiotics and Linguistics. London and New York: Longman. Tobin, Yishai. 1995. Invariance, Markedness and Distinctive Feature Analysis: A Contrastive Study of Sign Systems in English and Hebrew. Amsterdam/Philadelphia: John Benjamins and Be’er Sheva: Ben Gurion University Press. Tobin, Yishai. 1997. Phonology as Human Behavior: Theoretical Implications and Clinical Applications. Durham: Duke University Press. Tobin, Yishai. 2002. “Phonology as human behavior. Initital consonant clusters across languages.” In Reid, Otheguy, and Stern (eds.), 191–255. Tyler, Andrea and Evans, Vyvyan. 2003. “Reconsidering prepositional polysemy networks: the case of over.” In Nerlich, Tood, Herman, and Clarke (eds.), 95–159. Verhagen, Arie. 2002.“Praxis of linguistics: Passives in Dutch.” Cognitive Linguistics 3(3): 301– 342. Williams, Bernard. 1995. “Which slopes are slippery?” In Making Sense of Humanity and Other Philosophical Papers 1982–1993. Williams, Bernard, 213–223. Cambridge: Cambridge University Press. Wolpert, Lewis. 1992. The Unnatural Nature of Science. London: Faber and Faber Ltd.

PART I

Cognitive Grammar

CHAPTER 1

Form, meaning, and behavior The Cognitive Grammar analysis of double subject constructions Ronald W. Langacker University of California, San Diego

Having long been aware of Columbia School work and its insights into language, I am pleased to have this occasion to assess its relationship to Cognitive linguistics. I will first outline some basic notions of Cognitive Grammar and then show its application to a particular grammatical phenomenon, namely double subject constructions. This will function as a substantive background for discussing similarities and differences between the Columbia School (CS) and Cognitive Grammar (CG) frameworks.

1 Cognitive Grammar The basic framework of CG has existed for well over two decades and endured without essential change. It has however been progressively refined, articulated, and empirically tested by application to a wide range of representative phenomena in diverse languages.¹ Here I can offer only a brief introductory sketch.

1.1. Basic principles My title reflects the theme of this volume: form, meaning, and behavior. Form and meaning figure in the central claim of CG, namely that lexicon, morphology, and syntax make up a continuum consisting solely of assemblies of symbolic structures. A symbolic structure is defined as the pairing between a form (i.e. a phonological structure) and a meaning (a conceptualization, in the broadest sense). It follows that all elements validly posited in grammatical description have some kind of meaning (often quite schematic). Note that the form referred to here is specifically phonological. What is commonly called “grammatical form” is analyzed in this framework as consisting in schematic symbolic assemblies, i.e. abstracted patterns of symbolization that are immanent in particular complex expressions. Grammatical markers, grammatical classes, and grammatical constructions are all inherently meaningful. Hence the architecture of the framework directly reflects the semiological func-

22

Ronald W. Langacker

tion of language, that of allowing conceptual structures to be symbolized by sound structures. Grammar is the schematic systematicity inherent in this symbolizing activity, patterns for combining symbols into complex expressions, characterized as symbolic assemblies. I take this view as being both natural and desirable from the methodological standpoint. To fulfill its semiological function, a language must at least comprise semantic structures, phonological structures, and symbolic links between the two. CG claims that only these elements are necessary. It thus achieves a substantial conceptual unification — the unification of grammar with lexicon — as well as their reduction to something more fundamental. This reduction does not however imply that grammar does not exist.² Conventional grammatical patterns have to be specifically learned by speakers and explicitly described by linguists. The framework is constrained and theoretically austere owing to the content requirement. This requirement specifies that the only elements ascribable to a linguistic system are: (i) semantic, phonological, and symbolic structures that are (part of) overtly occurring expressions (hence directly apprehended); (ii) abstractions (schematizations) of permitted structures; and (iii) categorizing relationships between permitted structures (e.g. the relationship between a schema and a specific structure that instantiates it). In phonology, for example, particular sounds like [f], [θ], and [s] can be posited by virtue of occurring in actual expressions; the schematic characterization [voiceless fricative] is permitted as an abstraction of these elements; and also permitted is the categorization of each sound as an instantiation of this schema. Options allowed in the postulation of semantic and grammatical elements are similarly constrained. Thereby precluded are such devices as “filters” (reflecting what cannot occur), derivations from underlying structures, and purely formal devices with no semantic or phonological content (e.g. syntactic tree structures as viewed in the generative tradition). Consequently, nothing is posited unless it is directly manifested in actual expressions, or else is capable of emerging from this data by the basic cognitive phenomena of abstraction (schematization) and/or categorization. This is one respect in which CG merits the label “cognitive”. Cognitive linguistics and CG are also “cognitive” in the sense that, insofar as possible, linguistic structure is analyzed in terms of more basic systems and abilities (e.g. perception, attention, categorization) from which it cannot be dissociated. This point can be illustrated by the strategy I have used in adopting and motivating some basic descriptive constructs. The strategy consists of seeking converging evidence from three sources: (i) a particular construct is shown necessary for the adequate semantic description of multiple phenomena in various languages; (ii) this construct is related to an independently observable cognitive ability; and (iii) this same construct proves critical for the explicit characterization of varied grammatical phenomena. With respect to the volume’s theme, this strategy assumes that some of the same cognitive abilities manifested in non-linguistic behavior are also crucial factors in linguistic meaning (and hence in grammar). Let me give three examples, which will serve to introduce some basic constructs I will need in the subsequent analysis.

Cognitive Grammar and double subject constructions

1.2. Profiling The first construct is what I call profiling. An expression evokes a certain array of conceptual content, called its conceptual base, as the basis for its meaning. Within this base, the expression profiles (i.e. refers to) a particular substructure. For instance, as sketched in Figure 1, the term hypotenuse evokes as its base the conception of a right triangle, within which it profiles the side opposite the right angle. Likewise, aunt profiles a human female characterized by position in a kinship network relative to a reference individual (ego). Note that profiling is indicated diagrammatically by heavy lines. a. hypotenuse

b. aunt

ego Figure 1.

Just from the standpoint of semantic description, there is ample motivation for adopting this construct — I see no way to properly characterize and distinguish linguistic meanings without it (or something with comparable effect). This is especially clear in cases where alternate profiles are imposed on the same base, so that the difference in meaning cannot be ascribed to a difference in conceptual content. Some examples are presented in Figure 2. Husband and wife have the same conceptual base, comprising a male (M) and a female (F) united in a particular way (represented by a horizontal line). Since the terms share this conceptual content, the semantic contrast between them must reside in their choice of profile within it.³ In Figure 2(b), the dashed arrow represents a mental relationship in which a sentient creature bears a particular positive attitude toward another entity.⁴ The

M b

husband

Base

a.

F

M

wife F

admire

Base tr

Figure 2.

M

F admirer

lm

23

24

Ronald W. Langacker

verb admire profiles the relationship itself (and since the conception of a relationship presupposes and incorporates the conception of its participants, these are included in the profile). By contrast, the noun admirer profiles only the participant who maintains the attitude in question. The two expressions have the same overall conceptual content: admire makes intrinsic conceptual reference to the individual holding the attitude, and admirer necessarily evokes the relationship coded by the verb. They are non-synonymous by virtue of the alternate profiles they impose on the same conceptual base. Considerations like these provide semantic evidence for adopting profiling as a descriptive construct. What about psychological evidence? It seems apparent that profiling is a focusing of attention within a conceived situation. In non-linguistic cognition, it has both perceptual and more abstract mental analogs: looking at one particular person in a group; focusing on a particular lineman making a block (amidst the welter of activity constituting a play in football); or — in the case of linguists — thinking of one particular node in a syntactic tree structure. The focusing of attention is such a basic psychological phenomenon that it is hard to imagine it not having a basic linguistic manifestation. While there may be many kinds and levels of attention, it is clear at the very least that profiling is reasonably considered one of them. So just for purposes of semantic description — independently of grammar — one could justify the construct of profiling and see it as the linguistic application of a broader, more fundamental cognitive phenomenon. Grammar provides a further check, a third source of evidence. It turns out that, once expressions are properly described in purely semantic terms, profiling is just what is needed for the characterization of various grammatical notions. Let us briefly consider four applications. Each deserves extended discussion, but here I can only offer succinct illustration. First, an expression’s grammatical class is determined by the nature of its profile (not by its overall conceptual content). Expressions can profile either things or relationships, these terms being defined quite abstractly.⁵ A noun profiles a thing. A verb profiles a process, defined as a relationship whose evolution through time is rendered salient. Classes like adjectives and prepositions involve the profiling of non-processual relations. With respect to Figure 2(b), for example, admire is a verb because it profiles a relationship portrayed as extending or unfolding through time, whereas admirer is a noun because it profiles a thing; admirer evokes the process of admiring as its conceptual base but profiles only the sentient participant. By the same token, a relationship is crucial to the semantic content of aunt, which is nonetheless a noun because its profile (or conceptual referent) is a thing (Figure 1(b)). Second, the imposition of a particular profile constitutes the essential meaning of certain grammatical markers. For instance, the effect of the derivational suffix -er (in its prototypical sense) is to profile a thing characterized by its active role in the process designated by the verb stem it combines with. Admirer thus designates the active participant in the process admire. The meaning of -er resides precisely in the imposition of this profile (and the concomitant change of grammatical class).

Cognitive Grammar and double subject constructions

Third, a head (at a given level of structural organization) can be characterized as the element whose profile is inherited by the structure as a whole (at that level). By this definition, -er is the head within the V+er construction, since its profile (rather than that of the verb) is inherited as that of the expression as a whole, making admirer a noun rather than a verb. Likewise, a preposition is the head in a prepositional phrase construction, since the phrase as a whole profiles the non-processual relation designated by the preposition, rather than the thing profiled by its nominal object. The conceptual referent of near Chicago, for example, is a relationship of spatial proximity, not the city of Chicago. Fourth, a subordinate clause is one whose profile is overridden at a higher level of grammatical organization. Here I will merely observe that (1)a — taken as a whole — profiles the process of admiring, not that of pole-vaulting, while (1)b profiles an act of deciding (not buying) at the highest level of organization. (1) a. I admire anyone [who can pole-vault 16 feet]. b. She decided [to buy a new car]. The examples in (2) further illustrate how the constructs profile and base prove useful for both semantic and grammatical description. They make it possible to specify, in an explicit and principled manner, the precise nature of the similarities and differences among certain closely related but non-equivalent expressions. The diagrams in Figure 3 correspond respectively to the boldface elements in (2), i.e. the verb melt, the participle melted, and the adjective liquid. (2) a. b. c. d. e.

I melted it. It melted easily. It melted in the heat. It is finally melted. It is now liquid.

The verb melt in (2)a profiles a force-dynamic interaction (Talmy 1988) resulting in a change of state. One participant (represented by a small circle) functions as a source of energy or influence (double arrow) directed at a second participant (large circle), which thereby undergoes a transition (single arrow) into the state (given as a box) of a. tr c.

L tr

L

b.

L

lm d.

tr L tr Figure 3.

e.

L tr

25

26

Ronald W. Langacker

being liquid (L). The transitive melt profiles this entire chain of events. By contrast, the intransitive melt in (2)b–c profiles only the second participant’s change and resulting state. The force responsible for this change, as well as its source, are evoked with varying degrees of salience as unprofiled facets of the expression’s conceptual base. In (2)b, the adverb easily evokes the agent’s role by specifying the degree of difficulty the agent encounters. In (2)c there is no agent, although the heat is readily construed as a causing influence in addition to its portrayal as an abstract location. At the extreme, the conception of external influence may be entirely absent. As a stative-adjectival participle, melted in (2)d does not profile any change, but merely the relationship of its single participant exhibiting the property of being liquid. However, it does evoke the change-of-state process as part of its base: something can only be described as melted by virtue of its having undergone the process of melting. Grammatically, this type of participle qualifies as an adjective because it conforms to the schematic conceptual characterization of this class (namely, it profiles a non-processual relation whose single profiled participant is a thing). The same is true of the adjectival sense of liquid, in (2)e. The contrast between melted and liquid is that the latter does not necessarily evoke a change-of-state process. Something can perfectly well be described as liquid even if it has never melted (e.g. it might result from the liquefaction of a gas).

1.3. Trajector and landmark To further illustrate the strategy of converging evidence, we can best start with a contrasting pair of relational expressions, e.g. above and below, sketched in Figure 4. Above profiles a non-processual relationship involving two participants in the domain of space organized in terms of horizontal and vertical dimensions. Given as a double-headed arrow, the profiled relationship specifies that one participant occupies a higher position along the vertical axis than does the other. Note, however, that exactly the same is true for below, which evokes the same conceptual elements and profiles the same relationship (referentially, an above relationship is also a below relationship). The semantic contrast does not reside in either the conceptual content evoked (the base) or the profiling imposed on it. We therefore need some other a. above

b. below tr

lm

lm

tr Figure 4.

Cognitive Grammar and double subject constructions

descriptive construct, just to achieve a minimally adequate semantic characterization — one that successfully distinguishes the meanings of above and below (and likewise for many other such pairs). It seems evident that the contrast has to do with the relational participants. When a relationship is profiled, its participants are accorded varying degrees of prominence. There is usually a primary focal participant, termed the trajector (tr), and often a secondary focal participant, called a landmark (lm). The trajector is the entity the expression seeks to locate, track, assess, or characterize, and a landmark (when present) is another salient entity evoked for this purpose. In the case of above, the trajector is located in the vertical dimension with respect to a landmark found at a lower point along this axis, whereas below reverses these specifications and situates its trajector with respect to a landmark found at a higher point. Trajector/landmark alignment was also shown in previous diagrams. In Figure 2(b), the sentient individual functions as a trajector characterized as having a mental attitude toward some other entity, the landmark. In Figure 3(a), the agent or energy source functions as trajector, since the transitive melt tracks its activity, in particular the effect it has on the patient, which is therefore put in focus as a landmark. Observe that if there is only one participant, it must have primary focal prominence, making it a trajector; the secondary focal prominence of a landmark is only possible relative to another element with primary focus. Consequently, since there is only one profiled participant in Diagrams (3)b–e, it is necessarily the trajector even when construed as a patient. What about non-linguistic manifestations of this ability to variably impose focal prominence in conceptualizing a relationship? I suggest that the relative prominence accorded to relational participants is very much akin to the figure/ground organization observed in Gestalt psychology, with alternate choices of trajector being analogous to figure/ground reversal. But can we speak of two degrees of figure status, a “primary figure” and a “secondary figure”, corresponding to the constructs trajector and landmark? I believe so. We certainly do have the ability — independently of language — to focus selectively on one participant in an interaction. If I watch a boxing match, for example, I can focus on one particular boxer, observing his techniques, which hand he favors, his footwork, even the number of blows he absorbs (this is analogous to a passive construal). Moreover, as I do so the other boxer naturally becomes a kind of secondary focus: I am following boxer1 mostly in relation to his interaction with boxer2, and am only in a peripheral way concerned with his corner, the referee, or the audience. We thus have good reason to think that the kind of cognitive ability being appealed to does in fact exist. I therefore posit trajector/landmark alignment for purposes of viable semantic description, and would argue for its psychological validity. Does it also have grammatical import? I claim that a subject, at a given level of grammatical organization, is properly characterized as a nominal element which specifies the trajector of the relationship profiled at that level. Likewise, an object is a nominal element which specifies the landmark at a given level. I have argued extensively that this concep-

27

28

Ronald W. Langacker

tual characterization is both viable and well supported (see Langacker 1991: 7.3, 1999c, 2001). As a first step in making it more precise, we need to consider the nature of grammatical constructions. A construction is a symbolic assembly comprising a set of component symbolic structures and a composite symbolic structure, linked by correspondences. An initial (simplified) example is given in Figure 5, representing constructions at two levels of organization. At the lower level of organization, the component structures admires and Bill are integrated to form the composite structure admires Bill. Observe that the nominal expression Bill profiles a thing (B stands for its various semantic specifications, e.g. that it refers to a male human), which corresponds to the schematic thing functioning as the landmark of admires (correspondences are indicated by dotted lines). By merging the specifications of the corresponding elements, we obtain the composite structure admires Bill. At this level admires is the head, or profile determinant (marked by a heavy-line box), because its profile — not that of Bill — is inherited as the profile of the composite expression. In the second construction, at a higher level of organization, the component structures are Alice and admires Bill, the profile of the former corresponding to the schematic trajector of the latter. The profile determinant at this level is admires Bill, since the entire clause designates the process of admiring rather than Alice. Observe that a kind of constituency emerges when the composite structure at one level of organization happens to be invoked as a component structure at a higher level. Constituency is not, however, either fundamental or essential in this framework (Langacker 1997). In particular, it is not used for the characterization of grammatical dependencies, such as the subject and object relations. Instead, a nominal Alice admires Bill A

Alice

tr

B

lm

tr

A

tr

B

lm admires Figure 5.

lm

admires Bill

B Bill

Cognitive Grammar and double subject constructions

expression qualifies as a subject if its profile (a thing) corresponds to the trajector of a profiled relationship, and as an object if its profile corresponds to a landmark. With respect to Figure 5, therefore, Alice is the subject of admires, admires Bill, and Alice admires Bill, as seen by tracing both “horizontal” and “vertical” correspondences. Bill is likewise the object at all three levels. Part of the conceptual unification achieved in CG is that grammatical patterns (or “rules”) have the same basic nature as the expressions which instantiate them. Like expressions, they consist in assemblies of symbolic structures, except that some or all of their elements are schematic rather than specific. Grammatical patterns can thus be described as constructional schemas, which serve as templates governing the formation of novel expressions. An expression like Alice admires Bill, as seen in Figure 5, instantiates a number of constructional schemas pertaining to different levels and facets of linguistic structure.

1.4. Reference points We come now to the final example of using converging evidence to justify a descriptive construct. It involves our basic cognitive ability to invoke the conception of one entity as a conceptual reference point (R) to establish “mental contact” with another entity (i.e. to single it out for individual conscious awareness). An entity accessed via a reference point is called a target (T). A reference point’s dominion (D) is the set of potential targets to which it gives access. The diagrammatic representation in Figure 6 will prove convenient. This general cognitive ability has both perceptual and more abstract manifestations. It provides the coherence of sequences like the following: (3) a.

Do you see that large boulder about half-way up the mountain? There’s a climber just above it. b. Do you remember the lawyer who got disbarred? His daughter is in my ethics class.

In each case the first sentence evokes a reference point (the boulder or the lawyer), which the second sentence uses as a basis for directing attention to the intended target (the climber or the daughter).

R D C Figure 6.

T

29

30

Ronald W. Langacker

Semantically, a reference point relationship is my candidate for a viable description of possessive constructions (Langacker 1993, 1995a; Taylor 1996). It is sufficiently abstract to account for the wide range of specific possessive uses, such as those in (4)a. At the same time, owing to its inherent directionality (the target being accessed via the reference point), it accounts for the usual irreversibility of possessive expressions, as seen in (4)b. (4) a.

the doctor’s wallet; the dog’s tail; Bill’s uncle; the cat’s fleas; her anxiety; the teacher’s situation; our bus; your candidate; Lincoln’s assassination b. *the wallet’s doctor; *the tail’s dog; *the situation’s teacher; *the assassination’s Lincoln

Grammatically, I take a reference point relationship as providing the abstract conceptual content of the possessive element ’s. Figure 7 illustrates its use in a basic possessive construction. Following Taylor, I assume that ’s profiles the target (although this is not essential). At the first level of organization, Bill elaborates the schematic reference point of ’s, producing Bill’s. The schematic target is then elaborated by knife, which functions as head (profile determinant), to form the noun phrase Bill’s knife. Bill’s knife R

T K

B D

Bill’s

B

T

R

K D

R

T

B D Bill

’s Figure 7.

knife

Cognitive Grammar and double subject constructions

T R

C Figure 8.

I believe our reference point ability has numerous linguistic manifestations pertaining to different levels and dimensions of organization. Within a noun phrase, a possessive relation holds between two things. At the discourse level, I consider a topic to be a reference point whose target is a proposition, as sketched in Figure 8.⁶ There are topics at various levels of discourse organization. In (50)a–c, we observe a topic in force over the span of a “paragraph”, a complex sentence, and a single clause. As the stretch of discourse over which a topic holds sway becomes smaller, its degree of structural integration with the proposition becomes tighter. In (5)a, the topic is introduced in a separate sentence, and is not explicitly mentioned in every proposition brought into its dominion. The topic in (5)b is internal to the sentence but set off as a separate intonational unit. In (5)c, where the sentence in question has just one clause, the topic is still clause-external both structurally and intonationally, yet normally the clause contains a pronoun referring back to it. Finally, in (5)d that wedding is clause-internal, functioning simultaneously as discourse topic and clausal object. (5) a.

Let’s talk about the wedding. June would be a good month. I don’t want to invite too many guests. Do you think we should have it outdoors? We really ought to have it recorded on video tape b. That wedding, I’m really looking forward to it, but I don’t want to buy a tuxedo, since I would never use it again. c. That wedding, I’m really looking forward to it. d. That wedding I’m really looking forward to.

More controversially, I suggest that subject and object are also reference point notions. This is a further elucidation of their characterization as nominal expressions that respectively correspond to the trajector and the landmark of a profiled relationship, i.e. the primary and secondary focal participants. To see this, observe that a relationship is conceptually dependent, in the sense that its own conception presupposes and incorporates the conception of its central participants. To the extent that this is so, the participants provide mental access to a relational conception, and as such are reference points with respect to it. A certain amount of evidence suggests that the primary and secondary focal prominence of trajector and land-

3

32

Ronald W. Langacker

a.

b. T

T R

R

C

C Figure 9.

mark reside in their being the first and second reference points accessed in building up to the full conception of a relationship (see Langacker 1999c, 2001). Figure 9(a) depicts the reference point relation between a participant and a relationship mentally accessed through it. The latter is given by a solid arrow, the box indicating (as in Figure 8) that the relationship as a whole — including its focal participant — functions as target. It is by virtue of “anchoring” the conception of the profiled relationship that this participant constitutes its trajector. The dominion (D) comprises the innumerable relationships that this reference point could potentially anchor, the target being the one that happens to be evoked and profiled. Let me offer just one bit of evidence for this characterization, out of many. We need to explain the cross-linguistic prevalence of possessive constructions being used to specify the central participants of nominalized verbs, e.g. Lincoln’s assassination [by Booth] or Booth’s assassination [of Lincoln]. This common periphrastic device is actually predicted given the conceptual characterizations independently proposed for possessives, for trajector/landmark alignment, and for nominalization. This latter involves conceptual reification, such that a relationship is reconceptualized as an abstract thing. The effect of nominalization is shown in Figure 9, where the box in 9(a) becomes an ellipse (representing a thing) in 9(b). If — as claimed — an event’s participants function as reference points with respect to it, and nominalization creates an abstract thing, then the relation a participant bears to the reified event constitutes a reference point relation between two things. That is precisely the definition proposed for possessives (Figure 6). How does a subject relate to a topic? Both are characterized as reference point relationships between a thing and a proposition (i.e. a profiled process). The difference is seen by comparing Figures 8 and 9(a). In the case of a topic, the reference point relationship is extrinsic to the clausal conception, a matter of discourse contingency. It is possible to conceptualize the profiled relationship in the absence of a reference point relation at the discourse level. In the case of a subject, on the other hand, the reference point relationship is intrinsic, i.e. inherent in the very conception of the clausal proposition. This follows from the conceptual dependence of a relationship on its focal participants. Although I will not pursue the matter, I believe these characterizations give us a good handle on how topic constructions might

Cognitive Grammar and double subject constructions

evolve historically into subject constructions, and why it is sometimes hard to distinguish them.

2 Double subject constructions The intimate association of topic and subject is especially evident in double subject constructions (Langacker 1999a). I will first present some typical data, to be followed by a description and then some discussion.

2.1. Data Double subject constructions are not at all uncommon cross-linguistically. Some cases I am aware of are exemplified in (6)–(15).⁷ (6) Taroo-ga hana-ga hikui. Taro-SUBJ nose-SUBJ low ‘Taro has a flat nose.’

[Japanese]

(7) Nihonzin-ga kome-ga syusyoku-da. Japanese-SUBJ rice-SUBJ staple.food-be ‘The Japanese have rice as their staple food.’ (8) Na pay aphu-ta. I stomach ache-ASSR ‘My stomach aches.’

[Korean]

(9) Na ku salam coh-ta. I the man likable-ASSR ‘I like the man.’ (10) Tā dùzi è. he stomach hungry ‘He is hungry.’

[Mandarin]

(11) Tā tóu téng. he head painful ‘He has a headache.’ (12) Wa khicaa-yaake bhugin chama du. the dog-COM fly one exist ‘There’s a fly on the dog.’ (13) Ji-ta wa baanlaa. I-DAT she beautiful ‘I think she’s beautiful.’

[Newari]

33

34

Ronald W. Langacker

(14) Noo=n no-puush konoknish. I=1s:PRES my-eye green ‘I have green eyes.’

[Luiseño]

(15) Noo=up no-te’ tiiwu-q. I=3s:PRES my-stomach hurt-PRES ‘I have a stomach ache.’ The basic phenomenon is the existence of sentences having roughly the form in (16). They are headed by two NP’s, each of which has some claim to being a subject. NP2 and the predicate appear to form a clause by themselves, yet the entire sequence also functions as a single clause.⁸ NP1 appears to be a topic with respect to the inner clause. It is often a possessor with respect to NP2 or some other element of the inner clause. Body-part relationships between NP1 and NP2 are the most frequent, evidently the prototype for this kind of construction, which can however be extended to many other sorts of situations. It is common for NP1 to be an experiencer with respect to the inner clause. Finally, there is a strong tendency for these sentences to be stative. Often they translate naturally with have even though they contain no such verb. (16) [ NP1 [NP2 PREDICATE]] ________________ inner clause _____________________ full clause

2.2. Analysis Though problematic from certain theoretical perspectives, this type of construction is quite straightforwardly described in CG. Its description requires nothing whatever that is not independently established, and most points have already been presented. Just three additional observations are needed, all independent of double subject constructions: (i) Reference point relationships are often signalled by simple juxtaposition (not by any separate morphological element). (ii) A verb or a clause can profile a reference point relationship. (iii) An expression’s composite structure can profile an entity not profiled by either component structure. In regard to point (i), we can observe that certain kinds of relationships are so basic and conceptually non-elaborate that they are commonly signalled linguistically by indirect means, rather than by specific morphological material. Two such cases are identity and reference point relations. These are exemplified by her and president in (17)a, and by them and a package in (17)b, respectively. (17) a. They elected her president. b. We sent them a package.

Cognitive Grammar and double subject constructions

Marking by mere juxtaposition (which can be considered iconic) is especially common for nominal possessive constructions, as in (18): (18) jakare ruguai crocodile tail ‘the crocodile’s tail’

[Guaraní]

If English worked this way, so that we said Bill knife instead of Bill’s knife, it would have the structure shown in Figure 10 (cf. Figure 7). The reference point relation need not be strongly invoked by either of the juxtaposed component elements, considered separately (though it may be in particular cases). It is rather a function of the overall construction, evoked as conceptual base at the composite structure level. Bill corresponds to the reference point (making it the possessor), and knife to the target, which is profiled by the overall expression. Note that despite their different modes of assembly, the composite structures of Figures 7 and 10 are the same. Bill’s knife R

T K

B D

B

K

Bill

knife Figure 10.

Point (ii), that a reference point relation can be profiled, is illustrated by a possessive verb like have. While such a verb has many uses and may have more specific senses, the schematic value — that of profiling a reference point relationship — is what it arguably shares in all its uses. Examples like (19)b-e show that any notion like ownership or control is too specific to apply to all uses. At the extreme, the subject comes very close to merely serving as a point of reference for specifying the object’s location. (19) a. b. c. d. e.

She has three dogs. I have the spaghetti. [customer to waiter holding the spaghetti] The US has a high ratio of prisoners to total population. He has freckles. We have a lot of skunks around here.

35

36

Ronald W. Langacker

have tr

lm K

B D Figure 11.

In its most schematic value, a verb like have is diagrammed in Figure 11, where the dashed arrows represent the conceptualizer’s path of mental access. On occasion, even a clausal reference point relationship can be coded by mere juxtaposition, as exemplified in (20): (20) Inepo kari-ne. I house-FUT ‘I will have a house.’

[Yaqui]

In this Yaqui example, there is no overt verb, yet a clausal process is evoked and marked for future tense. Although the tense suffix appears on the possessed noun, it is the process of possessing it (a specific instantiation of a reference point relation) that occurs in the future. This expression also illustrates point (iii), namely that a composite structure sometimes profiles an entity not profiled by either component structure (i.e. some constructions are exocentric). In the case at hand, diagrammed in Figure 12, the component structures profile things, while the expression as a whole designates the possessive relationship.

inepo kari tr I R

D

lm K T

I

K

inepo

kari Figure 12.

Cognitive Grammar and double subject constructions

Further illustrating the marking of a reference point relation by simple juxtaposition is a regular topic construction, e.g. (21): (21) The lottery, I never have any luck. To keep things simple, I have taken an example where the comment clause makes no overt reference to the topic. The basic organization of such an expression is sketched in Figure 13. The two component structures respectively profile a thing (X) and a proposition (Y V). The topic relationship emerges only at the composite structure level, being signalled by the adjacency and temporal ordering of these two elements. My assumption is that the overall expression inherits its profile from the comment clause. X, Y V R X

T

tr

Y

D

X

Y

X

YV

tr

Figure 13.

Given the points just established, a double subject construction can be seen as resulting from a specific way of applying them. In particular, a double subject construction results when (i) a reference point relationship is signalled by mere juxtaposition; (ii) the target is a clause (rather than a noun phrase); and (iii) the reference point relationship — not profiled by either component structure — is nonetheless profiled at the composite structure level. The basic structure is diagrammed in Figure 14. At the lower level of organization, NP2 combines with a predicate to form the inner clause. At the higher level, NP1 and the inner clause are juxtaposed, producing a composite structure that profiles a reference point relation between them. Not profiled by either component structure, this reference point relation has a clausal proposition for its target. This is very much like a topic construction, as in Figure 13, except that the reference point relationship created at the composite structure level is profiled, resulting in a single-clause structure with a clausal complement. It also resembles the possessive construction in Figure 12, except that the target is a proposition rather than a thing. Like the possessive construction, it is exocentric, since the reference point

37

38

Ronald W. Langacker

Full clause tr

T

X R

NP1

lm Y

D

X

Y

tr

Inner clause

tr

X

Y

NP2

Predicate

Figure 14.

relationship it profiles is not inherited from either component structure. Given this profiling, it is automatic that the reference point itself (NP1) functions as trajector, hence subject. Recall that a trajector was characterized as the first reference point accessed in building up to the conception of a profiled relationship. It follows that when the reference point relationship is itself put in profile, the reference point should be its trajector.

2.3. Discussion Why does this type of construction exist? Indeed, why is it cross-linguistically prevalent? It is not an idiosyncrasy of one language, or even one linguistic area, so it must have some functional motivation. Its function, I suggest, resembles that of so-called “raising” constructions, which provide a means for naturally salient entities to occur in grammatically salient positions (notably clausal subject): (22) a. For the prime minister to resign is unlikely. b. The prime minister is unlikely to resign. I have argued (Langacker 1995b) that “raising” is a matter of metonymy in the choice of relational participants, metonymy being another reference point phenomenon. It enables us to evoke one entity by explicitly mentioning another that is cognitively salient or easier to encode. It is well known, of course, that certain kinds of entities make “better” subjects than others. Certainly humans outrank clauses in regard to “topicality” and “subjectworthiness”.

Cognitive Grammar and double subject constructions

I believe double subject constructions serve a similar function, by means of a different grammatical device, albeit one that still relies on a reference point relation. Many grammatical phenomena conspire to ensure that elements with intrinsic cognitive salience appear in positions that are structurally prominent. People in particular are cognitively salient and naturally taken as reference points. Structurally, a subject is highly prominent, whereas a possessor — as a modifier, inside a noun phrase — is not. The double subject examples reveal that NP1 is typically human, whereas NP2, which would otherwise be the subject, is less topical. Often it designates a body part, which in any case is most naturally accessed via the conception of the person it belongs to. The double subject construction straightens things out and lets us access NP2 via NP1, which as trajector of the profiled relationship is the point of access for the entire complex expression. It allows an intrinsically salient possessor (often the speaker) to appear in a maximally prominent position, as subject of a higher-level clause. Let me summarize by reviewing the various properties of a double subject construction. These properties are reflected in the structure proposed and directly related to its function. • Double subject sentences are normally stative. Active sentences are likely to have agentive, inherently topical subjects. The functional desideratum of superseding a non-topical subject is more likely to arise in stative expressions. • NP2 and the predicate can themselves constitute a clause. This is because the construction builds on a clause, adding another level of organization with topic-like function. • The entire structure also has clausal status. It is by building a higher-level clause, with a new trajector, that subject status is shifted to an inherently more topical participant. • NP1 and NP2 both have some claim to subjecthood. They are subjects (i.e. correspondents of profiled trajectors) at different levels of organization. • NP1 is a topic with respect to the inner clause. This is because the higher-level relationship grafted onto the inner clause is precisely a reference point relationship between NP1 and that clause. • NP1 is typically a possessor with respect to NP2. This makes NP1 a reference point for NP2, which in turn — as trajector — is a reference point with respect to the predicate. Hence this circumstance is optimal for purposes of NP1 being adopted as reference point with respect to the inner clause. • Body-part relationships between NP1 and NP2 are prototypical. Such relationships are prototypical for possessives, and also involve the requisite asymmetry in cognitive salience (a person being inherently more topical than a body part). • NP1 is commonly an experiencer with respect to the inner clause. This goes along with body-part possessive relationships, as well as the construction’s stativity. • Double subject sentences often translate naturally with have, despite the absence of a possessive verb. That is because they are reference point constructions, and have itself profiles a reference point relation.

39

40

Ronald W. Langacker

3 Comparison of CG and CS I will now turn to some general observations and a few comparisons of Cognitive Grammar with the Columbia School approach. It should come as no great shock that I subscribe to the former rather than the latter. They do however have a lot in common, and the points of difference — though substantial — are not necessarily so drastic as they might appear.

3.1. In defense of CG By way of defending my preference, let me start by emphasizing the extent to which the architecture, the constructs, and the specific descriptions of CG are principled in nature. In the course of the discussion I have pointed out the following: (i) CG posits only semantic structures, phonological structures, and symbolic structures (pairings of the other two). It is thus congruent with the basic semiological function of language, that of allowing meanings to be symbolized by sounds. (ii) The theory achieves conceptual unification and theoretical austerity. (iii) It relies solely on mental abilities and phenomena whose existence is independently demonstrable. (iv) Converging evidence is required for the adoption of basic descriptive constructs. (v) CG avoids the imposition of artificial boundaries that, despite being traditionally assumed, have always proved problematic in practice. Examples include lexicon vs. grammar, morphology vs. syntax, and semantics vs. pragmatics. In each case the distinction is seen as being a matter of degree. (vi) Independently justified constructs and phenomena come together to provide straightforward descriptions of constructions that are problematic from other perspectives. I have shown this in detail for the particular case of double subject constructions. (vii) Such descriptions are successful in explaining, in conceptual and functional terms, the existence and specific properties of grammatical constructions. In his excellent monograph on French lui and le, Huffman (1997) expends considerable effort in contrasting CS and CG. Though his criticisms of the latter are thoughtful and substantial, I naturally believe them to be a bit off the mark. A brief response may be useful as a way of correcting some misconceptions that tend to exaggerate the actual differences between the two frameworks. I will examine four issues, in each case starting with a quote from Huffman’s book. • Unified analyses. Alternatively, lui might represent a network of interconnecting values, each one related at least indirectly to the others, and with one perhaps being ‘prototypical’, but with no central factor directly uniting them all; this kind of relationship is often posited by analyses in Cognitive Grammar. However, if an analysis can demonstrate that there is a unifying factor on the basis of which the appearance of lui in all cases can be accounted for, then this analysis clearly presents the stronger case. Analyses that overlook such a unifying factor miss the point of what actually motivates speakers to use the form. [p. 17]

Cognitive Grammar and double subject constructions

This pertains to my network model of complex categories (Langacker 1987: ch. 10), as applied to the meanings of linguistic elements, where polysemy is taken as being quite common. Huffman implies that a polysemous account is incompatible with finding and stating a unifying factor. But there is no such incompatibility. A highly schematic meaning unifying all uses is always to be sought, and is posited in CG when one can be found. The network model of complex categories recognizes the axis of schematicity and instantiation, as well as that of prototypicality and central tendencies; it makes no a priori claims as to which axis will prove most important, in a particular case or in general. CS-type meanings are often adopted, and the specific meanings posited in the CS literature generally seem plausible to me and very much on the right track. At the most schematic level, I would probably accept Huffman’s characterization of lui in terms of mid-level control, which can easily be incorporated in a CG analysis. Two critical issues remain, however: (i) Is a unifying schematic meaning exhaustive of what one has to say to adequately describe a particular element? (ii) To what extent does such a meaning actually “account for” the range of uses? We will return to these matters. • Distribution. The problem on which Columbia school focuses attention — occurrences of morphemes — is not directly addressed by their work [p. 22] This is in reference to Wierzbicka and myself, inter alia. In regard to Wierzbicka (e.g. 1988, 1996), I wonder how it squares with Huffman’s comment on the very next page that her goal is “to specify the precise set of semantic formulae which can generate no more and no less than the entire range of dative sentences acceptable in a given language”. This is representative of Wierzbicka’s strategy in general, and she takes it very seriously.⁹ As for myself, the whole point of positing complex categories (modeled as networks) is to handle the problem of distribution. There is elaborate discussion in my work (e.g. 1988, 1999b: ch. 4) of the usage-based conception, why it is necessary, and precisely how distribution and judgments of well-/ill-formedness arise. It might be argued that a simpler account is possible than Wierzbicka’s or mine (i.e. that a single schematic meaning will do the job), but it is unfair to say that our work does not directly address occurrences of morphemes. • Advance knowledge about cognition. Cognitive Grammar sets out to apply to language a body of preestablished knowledge about cognition . . . A great deal must be known in advance about cognition for grammatical analysis to be possible; so much, in fact, that the writing of a complete cognitive grammar is presently unfeasible . . . A Columbia-school analysis starts with much less: morphemes and their observed occurrences . . . Columbia-school linguists thus allow the feasibility of writing complete grammars, and in fact attempt to do so. [pp. 329–331] This gives a very inaccurate impression. Most of what I think I know about cognition was inspired by linguistic analysis, and the same is true of Cognitive linguistics

4

42

Ronald W. Langacker

generally.¹⁰ Cognitive linguistic research involves an intricate dialectic, using linguistic facts to suggest the need for particular descriptive constructs, and using independent knowledge about cognition as a check on the plausibility of what is posited. The test is whether a consistent, revelatory overall picture eventually emerges, one which is adequate and supported from both the linguistic and the psychological standpoints. As for the writing of complete grammars, CS linguists have the advantage because, as a matter of theoretical principle, they eliminate from what they call “language” the vast majority of what Cognitive linguists feel obliged to study. It is thus a wholly erroneous comparison. • Relation to traditional grammar. In spite of the promise of something new and different offered by Cognitive Grammar, its most fundamental categories are identical with those of the traditional canon . . . In Columbia-school theory, no category of grammar is ‘beyond dispute’. On the contrary, every category must withstand rigorous testing and earn its way into the grammar. [pp. 331–332] Adopted in CG are certain traditional notions like subject, object, noun, verb, possessive, count vs. mass, and subordinate clause. I did not retain such notions by virtue of conservatism, but rather because — from decades of work with diverse languages — I see the need to employ them in comprehensive linguistic descriptions. Morever, they are not just taken over from tradition. Rather, I have redefined them in what I believe are more viable terms (e.g. what I call a “noun” is far more inclusive than in traditional usage) and in each case offered a highly schematic conceptual characterization that I believe is compatible with all their manifestations. Is this not in the CS spirit? Indeed, my characterization of trajector (in terms of which subjects are defined) as “primary focal participant” is very much akin to the Diverian “central focus”, and not at all comparable to traditional definitions. If only because they could not understand it, most linguists recognized that CG was quite new and different when first proposed.¹¹ I do admit to believing that a few basic notions from the previous history of linguistics might prove potentially useful and thus worthy of retention (given proper clarification and characterization). So if it seems important I will gladly cede to CS (apparently not guilty of that sin) the privilege of being considered the most “radical”.

3.2. Some similarities and non-differences A number of similarities between CS and CG are evident enough that I will merely list them without discussion: (i) There is a common desire to describe language — and to describe each individual language — in its own terms, rather than imposing on it an elaborate set of preconceptions. (ii) They are both functionalist approaches concerned with the communicative role of language and how this function is achieved. (iii) Differences in form are not taken as arbitrary, but as having some communicative import that needs to be discovered and characterized. (iv)

Cognitive Grammar and double subject constructions

In contrast to some functionalist approaches, they are both interested in cognitive issues and actually attempt to characterize meanings. (v) In both cases, in fact, meaning is seen as the key to understanding grammar.¹² (vi) Both recognize the need to posit meanings that are highly abstract (or schematic). At a basic level, therefore, CS has much in common with CG (and with Cognitive linguistics generally). Even certain apparent contrasts turn out to be only apparent. Rather than differences in principle, they are best described as matters of emphasis or historical accident (in regard to which problems investigators have gotten around to tackling). My overall impression is that the true differences are rather few, and perhaps less vital than they seem. One apparent contrast which proves to be just a matter of degree is the CS emphasis on empirical evidence of a quantitative sort, notably from text counts. Cognitive linguists can certainly profit here from the methods developed by Reid (1991), Huffman (1997), and others. Quantitative methods, especially in the form of experiments and corpus searches, are however becoming increasingly common in Cognitive linguistics (as well as in the discipline at large). But regardless of their current prevalence, there is no disagreement about their importance. CS places considerable emphasis on systems of oppositions. This is admittedly a facet of linguistic organization that Cognitive linguistics and CG have not put much stress on. That it is a factor — that often an element’s value derives at least in part from its opposition to others — is something that is readily accepted.¹³ Systems of oppositions obviously have a far more central role in the CS perspective. Still, Cognitive linguistics should have little trouble accepting and accommodating them to whatever extent proves justified. A final point of non-disagreement is the distinction between “meaning” and “message” (Kirsner and Thompson 1976), and the role of inferencing (“communicative strategies”) in bridging the gap. On this point I am often misinterpreted and wrongly portrayed. For instance, Levinson (1997: 18–19) lumps me with Jackendoff, both of us being “stubborn ostriches” who “refuse to countenance the existence of pragmatics”. I would not presume to speak for Jackendoff, but in my case the characterization is erroneous if not irresponsible. What I actually said is the following: “The distinction between semantics and pragmatics (or between linguistic and extra-linguistic knowledge) is largely artifactual, and the only viable conception of linguistic semantics is one that avoids such false dichotomies” (Langacker 1987: 154). While these words do, unfortunately, leave room for alternate interpretations, Levinson’s confident and unqualified extrapolation is nonetheless quite gratuitous. My claim is that semantics and pragmatics form a gradation, there being no particular point where one can draw a sharp and rigid line of demarcation. What is being denied, then, is the situation depicted in Figure 15(a), where semantics and pragmatics are discrete, non-overlapping “components” with a specific boundary between them. That denial is compatible with the alternatives shown in 15(b)–(d). For his own polemic purposes, Levinson portrays me as subscribing to either of two

43

44

Ronald W. Langacker

a. b. c.

Semantics Semantics

Pragmatics Pragmatics

Gradation Non-existence

Semantics

d.

Separate components

Semantics/Pragmatics

Non-differentiation

Figure 15.

untenable positions: that pragmatics does not exist at all (15(c)), or that the entire spectrum of semantic/pragmatic phenomena is undifferentiated, with no distinction at all between the two sorts (15(d)). My actual position, however, is 15(b). Semantics and pragmatics grade into one another, so that any specific line of demarcation would be arbitrary. Phenomena drawn from opposite ends of the spectrum will of course be quite different in nature, and those lying toward either pole can for all practical purposes be considered exclusively semantic or exclusively pragmatic. Let me state quite explicitly, then, that what we understand from an expression — on the basis of context, implicature, or interpretive strategies — is often (if not always) more extensive than anything we could reasonably identify with its linguistic meaning, even in a broad sense. That is a definite point of agreement with CS, and for that matter with just about everybody else. As I understand it, however, CS practitioners would opt for Figure 15(a) instead of 15(b), with the semantics box being quite small in relation to the pragmatics box.

3.3. Divergence The place where disagreement sets in is when we ask how much of our total understanding of expressions — how much of the overall message — constitutes linguistic meaning. Here CS opts for a minimalist view of meaning. It is limited to what is derivable from the meanings of component elements, where these in turn are seen as quite schematic, the abstract characterizations valid for all uses of these elements. Any postulation of multiple senses (polysemy) is strongly resisted, and specific values which instantiate the abstract meanings are denied any status in the “language,” even if they are frequently and consistently used within a speech community. All aspects of an expression’s interpretation that go beyond the schematic meanings of its component elements are ascribed to the efforts of speakers employing particular “communicative strategies” as well as the full range of their interpretive abilities. By contrast, Cognitive linguists recognize polysemy as the usual state of affairs. Lexical and grammatical elements tend to have a variety of familiar and conventionally established values, or “senses,” often centered on a prototype. New senses

Cognitive Grammar and double subject constructions

arise as extensions from the prototype (or from other established senses) to the extent that specific interpretations — originally due to speakers’ interpretive efforts in context — recur with sufficient frequency that they become “entrenched” in the minds of individuals and conventional within a speech community.¹⁴ In the network model adopted in CG, extension to encompass a wider range of values engenders the extraction of a schema that embodies their abstract commonality. Often (but not necessarily always), a highly abstract schema emerges which all other senses instantiate as special cases. If we look beyond the labels to see what is actually being posited (or at least contemplated) as part of a comprehensive view of language and its use, the two models are not fundamentally incompatible. The abstract, monosemous meanings of CS are comparable to the highest level schemas posited in CG. Both frameworks recognize that, in actual use, expressions are understood in specific ways determined by context, communicative strategies, and interpretive abilities. Particular motivations for extension that are recognized in Cognitive linguistics, such as metaphor and metonymy, are presumably allowed in CS as possible strategies of interpretation. If it is further accepted in CS that particular interpretations are sometimes learned by speakers and established in a speech community, the two models can be seen as basically congruent (in principle, if not in practice). In comparing the two frameworks, therefore, a pivotal question is whether CS does in fact accept the learning and conventionalization of certain communicative strategies and interpretations. There does not appear to be unanimity in this regard. As I read Huffman (1997), he tacitly denies this possibility, describing even totally invariant applications of lui and le as being decided by speakers de novo on every occasion of their use. By contrast, Reid (1995: 149) describes the problematic, intermediate status of communicative strategies: [I]n García’s work a communicative strategy became the crucial mediating link between a meaning and the distribution of its signal. However the precise theoretical status of a communicative strategy is still an open question. Is a strategy a theoretical construct on a par with a meaning? It does not seem appropriate to regard strategies as a part of langue because they are not completely arbitrary. On the other hand to say they are part of parole makes them sound too idiosyncratic.

Reid has also described speakers as having decades-long experience in solving particular communicative problems, commonly resulting in packaged solutions. On this crucial issue, Contini-Morava (1995: 19) is quite explicit and fully in agreement with CG: A ‘strategy’ is a routinized exploitation of a given meaning, so that it is regularly used to suggest/infer a particular type of message. A postulation of such conventionalized patterns of inference is justified by appeal to the human preference for habit or routine . . .

In my view, there is no doubt at all about the matter. I fully concur with Janssen (1995: 266), who states it quite succinctly:

45

46

Ronald W. Langacker [T]he language user has highly detailed knowledge of when certain linguistic elements can be employed and when they cannot. And it is this very kind of knowledge that any language learner needs to have. The abstract description is obviously not enough to prevent him from making mistakes.

In other words, the set of maximally schematic meanings taken in CS as constituting the “linguistic system” come nowhere near being exhaustive of the conventionally established information required for a person to speak a language fluently and correctly. The function of the networks posited in CG is precisely to specify the requisite information. They specify the full range of conventional uses of elements, which are nothing other than packaged solutions to communicative problems. But is this really necessary? Is it not the case, as consistently maintained in the CS rhetoric, that a single schematic meaning accounts for all the uses of an element? That depends on what is meant by “account for”. What a CS analysis typically does — revealingly and persuasively — is to show how the abstract meaning posited can be reconciled with each attested use via some interpretive process. I regard this as a non-trivial accomplishment and a valid contribution. What a CS analysis does not, however, do (in my own experience and judgment) is to show how the specific, actual range of uses follows inexorably from the abstract characterization, so that this alone would exhaust the information a speaker needs in order to use the element correctly (i.e. in accordance with the conventional expectations of the speech community). Our interpretive capacities are extraordinarily flexible and multifaceted. Starting from any single meaning — especially one that is highly abstract — a speaker or a linguist can find some plausible rationale for using an element in an extremely wide variety of conceivable ways, only some of which are conventionally exploited. A substantial part of knowing how to speak a language is knowing which of these uses (or interpretive solutions) are in fact conventional, and which ones not.¹⁵ To be sure, speakers are always stretching the limits of convention, but to appreciate how and why they are doing this we need some idea of what those limits are. If it is granted that at least some communicative strategies and interpretive solutions are learned and conventionalized, the CS and CG positions on these issues are much less different than they first appear. Both acknowledge the entities and developments sketched in Figure 16. In 16(a), the larger box represents the linguistic system as it stands at a given time. One element of the system is an expression (say a lexical item) with a single basic meaning, A.¹⁶ These are also enclosed in boxes, indicating that they have been learned by speakers as conventional linguistic units. Further depicted in 16(a) is a novel usage in which the expression is understood — via some interpretive process — as having a distinct or more elaborate value, labeled B. The boxes surrounding B and the overall interpretive process are given with rounded corners, to indicate their novelty. Suppose, however, that this same expression — interpreted as having the contextual value B — proves efficacious for some purpose and is thus repeatedly used in this way. The interpretive process, as well as the extended value it produces, can then be learned by speakers and conventionally

Cognitive Grammar and double subject constructions

a. Basic meaning A expression

Interpretation

Contextual value B

Novel usage

Linguistic system b. Basic meaning A expression

Interpretation

Extended meaning B

Learned, conventional usage

Linguistic system (narrow definition) Linguistic system (broad definition) Figure 16.

established in the speech community. The boxes enclosing them are shown in 16(b) with square (rather than rounded) corners to mark their resultant status as learned, conventional units. Accepting this scenario, the pertinent difference between CS and CG comes down to the terminological issue of how we choose to use terms such as “language,” “linguistic system,” and “meaning.” CS adopts narrow definitions, thereby excluding from what it calls “language” both the interpretive process and the extended semantic value it creates, which would not be called a “meaning” but just a “message.” CG adopts broader definitions. A semantic, phonological, or symbolic unit is defined as part of the “language” or “linguistic system” to the extent that it is entrenched in the minds of speakers and conventional in a speech community. Once established, an extended semantic value is a “meaning,” no longer just a “message,” despite its pragmatic origin. The interpretation deriving this new meaning remains in the language as a representation of its motivation by the basic unit. It constitutes one link in the network describing the expression’s related senses. A difference in terminology — if it is merely terminological — has no substantive import. If all the phenomena depicted in Figure 16 are accepted as part of the

47

48

Ronald W. Langacker

total story of language and language use, no real importance attaches to what part of it we choose to call “language.”¹⁷ But perhaps additional considerations could be introduced that would motivate using the term in either a broad or a narrow sense. Let us examine a number of actual and potential arguments of this sort. A standard CS argument against the broad conception of a linguistic system, ascribing multiple (related) senses to a single element, is the difficulty in determining precisely which ones should be posited. For a certain range of uses, how many distinct senses are needed? Two? Three? Thirty-five? Are they really all discretely different? How can we be sure which particular sense is being employed on a particular occasion? We can best resolve these problems, the argument goes, by assuming that there is only one, highly abstract meaning which accounts for all the uses. This is not a valid argument. It is analogous to asserting that, since it is hard to count the stripes on a zebra, a zebra must have only one.¹⁸ There is to be sure a certain intellectual comfort in having a clear, definite answer to the question of how many senses to posit (the number one being especially attractive). We have no reason to believe, however, that language was designed with the comfort of the analyst in mind. Nor is it valid here to claim that positing one meaning is simpler than positing multiple senses. To the extent that speakers learn a conventionally sanctioned range of uses, the same complexity exists in the overall account of language and language use, regardless of how we label it. I would not deny that the network model faces real problems at both the descriptive and theoretical levels. Clear grounds for positing a specific number of distinct senses are not always evident. It seems quite likely, in fact, that the network metaphor is seriously misleading, precisely because it is too discrete. Instead, we might distort things less by comparing an element’s range of meanings to a mountain range, which occupies a continuous expanse but is very uneven owing to rises, depressions, peaks, and valleys. Counting the senses of a lexical item would then be analogous to counting the peaks in a mountain range: how many there are depends on how salient they have to be before we count them, and they appear discrete in the first place only because we ignore how they grade into one another at lower altitudes. The uncertainty we sometimes experience in determining which particular sense an expression instantiates on a given occasion is then to be expected. The uses in question are like points in the valley between two peaks. It is essentially an arbitrary matter whether we assign such points to one peak, to the other, to both, or to neither. Another CS argument against defining “language” broadly reduces to an oft-repeated slogan: “Where does one stop?” If we go beyond a single abstract meaning, allowing more specific senses into the “linguistic system,” how far down should we go? If we start accepting specific interpretations as linguistic meanings, do we not lose the vital distinction between meaning and message? Surely (the argument continues) we should not — as linguists — be expected to descend this slippery slope. How can we describe a language if we are forced to deal with the interpretive process in all its complexity, accommodating every conceivable interpretation in every conceivable context?

Cognitive Grammar and double subject constructions

Of course, by refusing to descend a slippery slope one does not succeed in making the landscape flat. The overall problem of language use and interpretation remains to be dealt with irrespective of how linguists write their job description. The more narrowly linguists define their own task (and CS scholars define it quite narrowly indeed), the less they contribute to the global enterprise of figuring out how language works.¹⁹ Still, it is not the case that by venturing down the slope we stumble into a bottomless pit. Even with a broad definition, there is a natural way of defining a “language” that does indeed offer a principled basis for stopping at a certain place. In CG, what counts as the “linguistic system” is delimited by the dual factors of psychological entrenchment and conventionalization within a speech community. A linguistic structure is reasonably considered part of a “language” to the extent that speakers learn it and can use it as a prepackaged unit (an established cognitive routine), and can count on other speakers knowing it as well. Because both entrenchment and conventionalization are matters of degree, we should not expect to find a sharp and rigid line of demarcation. There is rather a gradation, as in Figure 15(b), with indisputably linguistic structures and indisputably non-linguistic structures lying toward opposite ends of the spectrum, separated by a zone of intermediate cases. Under this definition, of course, the place one stops in describing a linguistic system is not a determinate point but an ever-fluctuating region reflecting the vicissitudes of ongoing language use. Nonetheless, the descriptive task does not go on forever — the defining factors impose both practical and theoretical limits. These factors are, moreover, subject to empirical verification. In principle, degree of entrenchment can be measured experimentally (e.g. in terms of response latency), and degree of conventionalization by corpus-based methods. In practice it may not prove easy to measure them, but neither is there any overwhelming need to do so. The point is that a linguistic system (however vast) is bounded (however fuzzily), and for a very good functional reason. Beyond a certain level, there is a trade-off between specificity and differentiation, on the one hand, and utility on the other: as structures become more and more specific and finely articulated, they have less and less likelihood of being used with enough frequency to be entrenched or conventionalized. A further CS argument is that specific senses would be different in nature from abstract meanings and thus inappropriate as part of the language per se. Recall, for instance, Reid’ s statement: “It does not seem appropriate to regard strategies as a part of langue because they are not completely arbitrary” (1995: 149). But a language comprises many kinds of elements whose nature is even more drastically distinct — sounds and meanings, for instance, or within phonology, a segment and a stress alternation. Who is authorized to say, on an a priori basis, that linguistic elements have to be homogeneous, or that a particular sort of difference is automatically disqualifying? With respect to Reid’s comment, can complete arbitrariness really be taken as criterial for inclusion in the linguistic system? If so there may be virtually nothing in the system. Even the form-meaning pairings of individual mor-

49

50

Ronald W. Langacker

phemes have some degree of non-arbitrariness. For example, the more schematic meanings of grammatical elements bear a natural affinity to their phonologically more attenuated symbolization (vis-à-vis lexical items). To support the privileged status of a system of highly abstract meanings, an obvious strategy is to look for independent evidence that these in particular have a special role in behavior and cognition. One might, for example, study language acquisition to see whether abstract meanings develop first, as a necessary basis for the later emergence of more specific senses. Or one might turn to psycholinguistic experimentation, to see what role schematic meanings play in language processing. At the very least there ought to be a coherent story of how processing, development, and structure fit together in a comprehensive account of language. I quite understand the hesitation of linguists, qua linguists, to get involved with the complex issues in these vast and difficult fields of inquiry. Even so, the apparent CS reluctance to engage them (if not the conscious decision to avoid them altogether) is rather striking. It precludes a possible source of external evidence that could support the framework and show its potential as a realistic model of linguistic knowledge.²⁰ Given their critical reliance on communicative strategies and “human factors” more generally, it is surprising that CS practitioners are not engaged in large scale efforts to study them systematically. Even if they lie outside of “language” (narrowly defined), these factors are being appealed to for bridging the very large gap between abstract “linguistic” meanings and the panoply of more specific senses that Cognitive linguists feel obliged to posit. I usually find plausible both the abstract meanings proposed in CS and the account of how human and communicative factors give rise to specific interpretations. These explanations sometimes give the impression, however, of being ad hoc and unconstrained. In the absence of rigorous, detailed investigation, how can one be sure that the factors appealed to will invariably function in just the required way, to produce the precise patterns of uses and interpretations found in the data? To validate their own ideas, CS scholars might consider the strategy of Cognitive linguists, who feel compelled to learn all they can about phenomena like metaphor, metonymy, mental spaces, blending, prototypes, and basic level categories. Even though these are not specifically or uniquely linguistic, they are crucial to understanding how particular interpretations arise, and thus to an overall account of language, wherever we draw the line.

3.4. Complex expressions If one purports to have a theory of language, and wants it to be taken seriously, something cogent has to be said about the related problems of grammar (i.e. morphology and syntax) and semantic composition. It is not enough to bemoan the inadequacies of traditional grammar and the sins of generative grammar, egregious though these may be. The task remains of describing the formation of complex expressions and how their meanings relate to the meanings of their parts. A conceivable position is that there is in fact nothing to describe at the level of langue, that

Cognitive Grammar and double subject constructions

a language comprises nothing other than individual elements and their abstract meanings. This may well be the CS position. If so, it nonetheless has to be shown in detail how grammatical patterns and composite meanings emerge extra-linguistically through interpretive processes, or alternatively, why virtually every other linguistic tradition is simply mistaken in believing that they exist. One aspect of grammar that CS does explicitly address is the abstract meanings of individual grammatical elements. I generally like the meanings proposed, and in the case of grammar I suspect that the most schematic values are often the most important. While I myself see a gradation, the elements linguists consider “grammatical” (as opposed to “lexical”) are precisely those whose meanings are more schematic. The semantic import of grammar resides primarily in the construal imposed on lexical content, and one dimension of construal consists of various kinds of prominence (e.g. profiling, trajector/landmark alignment). I would not disagree with the following description by Diver (1995: 103): . . . the discovery, in language after language, of a semantic substance that may be characterized in general as ‘Concentration of Attention . . .’ This communicative problem appears to be the single most important factor in determining the makeup of grammatical systems.

For a specific example of grammatical structure, let us focus on prenominal possessives in English, marked by ’s and its variants. The abstract meaning I ascribe to ’s consists in a reference point relation between two things, i.e. the dynamic process of directing attention first to a reference point and then to a target accessible through it (Figure 6). This schematic meaning is comparable (at least in degree of abstractness) to those adopted in CS. Moreover, it accounts in part for the morpheme’s range of uses. It is sufficiently abstract to accommodate the wide variety of possessive locutions, as in (4)a. Yet, owing to the inherent asymmetry of a reference point relationship, it accounts for the usual irreversibility of such expressions, as seen in (4)b. This too is in the CS spirit. But is that sufficient? Is this single abstract meaning all we need to say about ’s and the prenominal possessive construction in our description of English grammar? In the CG perspective, it is also necessary to include a range of more specific senses, reflecting certain kinds of uses common enough to become entrenched and conventionalized.²¹ Among these are prototypical instantiations of possessive relationships, notably ownership, kinship, and whole-part relations. Examples of more peripheral senses are those pertaining to modes of transport (e.g. our bus), emotional states (her anxiety), and the focal participants in a reified event (the army’s destruction [of the city], the city’s destruction [by the army]). As noted previously, how many senses we posit is largely dependent on how salient we require them to be, which in turn depends on our purpose. The important point is that a particular range of uses — out of all the conceivable ones — stand out as being familiar and conventional. As such they constitute packaged solutions to particular interpretive and communicative problems, one facet of knowing how to speak the lan-

5

52

Ronald W. Langacker

guage. Hence they are part of the “linguistic system,” under the broad definition (Figure 16). From the abstract meaning alone, one cannot predict exactly what the conventional range of uses might be (cf. Nikiforidou 1991). For instance, we can imagine a language just like English except that prenominal possessives could not be used for emotional states, or could only be used for the trajector of a reified event (not its landmark). But whether a single meaning is posited or a family of related meanings, we have not yet described the grammatical construction. The possessive morpheme’s meaning, in and of itself, will not tell us everything we need to know in order to use it correctly in complex expressions. For instance, it has to be specified that ’s occurs on the possessor noun phrase (e.g. Bill’s knife), and not on the possessed noun (*Bill knife’s), on both (*Bill’s knife’s), or on neither (*Bill knife). It has to be specified that the possessor nominal precedes rather than follows the possessed noun (*knife Bill’s). It has to be specified that, in the case of complex possessor nominals, ’s attaches phonologically to the last word of the nominal, which retains its contiguity (the King of France’s wig); it does not attach to the head noun (*the King’s of France wig), nor can a modifier be postposed (*the King’s wig of France). But even when the marker is borne by another element, e.g. France in the King of France’s wig, the possessor (or reference point) is still identified with the entity profiled by the head noun (and thus by the nominal as a whole). Suppose you knew nothing about English possessives other than the form and abstract meaning of the possessive marker. From this information, you could not predict all the details just listed, nor the countless other details required for conventional possessive use in the language. For this reason, a full description of English — even as langue — has to include an explicit characterization of the prenominal possessive construction. In CG this is done by means of constructional schemas, i.e. schematic symbolic assemblies which embody the abstract organizational commonality inherent over a range of complex expressions. Prenominal possessives, for instance, would be described by a symbolic assembly analogous to Figure 7, except for containing schematic nominal elements rather than specific ones.²² The one explicit discussion of this issue I have encountered in the CS literature actually amounts to a proscription of constructional schemas: . . . If combinations of signs can themselves be signs, then there is no principled way to put limits on the system and make it discrete . . . The limitless, infinitely recursive Saussurean sign, then, is the sponsor of maximalist linguistics . . . In both [Saussure and Langacker], the inventory of units is ill-defined, making it very easy to attribute to the conceptual inventory of the language what would be more reasonably confined to the ever-shifting conceptual set of communication. (Otheguy 1995: 232–233)

A symbolic structure is a sign, and an assembly of symbolic structures is a combination of signs. It is not, however, true that recognizing constructions and constructional schemas results in an infinite regress. It is wrong to claim that in a maxi-

Cognitive Grammar and double subject constructions

malist view there are no limits and no principled grounds for delimitation. The basis for delimitation is the same as for the senses of a lexical item: the joint factors of entrenchment and conventionalization. These are matters of degree, so there is no sharp cut-off point, but rather a gradation. But they do impose limits, since structures with zero entrenchment and conventionalization are totally “outside” the linguistic system.²³ It is true that the inventory of complex signs is nonetheless very large, so that actually writing a complete grammar is not a realistic task. This is not a valid argument against the maximalist view, however, any more than the impossibility of fully describing the brain (describing every neuron and every neural connection) is an argument that the brain does not exist. Large numbers of constructional schemas have to be recognized in a minimally adequate description of the conventional patterns speakers employ in putting together complex expressions. On grounds of theoretical preference, one can certainly make an a priori decision to exclude such patterns from the “linguistic system,” but only at the cost of omitting a large proportion of the conventional knowledge one needs to learn in order to speak a language. We cannot just attribute grammatical patterns “to the ever-shifting conceptual set of communication” and let it go at that — whatever their degree of conceptual and communicative motivation, the specific patterns conventionally employed in a language have to be learned by speakers and explicitly described by linguists. Nor does the fact that speakers often stray from these patterns, through negligence or creative innovation, constitute an argument against their existence. It is only by acknowledging and elucidating the patterns that we can recognize and characterize such behavior.²⁴ Grammatical patterns are further necessary for an account of semantic composition, which concerns the relation between the meaning of a complex expression and the meanings of its parts. I have not detected in CS any systematic attempt to deal with this fundamental linguistic problem, or even an explicit recognition that the problem exists. In practice, there often seems to be a tacit assumption that the meaning of a whole is merely the sum of the abstract meanings of its parts (i.e. the component lexical items), and that anything beyond this is attributable to contextual interpretation based on communicative strategies and other “human factors.” That is not however a viable assumption. If it were true, then a pair of sentences like (23)a-b would have the same linguistic meaning, since they consist of the same lexical items: (23) a. The big dog frightened a tiny cat. b. A big cat frightened the tiny dog. A complex expression’s meaning is not just a pile of semantic features, nor an unordered set of lexical meanings, but rather a structured conceptualization in which conceptual elements fit together in very specific ways. In (23)a, for instance, the property ‘big’ is ascribed to the dog, and the property ‘tiny’ to the cat — it is not the reverse, nor are these properties assigned to other entities evident in the discourse context.²⁵

53

54

Ronald W. Langacker

These specifications and others (e.g. the ascription of agentivity and of definiteness to the dog rather than the cat in (23)a) are totally determined by structural relationships enshrined in linguistic convention. They are not left for individual speakers to figure out de novo subject to the vicissitudes of an “ever-shifting conceptual set of communication”. Nor do they follow inexorably from human factors or functional principles. It might be suggested, for instance, that contiguity is such a strong factor in cognitive processing that big is necessarily attributed to dog in (23)a, and tiny to cat, as the invariable and automatic outcome of interpretive processes. This strategy is simply not adequate, however. There are languages which do not require contiguity between a head and its modifiers. More generally, cognitive, communicative, and functional considerations are notorious for their underdetermination of linguistic structure — that is why languages differ synchronically and diverge diachronically, even though such factors are largely universal. Without denying the strong and pervasive influence of functional motivation, the fact remains that language structure includes large inventories of learned, conventional grammatical constructions. In CG, these constructions take the form of symbolic assemblies which specify how component elements are integrated semantically and phonologically (see Figures 5, 7, 10, 12–14).²⁶ Although grammatical constructions are a central concern of CG, and are seemingly proscribed in CS, I would like to suggest that the contrast is not absolute. I believe in fact that the notion of a grammatical construction is already implicit in CS. It inheres in the recognition that certain abstract meanings are symbolized by position rather than by a separate morphological element. A case in point is central focus (or trajector) in a language which marks it by means of word order. In a sentence like Alice admires Bill, for instance, position before the predicate is what tells us that Alice, and not Bill, is the central focus. As it stands, however, this description is quite vague. It presupposes a number of very specific relationships that need to be made explicit in any full or rigorous characterization of the sign. Phonologically, what the sign consists of is the relative position in the speech stream of two other signs, namely Alice and admires Bill (or just admires — the choice is immaterial to the argument). If we state this in general terms (not just for the sentence in question), the sign consists of the relation between two other phonological structures, call them X and Y, such that X and Y are contiguous in the dimension of speech time (T) and X precedes Y. This is shown in Figure 17(a). Semantically, a central focus does not exist in isolation; we can only speak of a central focus with respect to the situation described by a verb or clausal predicate. Using CG terminology (for sake of ready comparison), we can call this situation a process and represent it with an arrow, as shown on the right in 17(b). The central focus — or trajector — is some participant in this process, represented as a circle. The predicate, however, does not itself identify this entity (except through its role in the process). Its identification is effected by the accompanying nominal expression, e.g. Alice in Alice admires Bill. In particular, the focused entity is the one that is designated — or in CG terms, profiled — by this nominal as a whole. This is

Cognitive Grammar and double subject constructions

shown on the left in 17(b). The dotted correspondence line makes it explicit that the predicate’s trajector and the nominal’s profile are equated. We also need to be explicit concerning the relation between the phonological structures in Figure 17(a) and the semantic structures in 17(b). Namely, the process whose central focus is at issue is specifically the one symbolized by the phonological structure given as Y, and the nominal whose referent functions as its central focus is specifically the one symbolized by X. These connections are shown in 17(c), which depicts the semantic and phonological relationships between two component signs, characterized as a nominal expression and a predicate. The overall configuration amounts to a complex sign consisting of a combination of signs — supposedly proscribed in CS. Yet there is nothing here that is not essential to making precise and explicit the notion that a concept like “central focus” can be symbolized by relative position in the speech stream. I suggest, then, that the proscription of signs consisting of other signs has to be abandoned even on grounds internal to CS, to avoid contradiction. This should not be regarded as problematic, since (pace Otheguy) it does not in fact lead to infinite recursion; complex signs need only be recognized as part of the linguistic system to the extent that they are entrenched and conventionalized. And if complex signs are allowed in CS in the guise of meanings marked by relative position, why not allow them for other purposes as well? In particular, they would permit CS to say something substantive about grammatical patterns and semantic composition (which are currently just ignored, if not denied). Accepting complex signs would lend CS greater credibility as a viable theory of language structure. This would also move CS in the direction of CG, because there is no difference in principle between a complex sign and an assembly of symbolic structures. It should be evident that a structure like Figure 17(c) is equivalent to a CG constructional schema describing a grammatical pattern. Using the representational format adopted in this chapter, 17(c) can be reformulated as shown in 17(d). a.

X

tr

b.

Y

T tr

c. X

tr

d.

XY

Y tr

T X Figure 17.

Y

55

56

Ronald W. Langacker

The two component signs in 17(c) appear as the component symbolic structures in 17(d). The composite structure, shown at the top in 17(d), simply makes explicit the result of combining the component signs. The arrow for speech time in 17(c) is omitted in 17(d) just for sake of diagrammatic convenience. The differences, in short, are merely notational.

4 Conclusion I hope this critique of CS has been fair and constructive. While I obviously disagree with CS doctrine on certain fundamental issues, there are many points on which we agree, and grounds for much broader agreement should CS practitioners be willing to reconsider some basic tenets adopted for reasons that might be viewed as suspect. Theory aside, however, I usually find the analyses proposed in CS to be both insightful and revelatory. Linguists of other persuasions will profit by carefully examining CS descriptions and the methodology used to support them. At the same time, CS would benefit from greater awareness of work in Cognitive linguistics dealing with problems crucial to an overall account of language. Even under a narrow definition of “language,” interpretive phenomena have to be carefully investigated and thoroughly understood, since these are being relied on to make up the difference. Only by studying such phenomena systematically (rather than just appealing to them) can one show a narrow construal of “language” to be principled and viable. The bottom line is that I perceive the differences between CS and CG as being less fundamental and less important than they might appear to be on superficial examination. This has the consequence — like it or not — that I consider CS to be part of Cognitive linguistics, and one place I can regularly look to for insight into language. I hope nobody objects too strongly.

Notes 1. See Langacker 1987, 1990, 1991, 1999b, and the references cited therein. 2. Analogously, the reduction expressed by the formula H2O fails to imply the non-existence of water. 3. We thus avoid the circularity of defining husband as ‘one who has a wife’, and wife as ‘one who has a husband’. 4. The relational participants are given as circles. The labels tr and lm stand for “trajector” and “landmark”, discussed in the following section. 5. In diagrams, I normally use circles to represent things, and various kinds of lines and arrows for relationships. 6. Expressed by a clause, the proposition is represented mnemonically by a circle and an arrow, to indicate a thing participating in a process. The box is simply meant to show that the proposition as a whole functions as target.

Cognitive Grammar and double subject constructions 7. Examples from the five languages represented are taken respectively from Kumashiro 2000; informal notes by Kee-Dong Lee; Teng 1974; Cook 1988 and Hung 1988; and Steele 1977. The following abbreviations are used: SUBJ = subject marker; ASSR = assertive; COM = comitative; DAT = dative; PRES = present; 1/3s = first/third person singular. In his very detailed description of this phenomenon in Japanese, Kumashiro provides strong evidence that there are two basic construction types, the type illustrated in (6) being reanalyzed as having a complex predicate with a single subject. For present purposes we can ignore this further development. 8. Evidence for these points depends on the language. In Japanese, for instance, both NP1 and NP2 can take the subject marker ga, at least under certain conditions. In Luiseño, the clitic agreeing with the subject appears as the first person singular =n in (14), agreeing with NP1, and as the third person singular =up in (15), agreeing with NP2. There is no intonation break, and since the clitic in Luiseño occurs in second position within a clause, its placement after NP1 indicates the single-clause nature of these examples. 9. It is suggestive that this highly eminent lexicographer should believe that long, elaborate definitions are needed (and also polysemy, sparingly used) if they are to do the job of properly predicting occurrences. 10. For instance, the theories of metaphor, mental spaces, and conceptual blending give a radically different view of cognition than anything preestablished or generally assumed in psychology. See Lakoff and Johnson 1980; Lakoff and Turner 1989; Fauconnier 1985, 1997; Fauconnier and Turner 1998. 11. It seems less radical today, since “mainstream” linguistics has steadily evolved in its direction. 12. Note the title Meaning as Explanation (Contini-Morava and Goldberg 1995). 13. It comes to the fore in my study (with Eugene Casad) of the u/a opposition in Cora locatives (Langacker 1990: ch. 2). Also, maximal opposition figures in the characterization of prototypes and basic level concepts (Rosch 1978). 14. A new sense can either deviate from the one that spawns it (implying some conflict in semantic specifications) or else elaborate it (by incorporating additional content or spelling it out in finer-grained detail). 15. Note that I am not demanding absolute predictability of an element’s range of uses. In fact, its non-predictability is precisely why I adopt the network model. I further believe that the method of ex post facto rationalization employed in CS descriptions is quite appropriate (notwithstanding the pretentious rhetoric of prediction standard among linguistic theorists). I am merely proposing what I consider a realistic assessment of what a single abstract meaning can do, even granting its correctness. 16. For present purposes this basic meaning can be taken as either a prototype (more likely in CG) or an abstract schema (as in CS). 17. I realize, of course, that proponents of every theory would like to impose their own definition of “language” in order to enhance the perceived importance of their own work. 18. Presumably the pattern of stripes we see on a zebra would not be a property of the zebra per se, but the result of using “visual strategies”. 19. This is not to deny the value of limited, focused contributions. The issue, rather, is whether there are sound arguments for the particular limitations imposed by CS doctrine.

57

58

Ronald W. Langacker 20. Of course, the evidence might point in the opposite direction. It seems clear, for example, that children first use words in fairly specific circumstances, and that the general direction of development is from concrete to abstract. In claiming that highly schematic meanings are the only ones which count as “language,” CS scholars will be more credible if they indicate how their view can be reconciled with the developmental facts. 21. It does not matter whether we talk about multiple senses of the morpheme or multiple variants of the grammatical construction in which it occurs. In CG these are two sides of the same coin (Langacker 1999b: ch. 4). 22. Naturally, Figure 7 is only illustrative, hence significantly oversimplified. Moreover, to fully describe a construction — and capture the kinds of details listed in the text — requires not just one but a family of related constructional schemas (analogous to the family of related senses of a lexical item). 23. I reiterate that these factors can in principle be scientifically investigated and measured (e.g. by text counts for conventionalization, by psychological experiments for entrenchment), and that the maximalist position is adopted (reluctantly) for the purpose of achieving adequate descriptions. Counter to Otheguy’ s statements, positing units is not done just on the basis of “felt stability” of a group (1995: 234) or because a group “struck the analyst as unitary” (232). 24. Beyond grammatical patterns, a vast number of specific complex expressions are learned and stored as units. Typically they incorporate a measure of idiosyncrasy that has to be recognized as part of their conventionally established value. If CS wishes to deny the existence of signs consisting of other signs, it needs to confront the massive amount of evidence to the contrary adduced in the construction grammar literature (e.g. Fillmore et al. 1988; Goldberg 1995; Michaelis and Lambrecht 1996). 25. I realize that word order is recognized in CS as a signalling device. However, it is posited for only a small fraction of the structural and semantic relationships that need to be accommodated (notably for what are traditionally called “subject” and “object” relations), and I see no way to make it work for the full range of cases. Moreover, positing a word-order sign does not per se make explicit the nature of the structured conceptualization emerging from the ordered elements. I suggest below that making this explicit (as it has to be) amounts to the adoption of grammatical constructions in the CG sense. 26. Cognitive linguists do not claim that semantics is fully compositional, only that patterns of composition (as embodied in constructional schemas) are one, crucial factor in “meaning construction”.

References Contini-Morava, E. 1995. “Introduction: On linguistic sign theory.” In Contini-Morava and Goldberg, 1–39. Contini-Morava, E., and Goldberg, B. S. (eds.). 1995. Meaning as Explanation: Advances in Linguistic Sign Theory. Berlin/New York: Mouton de Gruyter. Cook, K.W. 1988.“The semantics of Newari case-marking distinctions.” Linguistic Notes from La Jolla 14: 42–56. Diver, W. 1995. “Theory.” In Contini-Morava and Goldberg, 43–114. Fauconnier, G. 1985. Mental Spaces: Aspects of Meaning Construction in Natural Language. Cambridge, MA/London: MIT Press/Bradford.

Cognitive Grammar and double subject constructions Fauconnier, G. 1997. Mappings in Thought and Language. Cambridge: Cambridge University Press. Fauconnier, G., and Turner, M. 1998. “Conceptual integration networks.” Cognitive Science 22: 133–187. Fillmore, C. J., Kay, P., and O’Connor, M. C. 1988.“Regularity and idiomaticity in grammatical constructions: The case of let alone.” Language 64: 501–538. Goldberg, A. E. 1995. Constructions: A Construction Grammar Approach to Argument Structure. Chicago/London: University of Chicago Press. Huffman, A. 1997. The Categories of Grammar: French lui and le. Amsterdam/Philadelphia: John Benjamins. Hung, T. 1988. “Case and role in Newari: A Cognitive Grammar approach.” Linguistic Notes from La Jolla 14: 95–107. Janssen, T. 1995. “Deixis from a cognitive point of view.” In Contini-Morava and Goldberg, 245–270. Kirsner, R. S., and Thompson, S. A. 1976. “The role of pragmatic inference in semantics: A study of sensory verb complements in English.” Glossa 10: 200–240. Kumashiro, T. 2000. The Conceptual Basis of Grammar: A Cognitive Approach to Japanese Clause Structure. San Diego: University of California doctoral dissertation. Lakoff, G., and Johnson, M. 1980. Metaphors We Live By. Chicago/London: University of Chicago Press. Lakoff, G., and Turner, M. 1989. More Than Cool Reason: A Field Guide to Poetic Metaphor. Chicago/London: University of Chicago Press. Langacker, R.W. 1987. Foundations of Cognitive Grammar, vol. 1, Theoretical Prerequisites. Stanford: Stanford University Press. Langacker, R.W. 1988. “A usage-based model.” In Topics in Cognitive Linguistics, B. RudzkaOstyn (ed.), 127–161. Amsterdam/Philadelphia: John Benjamins. Langacker, R.W. 1990. Concept, Image, and Symbol: The Cognitive Basis of Grammar. Berlin/ New York: Mouton de Gruyter. Langacker, R.W. 1991. Foundations of Cognitive Grammar, vol. 2, Descriptive Application. Stanford: Stanford University Press. Langacker, R.W. 1993. “Reference-point constructions.” Cognitive Linguistics 4: 1–38. Langacker, R.W. 1995a. “Possession and possessive constructions.” In Language and the Cognitive Construal of the World, J. R. Taylor and R. E. MacLaury (eds.), 51–79. Berlin/New York: Mouton de Gruyter. Langacker, R.W. 1995b. “Raising and transparency.” Language 71: 1–62. Langacker, R.W. 1997. “Constituency, dependency, and conceptual grouping.” Cognitive Linguistics 8: 1–32. Langacker, R.W. 1999a. “Double-Subject Constructions.” In Linguistics in the Morning Calm 4, S.-Y. Bak (ed.), 83–104. Seoul: Hanshin. Langacker, R.W. 1999b. Grammar and Conceptualization. Berlin/New York: Mouton de Gruyter. Langacker, R.W. 1999c.“Assessing the cognitive linguistic enterprise.” In Cognitive Linguistics: Foundations, Scope, and Methodology, T. Janssen and G. Redeker (eds.), 13–59. Berlin/ New York: Mouton de Gruyter. Langacker, R.W. 2001. “Topic, Subject, and Possessor.” In A Cognitive Approach to the Verb: Morphological and Constructional Perspectives, H. G. Simonsen and R. T. Endresen (eds.), 11–48. Berlin/New York: Mouton de Gruyter.

59

60

Ronald W. Langacker Levinson, S. C. 1997. “From outer to inner space: Linguistic categories and non-linguistic thinking.” In Language and Conceptualization, J. Nuyts and E. Pederson (eds.), 13–45. Cambridge: Cambridge University Press. Michaelis, L. A., and Lambrecht, K. 1996. “Toward a construction-based theory of language function: The case of nominal extraposition.” Language 72: 215–247. Nikiforidou, K. 1991. “The meanings of the genitive: A case study in semantic structure and semantic change.” Cognitive Linguistics 2: 149–205. Otheguy, R. 1995. “When contact speakers talk, linguistic theory listens.” In Contini-Morava and Goldberg, 213–242. Reid, W. 1991. Verb and Noun Number in English: A Functional Explanation. London/New York: Longman. Reid, W. 1995. “Quantitative analysis in Columbia School theory.” In Contini-Morava and Goldberg, 115–152. Rosch, E. 1978. “Principles of categorization.” In Cognition and Categorization, E. Rosch and B. B. Lloyd (eds.), 27–47. Hillsdale, NJ: Lawrence Erlbaum. Steele, S. M. 1977. “On being possessed.” Berkeley Linguistics Society 3: 114–131. Talmy, L. 1988. “Force dynamics in language and cognition.” Cognitive Science 12: 49–100. Taylor, J. R. 1996. Possessives in English: An Exploration in Cognitive Grammar. Oxford: Oxford University Press/Clarendon. Teng, S.-H. 1974. “Double nominatives in Chinese.” Language 50: 455–473. Wierzbicka, A. 1988. The Semantics of Grammar. Amsterdam/Philadelphia: John Benjamins. Wierzbicka, A. 1996. Semantics: Primes and Universals. Oxford/New York: Oxford University Press.

CHAPTER 2

Cataphoric pronouns as mental space designators Their conceptual import and discourse function Michael B. Smith Oakland University, Rochester, Michigan

Building on Fauconnier’s mental space theory (1985), I argue that some languages have mental space designators, morphosyntactic elements which designate the mental spaces set up by linguistic expressions called space builders. Cataphoric pronouns can serve this function as forward referring devices that link with a discourse entity not yet overtly mentioned, as illustrated by the pronoun it in the sentence I despise it that John voted for the governor. Their discourse function is to anticipate the mental spaces set up by space builders (such as despise) by grammatically designating the spaces themselves, which are then elaborated with linguistically expressed propositions (e.g. that John voted for the governor). Evidence for this analysis includes English space building verbs evoking negative feelings which require cataphoric object it (cf. dislike, resent), where the cataphor’s overt mention of the space itself accentuates the conceptual distance (cf. Haiman 1983, 1985a, b) between the matrix subject and the undesirable relation(s) holding within the spaces by iconically distancing — in the grammar — the matrix subject from the subordinate clause describing those relations. The cataphor can also accentuate the salience of the space’s boundaries as a setting for an event, e.g. You know (it) when the train goes by. Cataphoric it forces a reading emphasizing the train’s physical effect on the setting prefigured by it, and thus accentuates the sense of know which evokes knowledge gained by the senses; omission of cataphoric it emphasizes someone’s intellectual knowledge of the time when the train goes by, and thus accentuates the sense of know which evokes purely intellectual knowledge. The discussion will support a view of grammar in which there is an iconic link between the forms and the meanings conveyed by the forms in these constructions.

1. Introduction to the issue: anaphora vs. cataphora This chapter is concerned with investigating a grammatical phenomenon which is only infrequently discussed in the linguistic literature: the meaning and function of

62

Michael B. Smith

cataphoric pronouns like those occurring postverbally in the following English (1a) and German (1b) sentences and after the adjective uveren “sure” in the Russian sentence (1c) (the optionality of some of these pronouns will be addressed presently). (1) a. I despise it that John voted for the governor. b. Wir bedauern (es), daß Hans so dumm ist. we regret it that Hans so dumb is ‘We regret (it) that Hans is so stupid.’ c. Ja uveren (v tom), čto on sderžit svojo slovo. I sure in that, that he keep his word ‘I’m sure that he will keep his word.’ Note that the pronouns in these examples, English it, German es “it”, and Russian tom “that” (the prepositional case form of the neuter singular demonstrative pronoun to) are not anaphors, because they clearly have no antecedents within the clause to which they can refer. Consequently, their occurrence seems, at least at first glance, grammatically and semantically arbitrary and/or superfluous, their functional motivation is unclear, and their meaning contribution to the clause (if any) is obscure. This is because, in contrast to backward referring anaphoric pronouns which are clearly linked with previously mentioned entities in the discourse, cataphoric pronouns are forward referring devices, i.e. their function is to anticipate some aspect of the discourse which follows their occurrence. As noted by Greenberg, In anaphora there is coreferentiality between the anaphoric substitute and the antecedent which precedes and is therefore identified. If we reverse this, the cataphoric substitute cannot be coreferential at the moment it is used because we do not yet know what it refers to. After the cataphor has been mentioned we can then in retrospect say that it is coreferential or that when the substitute is used it is proleptically cataphoric. (Greenberg 1985: 283)

The forward referring nature of cataphors means that a cataphor must, presumably, be linked with an entity of some sort which has not yet been overtly mentioned. The question before us, then, is to investigate the nature of this yet-to-be-mentioned entity, which will lead to a discussion of whether cataphoric pronouns like those illustrated in (1) are semantically motivated, or whether they are just arbitrary grammatical elements without semantic significance. I will argue that it is possible to support the claim that cataphors of this kind do have semantic content when a sufficiently sophisticated view of meaning is taken into account.¹

2. Some theoretical background It is not an exaggeration to say that linguistic theory has devoted relatively little attention to the syntactic and semantic properties of cataphoric pronouns in comparison to their more familiar anaphoric counterparts. “The difficulty of finding true examples of cataphora, and the commonplace nature of anaphora suggests the

Cataphoric pronouns as mental space designators

unmarked status of the latter”, notes Greenberg (1985: 284), which is probably related to the fact that cataphors are appreciably less common than anaphors. Few introductory textbooks on syntax and semantics even mention cataphoric pronouns, though they usually devote a great deal of space to accounting for the intricacies of anaphoric relationships in various languages (cf. such books as Cook and Newson (1996), Haegeman (1994), Sag and Wasow (1999), Van Valin, Jr. and LaPolla (1997), Radford (1997), and Culicover (1997) to name but a few). This neglect of cataphoric pronouns even extends to scholarly monographs on syntax and semantics. One need only reflect on the central importance of the binding theory in one recently (and still) popular syntactic framework as evidence for this (cf. Chomsky 1981, 1986), and a perusal of some other influential theoretical syntactic frameworks that have arisen in recent years confirms this general neglect (cf. the articles in Bresnan (1982) on lexical-functional grammar; studies in Perlmutter (1983) and Perlmutter and Rosen (1984) on relational grammar; and papers in Webelhuth (1995) on the minimalist program, etc.). A descriptive monograph by Pütz (1986) on the syntax of German es is a notable exception to this generalization in devoting several sections to a consideration of the function of es in constructions like those in (1b) above. But, unfortunately, Pütz does not explicitly refer to this kind of es as a cataphor, nor does he offer any kind of semantic characterization for its occurrence with some verbs but not with others (indeed he rejects the very possibility of a semantic explanation) (cf. Pütz 1986: 142). Some traditional German grammars also mention this sort of cataphoric use of es in their descriptive surveys, noting its anticipatory function in contexts like those exemplified in (1b) above, but they too fail to fully consider the role of this morpheme in the grammar (see Pütz (1986: 55–56) for a summary of these descriptions). Even recent functional approaches to language, including cognitive grammar (Langacker 1987, 1991a, b), have had little to say to date about the use and meaning of cataphoric pronouns, though they are potentially well-suited to the task. Nevertheless, Givón (1990, 1995) has investigated the differences between what he calls anaphoric (backward) grounding and cataphoric (anticipatory) grounding in discourse. Evoking Gernsbacher (1990), he notes that the former “has been studied much more extensively . . . [i]t involves connecting incoming new information chunks to some existing mental representation”, whereas the latter “has been studied in much less detail . . . [i]t involves the opening of pending connections in yetto-be completed structure, in anticipation of a text that is in the process of being constructed” (Givón 1995: 347). As will become clear in the following paragraphs, this conception of cataphora is compatible in spirit to the approach I will adopt and attempt to make more explicit in this chapter. The analysis I will propose will be grounded within the cognitive linguistic approach to language as described in works by Langacker (1987, 1991a, 1991b), Lakoff (1987), and Fauconnier (1985, 1997), all of which emphasize investigating the symbolic nature of grammar and the meaningfulness of grammatical constructs.² A theoretical construct in this framework which will prove particularly relevant in

63

64

Michael B. Smith

explaining the occurrence of cataphoric pronouns is the notion of mental spaces, which have been shown independently by Fauconnier (1985) and studies in Fauconnier and Sweetser (1996) to be insightful in analyzing a wide array of classic problems in semantics, including opaque-transparent ambiguities associated with propositional attitude verbs, the projection problem with presuppositions, and the interpretation of counterfactuals, to name but a few. Because the notion of a mental space will play a crucial role in this chapter, I will briefly outline its substance below and provide a couple of short examples adapted from Fauconnier’s work to illustrate the concept’s validity as a theoretical construct in linguistics. Space limitations preclude a detailed presentation of the many arguments in favor of mental spaces, which can be found in Fauconnier’s works. Let us first consider some preliminary background. The relevant literature (cited above) holds that the notion of a mental space evokes the intuitively natural idea that various aspects of sentence structure — both lexical and grammatical — set up “cognitive constructions” which help organize the processing of meaning within individual sentences and the flow of information between sentences in a larger discourse (cf. Fauconnier and Sweetser 1996). Fauconnier and Sweetser identify a number of such cognitive constructions, including metaphoric projection, frame organization (Goffman 1974, Fillmore 1985), roles, figure ground configurations, metonymic pragmatic functions (Nunberg 1978), cognitive schemas, and cultural models (Fauconnier and Sweetser 1996: 8). Mental space research, they note, is one line of investigation within cognitive linguistics which addresses the question of what kinds of cognitive entities or mental representations are involved in meaning construction in language. This research program represents an attempt to develop “a detailed model of the meaning construction that takes place under pressure from language forms, context, structured background knowledge, and other pragmatic factors” (Fauconnier and Sweetser 1996: 10). Fundamentally, mental spaces are theoretical constructs within cognitive linguistics posited to facilitate important aspects of meaning construction. In effect, they are conceived to be mental “places” or frames, usually triggered by certain lexical or grammatical devices, that are created in certain contexts and within which semantico-pragmatic information is located and organized relative to aspects of background knowledge (as mentioned above). Often the real world itself (or the speaker’s conception of what that world is like) is represented as reality space. The following quote from Fauconnier and Sweetser gives a representative idea of their conception of mental spaces: The dynamics of mental space construction and space linking are technically abstract, but conceptually straightforward. The basic idea is that, as we think and talk, mental spaces are set up, structured, and linked under pressure from grammar, context, and culture. The effect is to create a network of spaces through which we move as discourse unfolds. (Fauconnier and Sweetser 1996: 11)

From this perspective, then, note that part of the meaning of a propositional attitude verb such as believe involves opening up a new domain or knowledge struc-

Cataphoric pronouns as mental space designators

ture within the prevailing or current context that is “set up to contain structured information and inferences ‘about’ beliefs” (Fauconnier and Sweetser 1996: 10). This knowledge structure is a mental space. According to Fauconnier, mental spaces play an important role in the dynamic aspects of meaning construction, which he defines as “high level, complex mental operations that apply within and across domains when we think, act, or communicate” (1997: 1). These domains are mental structures (Langacker has called them “conceived worlds” (1991b: 71)), i.e. “partial structures that proliferate when we think and talk, allowing a fine-grained partitioning of our discourse and knowledge structures” (Fauconnier 1997: 11).³ Mental spaces are internally structured by frames and cognitive models which are akin to Lakoff ’s idealized cognitive models (1987). Much of Fauconnier’s work focuses on motivating mental space mappings, “which link mental spaces set up in discourse” (1997: 12) by means of connectors (pragmatically induced links or functions) that “relate . . . structures across spaces” (1997: 39) (see also Fauconnier (1985: 3ff.)).⁴ Fauconnier claims that mental spaces are “constructs distinct from linguistic structures [my emphasis] but built up in any discourse according to guidelines provided by the linguistic expressions” (1985: 16). They can be built up in discourse by certain kinds of linguistic expressions called space builders, such as prepositional phrases (in Dali’s painting), adverbs (possibly), connectives (if . . . then), subject-verb complexes (Harry believes), and also subject-adjective complexes in Russian (Ja uveren “I’m sure” in (1c)). Space builders are “overt mechanisms which speakers can use to induce the hearer to set up a new mental space” (Fauconnier and Sweetser 1996: 10); i.e. they “will typically establish new spaces, elements within them, and relations holding between the elements” (Fauconnier 1985: 17).⁵ Note that the criteria used to discern which kinds of linguistic elements set up mental spaces is primarily semantic in nature. In effect, we are making a theoretical assumption that certain linguistic expressions set up mental spaces by virtue of the kinds of meanings they evoke, as will be seen in more detail in the discussion to follow.⁶ As mentioned earlier, the positing of mental spaces can elegantly facilitate the interpretation of opaque-transparent ambiguities associated with propositional attitude verbs, as illustrated in sentence (2), in which the verb believe sets up a mental (belief) space M along side of reality space R (representing the world as it is) against which the interpretation of the indefinite noun phrase a governor can be disambiguated. (2) John believes that a governor will win the presidency. The nominal a governor can be interpreted as referring either to a governor in John’s belief space which is the same as a particular governor in R (the transparent or specific reading sketched in the top diagram in Figure 1), or it can be interpreted as referring to an entity in M (John’s belief space) that has no counterpart in R (the opaque or nonspecific reading sketched in the lower diagram in Figure 1). Fauconnier notes that, according to this analysis, the existence of two different readings for

65

66

Michael B. Smith

x1: a real governor

x2: a governor in John’s belief space

x1

x2

R

M (“John believes ______”)

Transparent (specific) reading x2: a governor in John’s belief space x2

R

M

Opaque (nonspecific) reading Figure 1. Transparent vs. opaque ambiguities associated with verbs of propositional attitude (cf. Fauconnier 1985: 28)

this sentence “is purely a consequence of the discourse processing involved (construction of spaces, connectors, etc.); it is not linked to any structural ambiguity of the linguistic form at deep, semantic, or logical levels” (1985: 28–29). Mental spaces also provide a straightforward way of dealing with intricate and complex ambiguities involving anaphoric pronouns and their antecedents which cannot be represented easily by coindexing, especially in cases where “a pronoun with an antecedent in one space can freely identify its counterpart in another, connected space” (Fauconnier 1985: 35). Consider as an example a scenario (adapted and simplified from one discussed by Fauconnier (1985: 36–37)) in which someone makes a movie about Alfred Hitchcock’s life, with an actor playing the role of Hitchcock, but that Hitchcock himself plays a minor role (as he always did in his own films). Suppose further that in the movie’s script the character of the movie Hitchcock accidentally hurts himself in some way and that the real Hitchcock also actually hurt himself while playing the minor role in the movie. Then sentence (3) in this context is multiply ambiguous with respect to what the reflexive anaphor

Cataphoric pronouns as mental space designators

himself refers to (i.e. designates in the real or some possible world), in contrast to the apparent single reference of himself suggested by the traditional coindexing of the reflexive anaphor with its linguistic antecedent. (3) Hitchcocki hurt himselfi in that movie. Though coindexing adequately represents the anaphoric relationship between the anaphor and antecedent, it fails to indicate that this relationship is open to multiple interpretations which are not overtly distinguished by the linguistic expressions themselves, since the anaphor’s interpretation depends upon which referent of the proper noun Hitchcock is intended. A mental space analysis clarifies the matter in the following way. Note first that this sentence involves two different, connected, mental spaces: speaker reality R, in which both the real Hitchcock exists (r1) and also an actor exists (r2) who is playing him in the movie, and the movie space M set up by the prepositional phrase in that movie, which contains the minor character played by Hitchcock himself (m1) and the character of Hitchcock (m2) which is being played by the other actor. Positing the existence of mental spaces provides a simple account of the range of possible interpretations for sentence (3). The anaphor could refer to (designate) either the real Hitchcock (r1) in R (with the interpretation that the real Hitchcock hurt himself while making the movie), or to the cameo character Hitchcock is playing in the connected M space (m1) (with the interpretation that the minor character Hitchcock is playing hurts himself as part of the movie’s plot). Or the anaphor could refer to (designate) the character of Hitchcock which is being played by another actor (m2) in M (Fauconnier suggested Orson Welles could have played the part). The latter reading would have the interpretation that the Hitchcock character being played by Welles hurts himself as part of the plot. The point is that these three different interpretations of sentence (3) can be readily disambiguated by setting up different counterparts for the proper noun Hitchcock in two different mental spaces which the anaphor can refer to. Now that we have gotten a glimpse into how the notion of mental spaces can prove useful in dealing with problems in linguistic semantics, I will henceforth take it for granted that mental spaces are plausible theoretical constructs and turn now to the main claim of the chapter.

3. Mental space designators Up to this point we have focused on the idea that certain linguistic expressions set up or build mental spaces (space builders), but have not yet considered the possibility that in some cases these spaces might be overtly represented in grammatical structure itself. My central aim in this chapter is to present evidence that some languages not only have space builders, but also expressions I will call space designators, i.e. elements which actually designate — in the morphosyntax — the mental spaces

67

68

Michael B. Smith

themselves which are set up by the space builders (as opposed to forms which describe relations holding within such spaces). Fauconnier apparently makes no reference to this kind of construct in his work, but I will claim that the facts from English, German, and Russian support its viability and that the theory of mental spaces should be expanded to include it. Specifically, I argue that at least some uses of cataphoric pronouns serve this space-designating function, as illustrated by the anticipatory object pronouns in (1) above and the additional English (4), German (6), and Russian (7) sentences below. Consider first the cataphoric it which functions as the direct object of the matrix verbs in the English examples in (4) below. (4) a. b. c. d.

We abhor it that the mayor absconded with the money. She resents it that the class starts so early. They dislike it that the weather is so cold here. He detests it that Clinton was impeached.

Note that the matrix verbs in each example express different kinds of mental attitudes about the propositions which follow in the subordinate clauses. These propositional attitude verbs are therefore assumed to create mental spaces that represent new domains of knowledge within the prevailing or current context. The subordinate propositions are located conceptually within these spaces and are thus interpreted relative to the structured information and inferences about each kind of propositional attitude residing within each space. The cataphor it is either obligatory or highly preferred with the above space building verbs in English, but it is generally not tolerated with such space builders as believe, promise, feel, hope, know, think, etc., as shown in (5) below. A reason for this difference will be suggested in Section 5 below. (5) a. b. c. d.

*Mary believes it that the senator will lose. *The governor promised it that she would sign the bill. *They know it that there will likely be a drought this year. *She thinks it that the baby is crying.

In German, cataphoric es “it” may also appear either optionally or obligatorily with certain space building verbs, as shown by the data in (6) below. (6) a.

Ich wage (es) nicht, meinen Vater zu fragen. I dare it not my father to ask ‘I don’t dare ask my father.’ b. Paul vergißt (es), daß Inge kommt. Paul forgets it that Inge comes ‘Paul forgets that Inge’s coming.’ c. Peter lehnt *(es) ab, daß die Sozialisten gewonnen haben. Peter leans it from that the socialists won have ‘Peter refuses [to admit] that the socialists won.’

Cataphoric pronouns as mental space designators

d. Ich kann *(es) kaum ertragen, daß Hans gewonnen hat. I can it hardly bear that Hans won has ‘I can hardly stand it that Hans won.’ Cataphoric es is generally not allowed after such German verbs as wissen “know”, annehmen “accept”, behaupten “assert”, empfehlen “recommend”, denken “think”, glauben “believe”, fühlen “feel”, etc. (see Pütz (1986: 70) for a more complete list). The evidence suggests that there is a basic distinction in German between space builders which require cataphoric es (cf. (6c) and (6d) above) and those with which its occurrence is either optional or generally not allowed (cf. (6a) and (6b) above) (cf. Pütz 1986: 58–88 for detailed discussion).⁷ The determining factors for its optionality likely reside in the idiosyncratic meanings of each verb and go beyond the scope of this chapter (indeed they have not yet been adequately studied, as is noted by Pütz). Thus, in many instances it is not possible to predict whether the cataphoric es will or will not appear with a particular space builder, as is the case with verbs like vergessen “forget” in (6b) above. But the data are usually quite clear with respect to which verbs require the cataphor vs. those which do not, so I will focus on trying to account for the difference between these two main sets. I will propose a reason for the basic distinction between the two classes of verbs in Section 5 below.⁸ As noted above, a similar kind of cataphoric function is apparently served in Russian by a form of the demonstrative pronoun to ”that” in the following correlative constructions (where the differences in the pronoun’s phonological form reflects different cases) (cf. Leed, Nakhimovsky, and Nakhimovsky 1982: 150–151).⁹ (7) a.

Mne nravitsja (to), čto on vypolnjaet svoi obeščanija. me pleases that that he keeps REFL promises ‘I like the fact that he keeps his promises.’ b. Ja somnevajus’ (v tom), čto on skazal eto. I doubt in that that he said that ‘I doubt that he said that.’ c. Petja privyk (k tomy), čto kritiki ego vsegda xvaljat. Pete used to that that critics him always praise ‘Pete is used to critics always praising him.’ d. Ivan načal s togo, čto lično poznakomilsja so vsemi. Ivan began from that that personally acquainted with all ‘Ivan began by getting personally acquainted with everybody.’

We see from these data that, in addition to verbs and the other kinds of linguistic expressions mentioned previously, short form adjectives such as privyk “accustomed, used (to)” and uveren “sure” (from (1c)) can also serve as space builders in Russian. What might motivate the appearance of cataphors as mental space designators in constructions of this type? A natural explanation for the function of the cataphoric pronouns in discourse in these examples is to anticipate the mental spaces set up by space builders, such as despise in (1a), bedauern “regret” in (1b), and uv-

69

70

Michael B. Smith

Propositional attitude predicate prefigures a mental space (cataphoric pronoun accentuates space’s boundaries by actually designating space) I despise . . . it Wir bedauern . . . es Ja uveren v . . . tom

that . . . S dass . . . S čto . . . S

Figure 2.

eren “sure” in (1c), by actually designating the spaces themselves in the grammar. They often link the propositional attitude (or some action) of a matrix clause entity — usually the matrix subject — with particular relationships associated with that entity. The mental spaces are then subsequently elaborated by propositions expressed by “linguistic clauses, which typically. . .predicate relations holding between space elements” (Fauconnier 1985: 17). The sketch shown in Figure 2 gives a rough idea of how this works: the mental space set up by the space building verb or adjective that is designated by the cataphor is represented by the circle, and the clauses which fill in the space are shown inside the circle to indicate that they are inside the mental space. Under this analysis, the English space builder despise in (1a) sets up a mental space anticipated and designated by it, in which the proposition that John’s vote for the governor is negatively appraised. Similarly, in (1b) the German matrix verb bedauern “regret” sets up a mental space designated by the pronominal es, which is then elaborated by the clause stating the proposition that Hans is stupid. And in (1c) the Russian space builder uveren “sure” sets up a mental space which is then elaborated by a proposition expressed by subordinate clausal elements which describe what the matrix subject is “sure” (uveren) about. Similar descriptions could be given for the other data in (4), (6) and (7) above. In a nutshell, the proposed model is that the cataphor designates a space set up by a space building verb or adjective which is then elaborated with a subordinate proposition expressed by a linguistic expression. This latter proposition is then interpreted relative to the knowledge structures evoked in each space (i.e. with respect to despising something (1a), being regretful about something (1b), or being sure about something (1c)). Before proceeding further, it is worth mentioning that an apparently similar kind of function is proposed for such English “abstract and nonspecific nouns” as fact, idea, problem, rumor, advantage, claim, thing, point, etc. in a paper by Schmid (1999: 111), as illustrated by sentences like the following (which I have constructed to take the place of his longer examples extracted from corpus data).

Cataphoric pronouns as mental space designators

(8) a. b. c. d.

The fact was that we ran out of time. The idea was that we would go together to the concert. The problem was that they forgot to bring along any money. The rumor was that they were having an affair.

Schmid suggests that such nouns, which he calls “shell nouns”, function to provide “conceptual shells” (1999: 113) in the discourse which are subsequently elaborated by different contents that are expressed by the subordinate clauses which follow them.¹⁰ He goes on to note that “isolated uses of shell nouns do not activate conceptualisations that are experienced as saturated thoughts . . . Therefore speakers must link them to clauses or even longer stretches of text” (1999: 114). But each noun’s lexical content, however abstract it might be, does “have some potential for the characterisation of cognitive content” which influences the way in which the information to be supplied in the shell is construed (1999: 114). The use of the shell nouns problem in (8c) and rumor in (8d), for example, shade the interpretation of the subordinate clauses which follow their occurrence by labeling the content as being either negative or hypothetical in some way, respectively. Schmid notes that the conceptual shells set up by shell nouns are remarkably similar to Fauconnier’s notion of mental spaces, though Fauconnier does not seem to consider nouns as possible space builders (Schmid 1999: 115). Among the more interesting “cognitive effects” or functions of shell nouns noted by Schmid include their use in conceptual packaging, whereby they effect the “temporary reification of complex cognitive content as a thing” (1999: 122). That is, shell nouns are a linguistic means of packaging information economically and compactly into “unsaturated” (conceptually unelaborated) concepts which are then elaborated by “saturated thoughts” (conceptually elaborated concepts) that are ordinarily packaged less economically by linguistic clauses describing events or relations (Schmid 1999: 120). By using a shell noun, a speaker can more easily link the information referred to by the noun to other kinds of conceptualizations, since nouns prototypically designate things, and things are conceptually simpler and more efficient to manipulate than clauses designating events or relations. Schmid also observes that the shell-content pattern like that illustrated in (8) above effects the interpretation that “the information provided by [the shell nouns] is construed as given, and marked as being accessible” or taken for granted (i.e. topicalized), and that “the shell contents are highlighted for attention” and construed as “new and particularly noteworthy” or in focus (1999: 127–128). Schmid’s paper is interesting in the present context because it convincingly argues that mental spaces can be constructed by relatively abstract nominal entities (shell nouns) which are apparently more semantically abstract than Fauconnier’s space builders, but several degrees less abstract than cataphoric pronouns, plausibly the most abstract nominal elements of all. I would argue that his analysis supports my own claim that cataphoric pronouns can designate mental spaces, in that it suggests the possibility of the existence of a continuum of linguistic entities ranging

7

72

Michael B. Smith

from highly concrete expressions such as verbs, which set up mental spaces but do not designate them, to the more abstract space building shell nouns, to cataphoric pronouns which, I argue, do not so much create spaces as designate the spaces set up by more concrete clausal elements. Crucially, Schmid does not claim that shell nouns designate the mental spaces which they set up, but that they act very much like other kinds of more semantically concrete space builders in creating such spaces. If my own analysis is correct, then it is likely that cataphoric pronouns represent an extreme degree of grammaticalization and abstraction of this general process.¹¹ At this point let us stop to consider possible objections to an analysis that treats cataphors as space designators and even consider a possible counter-analysis. Some readers might claim that a linguistic construct with this kind of function really does not make any substantial meaning contribution to the clause, since the cataphor’s lexical content is essentially null, and that its proposed use in these sentences is nothing more than an arbitrary grammatical marker without semantic significance. Indeed, Greenberg alludes to this possibility when he notes that “a normally anaphoric marker used in this way cannot usually be anymore than a sign that some identification will follow, very much as an opening quotation mark tells us that a quotation is coming without it yet being known. For this reason it is always redundant from the specifically referential point of view” (1985: 283). While cognitive grammar does not deny the possibility that “such elements are, so to speak, grammaticized” (Greenberg 1985: 283), their regular and predictable grammatical function and apparently “redundant” contribution to the referential meaning of the clause does not necessarily mean that they are completely devoid of meaning. Indeed, the suggested function of cataphoric pronouns in the examples above can be viewed as reflecting the importance of the cognitive grammar notion of imagery in linguistic semantics, whereby each grammatical construction is conceived to impose a particular kind of conceptual organization on a conceived situation (cf. Langacker 1987, 1991a, 1991b), and the meaning of a linguistic expression involves not only its objective content but also how that content is construed. What this means is that, in spite of their apparent meaninglessness, the presence of cataphors in some constructions might provide a particular way of construing a scene or of organizing the elements in a given conceptualization which, at least in some instances, is semantically relevant. I will present evidence for this later in the chapter. Alternatively, it might be claimed that these pronouns are not cataphors at all, but that they are in fact actually “anaphoric, since what will follow has already been mentioned and the apparently cataphoric element is really coreferential with something that preceded and is simply about to give us an additional characterization or identification” (Greenberg 1985: 283). While I know of no way to prove conclusively that the pronominal elements in question are cataphors and not just anaphors in disguise (so to speak), I would simply note that it seems perfectly acceptable to use any of the English sentences in (4) above to start a conversation about a relevant topic, without the information which

Cataphoric pronouns as mental space designators

appears in the subordinate clauses necessarily already having been mentioned in previous discourse or even being part of the common ground. For example, it would be completely natural for me to change the topic of a conversation about a mutual friend of ours which had nothing at all to do with a class’s starting time by uttering (4b), and a similar point could be made about (4c) (repeated below). (4) b. She resents it that the class starts so early. c. They dislike it that the weather is so cold here. Notice that neither sentence would sound right if it were uttered in a context which had previously had nothing to do with someone resenting or disliking something and the subordinate clause following the pronoun were omitted, but both sound fine in such a context when the subordinate clause follows the pronoun. This suggests strongly that it is the matrix space building verbs which create the context for the pronouns, not some previous discourse context, and that they cataphorically prefigure (i.e. designate) the mental spaces set up by these verbs. Finally, note also the possibility that these pronouns may refer cataphorically not only to the spaces set up by the space builders, but also to the clauses themselves which explicitly describe the relations involved within those spaces. But, even if this is true, when the pronouns are first encountered by the hearer, they can refer to nothing yet introduced into the context but the expectation of information to come (propositions in the examples under discussion) which is prefigured by the space builders, as was mentioned earlier. Interpreting this expectation as the establishment of a mental space that is designated by the cataphoric pronoun seems to be a natural explanation for its occurrence. Characterizing cataphoric pronouns (vs. some other kind of linguistic item) as space designators therefore seems intuitively appealing due, at least in part, to their highly schematic meanings in this kind of usage and the fact that the meanings of pronouns in general (if they are attributed any meaning at all) are usually considered to be more grammatical than lexical in nature. But, to support an analysis claiming that cataphoric pronouns can designate mental spaces, it is necessary to show that the pronouns in question are polysemous, i.e. they have a number of distinct but interrelated senses — usually radiating from a prototypical sense — which can be plausibly related to one another via meaning chains motivated by semantic extensions of various kinds (see Taylor (1995) and Lakoff (1987) for extensive discussion). The reason for this is that in its use as a space designator we are attributing to the pronoun a sense/use which is clearly distinct from its more familiar anaphoric function. Unless these different uses can be linked in some way, there is no explanation for why a pronoun can be used as a space designator, and thus the use of what otherwise looks like an ordinary anaphoric pronoun to designate mental spaces would be arbitrary. In the next section, I will discuss how the use of cataphoric pronouns as abstract mental space (abstract setting) designators follows naturally from some of their more prototypical uses/senses as anaphors, focusing on the evidence from English and German.¹²

73

74

Michael B. Smith

4. Evidence for the mental space senses of English it and German es In this section I show how the various senses of German es can be related to each other as members of a complex conceptual category centering around its prototypical sense as a pronominal anaphor. For the most part, the senses of English it are essentially parallel to those of es, as will be clear from the English translations of the German sentences, so I will combine the discussion for both languages where possible (cf. Smith (2002) for more detailed argumentation).

4.1. German es and English it as pronominal anaphors Prototypically, these pronouns refer to an earlier 3rd person singular entity (thing) in the discourse (which is also specified as grammatically neuter in German), as shown in (9) below. The anaphoric relation is indicated by the usual subscript notation in the following examples. (9) a.

Ich kaufte [ein Buch]i und habe esi mit mir gebracht. I bought [a book and have it with me brought ‘I bought [a book]i and I’ve brought iti with me.’ b. Ist das [Ihr Kind]i? Bitte nimm esi weg von hier! is that [your child please take it away from here ‘Is that [your child]i? Please take iti away from here!’

It is evident that English it has an analogous function in this kind of sentence which requires no further discussion.

4.2. German es as an anaphor for other types of things German es, though apparently not English it, can also refer to non-neuter noun phrases (10), have a plural noun phrase antecedent (11), or be used as a general kind of indeterminate accusative object pronoun (called a “bloßes formales Objekt” in many German grammars), in which it does not refer to any particular antecedent, but rather to some unspecified thing or concept (12). Clearly, each of these uses represents a separate extension (and resulting sense) for es which is slightly different from the prototype exemplified in (9) above.¹³ (10) Ich kannte [seinen Bruder]i. Esi/eri war ein guter Arzt. I knew his brother it/he was a good doctor ‘I knew [his brother]i. Hei was a good doctor.’ (11) An der Ecke standen [ein Mädchen und ein Junge]i. Esi/siei waren seine on the corner stood a girl and a boy it/they were his Kinder. children ‘On the corner stood [a girl and a boy]i. Theyi were his children.’

Cataphoric pronouns as mental space designators

(12) a.

Er nimmt es mit jedem auf. he takes it with each up ‘He’s a match for everyone.’ b. Er hat es weit gebracht. he has it far brought ‘He’s gotten on (succeeded) in the world’.

These data show that German es is susceptible to semantic bleaching even in its basic anaphoric sense (cf. Sweetser 1988), in which, first, its erstwhile neuter gender specification is lost (10), then its number specification may be lost (11), and then, in a further extension, it may come to designate no more than some unspecified thing which has no apparent antecedent in the discourse (12).

4.3. German es and English it as anaphors for relations and reified clauses As a crucial way-station toward its more grammaticalized uses (as a cataphor, for example), German es can be anaphoric for relational predications (and not just for individual noun phrases mentioned previously in the discourse). In both (13) and (14) below, es refers back to the atemporal relations (i.e. adjectives) arm “poor” and angenehm “pleasant”, respectively. English it doesn’t seem to have this particular function. (13) Er ist armi, du bist esi auch. he is poor you are it [i.e. poor] too ‘He’s poor, (and) you are too.’ (14) Peter ist mir angenehmi. Aber Paul ist esi mir nicht. Peter is me pleasant but Paul is it me not ‘Peter is pleasant to me. But Paul is not (i.e. pleasant to me).’ And, in a further semantic extension, in (15) and (16) below, German es and English it appear to be anaphoric for an entire proposition, in which a clause (a linguistic representation of an entire proposition) has been reified as a kind of abstract thing. For example, in sentence (15) below the process of someone’s reciting has been reified (as indicated in the gloss) in order to serve as the antecedent for es and it in the following clause. The reification of an act of theft is likewise the antecedent for es and it in sentence (16). (15) [Wir haben ihn wieder deklamieren gehört]i. [we have him again recite heard Esi war ein verbales Schlachtfest. it was a verbal slaughter ‘[We’ve heard him recite again]i. Iti [the reciting process] was a verbal slaughter (lit. a ‘slaughter-fest’ in German).’

75

76

Michael B. Smith

(16) [In der Klasse ist gestohlen worden]i, [in the class(room) is stolen become aber keiner will esi getan haben. but no.one wants it done have ‘[Something was stolen in the classroom]i, but no one claims to have done iti.’ The evidence shows so far that es and it can be used as anaphors for anything from a thing construed as a participant in a given event to a reification of a full clause. I claim that this latter sense provides the bridge for its use to designate settings of various sorts.

4.4. German es and English it profiling mental spaces (abstract settings) There is a plausible conceptual mechanism by which es and it can be extended in use from designating individual entities (or relations) previously mentioned in the discourse to designate mental spaces or abstract settings (cf. note 4) of various kinds within which participants are conceived to interact. The conceptual mechanism in question is facilitated by the use of the pronouns in data like that in (15–16) above, where they designate reified clauses previously mentioned in the discourse. The semantic extension in question probably comes about via a kind of metonymy in which the pronouns lose their association with individual participants (however indeterminate) or reifications of an event itself (cf. 15–16) and instead come to be associated with the setting or space in which the action or state transpires.¹⁴ This may be a kind of whole for the part metonymy, as opposed to the more frequent part for the whole metonymy illustrated in a sentence like (17), in which the person who ate the sandwich is indirectly designated by referring to an entity with which the person was closely identified in the given context (cf. Lakoff and Johnson 1980: 35–40). (17) The ham sandwich left without paying. A metonymic analysis of the proposed semantic extension is completely reasonable when one considers that events are integrally associated with the settings in which they occur. Indeed, a setting often serves as a kind of reference point with respect to which the event in question can be located (cf. Langacker 1991b: 351, Langacker 1993, and note 4). Langacker notes, for example, that elements like it and there in English, in their use in designating abstract settings (cf. the setting subject designated by there in the sentence There’s a lamp on the table), “should probably be thought of as abstract presentational frames serving to announce the intention of subsequently introducing something in those frames. With such an analysis, it is perfectly understandable that it and there should serve as reference points: a setting constitutes a natural and obvious reference point for the elements that occur within it” (Langacker 2000: 347).

Cataphoric pronouns as mental space designators

It is not surprising, then, that the sense exhibited by the pronouns when they are used as anaphors for a reified clause could be extended to a related sense in which the pronouns designate the setting in which the clause is construed to take place. The additional conceptual leap required to go from the one use to the other is not appreciably greater than that needed previously to justify the extension of their meaning from an anaphor for a relation (13–14) to an anaphor for a reified clause (15)–(16). Each of these semantic extensions from the prototype reflects a further step toward abstractness, with the setting (mental space) sense representing the most abstract use of all, i.e. the use which is usually viewed as a meaningless grammatical marker in many frameworks. On the basis of the material in Section 4, I will now assume that cataphoric pronouns can be used to designate mental spaces prefigured by certain space builders.

5. Additional evidence that cataphors can designate mental spaces In this section I will present three kinds of additional evidence that cataphoric pronouns can designate mental spaces prefigured by space building predicates of various kinds.

5.1. The use of cataphors to accentuate conceptual distance As shown in (1a) and (4) above, the class of space building verbs in English requiring cataphoric object it is, in general, semantically restricted to those involving negative, uncertain, or otherwise undesirable feelings (as expressed by such verbs as dislike, resent, loathe, abhor, detest, hate, etc.). The same is true for German (6), though the class of German space builders either allowing or requiring the space designator es is apparently broader than that in English. German seems to be more tolerant of allowing cataphoric es to prefigure spaces construed either as important or especially desirable (18), or not yet accomplished or incomplete (19).¹⁵ (18) a.

Wir loben es, daß Sie das gemacht haben. we praise it that you that done have ‘We praise (it) that you did that.’ b. Meine Eltern lieben (es), am Wochenende an einen See zu my parents love it on-the weekend on a lake to fahren. drive ‘My parents love to drive to the lake on the weekend.’

(19) a.

Er kann es kaum erwarten, seine Braut wiederzusehen. he can it hardly wait his bride again-to-see ‘He can hardly wait to see his bride again.’

77

78

Michael B. Smith

b. Er gönnt es mir nicht, im Sommer nach Italien zu fahren. he allows it me not in-the summer to Italy to travel ‘He grudges my traveling to Italy in the summer.’ Also, cataphoric es is possible in German after verbs of permitting (something not yet done). (20) a.

Ich erlaube es dir, meine Schreibmaschine zu benützen. I permit it you my typewriter to use ‘I permit you to use my typewriter.’ b. Ich überlasse es dir, morgen zu kommen oder nicht. I leave it you tomorrow to come or not ‘I leave it to you to come tomorrow or not.’

In all such cases, I suggest that overt mention of the space itself by the pronoun accentuates (for whatever reason) the conceptual distance judged to exist between the matrix subject and the relation(s) holding within the spaces by iconically distancing — in the grammar — the matrix subject from the subordinate clause describing those relations (cf. Haiman (1983, 1985a, 1985b) for detailed discussion of the notion of conceptual distance and examples of the phenomenon from other languages). The use of cataphoric pronouns to accentuate the conceptual distance between the matrix subject and the proposition in the subordinate clause is likely a manifestation of a spatialization metaphor closely related to one identified in Lakoff and Johnson (1980) to the effect that closeness is strength of effect, as illustrated by the following examples (from Lakoff and Johnson (1980: 130)). (21) a. Harry is not happy. b. Harry is unhappy. (22) a. I taught Greek to Harry. b. I taught Harry Greek. Lakoff and Johnson note that (21a), in which the negative morpheme not is a separate word, could be neutral with respect to whether Harry is fundamentally happy or sad, even though it indicates that at the present time he is not happy. (21b), on the other hand, in which the negative morpheme is more closely attached to the word happy as a prefix, implies a stronger negative in which Harry is definitely sad. And, in (22a), in which Harry is separated from Greek by the preposition to, Harry may not have learned what I tried to teach him, whereas in (22b) the lack of any element separating Harry and Greek yields the clear implication of a stronger effect in which Harry did actually learn Greek. Lakoff and Johnson observe that spatial metaphors can “apply directly to the form of a sentence. . .. [t]his can provide automatic direct links between form and content, based on general metaphors in our conceptual system” (1980: 126). Thus, part of a sentence’s meaning can stem from the precise form the sentence takes due

Cataphoric pronouns as mental space designators

to this nonarbitrary linking between form and meaning, and “[l]inguistic forms are themselves endowed with content by virtue of spatialization metaphors” (Lakoff and Johnson 1980: 126). It should be stressed that there may be multiple semantic motivations for using cataphoric pronouns to accentuate conceptual distance in the grammar. Many occurrences of the German and English cataphoric pronouns under investigation here seem to reflect a related (and possibly inverted) form of the above spatialization metaphor in which distance is weakness of effect. In other words, the appearance of a cataphoric pronoun space designator after space builders signifying negative feelings about events or states (and in other circumstances in German) serves to distance the matrix subject from those events or states, which is not surprising, since it is natural for humans to avoid or distance themselves from negativelyconstrued situations in the hope that their effect will be weaker if they are kept as far away as possible. Alternatively, there is ordinarily no need to distance states or events from the matrix subject which are construed in mildly favorable or neutral terms, which explains why cataphors are generally not found after such English space builders as believe, think, hope, know, etc., or after such German space builders as wissen “know”, annehmen “accept”, behaupten “assert”, empfehlen “recommend”, denken “think”, glauben “believe”, fühlen “feel”, etc. (a fuller list can be found in Pütz 1986: 70). But extending this metaphor to other situations that can be construed as distant in some way from the matrix subject is also natural and not necessarily contradictory when the motivating factors are properly taken into account. Thus, space designators can also be used in German not only to distance an undesirable event from the matrix subject, as discussed in the above paragraph, but also to iconically accentuate the separation between this subject and something conceived by that entity to be especially important or desirable beyond ordinary circumstances (18) (as though putting it on a pedestal). The space designating function of cataphoric pronouns can also accentuate the separation between the matrix subject’s expectation of some event and its fulfillment (19), or his permitting something and the subsequent realization of what is permitted (20), all of which are construed as distant from the state of affairs assumed to hold for the matrix subject.¹⁶ The following additional data illustrate the meaning differences conveyed by the same space building German verbs when the cataphoric anticipatory-object pronouns are present vs. absent (from Pütz 1986: 88). (23) Ich schätze (*es), daß Peter wohl bald wieder gesund wird. I think (it that Peter definitely soon again healthy become ‘I think that Peter will definitely be well again.’ (24) Ich schätze *(es), daß du gekommen bist. I value (it that you come are ‘I value (esteem highly) that you have come.’

79

80

Michael B. Smith

In (23) the more neutral sense of schätzen “think” does not allow es, whereas its more positive sense in (24) “value, esteem” does require it, thereby accentuating what is valued by distancing it from the matrix subject by overt designation of the mental space by the pronoun. Note that the use of cataphoric pronouns in (18) and (24) above after matrix verbs designating highly positive feelings or emotions is just as motivated as its use after verbs designating highly negative feelings and emotions (though each for different reasons), because it is natural in either case to accentuate the distance between the matrix subjects and the subordinate propositions. The different reasons for either motivation have already been discussed above, and it is important to see that they are not contradictory from the perspective of cognitive grammar. What is likely going on is that in German (and to some extent also in English, as mentioned in note 15) there is a scale along which matrix verbs can be seen to lie with respect to their proclivity to co-occur with a following cataphoric pronoun. At the one extreme we find those designating highly negative feelings (detest, hate) and at the other we find those designating highly positive feelings (love, admire). As one moves away from either extreme it is less likely for the pronoun to occur, and for verbs in the middle which do not evoke strong feelings either way (think, believe) the cataphor generally does not occur, because there are no strong motivating factors for separating the matrix subject from the subordinate proposition with these verbs. It is not possible to strongly predict where the cut-off point for using the pronoun is from either extreme, and indeed, in some instances, as with the verb enjoy in English, the use of the cataphor is optional, with its optionality an indication of the gradation inherent in the scale.

5.2. The use of cataphors to indicate difference in emphasis Now that we have examined evidence based on conceptual distance that cataphoric pronouns can designate mental spaces in the grammar, let us turn to another kind of closely related evidence: the optional occurrence of cataphors after verbs (with which they usually do not occur) when the speaker wishes to emphasize the propositional content of the subordinate clause. The following set of sentences shows how a difference in emphasis can be accentuated by es after wissen “know” in German, where es normally does not occur (cf. Lederer 1969: 202). (25) Ich weiß, daß er morgen kommt. I know that he tomorrow comes ‘I know that he’s coming tomorrow.’ Apparently, wissen can be used with object es if special emphasis is placed on the clausal object.

Cataphoric pronouns as mental space designators

(26) Ich weiß es sicher, daß er morgen kommt. I know it certain that he tomorrow comes ‘I know for certain that he’s coming tomorrow.’ (27) Peter hat es doch gewußt, daß ich das getan hatte! Peter has it EMPHATIC known that I that done had ‘Peter knew [emphasis] that I had done that!’ Note that there is also a possible English translation of (26) in which cataphoric it can be used for emphasis, which suggests that this use of the cataphor is relevant in English as well, since it can also be used after other verbs of knowing, such as understand, realize, etc. (28) I know (it) for certain that he’s coming tomorrow. The cataphor’s evocation of special emphasis in these situations is likely due to the fact that overt designation of the mental space set up by the matrix verb draws more attention to the proposition contained in and characterized specifically by the knowledge structures inherent in that space than if the pronoun were absent. Also, by momentarily delaying mention of the subordinate clause by the use of the pronoun the speaker creates an air of anticipation in the flow of discourse about what is to follow which can heighten the hearer’s interest in the subsequent information (again evoking another kind of conceptual distance). Related to this is that the use of the cataphor in effect results in a kind of double-mention in which the space designated by the pronoun metonymically relates to the proposition located conceptually within that space by prefiguring the space grammatically. Finally, in each of (26–28) it is also evident that the information construed as known, which is specified within the mental space designated by the cataphoric pronoun, is already in the realm of discourse, thus accentuating its independent existence apart from the matrix subject’s musings, and motivating the use of es or it to anticipate its specific mention in the following subordinate clause.¹⁷ Another pair of German sentences illustrating how a difference in emphasis can be characterized by including the cataphor is given below (from Pütz 1986: 60). (29) Ich sage nicht, daß ich Hans verleumdet habe. ich say not that I Hans slandered have ‘I’m not saying that I slandered Hans.’ (30) Ich sage es ja auch gar nicht, daß ich Hans verleumdet habe. I say it yes (EMPH.) also at-all not that I Hans slandered have ‘I’m not at all saying that I slandered Hans.’ Sentence (29) is relatively neutral with respect to the speaker’s saying that he did not slander someone, whereas (30) more forcefully emphasizes this claim, as shown by the addition of the emphatic particle ja.

8

82

Michael B. Smith

5.3. The use of cataphors to accentuate the importance of a space’s physical boundaries The evidence presented so far has suggested that overt mention of a mental space itself by the cataphoric pronoun in English and German can be exploited to play up the importance of the space’s boundaries with certain matrix verbs (cf. Figure 2 above), which further accentuates the “mental” distance between the matrix subject and what is conceptualized by that subject. We have also seen how the cataphor’s designation of the mental space set up by a space building matrix verb can evoke emphasis in some cases. A slightly different kind of cataphoric it in English which evokes the enhanced relevance of a space’s boundaries is especially evident in minimal pairs like the following, with matrix verbs which usually evoke mental knowledge and/or some kinds of typical sensory experience. In these examples, rather than invoking a mental space’s boundaries to accentuate the conceptual distance between the matrix subject and what is conceptualized (or possibly intellectual emphasis on the subordinate concept per se), this kind of it highlights the relevant space’s boundaries to play up the importance of the space as a physical setting for some kind of event which is construed to have a physical rather than intellectual effect on the matrix subject. (31) a. You know it when the train goes by. b. You know _____ when the train goes by. (32) a. Did you hear it when the train arrived? b. Did you hear _____ when the train arrived? (33) a. She was aware of it when the plane landed. b. She was aware of _____ when the plane landed. (34) a. The psychic could feel it when the UFO landed. b. The psychic could feel _____ when the UFO landed. In each sentence in (31–34) the use of cataphoric it in the (a) examples forces a reading that clearly accentuates the effect of the vehicle’s action within a physical space or setting prefigured by it and that action’s effect on the matrix subject within that setting, whereas the lack of it in the (b) examples more likely evokes a reading in which a statement is made about a person’s intellectual knowledge or awareness of the time at which the vehicle went by, with little if any emphasis on the action’s strong physical effect on the setting. This can be seen in (31), for example, where the most likely interpretation of (31a) evokes the sense of the train’s having a physical effect on the surrounding environment and the people in it as it speeds by. Such an interpretation, while possible, is unlikely in (31b), where the lack of a cataphor and its evocation of the setting emphasizes a reading in which the train’s physical effect on bystanders is less important than the matrix subject’s mental awareness of the time at which it goes by. Note also how (34a) accentuates the psychic’s

Cataphoric pronouns as mental space designators

physical awareness of the effect of the UFO’s landing (as if he were actually there), whereas (34b) plays up the psychic’s mental ability to perceive the time when the UFO landed, without his being physically present or affected by the landing (which is presumably possible if the person is a decent psychic).¹⁸ Before going on it is imperative to understand and reiterate the patterning in these data, which gives an important clue about their explanation with respect to the function of the cataphor. The (a) examples in (31)–(34) strongly evoke the sense that the actions described in the subordinate clauses prefigured by cataphoric it have some kind of physical effect on the matrix subjects above and beyond whatever mental effects they might produce; it is difficult, if not impossible, to interpret them as meaning only that the matrix subjects have some kind of intellectual or mental understanding of the subordinate clause events. The (b) examples, on the other hand, while tolerating the interpretation that the subordinate clauses exert a physical effect on the matrix subjects, are much more naturally interpreted as meaning that the subordinate events are comprehended intellectually by the matrix subjects. Significantly, the (b) examples do not contain cataphoric pronouns which separate the matrix clause subjects from the subordinate clause events. We have assumed within mental space theory that emotionally neutral verbs evoking intellectual activity, like know and believe, set up mental spaces which are not ordinarily designated by cataphoric pronouns except under circumstances of special emphasis, as in example (28) above. And, even in cases of emphasis, the emphasis is usually of a purely intellectual kind. Nor do the other matrix predicates in (31–34) above ordinarily occur with cataphoric it when their meanings evoke purely intellectual activity. A moment’s reflection, however, confirms that in these data the meanings of the matrix verbs when used with cataphoric pronouns in the (a) examples are somewhat different from the meanings of these verbs in the (b) examples (when cataphors do not occur): they have been extended in the (a) examples to the perception or understanding of events in the concrete physical domain from their likely more basic senses in the abstract domain of purely intellectual activity which is typically reflected in the (b) examples. But my claim is that even in this extended meaning the matrix verbs are still space builders, with the proviso that the mental spaces set up by the matrix verbs now evoke knowledge structures relevant to the perception of physical events in addition to the usual knowledge structures associated with purely intellectual events. Thus, when the speaker wants to focus attention on perception of an event in the physical domain or setting (by using know, be aware of, etc.) he can accentuate the influence of that event’s occurrence in (and possibly on) the setting in which it occurs by designating the mental space set up by that verb (or verbal construction in the case of be aware of) with cataphoric it. The use of it to designate the space set up by the space builder highlights the importance of the physical location of the event with these verbs and thus the event’s effect on the matrix subject. Clearly, this sense can also be obtained without the use of the pronoun, but this latter reading is dispreferred in favor of the one evoking perception in the mental domain. For example,

83

84

Michael B. Smith

in the context of (31a) the matrix verb know means something like “be physically (as opposed to intellectually) aware of something in the physical domain”, and so the cataphoric pronoun it‘s designation of the boundaries of the space set up by the verb is now understood as much in physical terms as intellectual terms in its evocation of the actual boundaries of the physical location in which action occurs. This is why the (a) examples in (31–34) evoke so clearly the idea that the events described in the subordinate clauses have some kind of physical effect on the matrix subject, whereas in the (b) examples the relationships between the entities is purely mental in nature (under the usual interpretation). These observations are confirmed in the following additional examples. Note that in (35a) emphasis is placed on the traveler’s actually being present when the plane landed (with the traveler’s physical awareness of the landing event accentuated), whereas in (35b) the cataphor is impossible if the seeing focuses on the traveler’s intellectually ascertaining, by visual means, the time at which the flight landed. (35) a. The traveler could see it when the flight landed. b. By using the timetable the traveler could see (*it) when the flight landed. A similar distinction is found in another example with hear, as shown below in sentence (36): (36) a. The passengers were able to hear it when the train arrived. b. By listening to announcements over the loudspeaker the passengers could hear (*it) when the train arrived. Here, too, the cataphor is only possible when emphasis is placed on the physical effect of the train’s arrival on the passengers (36a). When the context is changed to emphasize the intellectual understanding of the time at which the train arrives, cataphoric it is impossible, since the relevance of the physical setting in this case (prefigured by the mental space designated by it) is less important.

6. The potential space designating function of another kind of cataphor As is well known, both English it and German es occur as so-called expletive (extraposition) syntactic-subject pronominals in constructions like those illustrated in (37) and (38) below. (37) a. It is likely that the governor will run for president. b. It is certain that Gore and Bush will have a close race. (38) a.

Es ist sicher, daß Inge morgen ankommt. it is certain that Inge tomorrow arrives ‘It’s certain that Inge will arrive tomorrow.’

Cataphoric pronouns as mental space designators

b. Es ist interessant, daß Hans sein Auto verkauft hat. it is interesting that Hans his car sold has ‘It’s interesting that Hans sold his car.’ I think that analyzing these cataphoric pronouns as mental space designators similar to those which prefigure spaces built up by propositional attitude predicates (such as verbs and adjectives) helps explain their semantic function in these clauses, and provides an argument against the usual view that they are just meaningless syntactic place-holders. While it is likely true that their syntactic-subject place-holding function is due in part to heavy noun phrase shift (reflecting phonological constraints on the length of sentential subjects), these pronouns may serve the additional semantico-discourse function of delaying overt mention of a clause designating a proposition which describes the relationships holding in a mental space (which might otherwise appear as a sentential subject) by setting up mental spaces (which they also designate) as semantic placeholders that remain empty until the speaker’s attitude toward or belief about those relationships is specified by the following adjectives. For example, in (37a) above the cataphor it would be analyzed as setting up a mental space which is then “tagged”, so to speak, with the speaker’s judgment about the likelihood of the following proposition taking place. In effect, a kind of “likelihood” space is created. By affording speakers the option of mentioning their attitude about a proposition before uttering the proposition itself, this construction type may reflect — in the grammar — a self-centered human tendency to rank one’s opinions about events as just as important (if not more so) as the events themselves. Undoubtedly, much more can and should be investigated along these lines, but space limitations preclude this here.

7. Concluding remarks I have argued in this chapter that the idea of analyzing cataphoric pronouns as mental space designators in English, German, and Russian leads to a revealing account of their meaning and function in grammar and discourse which has eluded discovery until now. The resulting conclusion is that cataphors are not arbitrary, meaningless grammatical devices, but are rather semantically motivated elements in the grammar. The chapter thus contributes additional evidence for the growing realization that many hitherto unnoticed aspects of grammar manifest iconic links between the linguistic forms and the meanings conveyed by the forms (cf. Haiman 1983, 1985a, 1985b, Givón 1990, 1995, and Greenberg 1985 for more discussion). Clearly, more work needs to be done to make the analysis begun here more detailed and convincing. This would include finding additional data in English and German to further support the points made in the earlier sections of the chapter, and to find arguments for the analysis in Russian which go beyond its intuitive

85

86

Michael B. Smith

appeal and apparent fit with the meanings of the matrix predicates presented in (7) above. More work also needs to be done to further justify applying the analysis to the cataphoric subject pronouns exemplified by the English and German extraposition data in the preceding section. And, finally, data from other languages with cataphoric pronouns should be investigated to see whether they, too, might be insightfully characterized in the way I have suggested here.

Notes 1. I would like to acknowledge the valuable contribution of both the editors of this volume and several anonymous referees, whose suggestions have greatly improved the final version of this chapter. Unfortunately, time and space limitations have prevented me from following all of their advice. Consequently, I assume full responsibility for any errors remaining in the finished product. 2. In this respect, cognitive linguistics owes a lot to the Saussurean approach to grammar known as the “Columbia School” and is largely compatible with its basic tenets. Some representative works in this framework include Diver (1975), Garcia (1977), and Kirsner’s monograph on Dutch presentative sentences (1979). More recently, Huffman (1997) deals specifically with pronouns. 3. This quotation is the closest language approximating an actual formal definition of a mental space that I can find in Fauconnier’s work. I think he felt, as do I, that the notion is an intuitively natural one within the general cognitive approach to linguistics. 4. Mental spaces are theoretically similar to Langacker’s notion of abstract settings(1991b), which have been shown to play an important role in linguistic semantics (cf. Langacker’s discussion of existential there-clauses in English (1991b: 352–355), analyses of German and Russian impersonal setting-subject constructions found in Smith (1985a, 1985b, 1993a, 1993b, 1994) and parts of Smith (1987)). These analyses build on Langacker’s basic insight that the setting in which a state or event is conceived to occur can sometimes be accorded grammatical status via coding as the clausal subject (Langacker 1991b: 345ff.), as in such English sentences as Tiger Stadium has witnessed many exciting baseball games and My dog is crawling with fleas, in which the setting subjects are relatively concrete. A somewhat more abstract setting subject is exemplified by the German sentence Es regnet heute and its English translation “It’s raining today”, in which the pronominal subjects es and it, respectively, designate the setting construed as the surrounding environment within which the weather is conceived to occur. Mental spaces extend this notion to the logical extreme in which the setting and/or space is a purely mental construct. See also Kirsner’s classic work on Dutch presentative sentences (1979). 5. Note that mental space researchers typically use any of the following terms to refer to the creation of a mental space by a space builder: build, create, set up, construct, etc. These terms and their synonyms should be interpreted by the reader as signifying the same kind of process, i.e. the creation of a mental space by a space builder. 6. An anonymous reviewer wondered if there are any kinds of objective tests or evidence that certain expressions build mental spaces. My understanding of the concept as employed in mental space research is that the theory assumes that an expression which evokes an appropriate meaning involving some kind of relevant knowledge structure or domain builds

Cataphoric pronouns as mental space designators a corresponding space to represent that structure, and that the validation of invoking the notion of mental spaces resides in their usefulness in explicating a wide range of linguistic phenomena in an intuitively natural way. 7. As noted earlier, Pütz (1986) presents evidence for just such a division, but does not explain the factors determining which verb belongs to which class. I will return to this question later in the chapter. 8. Readers who are familiar with German grammar might wonder whether the phenomenon under investigation in this chapter extends to constructions in which the subordinate clause functions syntactically as a prepositional object. German has many complex verbal constructions in which the verb co-occurs with a lexically-determined preposition (and in some cases also a reflexive (REFL) pronoun that is coreferential with the subject), e.g. sich freuen auf “look forward to”, sich verlassen auf “rely on”, sich erinnern an “remember”, denken an “think of ”, etc. When these verbs take sentential objects, the pronominal element da (not es) cataphorically anticipates the subordinate clause and is attached phonologically to the preposition in question. Here is an example of such a da-compound: (i) Ich habe mich darüber sehr gefreut, daß Peter gekommen ist. I have REFL it-over very been-happy that Peter come is ‘I was very happy (about the fact) that Peter arrived.’ It would appear that da in this kind of prepositional object construction is serving the same function as es in data like that in (6) above, with the main difference the fact that the da-compounds are generally not omissible, in contrast to es in some circumstances. Due to space limitations I will not mention da-compounds further in this chapter. 9. While it is true that the pronoun in example (7d) seems to be required by the presence of a preposition which would otherwise be stranded without an object, this specific grammatical function of the pronoun need not preclude its possible cataphoric discourse function, as well. It should also be noted that, as is often the case, there may be differences of opinion among native speakers as to whether the pronouns (and their sometime accompanying prepositions) in (7a–c) are required, or not. My claim is that, if the pronouns appear, they are cataphoric and designate mental spaces. 10. Schmid considers a larger variety of construction types containing shell nouns than I will discuss in this chapter. My use of the basic type illustrated in (8) will be sufficient to relate this concept to my own notion of mental space designators. 11. An anonymous reviewer suggests that the cataphoric pronouns also seem to effect the kind of temporary reification like that mentioned by Schmid and I would tend to agree. 12. I will not deal further with the Russian data, primarily for reasons of space, but also because I do not yet have arguments for the mental space designator analysis in Russian which are as convincing as those I have found for English and German. Nevertheless, I am convinced that the analysis can be extended to Russian and to other languages, as well, and that the Russian data presented here exemplify cataphoric pronouns that function as mental space designators. 13. An anonymous reviewer notes that “there will be discourse differences between Es war ein guter Arzt and Er war ein guter Arzt” in example (10) (and similarly between es and sie in example (11)). While this may be true, the difference is not relevant to my present argument. The point of the possibility of using es in place of er or sie in these examples is that the meaning of es has been extended to include these non-prototypical kinds of reference.

87

88

Michael B. Smith 14. Cf. one of the senses of -er in a word like kegger, where the usually agentive suffix is used to derive a noun from a noun whose meaning is associated — by a similar process — with a reified event, not a participant in the event. (A kegger is a kind of party event often put together by college students in the United States in which the consumption of beer — which comes in kegs — is conspicuously prodigious.) An example involving an actual setting sense of -er would be the noun sleeper in its sense as a car on a passenger train (hence a location or setting) in which people sleep. 15. English also occasionally tolerates cataphoric it when the following subordinate clause is construed as especially desirable, as in sentences like I love it when she really prepares her speeches and He enjoys it when the students participate. 16. It should be clarified here that, under the proposed analysis, space builders invariably set up mental spaces, whether the spaces are designated by cataphoric pronouns or not. What the cataphors are proposed to do is to highlight, for some reason, the salience of a mental space which has been introduced into the discourse by the space builders. 17. Interestingly, this suggests that in (26–28) the pronouns may be serving both anaphoric and cataphoric functions in the same sentence. It should also be kept in mind that there may often be multiple motivations for conceptual distancing which will depend on the meanings of the matrix verbs. Note, for example, that the distancing of a subordinate clause with es because it is construed as particularly desirable or noteworthy (as if putting it on a pedestal), which was described in the context of (18) above, could also be motivated as a kind of emphasis, as well. In most cases it is difficult (and perhaps not all that important) to tease apart these mutually supporting motivations for using the cataphor to distance the subordinate proposition from the matrix subject, since they mutually support one another. 18. Apparently the only wh-word that can be used to introduce the subordinate clause in these examples is when, which creates a context which, together with the cataphoric pronoun, brings about the interpretation that the action designated by the clause has an effect on the matrix subject. Note that substituting any other wh-word in place of when in the sentences in (31–34) does not have this effect. While this may be a minor, possibly idiosyncratic, fact about English, the examples can still be used to make the point that the cataphors in these expressions designate mental spaces.

References Bresnan, Joan (ed.) 1982. The Mental Representation of Grammatical Relations. Cambridge, MA: MIT Press. Chomsky, Noam. 1981. Lectures on Government and Binding. The Pisa Lectures. DordrechtHolland/Cinnaminson-USA: Foris. Chomsky, Noam. 1986. Knowledge of language. Its nature, origin, and use. New York: Praeger. Cook, V.J. and Mark Newson, (eds.). 1996. Chomsky’sUniversal Grammar. An Introduction. 2nd edition. Oxford: Blackwell. Culicover, Peter W. 1997. Principles and Parameters: An Introduction to Syntactic Theory. Oxford: Oxford University Press. Diver, William. 1975. “Introduction.” In Columbia University Working Papers in Linguistics 2, William Diver (ed.), 1–20. New York: Columbia University. [Corrected reprint 1980]

Cataphoric pronouns as mental space designators Fauconnier, Gilles. 1985. Mental Spaces: Aspects of Meaning Construction in Natural Language. Cambridge, MA: MIT Press/Bradford. Fauconnier, Gilles. 1997. Mappings in Thought and Language. Cambridge: Cambridge University Press. Fauconnier, Gilles and Eve Sweetser (eds.). 1996. Spaces, Worlds, and Grammar. Chicago: University of Chicago Press. Fillmore, Charles J. 1985. “Frames and the semantics of understanding.” Quaderni di Semantica 6(2): 222–254. Garcia, Erica. 1977. “On the practical consequences of theoretical principles.” Lingua 43: 129– 170. Gernsbacher, M.A. 1990. Language Comprehension as Structure Building. Hillsdale, NJ: Erlbaum. Givón, T. 1990. Syntax: A Functional-Typological Introduction, vol. 2. Amsterdam/Philadelphia: John Benjamins. Givón, T. 1995. Functionalism and Grammar. Amsterdam/Philadelphia: John Benjamins. Goffman, Erving. 1974. Frame Analysis. New York: Harper and Row. Greenberg, Joseph H. 1985. “Some iconic relationships among place, time, and discourse deixis.” In Haiman (ed.) 1985a, 271–287. Haegeman, Liliane. 1994. Introduction to government and binding theory. 2nd edition. Oxford/Cambridge: Blackwell. Haiman, John. 1983. “Iconic and economic motivation.” Language 59: 781–819. Haiman, John. (ed.). 1985a. Iconicity in Syntax. Amsterdam/Philadelphia: John Benjamins. Haiman, John. 1985b. Natural Syntax. Iconicity and Erosion. Cambridge: Cambridge University Press. Huffman, Alan. 1997. The Categories of Grammar: French lui and le. Amsterdam: John Benjamins. Kirsner, Robert S. 1979. The problem of Presentative Sentences in Modern Dutch. [North-Holland Linguistic Series43]. Amsterdam/New York: North-Holland. Lakoff, George. 1987. Women, Fire, and Dangerous Things: What Categories Reveal About the Mind. Chicago: The University of Chicago Press. Lakoff, George and Mark Johnson. 1980. Metaphors We Live By. Chicago: The University of Chicago Press. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, vol. 1: Theoretical Prerequisites. Stanford: Stanford University Press. Langacker, Ronald W. 1991a. Concept, Image, and Symbol. The Cognitive Basis of Grammar. [Cognitive Linguistics Research1]. Berlin: Mouton de Gruyter. Langacker, Ronald W. 1991b. Foundations of Cognitive Grammar, vol. 2: Descriptive Application. Stanford: Stanford University Press. Langacker, Ronald W. 1993. “Reference-point constructions.” Cognitive Linguistics 4: 1–38. Langacker, Ronald W. 2000. Grammar and Conceptualization. [Cognitive Linguistics Research 14]. Berlin: Mouton de Gruyter. Lederer, Herbert. 1969. Reference Grammar of the German Language. New York: Charles Scribner’s Sons. Leed, Richard, Alexander Nakhimovsky and Alice Nakhimovsky. 1982. Beginning Russian, Volume 2. Columbus, OH: Slavica Publishers. Nunberg, Geoffrey. 1978. The Pragmatics of Reference. Bloomington, IN: Indiana University Linguistics Club.

89

90

Michael B. Smith Perlmutter, David M. (ed.) 1983. Studies in relational grammar 1. Chicago and London: The University of Chicago Press. Perlmutter, David M. and Carol G. Rosen (eds.) 1984. Studies in Relational Grammar 2. Chicago and London: The University of Chicago Press. Pütz, Herbert. 1986. Über die Syntax der Pronominalform es im modernen Deutsch. Tübingen: Gunter Narr Verlag. Radford, Andrew. 1997. Syntactic Theory and the Structure of English: A Minimalist Approach. Cambridge: Cambridge University Press. Sag, Ivan A. and Thomas Wasow. 1999. Syntactic Theory: A Formal Introduction. Stanford: CSLI Publications. Schmid, Hans-Jörg. 1999. “Cognitive effects of shell nouns.” In Discourse studies in cognitive linguistics, Karen van Hoek, Andrej A. Kibrik, and Leo Noordman (eds), 111–132. Amsterdam/Philadelphia: John Benjamins. Smith, Michael B. 1985a. “An analysis of German dummy subject constructions in cognitive grammar.” In Proceedings of the First Annual Meeting of the Pacific Linguistics Conference, Scott Delancey and Russell S. Tomlin (eds), 412–425. Eugene, OR: Department of Linguistics, University of Oregon. Smith, Michael B. 1985b. “Event chains, grammatical relations, and the semantics of case in German.” Papers from the Twenty-First Regional Meeting of the Chicago Linguistic Society, 388–407. Chicago: Chicago Linguistic Society. Smith, Michael B. 1987. The Semantics of Dative and Accusative in German: An Investigation in Cognitive Grammar. Ph.D dissertation, University of California at San Diego. Smith, Michael B. 1993a. “Aspects of German clause structure from a Cognitive Grammar perspective.” Studi Italiani di Linguistica: Teorica e Applicata 22: 601–638. Smith, Michael B. 1993b. “Cases as conceptual categories: Evidence from German.” In Conceptualizations and Mental Processing in Language. [Cognitive Linguistics Research 2], Richard A. Geiger and Brygida Rudzka-Ostyn (eds), 531–565. Berlin/New York: Mouton de Gruyter. Smith, Michael B. 1994. “Agreement and iconicity in Russian impersonal constructions.” Cognitive Linguistics 5: 5–56. Smith, Michael B. 2002. “The Polysemy of German es, iconicity, and the notion of conceptual distance.” Cognitive Linguistics 13: 67–112. Sweetser, Eve. E. 1988. “Grammaticalization and semantic bleaching.” Berkeley Linguistics Society 14: 389–405. Taylor, John R. 1995. Linguistic Categorization: Prototypes in Linguistic Theory. 2nd edition. New York: Clarendon Press/Oxford. Van Valin, Jr., Robert D. and Randy J. LaPolla. 1997. Syntax: Structure, Meaning, and Function. Cambridge: Cambridge University Press. Webelhuth, Gert (ed.) 1995. Government and Binding Theory and the Minimalist Program. Principles and Parameters in Syntactic Theory. Oxford: Blackwell.

PART II

Theoretical issues in classical sign-based linguistics

CHAPTER 3

Monosemy, homonymy and polysemy* Wallis Reid Rutgers University

1. Introduction Linguistic tradition has a three-way typology for describing the outcome of the semantic analysis of a lexical or grammatical form. The analysis may posit a single meaning for the form, as one would do for the English word barn, ‘a large wooden structure on farms for animals’. Charles Ruhl (l989) has dubbed this monosemy. Alternatively, the analysis may treat the form as two or more independent linguistic units and posit distinct and unrelated meanings for each. The traditional name for this is homonymy, and a classic example is the two English words seal the animal, and seal the stamp. Finally, the analysis may treat a form as a single linguistic unit and posit a series of semantically related meanings, or senses, for it. John Taylor (1995: 103) cites the word neck, meaning both ‘neck of a human body’ and ‘neck of a bottle’ as a clear example of what is traditionally called polysemy. Columbia School analysts have posited monosemy and homonymy, but not polysemy.¹ How so? In this chapter I will sketch out an answer. But explaining why one is not doing something puts one in an unavoidably defensive position. Better to keep theory to a minimum at first and illustrate how a Columbia School analysis handles as monosemy data that would seem to call for polysemic treatment. The body of this chapter will thus consist of a monosemic analysis of three semantically related English words. I will make theoretical points foreshadowing my case against polysemy in passing, but will postpone a direct discussion of the issue until the end of the chapter. One theoretical point must be made at the outset, however, precisely because the existence of polysemy is sometimes not recognized as a theoretical issue. “Any dunce can see that neck has two senses” some might say. But this is because we have internalized the practical but unprincipled dictionary treatment of words at an early age. Embracing that tradition as a linguist, however, commits one to a particular view of language. To be sure, speakers can interpret an utterance holistically without theoretical commitments. But determining the precise semantic contribution of each lexical and grammatical form to that interpretation is an act of linguistic analysis, not observation.²

94

Wallis Reid

Polysemy appears to be a directly observable phenomenon only if one makes two familiar assumptions about the essential nature of language. The first is that language is a representational system of sentential meaning.³ The second assumption is the compositional view of sentence meaning. In the compositional view, the literal meaning of a sentence is assumed to consist of the meanings of the individual words amalgamated in a way determined by the grammatical structure of the sentence. In short, the semantic whole must equal the sum of its semantic parts.⁴ Adhering to the representational view of language and the compositional view of meaning obliges the analyst to attribute every notional fraction of a completed interpretation to some specific form. This requirement can only be met by assigning multiple semantic values to most forms. So, for example, in his wrinkly neck was sunburned and she grabbed the bottle by its neck, neck must necessarily be assigned two different values if the meaning of those sentences is to be totally derived from their contributing words, among them neck. By contrast, Columbia School makes neither of these two assumptions. It begins with the fact that language functions as a communicative system but leaves open to empirical investigation whether or not it does so by virtue of semantically representing individual communications. This means the Columbia School analyst is not committed at the outset to making the semantic parts equal a semantic whole; they may or they may not. Analytical experience has led us to the conclusion that the communicated whole — the message — is always more than its semantic parts. Language functions not by formally representing the message but by providing hearers with notional fragments, clues, hints, and prompts for them to synthesize against the background of their unique cognitive, linguistic and life experience. This synthesis is a creative interpretive act, one that is not reducible to a mechanical algorithm. Alan Huffman (1997: 16–19) has dubbed this an instrumental concept of meaning. The meanings of linguistic forms contribute in various ways to the communication of a message without collectively composing it. Returning to the problematic word neck, the Columbia School analyst must decide whether a meaning can be posited for it that could serve to refer to both a body part and a bottle part; or, alternatively, decide to treat the word as two (or more) independent semantic units — two linguistic signs — each with its own meaning. But this decision is guided not by representational and compositional assumptions but by the general principle underlying all theory construction: namely, posit only as many theoretical constructs as required to account for the observations.

2. English IN, AT, and ON Regretfully, I currently have no solution to offer for neck. Instead, I will present an analysis of in, at and on, three words that allow a more extensive exploration of the issues raised in the introduction above. Appearing below is an initial hypothesis of the semantic values of in, at and on.⁵

Monosemy, homonymy and polysemy

in = three-dimensional location at = point-like location on = linear or planar location In, at and on all signal spatial location, but with differences. In designates location in three-dimensional space. At designates location at a point. And on designates location on a line or plane. Consider examples (1), (2) and (3); (1) He lives in New Brunswick. (= three dimensional location) (2) He stopped at the traffic light. (= point-like location) (3) He was shopping on Main Street. (= linear or flat location) A city is clearly three-dimensional; a traffic light is reasonably point-like; and a street is flat and linear. In examples (1), (2) and (3), in, at and on seem to be echoing information already implied by the words light, street and New Brunswick. In examples (4), (5) and (6) the three words function non-redundantly, making us focus on different aspects of a car all by themselves: (4) She left her briefcase in the car. (5) She left her briefcase on the car. (6) She left her briefcase at the car. We understand a briefcase left in the car to be inside the car, a three-dimensional location. On the car suggests the hood or roof of the car, both planes. At the car suggests next to the car, on the ground perhaps, but not inside and not on the roof. Why so? The explanation begins with the fact that an automobile is not actually pointlike; it’s an object unmistakably occupying three-dimensional space. At, by designating a point-like location, implies that the dimensional extensions of the car are irrelevant to the message being communicated, thus disfavoring the ‘inside’ and ‘hood’ interpretations. Operative here is the systemic aspect of these three words. Because of the availability of all three options, the choice of one implies the inappropriateness of the other two. We don’t understand ‘at the car’ as including the seat or roof, because if the briefcase were in either of those two places, one would have said so. At work here is the Gricean Maxim of Quantity: “make your contribution as informative as is required (for the current purposes of the exchange)” (Grice 1967: 45). In brief, at implies ‘not an in-like location and not an on-like location’. This interrelatedness of at, on and in is best captured by putting them into a system, shown in Figure 1. At, on and in are in a single system because their meanings contrast with each other along a single semantic parameter: number of dimensions; and because the existence of each affects the use and interpretation of the others. Note that both at and in are precise in their dimensional meanings. But on is imprecise; it designates a range of dimensions, those falling between zero and three. And since zero and three dimensions are the meanings of at and in, this makes on quite literally ‘not at

95

96

Wallis Reid

Number of DIMENSIONS OF LOCATION

at: ZERO (0) on: MORE THAN ZERO AND LESS THAN THREE (0 < _< 3) in: THREE (3) Figure 1.

and not in’. Defining on relationally this way would not be feasible if on were being treated in isolation, independent of its relation to at and in. Here, of course, one could question the attractiveness of such a definition in the first place. Its appeal for the polysemy-shunning Columbia School analyst is clear. In my first, informal characterization of on I said it designated location on a line or a plane. That made it sound as if on has two meanings. My more formal characterization — a location falling between zero and three dimensions — succeeds in positing a single, albeit imprecise, meaning for on. But why should one imprecise meaning be preferable to two precise meanings? Let me answer somewhat obliquely for the moment, postponing the fuller discussion for later. Imprecision is inescapable in language because no definition specifies along all possible descriptive parameters. Consider (7) and (8): (7) He threw the keys on the grass. (8) He threw the keys in the grass. Asked how they differ, informants usually say that the grass in (7) is short, a lawn perhaps, while the grass in (8) is tall, a field perhaps. Does the word grass have two senses, ‘tall grass’ and ‘short grass’? I suspect most linguists would feel uneasy about positing two different senses for grass and opt instead to say that grass doesn’t specify any particular height as part of its meaning; it is neutral for height. Moreover, my analysis of on and in offers a plausible explanation for the different interpretations of grass. Tall grass is unmistakably three-dimensional, whereas the height of short grass can be overlooked, making a well-mown lawn appear to be flat. And the difference between a three-dimensional location and a two-dimensional location is the semantic boundary between in and on according to Figure 1. What this means then, is that specifications of ‘grass’ come from elsewhere in the context. The meanings of in and on are shaping its interpretation, fleshing out detail that is not inherent in the word grass itself. In the case of grass, at any rate, an imprecise meaning not marked for height seems intuitively right. To be sure, an alternative formal treatment is possible. Posit two senses for grass and then have the meanings of in and on select the appropriate sense. This is an ideal solution if one were writing a mechanical translation program because it capitalizes on something a computer can do very well, store and select. But as a picture of how natural language works it is less satisfactory because it fails to capitalize on something human beings can do very well: solve problems and make sense of things, such

Monosemy, homonymy and polysemy

as figuring out that one way grass can be a three-dimensional location is to be tall. By contrast, the mechanical solution, so appropriate for the computer, is quite literally ‘mindless’. Returning now to the issue at hand, the meaning posited for on in Figure 1 does not distinguish between one and two dimensions. The rationale is the same in spirit as that for making grass unspecified for height: assigning two different meanings, or senses, to on would be building precision into the language at the wrong place. Whether an on-like location is interpreted as a line (‘the bird on the wire’) or a plane (‘the cat on the mat’) is always decided by something else in the context. On by itself merely signals the semantic outer boundaries of its interpretation — constraining its interpretation without determining it — which is precisely the value assigned to it in Figure 1.

3. The Role of conceptualization In examples (1)–(8) in, at and on have functioned as close as language gets to objective description of the physical world. But in truth, the connection between language and the world is always mediated by a particular conceptualization by the speaker, even when speaking about the physical world.⁶ The role of speaker conceptualization is illustrated in the examples below: (9) Ricardo grew up in Cuba. (10) Ricardo grew up on the island of Cuba. Example (9) employs in because a country is clearly three dimensional. So why on in (10)? An island is objectively as three-dimensional as a country; but what makes an island an island is being surrounded by water, and that defining property is often envisioned in a two-dimensional, map-like representation. On specifies how many spatial dimensions are relevant to the message. (See Leech [1969, p. 162] for this point that the assessment of dimensions is subjective.) Usually cities are categorized as three-dimensional locations: I live in Boston. But if the message is one of contrastive location — here rather than there — then a city will be shrunk to a point with at. The message is simply ‘this place rather than that place’ and the actual dimensions of the place are irrelevant. Thus we say ‘the State University of New York at Plattsburgh’ because there are branches in other cities as well (e.g. at Stonybrook). Example (11), taken from the Brown Corpus, nicely illustrates this contrast between in and at with cities. (11) Arnold Palmer has been a blazing figure in golf over the past twelve months . . . He was heralded as “Sportsman of the Year” by Sports Illustrated, and last night was acclaimed in Rochester as the “Professional Athlete of the Year” . . . But he also achieved something that endeared him to every duffer who ever flubbed a shot. A couple of weeks ago, he scored a

97

98

Wallis Reid

monstrous 12 on a par-5 hole. It made him human. [Episode recounted] . . . He’ll even bull head-on into the rules when he is sure he’s right. That’s how he first won the Masters in 1958. [Episode recounted] . . . Until a few weeks ago, however, Arnold Palmer was some god-like creature who had nothing in common with the duffers. But after that 12 at Los Angeles he became one of the boys, a bigger hero than he ever had been before. Brown 34101 Los Angeles is categorized as a point-like location with at because that highlights its spatial contrast with other host cities in which Palmer has played. None are actually named in this story, but the first sentence speaks of Palmer’s performance over the past twelve months and the paragraph prior to the phrase at Los Angeles has recounted an episode at another location. On the other hand, there is no hint of contrast in the location of the awards ceremony, so the three-dimensional experiential reality of the location comes to the fore: in Rochester. Capturing such subjective ways of using in, at and on via a multiple sense treatment is awkward at best. Consider examples (12) and (13). (12) He put the leaf on the table. (13) He put the leaf in the table. Now the referential difference between a ‘tree leaf ’ in (12) and a ‘table leaf ’ in (13) may well call for a homonymic analysis of leaf. By contrast, no referential difference exists for the word table. But consider the striking difference in conceptualizations: the table is a flat supporting surface in (12) and a three-dimensional object in (13). And a table can function as a point-like object as well, as in ‘he sat down at the table’. To enter these three interpretive possibilities as part of the semantic description of the word table seems a rather pointless exercise since it is adding nothing semantically beyond what at, on and in already provide. Examples (12) and (13) specify what in and on apply to: a ‘table’. Examples (14) and (15), on the other hand, leave that feature of the message entirely for the hearer to figure out: (14) He put his clothes in. (15) He put his clothes on. The three dimensional location for the clothes in (14) is likely to be understood as a washing machine (if interpreted in isolation), while the two dimensional location for the clothes in (15) is the surface of the body. The reason these features can be left out linguistically is that washing machines and bodies are two familiar places for clothes to be put, and they differ from each other in terms of dimensions. In and on are simply semantic nods, as it were, in the direction of one or the other.⁷

Monosemy, homonymy and polysemy

4. ON for linear location I want to turn now to examine on in more detail. Recall that the meaning of on — LOCATION IN MORE THAN ZERO AND LESS THAN THREE DIMENSIONS (0

male is inherently > powerful

Meaning chosen

Message communicated

(padre) LESS active (le) (hizo) (cambiar)

Son allows himself to be influenced by, i.e., cooperates with, father in a change.

predictions that we then make, as part of the task of validation, may take several forms. They may make reference to other linguistic elements in the context (English a and be favoring EP); they may call upon some characteristic of the referent on the scene (males favoring Spanish le); they may make use of non-semantic aspects of the context (Dutch demonstratives within or across sentences); or they may even exploit different types of discourse.²⁵ Columbia School must resist the siren call of the foolproof discovery procedure, the validation algorithm. Our analytical procedures cannot be formalized; each analysis will call for slightly different techniques, will motivate unforeseen predictions that suggest any variety of tests. As far as counts go, if a count is well motivated by a hypothesis and our orientations, then it offers some support for the hypothesis. But that is all; on its own, a count proves nothing. It certainly does not demonstrate the psychological reality of a categorical causal or inferential pathway to or from a linguistic meaning. To address an issue raised by Kirsner (1989: 161–162): What would it take, if not an unconfirmed quantitative prediction, to falsify a Columbia School hypothesis? The question is malapropos. A CS hypothesis is not, like a geometric theorem, a statement participating in a closed system of statements that is completely internally consistent, based on a finite number of postulates, and subject to a formal logic. Nor does it, like a chemical equation, describe the behavior of independently identifiable elements. A CS signal-meaning hypothesis can be neither proved nor disproved; it can be only accepted or rejected. If accepted, it is accepted provisionally, pending the devising of a better hypothesis. (It cannot be accepted by someone who operates with incompatible primitives, such as the sentence.) A better hypothesis will typically be one that accounts for a wider range of data or that provides a deeper understanding of existing data. Such is the history of Diver’s revision (forthcoming²⁶) of his own analysis of the Latin systems involving traditional case and voice. His eventual coming to grips with, inter alia, the deponent verbs led him to a deeper understanding of the meaning of passive verb morphology, which in turn revolutionized his treatment of the case system. He rejected his own hypothesis. The notion of communicative strategy will not furnish Columbia School with a quick and easy, surefire technique for validating our hypotheses. We shall simply have to do the hard work of studying examples, interpreting quantitative data, and

The gap between meaning and message

writing up our results in a way that causes us to believe we are making progress.

Acknowledgement I wish to express my deepest gratitude and respect for Wallis Reid, who labored intensively and at length with me and other colleagues on the question of the status of strategy in Columbia School. As this article goes to press, I must point out that Reid (1995) did not speak of “inferential process,” as used here (see p. 156). In using this term, I have essentially, and perhaps recklessly, conflated the activities of speaker and hearer as each bridges the meaning– message gap. Reid (1995), as nevertheless accurately characterized here, considered only the speaker’s perspective. My thanks to Wallis Reid for bringing this to my attention.

Notes 1. The New York Times, Dec. 8, 1999, pp. A1, A15. 2. Diver was well aware of such examples. One of his favorite examples of semantic incompatibility was a crossroads, juxtaposing singular with plural. 3. Diver often spoke of the phrase a certain person in this light. 4. See Davis (2002) for further discussion of texts that go against predictions. 5. Reid (1995: 117–128) identifies a third rationale for Diver’s counts, the “meaning suitability rationale.” Reid says this rationale, not requiring the “forced marriage” assumption, would have been acceptable, but it did not catch on in the development of Columbia School. This rationale would require that the analyst know in advance which messages are more and less frequently expressed. It is not clear how the analyst should come to possess such knowledge. Moreover, if one did possess it, then a count would be superfluous. 6. I notice, by the way, that the grammar check on my word processor does not highlight the sentence as being problematic. 7. One might well question whether it is wise in the first place to base the establishment of a fundamental theoretical construct solely upon one person’s analysis of one non-occurring construction (as we shall see) in one language. 8. Current research being conducted by the present writer on the Italian cognate si suggests that it may be a mistake to equate the Spanish message with the English Pedro bathed himself; closer to the Spanish would be Pedro bathed. 9. Reid (1995: 149–150) rightly foresees unhappy consequences of the elevation of strategy to an equal status with meaning, chiefly a loss of unity in analysis. 10. See Menn and MacWhinney (1984) for a treatment of the general problem, including *se se among many others cross-linguistically; theirs, too, is a treatment that appeals to “inconvenience for language processing.” See also Grimshaw (1977). 11. In 1975, CS theory as we know it was all of six years old and still very much trying to fight its way out of the cradle of traditional grammar. Traditional and neo-traditional generative grammar was as much García’s target as was Spanish se, as is evident both from the text and from the title. Later, García (2002) would explore the general phenomenon of nonoccurring combinations of clitics, and would do so without employing the (reified) construct of strategy.

7

72

Joseph Davis 12. See Davis (1995) for justification of these morphemic analyses. 13. This analysis is tentative. The Referent-Number interlock is independently justified by other clitics (mi, ti, lo/la/li/le). The adverbial uses here attributed to the system of Attention to a Circumstance are not likely to be successfully ascribable to Discourse Referent, since ‘us’ and ‘you’ are often not applicable in messages of ‘here’ and ‘there.’ 14. Phonology per se must not be the problem but rather the processing of phonologically identical, closely satellite signals; witness the Latin reduplicated perfects such as pepigi ‘I fixed.’ 15. In English, we have to insist, with the -self forms, that we really do mean two roles (Stern 2001). 16. Garcia (1975) did treat *le lo in terms of inferential complexity and processing overload (cf. pp. 414, 463), so her treatment of *se se, even at this stage, must be seen in that context. Garcia (2002) explicitly integrates *se se into the general phenomenon. 17. Kirsner’s (1989: 165) first prediction rests upon “the plausible assumption that the beginning of the sentence acts as a point of ‘natural discourse focus’ where [writers] would put noun phrases whose referents are especially deserving of attention.” The assumption may be unfounded. In a short text analyzed by the present writer (“Hercules” in Thomas Bulfinch, Bulfinch’s Mythology, New York: Avenel, 1979, 143–150), the median position of the first reference to the hero in a sentence is 3, not 0, words away from the beginning of the sentence. A more useful understanding of sentence may be that it is a unit of text; this view is consonant, moreover, with Kirsner’s competing, subsuming prediction. In this case, initial position would be a point of re-orientation of the reader. Two common ways to re-orient are to provide transitions (but, so, another) and to re-introduce something familiar. Of course, the most familiar element in the story will be the hero (Hercules claims initial position in 21 of 64 sentences). 18. Note too, in passing, that for Reid, a strategy is a hypothesis. 19. Certainly we should be suspicious when we see that the number of excess tokens in the two “+” cells will always be equal. Why should there always be exactly the same number of instances when the speaker chooses on the basis of maleness and on the basis of femaleness? Reid fails to see that the ±30 merely reflects the strength of the skewing, nothing more. 20. The rationale would be that something known is easier to focus on, while the hearer will be unprepared to focus on something unfamiliar; and we know independently that the correlates with discourse-familiar entities and a with unfamiliar. 21. The rationale is that activity is more likely than stasis to draw attention. This count appears in an earlier, unpublished version of the study. 22. “A Horseman in the Sky,” “An Occurrence at Owl Creek Bridge,” “Chickamauga,” and “A Son of the Gods” in Ambrose Bierce, The Best of Ambrose Bierce, 1946, Secaucus: Castle. 23. The data: PE EP TOTAL the 144 14 158 a 43 22 65 TOTAL 187

36

223

PE SbV 125 be 62 187

EP TOTAL 17 142 19 81 36

223

P(a) = 65/223 = .291; P(EP) = 36/223 = .161. Calculated joint probability is then P(a)P(EP) = .047. Observed P(a, EP) = 22/223 = .099 > .047.

The gap between meaning and message P(be) = 81/223 = .363; P(EP) = .161. Calculated joint probability is P(be)P(EP) = .059. Observed P(be, EP) = 19/223 = .085 > .059. Calculated joint probability P(a)P(be)P(EP) = .017. Observed P(a, be, EP) = 13/223 = .058 > .017. 24. An analysis of Italian si, forthcoming from the present writer, was originally motivated (Davis 1999) in part by the goal of demonstrating how a CS analysis might be accomplished without an appeal to strategies. 25. As is well known, French passé composé tends towards speech, passé simple towards literature (Reid 1979: 265–276). Also well known, Italian lui tends towards speech, egli towards literature (Davis 1992: Chapter VI). 26. The relevant unpublished manuscripts are “The Focus-Control Interlock in Latin,”“Latin Is a Language,” and “Latin-T and Latin-TUR.”

References Chomsky, N. 1957. Syntactic Structures. The Hague: Mouton. Contini-Morava, E., and Goldberg, B. S. (eds). 1995. Meaning as Explanation: Advances in Linguistic Sign Theory. Berlin: Mouton de Gruyter. Davis, J. 1992. Italian egli and lui: Grammatical Meaning and Inference. Diss. Columbia U, New York. UMI 9313892. Davis, J. 1995. “The linguistic sign as unifying principle behind syntactic peculiarities: The Italian clitic ne.” In CLS 31: Papers from the 31st Regional Meeting of the Chicago Linguistic Society. Vol. 2: The Parasession on Clitics, Dainora, Audra, Rachel Hemphill, Barbara Luka, Barbara Need, Sheri Pargman (eds), 79–90. Davis, J. 1999. “On abstinence and abdication: Italian si.” Presented at the Sixth International Columbia School Conference on Linguistics, Rutgers U, New Brunswick, N. J. Davis, J. 2002. “Rethinking the place of statistics in Columbia School analysis.” In Otheguy, Reid, and Stern (eds), 65–90. Diver, W. 1969. “The system of Relevance of the Homeric verb.” Acta Linguistica Hafniensia 12: 45–68. Diver, W. 1990. “And now for something completely different.” Unpublished MS. Diver, W. 1995. “Theory.” In Contini-Morava and Goldberg (eds), 43–114. Diver, W. Forthcoming. Language as Human Behavior: The Linguistic Writings of William Diver. A. Huffman, R. Otheguy, and J. Davis (eds.) García, E. 1975. The Role of Theory in Linguistic Analysis: The Spanish Pronoun System. Amsterdam: North-Holland. García, E. 2002. “Deconstructed morphology vs. con-structive syntax.” Neuphilologische Mitteilungen 1 (103): 73–102. García, E., and Otheguy, R. 1983. “Being polite in Ecuador.” Lingua 61: 103–132. Gorup, R. 1992. “The motivation for the fixed order of clitics in Serbo-Croatian.” Juznoslovenski Filolog XLVIII: 25–34. Grimshaw, J. 1977. “The best clitic: Constraint conflict in morphosyntax.” Elements of Grammar, L. Haegeman (ed.), 169–196. Dordrecht: Kluwer. Huffman, A. 1985. The Semantic Organization of the French Clitic Pronouns: lui and le. Diss. Columbia U, New York. UMI 8511512.

73

74

Joseph Davis Huffman, A. 1997. The Categories of Grammar: French lui and le. Amsterdam: John Benjamins. Huffman, A. 2001. “The linguistics of William Diver and the Columbia School.” Word 52 (1): 29–68. Huffman, A. 2002. “Cognitive and semiotic modes of explanation in functional grammar.” In Otheguy, Reid, and Stern (eds), 311–337. Kirsner, R. 1989. “Does sign-oriented linguistics have a future?: On sign, text, and the falsifiability of theoretical constructs.” From Sign to Text: A Semiotic View of Communication, Y. Tobin (ed.), 161–178. Amsterdam: John Benjamins. Menn, L., and MacWhinney, B. 1984. “The repeated morph constraint: Toward an explanation.” Language 60 (3): 519–541. Otheguy, R. 1993. “The role of prediction in sign-based grammatical analysis.” Unpublished MS. Otheguy, R., Reid, W., and Stern, N. (eds.). 2002. Signal Meaning and Message: Perspectives on Sign-Based Linguistics. Amsterdam: John Benjamins. Reid, W. 1979. The Human Factor in Linguistic Analysis: The Passé Simple and the Imparfait. Diss. Columbia U, New York. UMI 7924964. Reid, W. 1991. Verb and Noun Number in English: A Functional Explanation. London: Longman. Reid, W. 1995. “Quantitative analysis in Columbia School theory.” In Contini-Morava and Goldberg (eds), 115–152. Stern, N. 2001. The Meaning and Use of English -self. Diss. City University of New York. UMI 3024839.

PART III

Analyses on the level of the classic linguistic sign

CHAPTER 6

The givenness of background A semantic–pragmatic study of two modern German subordinating conjunctions Zhuo Jing University of California, Los Angeles and Universität zu Köln

Most grammars of German conceive of the contrast between the conjunctions wenn and als in mixed terms, whereby the two items are considered as having a variety of discrete meanings that are related to grammatical tense, conditionality, habituality, etc. This conception is confusing on the descriptive level and non-explanatory on the analytic level. It fails to account for the actual uses of the two items in a unitary manner. This chapter proposes a semantic-pragmatic model and suggests that the two conjunctions are linguistic signs the speaker employs not to provide objective descriptive detail but rather to give instructions to the hearer about how to interpret information in the ongoing communication. This model enables us to show that the different distributional patterns of the two subordinating conjunctions in texts can be attributed to their respective instructional meanings. One invariant meaning is postulated for each conjunction. The unitary postulation at the semantic level allows a wider range of semantic-pragmatic interpretations in varying contextual situations, a problem traditional grammars are unable to cope with. This chapter shows that descriptions as they exist in German grammars can be improved on the basis of principles of “Columbia School linguistics”, formerly known as form-content analysis.*

1. Introduction This chapter focuses on two subordinating conjunctions in modern german, wenn and als, with respect to the meanings they contribute to the sentence in which they occur, as in (1)–(5). (1) Was erfahren wir, wenn wir dies genügend bedenken? ‘What will we learn if/when we consider this sufficiently?’ (Heidegger 1959: 175)

78

Zhuo Jing

(2) Ich hätte es auch nicht getan, wenn du dich nicht geweigert hättest, mir nur ein einziges Glas Wein auszuschenken. ‘I wouldn’t have done it if you hadn’t refused to give me but one glass of wine.’ (Ende 1973: 19) (3) Wenn man zwischen vier und sechs hinkam, war man sicher, die Mutter zu finden, und fast nie ohne Gesellschaft. ‘Whenever one got there between four and six, one would certainly find the mother, and almost never out of company.’ (Hofmannsthal 1910: 39) (4) In alten, alten Zeiten, als die Menschen noch in ganz anderen Sprachen redeten, gab es in den warmen Ländern schon große und prächtige Städte. ‘In old, old times, when humans still talked in very different languages, there were already big and glorious towns in the warm countries.’ (Ende 1973: 7) (5) Als die Kaiserin Strapazia schließlich von der Sache erfuhr, rief sie die bekannten Worte . . . ‘When the Queen Strapazia finally learned about the matter, she shouted the famous words . . .’ (Ende 1973: 47) The subordinating clauses introduced by wenn and als are known as “adverbial clauses”, which “are employed to provided the situational context for the event or state that is described in a main clause” (Whaley 1997: 247). Both pedagogic and traditional grammars of German contrast the two conjunctions in mixed terms: als is described as indicating single events in the past whereas wenn is described as denoting habitual and repeated actions in the present or the past. In addition, wenn is said to have a conditional meaning. The major aim of this study is to seek a unitary semantic-pragmatic explanation for the actual uses of the two items in texts. Attention is paid to the invariant meanings that motivate the concrete messages in varying communicational contexts. This study draws on the “Columbia School” approach to linguistic analysis, which treats linguistic units (morphemes and constructions) as signs of one-formone-meaning iconicity in a somewhat Saussurean sense. Each sign has an invariant meaning that motivates varying messages in varying communicational contexts. Contrastively related meanings form the cognitive basis of a certain semantic substance and are organized into a certain grammatical system (in the sense of Diver 1969: 47) that serves a certain pragmatic function. For example, Kirsner (1979a) argues that the Dutch demonstrative adjectives deze and die, rather than denoting proximity and distance, form a system of deixis, which serves the pragmatic function of urging the hearer to locate the referent. Within this system, deze signals high deixis and die signals low deixis such that the choice between them is made according to the level of urging as judged by the speaker. The present study offers a data-driven analysis. Textual data are examined in support for the hypothesis. It will be manifest that the descriptive account given

The givenness of background

by traditional grammars cannot explain why certain linguistic phenomena present themselves exactly the way they do in actual communication. It will also be shown that textbooks of German as a foreign language have unduly complicated the issue.

2. A review of the explanations found in pedagogical and traditional grammars I shall begin by considering Terell at al. Kontakte (3rd Edition, 1996), one of the textbooks currently used in college elementary German courses in the United States and which I myself used in teaching German to UCLA undergraduates. With regard to the two conjunctions in question, Kontakte tries to exhaust all possible cases in which the conjunctions occur by describing the events which the conjunctions used in individual cases communicate. “Als refers to a circumstance (time period) in the past or to a single event (point in time) in the past or present, but never in the future” (p. 297). Three examples are given to illustrate the idea of temporality denoted by als : Als ich 15 Jahre alt war, sind meine Eltern nach Texas gezogen. ‘When I was 15 years old, my parents moved to Texas.’ b. Als wir in Texas angekommen sind, war es sehr heiß. ‘When we arrived in Texas, it was very hot.’ c. Als Veronika ins Zimmer kommt, klingelt das Telefon. ‘When (As) Veronica comes into the room, the phone rings.’

(6) a.

No explicit explanation is given as to why als can be used in the past or present but never in the future. Nor is any suggestion provided as to what pragmatic function als serves when it “refers to a circumstance in the past or a single event in the past”. If language is used to communicate information for social purposes, merely referring to things is certainly less than sufficient to keep communication going. Note that in all three sentences in (6), the speaker relates the events being communicated from an offstage viewpoint, such that he is detached from the events, whether these occur in the past or in the present. Naturally, one cannot detach oneself from something that has not happened, something that is not happening, and something that may or may not happen. That is, one cannot detach oneself from a non-existent, i.e. future, event. In the present study, we will pursue the semantic cause that enables such narrative detachment as observed above. Turning now to wenn, Kontakte proceeds as follows: “Wenn has three distinct meanings: a conditional meaning and two temporal meanings. In conditional sentences, wenn means if. In the temporal sense, wenn may be used to describe events that happen or happened one or more times (when[ever]) or to describe events that will happen in the future (when)” (p. 298, emphasis in original). This claim is followed by four examples:

79

80

Zhuo Jing

Wenn man auf diesen Knopf drückt, öffnet sich die Tür. ‘If you press this button, the door will open.’ b. Wenn Herr Wagner nach Hause kam, freuten sich die Kinder. ‘When(ever) Mr. Wagner came home, the children would be happy.’ c. Wenn Herr Wagner nach Hause kommt, freuen sich die Kinder. ‘When(ever) Mr. Wagner comes home, the children are happy.’ d. Wenn ich in Frankfurt ankomme, rufe ich dich an. ‘When I arrive in Frankfurt, I’ll call you.’

(7) a.

By treating wenn as polysemous, with three discrete meanings, the grammar ignores the semantic interrelation between the so-called “meanings” (which are actually messages inferred from one single meaning in terms of the present study) and rips the individual “meanings” out of the semantic perspective where they belong. As one can sense by comparing what the textbook has to say regarding als with how it describes wenn, there seems to be some murky zone of possible overlap in the usage of the two. In other words, the description of wenn referring to “events that happen or happened one or more times” is too loose to exclude the suggestion that wenn and als may be somewhere interchangeable, which is absolutely untrue of any of the examples without change of message in the given sentence. The many unfortunate ors employed in the statements to juxtapose the different possible occurrences of each conjunction allow too much leaking for the grammar to hold. The textbook further singles out the usage of wenn versus als in the simple past tense. It points to the distinction that “wenn refers to a habit or an action or event that happened repeatedly or customarily; als refers to a specific action or event that happened once, over a particular time period or at a particular point in time in the past”. Two examples are given: Wenn ich nicht zur Schule gehen wollte, habe ich gesagt, ich bin krank. ‘When(ever) I didn’t want to go to school, I said that I was sick.’ b. Als ich mein erstes F bekommen habe, habe ich geweint. ‘When I got my first F, I cried.’

(8) a.

However, continually trapped in the clouds of temporality, the grammar is not able to come up with a unitary explanation of all possibilities beyond any one tense. The tendency to chop up the two signals into multiple meanings seen here is not restricted to textbooks for foreign language acquisition. Many monolingual German grammars for native German speakers exhibit the same symptom. The most reputable and authoritative German grammar, Duden Grammatik (1995, 5th edition) describes wenn as having three semantic “connections” (drei semantische Anschlüsse): 1) the conditional, 2) the temporal, and 3) the hypothetical. The temporal meaning has three sub-meanings: simultaneity (Gleichzeitigkeit), anteriority (Vorzeitigkeit) and subsequency (Nachzeitigkeit). The distinction of the three sub-meanings is made according to the temporal relation between the subordinate clause and the main clause. Apart from the temporal meaning,

The givenness of background

Duden attempts mightily to differentiate the hypothetical meaning from the conditional meaning; but its wording is deliberately and elaborately murky, as though it were striving to avoid the concept ‘conditional’ by using the less conspicuous word bedingend ‘conditioning’. But this attempt to make a qualitative distinction between what it calls ‘hypothetical’ and what it calls ‘conditional’ must be judged a failure: “Dieser [hypothetische] Anschluss ist nicht zu verwechseln mit konditionalem Anschluss bei den Verhältnissätzen, mit dem er allerdings die bedingende Grundprägung gemeinsam hat. (p. 745)” This [hypothetical] connection should not be confused with a conditional connection as in relative clauses, although it shares the basic conditioning factor with the latter. [Translation mine; Z. J.]

According to Duden, the hypothetical connection is solely associated with the subjunctive context in which the wenn-clause can be replaced by a complement clause, the dass-clause. The substitute dass-clause, however, must be in indicative mood. The same kind of substitution is in no way possible in conditional sentences with wenn. For example, the hypothetical sentence (9a) can be restructured in the form of (9a′): Es würde mich sehr freuen, wenn ich ihn sehen würde. ‘I would be very much pleased, if I would see him.’ a′. Dass ich ihn sehen werde, wird mich sehr freuen. ‘That I would see him would please me very much.’

(9) a.

But the conditional sentence (9b) is unable to undergo the same kind of paraphrase relationship: b. Wenn das wahr ist, dann müssen wir uns beeilen. ‘If this is true, then we have to rush.’ b′. *Dass das wahr ist, dann müssen wir uns beeilen. Apparently, Duden justifies its labeling by appealing to syntactic paraphrase. This justification is based on the assumption that the notion of “syntactic paraphrase relationship” is valid in semantic analyses and that the transformed sentence is identical in message to the original. Furthermore, Duden fails to address the question whether the inhibited paraphrase relationship in (9b’) determines the conditional meaning of wenn or the other way round and the same question regarding the hypothetical sentence (9a). If meaning rules over form, the question arises: how does hypothetical differ semantically from conditional? In fact, the distinction between hypothetical and conditional cannot be made in terms of mutual exclusion, because the two concepts do not occupy the same logical level. Although something hypothetical does not necessarily bear a consequence as conditional statements do, a condition is always expressed in the form of a hypothesis. In short, hypothetical includes conditional. On the other hand, suppose that syntactic paraphrase relationships were relevant

8

82

Zhuo Jing

to the distinction between hypothetical meaning and conditional meaning. Then we would need to decide whether it could be used to validate the distinction. In fact, the description by Duden might make sense if one ignores not only the reality of word order and its relation to other morphological realities, but also the reality of mood. Different arrangements of word order convey different messages. What appears in the initial position of a sentence carries more weight in terms of salience or emphasis than what comes later in the sentence (cf. Chafe 1976 and Lambrecht 1994 on information structure). Sentence (9a) emphasizes the content of the main clause (‘I would be very much pleased’) while (9a’) emphasizes that of the subordinate clause (‘That I would see him’). On the other hand, the compatibility between a certain mood and a certain subordinating conjunction must be taken seriously as well. The trouble with any syntactic paraphrase relationship that is claimed legitimate lies in the unfortunate fact that the message necessarily changes once any change in word order and/or mood is involved in the paraphrase relationship. Thus even though both alternatives are possible, they are not semantically equivalent. Since they are two different morphological sets, they necessarily bear different presuppositions to serve different pragmatic functions in communication. It is crucial to take into consideration all the formal cues in the input to reach an explanation as why the input is designed the way it is. In the current case, the interactions between the meaning of the subordinating conjunction, the meaning of the word order, and the meaning of the modality representation must be examined. Isolation of any one factor from the others will lead to a wrong interpretation. As far as als is concerned, the Duden Grammatik suggests two meanings: 1) temporal and 2) factual. Again the temporal meaning is divided into three sub-meanings as with wenn. Two other important German grammars, Buscha (1989) and Hentschel & Weydt (1990) share this tripartite approach to temporality in their account of wenn and als in addition to their common tendency to distinguish conditional and temporal meanings of wenn. The tripartite description of both wenn and als in terms of temporality causes the same sort of murkiness as the American textbook. Examples by Duden: (10) Temporal a. Tress L. hat nur gelacht, als sie die Geschichte hörte. ‘Tress L. just laughed when she heard the story.’ b. Als er geendet hatte, brach der Beifall los. ‘When he had finished, (the) applause broke out.’ c. Als ich nach Hause kam, hatte der Vertreter meine Frau bereits überzeugt. ‘When I came home, the sales representative had already convinced my wife.’ Factual:¹ d. Als er plötzlich auftauchte, freute mich das ungemein. ‘When he suddenly appeared, I was tremendously pleased about it.’

The givenness of background

It will be demonstrated later in this chapter that the separation of the temporal from the factual meaning is superfluous because the temporal meaning is in no case nonfactual, whether it is about simultaneity, anteriority or subsequency. Genzmer (1995), in his Deutsche Grammatik [German Grammar], treats the difference between als and wenn in terms of temporality and repetitiveness. He states the contrast between als and wenn in the following terms: Als drückt eine Vorzeitigkeit aus und bezieht sich auf ein einmaliges Ereignis in der Vergangenheit. Im Gegensatz dazu steht wenn, das Wiederholungen in der Gegenwart oder Vergangenheit bezeichnet und oft mit immer kombiniert wird. (p. 303) Als expresses anteriority and refers to a single event in the past. Unlike als, wenn denotes repetitions in the present or in the past and is often used in combination with immer ‘always’. [Translation mine, Z.J.]

Genzmer’s approach, though appearing unitary as far as temporality is concerned (conditionality is a separate meaning assigned to wenn), poses the following problems. First, anteriority does not describe occurrences in which als denotes non-anteriority (simultaneity, subsequency, or worse yet, an unspecified temporal relation), as in the following examples (examples are taken from Genzmer 1995: 303–304, translations mine; Z.J.): Zu dem Zwischenfall kam es, als der Angeklagte den Richter ohrfeigte. ‘Up came the incident as the defendant slapped the judge in the face.’ b. Als Gala nach Cadaques kam, rasierte sich Dalí die Achselhöhlen aus, um sie zu beeindrucken. ‘When Gala came to Cadaques, Dalí shaved his armpits to impress her.’

(11) a.

Sentences (11a–b) are cases where als cannot be taken to indicate temporal anteriority. One may argue that (11a) is about simultaneity because the slapping is the incident itself, yet even this argument is beside the point. The first clause of (11a) is structured deictically such that the hearer anticipates its referent, which is introduced later by als. In this sense, the situation in the first clause does not precede the situation in the second in terms of truth-conditional temporality, but in terms of the speaker’s viewing arrangement, to use Langacker’s words.² This special “viewing arrangement” is entirely at the speaker’s disposal. As such, it has the effect that the situation being described is introduced as a distinct episode of narrative tension whereas the narrator/speaker himself remains offstage the whole time. Here, the truth-conditional temporal relationship between the two clauses is irrelevant because there isn’t such a thing as far as this situation is concerned. When Wolf Schneider, the celebrated provider of good style of German, judged (11a) as ill-formed (Unfug ‘mischief ’, in his wording) and rewrote it as . . . kam es zu einem Zwischenfall: Der Angeklagte ohrfeigte den Richter (Genzmer1995: 304), he certainly did not approve of narrative tension at the expenses of truth-conditionally oriented

83

84

Zhuo Jing

style. By erasing the instructional conjunction als, the improvement he made virtually disjoins the two propositions and thereby renders the special viewing arrangement undone. As for sentence (11b), it is hard to determine the sequential order of the two events in strict temporal terms. Gala’s actual coming may just as well coincide with or follow Dalí’s shaving instead of preceding it. What als does here is to offer background information such that the main line event to be introduced is anchored and will not appear abruptly. In other words, the speaker sets up the stage, a background against which the zooming-in on the main event takes place. Again, the truth-conditional temporal sequence is uninteresting. Another problem with Genzmer’ s analysis concerns the viability of the notion of “repetition” with regard to the temporal wenn. The key point is that wenn indicates the potential to repeat an action rather than entailing the actual repetitions. In this respect, Genzmer’ s analysis shares the same weakness exhibited by Kontakte, which I discussed in the earlier part of this section.

3. An alternative approach to wenn versus als As an alternative to the treatments considered above, the present study postulates one single instructional meaning for each of the two subordinating conjunctions and places the two signs in a system where they contrast with each other ACROSS TENSES. In other words, their semantic contrast is real regardless of the tense because their meanings contrast with each other in all tenses. Moreover, the two conjunctions exhibit significantly different degrees of compatibility with different moods. It will be demonstrated how the single meaning that is relatively imprecise is able to form a community of semantic relatedness among the seemingly disparate messages in different contexts. And it will be shown how the gap between the super-ordinate meaning and the messages in concrete situations is filled by inferences based on all syntactical, morphological, semantic, lexical and supra-linguistic cues that are present in the input and the hearer’s general knowledge of the world.

3.1. Hypothesis We hypothesize that wenn and als form the cognitive basis of a semantic substance that I call the GIVENNESS OF BACKGROUND. The word “background” is a relational notion and in the present context it pertains to the status of the subordinate clause introduced by the respective conjunctions in relation to the main clause in terms of discourse focus (in Reid’ s sense, 1976). In contrast to the background, or “shunted” situation, denoted by the subordinate clause, the situation denoted by the main clause is in the foreground of the discourse and is also known as the “main line” situation, to use Hopper’ s (1979) terms. The meanings of wenn and als are organized into what could be called a system of the speaker’s instruction to the hearer

The givenness of background

GIVENNESS OF BACKGROUND

als = BACKGROUND GIVEN wenn = BACKGROUND NOT GIVEN

Figure 1. Invariant meaning for wenn versus als

with respect to the givenness of the background. The conjunction als has the instructional meaning BACKGROUND GIVEN while wenn has the instructional meaning BACKGROUND NOT GIVEN, as illustrated by Figure 1. Given the hypothesis, it is obvious that whenever als is used in a sentence containing two clauses, an instruction is provided to the hearer that the background is given and can be taken for granted. Whenever wenn is used in a sentence, it is meant to instruct the hearer that the background is not given and that the speaker is providing an imaginary or hypothetical situation as background against which the main line situation is introduced. In general, we anticipate that the speaker will employ the meaning BACKGROUND GIVEN most appropriately when he possesses sufficient evidence to provide a factual account of the background that can be taken for granted. A given background that can be taken for granted is likely to be factual. In contrast, the speaker will employ the meaning BACKGROUND NOT GIVEN most appropriately either because he (a) does not possess sufficient evidence to provide more than a tentative or hypothetical statement about the situation that he chooses as the background; or because he (b) assumes that the hearer is unaware of the background he chooses. For an illustration of the use of wenn, consider (9), reproduced here as (12): Es würde mich sehr freuen, wenn ich ihn sehen würde. ‘I would be very much pleased, if I would see him.’ a′. Dass ich ihn sehen werde, wird mich sehr freuen. ‘That I will see him will please me very much.’

(12) a.

In (12a), the speaker is not certain whether or not he will see the other person. He uses wenn to signal that he recognizes this lack of certainty and draws the hearer’s attention to the contingency of the discourse situation. That is, he recognizes that the circumstance of seeing the person is unpredictable and makes the hypothesis: given the circumstance of seeing the person, his being pleased would be true. (12a’), on the other hand, states that seeing the person is predictable and this prediction makes being pleased predictable. The contrast between (12a) and (12a’) demonstrates the untrustworthiness of syntactic paraphrasing (typically invoked in traditional grammars) as any sort of semantic explanation. Sentences in (6), reproduced here as (13), illustrate the use of als: (13) a.

Als ich 15 Jahre alt war, sind meine Eltern nach Texas gezogen. ‘When I was 15 years old, my parents moved to Texas.’

85

86

Zhuo Jing

b. Als wir in Texas angekommen sind, war es sehr heiß. ‘When we arrived in Texas, it was very hot.’ c. Als Veronika ins Zimmer kommt, klingelt das Telefon. ‘When (As) Veronica comes into the room, the phone rings.’ In all the sentences in (13), als introduces a background situation (subordinate clause), via which the main line situation (main clause) is approached. The actual temporal sequentiality is irrelevant here; it is rather the structuring of the background-foreground distinction which makes the narration possible. Furthermore, since a given background must be something that has reality in the speaker’s experience, the low probability of co-occurrence of als with future is unsurprising. The examples in (13) below reveal the semantic contrast between wenn and als in terms of givenness as signaled by the givenness of the subject of the backgrounded clause. Both wenn and als are compatible with a definite personal pronoun. But only wenn is compatible with the indefinite personal pronoun man ‘one’ (not Mann ‘man, husband’), which signals that although the referent is not “pinned down”, it is nevertheless to be taken for granted. Wenn man krank ist, isst man keine Wurst. ‘When one is ill, one doesn’t eat sausage.’ b. Wenn ich krank bin, esse ich keine Wurst. ‘When I’m ill, I don’t eat sausage.’ c. *Als man krank war, hat man keine Wurst gegessen. ‘When one was ill, one didn’t eat sausage.’ d. Als ich krank war, habe ich keine Wurst gegessen. ‘When I was ill, I didn’t eat sausage.’

(13) a.

The use of wenn matches closely with the use of man because the nongivenness signaled by wenn is compatible with the non-differentiation of the referent as signaled by man. The effect here is a general applicability of the hypothetical background. The reason why als is incompatible with man is obvious: als claims that the speaker possesses sufficient knowledge of the situation and his account of the situation can be taken for granted. Yet the use of man makes it impossible for this account to be taken for granted because the hearer cannot identify the exact referent of man. According to one informant, (13c) is odd because man can refer to anyone, which undoes the effort made by als to relate to an specific episode in the past, which presumably involves a specific subject. Here, the role played by tense is indicated. We assume that the past tense is a marked paradigmatic variation of the unmarked present tense within the grammatical category of tense (cf. Croft 1990 on the criteria for markedness) and as such it claims that the situation being communicated is specific in time and therefore already part of the reality. Thus the contrast between (13a–b) and (13d) in tense correlates to the contrast between wenn and als in the givenness of background. (14) below further illustrates this point:

The givenness of background

Wenn man krank ist, isst man nicht viel. ‘When one is ill, one doesn’t eat much.’ b. *Wenn man krank war, aß man nicht viel. ‘When one was ill, one didn’t eat much.’ c. *Wenn ich krank war, aß ich nicht viel. ‘When(ever) I was ill, I didn’t eat much.’ d. Immer, wenn ich krank war, aß ich nicht viel. ‘Whenever I was ill, I didn’t eat much.’ e. Als ich krank war, aß ich nicht viel. ‘When I was ill, I didn’t eat much.’

(14) a.

The infelicity of (14b–c) lies in the semantic incompatibility between the ungiven but hypothetical background signaled by wenn and the specification and givenness of time marked by the past tense morphology. The givenness of time collides with the hypothesis claimed by wenn. The acceptability of (14d), however, shows that the addition of an explicit hint in the sentence, namely the use of immer ‘always’, denoting universality and timelessness, serves to neutralize the potential infelicity caused by the mismatch between wenn and the marked time morphology. It makes possible the inference that a concrete instance of being ill is not given as the background. This function of restoring contingency explains the correlation between immer and wenn. (14e), in turn, is well formed because a semantic balance, a kind of pragmatic coherence, holds among all the three signs involved. All point, as it were, in the same direction: Als denotes that the background is given and can be taken for granted, the subject referent is pinned down by the “situationally evoked” (Prince 1982) referent of ich ‘I’, and the time is given by the specifying past tense. These examples show that the speaker can bring into play extra grammatical devices such as the indefinite personal pronoun man, the extremity word immer, and marked tense morphology to accentuate the givenness or ungivenness of the background. A further example is the use of mood and modality in this function. Let us look at sentence (2), reproduced as (15) below, and sentence (16): (15) Ich hätte es auch nicht getan, wenn/*als du dich nicht geweigert hättest, mir nur ein einziges Glas Wein auszuschenken. ‘I wouldn’t have done it if you hadn’t refused to give me but one glass of wine.’ (16) Wie sollen wir dem Wesen nachfragen, wenn/*als sogleich strittig werden kann, was Wesen heißt? ‘How should we inquire of the essence, if it can immediately become controversial as to what essence means?’ (H, p.174) Sentence (15) is produced in the subjunctive mood as marked by the umlaut on the vowel of the finite verb in both the main clause and the subordinate clause. Subjunctive, by definition, relates to or constitutes “a verb form or sets of verb forms that

87

88

Zhuo Jing

represents a denoted act or state not as fact but as contingent or possible or viewed emotionally (as with doubt or desire)”, according to Webster’s Third New International Dictionary (1986). The use of wenn, a signal of BACKGROUND NOT GIVEN, exactly matches the counterfactual situation denoted by the subjunctive. The foreground situation is also in subjunctive — a counterfactual background supports nothing more than a counterfactual scenario. The use of als in place of wenn, however, would be illicit, as declared by the asterisk. This is because the meaning BACKGROUND GIVEN signaled by als claims that the speaker is certain about the situation being communicated and that the certainty is to be taken for granted by the hearer, which is exactly opposite to the meaning of the subjunctive. Sentence (16) illustrates the same strategy of using more than one means of signaling the tentativeness of the situation considered as background: The use of wenn as a sign of hypothetical background is coupled with the use of the modal verb kann, a signal of possibility according to the speaker’s evaluation on the basis of insufficient evidence or certainty. Note also that the foreground clause with another modal verb sollen ‘should’ is in an interrogative mood, i.e. uncertainty par excellence. It is by means of this joined effort that the ungivenness is reinforced. In general, our hypothesis predicts the following strategies in making choices between wenn and als: 1. Framing discourse. The speaker will use the two conjunctions to form different discourse frames:³ The speaker will use als to frame the discourse as a story, or an episode within a story, of which the speaker claims exclusive knowledge; the speaker will use wenn to form a hypothesis frame in which he makes contingent statements that are open to the hearer’s questioning. 2. Givenness vs.ungivenness. As for a given background, the speaker will specify its time and its subject. As for a hypothetical background, its location in time is more likely to be unmarked than marked and the subject may not be specific. 3. Reinforcing. The speaker will use more than one grammatical means of claiming the state of the givenness of the background being introduced. Givenness will be coupled with devices for reinforcing the speaker’s certainty about the situation and ungivenness will be coupled with devices for recognizing contingency or tentativeness.

3.2. Validation of the hypothesis Our task for the following part of the study is to explain the different occurrences of the two conjunctions based on the above hypothesis and strategies. Two questions need to be answered: (1) In what contexts does one occur instead of the other? (2) Why does the occurrence take the shape it does? To show the validity of our hypothesis that als signals BACKGROUND GIVEN and wenn signals BACKGROUND NOT GIVEN, we provide quantitative evidence to show that the two items are used in accordance with their respective instructional mean-

The givenness of background

ings. Quantitative data are especially important where the exact pattern of correlation between the use of a conjunction and other linguistic mechanisms is under investigation. Let us take the grammatical category of tense as an example and look at the compatibility between a conjunction and the tense in which the conjunction is used. As described by textbooks, wenn can be used in both the past tense and the non-past tense. The non-past tense includes the present tense, the present perfect tense⁴ and the future tense.⁵ In short, wenn is compatible with all tenses in German. It is understandable that something hypothetical can be related to the past, the present or the future, because hypotheticalness is essentially orthogonal to time, on a different level than time. The “null hypothesis” would be that wenn, being equally compatible with all tenses, would occur with roughly the same frequency in each tense. However, when we perform frequency counts on the use of the conjunctions in texts, we confront the fact that wenn and als are not equally compatible with all different tenses in German. The frequency counts that follow are taken from two corpora. Corpus E consists of the first eight chapters of Ende (1973), a novel about a little girl named Momo and the world viewed from the perspective of Momo and her friends. Corpus H consists of three essays from Heidegger (1959), all of which are philosophical discussions of language interspersed with literary analyses of poetry as illustrations. Poetic examples are not included in the counting. As predicted, als does not occur at all in H, the philosophical analyses of poetic language. We argue that the absence of als in H reflects the lack of compatibility between the meaning of als, BACKGROUND GIVEN, and the communicational goal of the philosophical writing being surveyed. In other words, if everything were factual and to be taken for granted, there would be no room for thinking and analyzing, whereby ideas and arguments generate from hypotheses. Since wenn occurs in both E and H, and als occurs only in E, we need more evidence to determine the exact discourse frame they each favor. We consider it crucial to examine the particular contexts in which they occur in a text with potential sub-discourse frames. Corpus E is of advantage in this respect because it is richly filled with constructed conversations in the form of first-person dialogues (henceforth FD), which clearly differ from the writer’s narration in terms of discourse frame. Table 2 provides data on the distribution of wenn vs. als across discourse frames in the corpus E. Table 2 shows that 65.9% of wenn’s occurrences in E are found in constructed first-person dialogues instead of the writer’s narration. All occurrences of als, on Table 1. Total numNber of occurrences of wenn and als in the two corpora

Corpus

wenn

als

E H

85 57

24 0

89

90

Zhuo Jing Table 2. Choice of conjunction versus discourse frame in Corpus E Conjunction

FD

Narration

Total

wenn

54 65.9%

28 34.1%

82 100%

als

0 0%

24 100%

24 100%

the other hand, are part of the writer’s narration. The prediction about als as introducing a narrative background is confirmed. The question now is what does it mean that wenn is favored more by first-person dialogues than by the writer’s narration? Or, more precisely, within one and the same narrative text, other things being equal, in what way do constructed first-person dialogues differ from the writer’s narration? We argue that constructed conversations are verbal exchanges in mock interactions and therefore show the interactive quality and illocutionary force of real face-to-face speech communications (cf. Searle1969). In Tannen’ s (1988) words, it is about “the creation of voices” that “fires the individual imagination”. Tannen argues: . . . [The activation of the individual imagination is what makes it possible to understand another’s speech. This understanding — communication — simultaneously establishes the sense of rapport, of individual involvement with other individuals, that makes communication a social activity. As a culturally familiar figure, constructed dialogue in conversation and in fiction is a means by which experience surpasses story to become drama — a drama staged in the speech of one individual and enacted in the mind of another. The creation of drama from personal experience and hearsay contributes to the emotional involvement that is crucial for understanding and becomes the basis of human interaction”. (Tannen 1988: 91).

Clearly, the wonder of constructed dialogue is the dramatic emotional involvement, or subjectivity, to use Langacker’ s words, it presents. Compared to constructed dialogue, which is subjectively charged, narration is less so. Having recognized this difference between the two discourse frames, we can now explain the greater utilization of wenn in FD than in narration in terms of the compatibility between the meaning of wenn and the subjective nature of FD. If what wenn introduces into the discourse is a hypothetical background, then the nature of this piece of discourse is nonfactual and therefore open to individual evaluations, imaginations, and attitudes shaped by individual experience. In short, wenn enables subjective discourse that is above the level of facts. Because of this, wenn fits perfectly into the dramatic discourse in first-person dialogues. FD disfavors als because it is factually oriented and speaker-oriented: Such an orientation is motivated by the invariant meaning BACKGROUND GIVEN. The factual background it provides serves the speaker’s structuring of discourse while contrib-

The givenness of background

uting little to the exchange of ideas in communication that is also hearer-oriented. The factual orientation exhibited by als explains its aptness as a zoom-in device used for the introducing of an episode. Actually, on 12 out of a total of 24 occasions in the corpus E als is used to begin a new paragraph in text. On another note, the fact-oriented use of als in introducing episodes and stories in the past may create the impression that als is used to refer to single events, as observed by the pedagogic and traditional grammars. The second strategy, GIVENNESS VS UNGIVENNESS, is measured in two ways. First, we examine the types of the grammatical subject NP of the background clause introduced by the respective conjunctions in search for cues that indicate the givenness or ungivenness of the subject. The criterion here is the assumption that when the speaker introduces a given background, the referent of the grammatical subject is already differentiated or “pinned down”.⁶ That is, the subject is given as definite, or DIFFERENTIATION REQUIRED AND MADE, according to Kirsner (1972), rather than indefinite (where the differentiation of an intended referent from other, potential referents is either not achieved, as in “DIFFERENTIATION REQUIRED AND NOT MADE” or is irrelevant to the speaker’s purpose, as in “DIFFERENTIATION NOT REQUIRED”). In the present study, a subject NP is considered definite when it is any of the following: (1) an NP with a definite article (henceforth D-NP); (2) a proper name; (3) a definite personal pronoun (henceforth D-PP); (4) deixis such as dies (e/ er/es) ‘this/these’, solch (e/er/e) ‘such’. A subject NP is considered indefinite when it is marked by any of the following: (1) an NP with an indefinite article (ein/einer/eine ‘a’) or an indefinite adjective (manch ‘some’); (2) an indefinite pronoun (man ‘one’, jemand ‘somebody’, es ‘it’); (3) an NP with no determiner, neither article nor attribution. Now look at Table 3. All occurrences of als are found with a definite subject NP and no indefinite subject NPs are found in the als-clauses. By contrast, wenn is more versatile and exhibits high compatibility with indefinite referents. The prediction is thus confirmed. The high tolerance for indefinite subject NP exhibited by wenn is also confirmed by a survey of corpus H (see table 4). Another way of measuring definiteness is based on the assumption that marked tenses (non-present) are linguistic devices for marking a situation given by specifying its location in time. We counted the number of occurrences of wenn versus als in the present tense and in the past tense in the indicative mood. Subjunctives are not

Table 3. Choice of conjunction versus type of subject NP in Corpus E ConjunctionD-NP

D-PP

Deixis

Proper N

Indefinite

Total

wenn

6 7.3%

46 56.1%

2 2.4%

2 2.4%

26 31.7%

82 100%

als

10 41.7%

10 41.7%

3 12.5%

1 4.1%

0 0%

24 100%

9

92

Zhuo Jing Table 4. Use of wenn versus definiteness of subject NP in Corpus H Definite

Indefinite

Total

27 47.4%

30 52.6%

57 100%

included. The present tense includes present (inferably future) and present perfect and the past tense includes simple past and past perfect. Consider Table 5. The data on als demonstrate an absolute correlation between the meaning BACKGROUND GIVEN and the definiteness of temporal reference — past tense. The data on wenn show that wenn is not evenly distributed across the two tenses it matches theoretically. Wenn is used two and a quarter times as frequently in the present tense as in the past tense. This fact is clearly a direct consequence of wenn’s meaning BACKGROUND NOT GIVEN. When wenn is present in a piece of discourse, it hints that the situation is tentative and hypothetical and therefore may not be elaborated upon with regard to its temporal location. To discourse on something hypothetical in the past is by nature of the term and by default less common than to do so in a non-past environment. Because the unknown is hypothetical in nature, it is probably not beyond common sense that people tend to associate the unknown with hypothesis more than they do the known. The distinction between the unknown and the known is a function of time. The same survey is done with wenn in corpus H (where no als occurs). Here all the occurrences of wenn in the indicative mood are found in the present tense, which suggests that given facts associated with the past are uninteresting for the analytic purpose of philosophy. Thus the prediction that marked tense morphology is a correlate of the givenness of the background is confirmed. In the third strategy, REINFORCING, we assume that the speaker will employ additional linguistic means of representing the state of givenness of the background introduced by the respective conjunctions. Here we explore the following two linguistic mechanisms: (1) the use of mood and modality in expressing the degree of the speaker’s certainty about the situation. We assume that the degree of certainty marked by mood and modality immediately reflects the state of givenness as judged by the speaker. (2) The use of lexical means in reinforcing the state of givenness. As Table 5. Choice of conjunction versus grammatical tense in Corpus E Conjunction

Present

Past

Total in indicative

wenn

47 69.1%

21 30.9%

68 100%

als

0 0%

24 100%

24 100%

The givenness of background Table 6. Choice of conjunction versus mood and modality in Corpus E Conjunction

Subjunctive

Modal verb

Total

wenn

16 19.5%

14 17%

82 100%

als

0 0%

1 4.2%

24 100%

for the first mechanism, we measure the degree of certainty by counting the number of occurrences of wenn versus als in the subjunctive mood and in cases where the predicate is a modal verb. Here, subjunctives and modal verbs in both the subordinate clause and the main clause are counted on account of the grammatical agreement between finite verbs in the two clauses. As for the second mechanism, the criterion is that degree words (in Bolinger’ s sense 1972) and modal particles are further investments made by the speaker in reinforcing the claim of givenness or ungivenness. Observe the data in Table 6. Table 6 shows that while the occurrences of wenn in subjunctive make up 19.5% of its total number of occurrences, als does not occur at all in this mood. The correlation between the subjunctive and wenn provides sound evidence that wenn has a meaning that is compatible with more than just one single grammatical facet. The compatibility cannot be accounted for in terms of either temporality or conditionality alone. The wenn-subjunctive is a case that most of the time does not fall into either category. Consider the following example: (16) Sie hatten niemand, der ihnen so zuhören konnte, dass sie davon gescheit, versöhnlich, oder gar froh geworden wären. Aber selbst, wenn es dort so jemand gegeben hätte, es wäre doch höchst zweifelhaft gewesen, ob sie je zu ihm hingegangen wären — es sei denn, man hätte die Sache in fünf Minuten erledigen können. ‘They had no one who could listen to them like that such that they would have become smart, conciliatory, or even happy. But even if there had been some such person, it hwould be very doubtful whether he would have ever gone to him — unless they could get the matter done within five minutes.’ (E, pp. 70–71) The use of wenn in (16) cannot be accurately described in terms of tense, nor in terms of conditionality. Note that the circumstance stated by the wenn-clause does not provide a condition, sufficient or necessary, for the situation stated in the following main clause to hold. In fact, wenn introduces a hypothetical, or, to be precise, counterfactual, situation the potential effect of which is immediately rejected through the proposition stated in the main clause. Such a sentence is known as the concessive sentence, the function of which is to contradict the effect of a madeup instance as the potential alternative to reality. The notion conditional, a notion

93

94

Zhuo Jing

traditional grammars — including the Duden Grammatik — adopt to describe the meaning of wenn, proves useless in this case. This example makes the notion of ungiven background more conceivable in terms of hypotheticality. Moreover, because something made-up rather than real cannot be referred to as given, the logical subject of the wenn-clause can only be as fuzzy as so jemand ‘some such person’, a blurred image in a hypothesis. It is interesting to note that the subjunctive is not just used in one sentence; it is all over the place in the entire discourse. The dense use of the subjunctive in the discourse environment surrounding wenn makes the meaning of this conjunction even more obvious. The fact that als finds itself in the position of being completely disfavored by the subjunctive reflects the high degree of factuality associated with the meaning of als. The subjunctive form as one morpheme serves a semantic function that is the opposite of the conjunction als as another. Thus the two morphemes will not coexist within the span of one meaningful sentence. The quantitative observation made with wenn versus als with regard to their respective relationships with modal verbs raises further analytic interest: The combination of wenn with modal verbs achieves a percentage roughly four times as high as that of als. The explanation can be found in the semantic interaction between the two grammatical components in question, the conjunctions and the modal verbs. Let us now explore the semantic nature of modal verbs. These verbs normally distinguish themselves from non-modal verbs by their “modal markedness” among other qualities. This is a term used by Joos (1968) in describing the unique characteristic of modal verbs. Joos differentiates between “relative assertion” and “factual assertion”. The former refers to a clause with a modal and the latter refers to one without a modal. As opposed to factual assertion that has truth-value, relative assertion assumes no such truth-value with respect to the event. It rather suggests “the relation of that event in terms of its potential for realization or fulfillment (relationship to reality)”.⁷ Halliday (1970) elucidates the non-factual and subjective nature of modal verbs with his straightforward definition of modality: “modality, then, is the speaker’s assessment of probability and predictability. It is external to the content, being a part of the attitude taken up by the speaker: his attitude, in this case, towards his own speech role as ‘declarer’” (Halliday 1970: 349)

Whether it is about potentiality or probability or predictability, one thing should be certain. That is, modal verbs are used to compose the speaker’s “comments” on and evaluation of the events instead of providing objective information about the occurrence of the events. They are of subjective nature and are by and large emotionally charged. The subjectivity of modality is unambiguously asserted by Lyons (1995) who argues “subjective modality is more common than objective modality in most everyday uses of language”. He maintains that speakers (locutionary agents, in Lyons’ terms) tend to use modal verbs to express “either their own beliefs and attitudes or their own will and authority, rather than reporting, as neutral observers, the existence of this or that state of affairs. (p. 330)”

The givenness of background

An observation that additionally supports the subjective use of the modal verbs is provided by the fact that all the modal verbs occurred in sentences with wenn in the corpus E are also in constructed first-person dialogues, the subjective nature of which has been already discussed in the previous part of the study. Having explained the use of the modal verbs in the wenn-sentences in terms of subjectivity, it is clear that the correlation between the use of modal verbs and the conjunction wenn with the meaning BACKGROUND NOT GIVEN is nonarbitrary. The common subjectivity determines a closeness of fit of the two. Unlike wenn, als has a meaning that motivates the message of non-modal, i.e. factual. A given background must be known and familiar to the speaker. Nothing is more so than facts whose actual occurrence concerns the speaker. Thus it should be no wonder that als shows a lower compatibility with modal verbs that denote the speaker’s belief state or attitudes rather than facts. Unsurprisingly, the only occasion on which als co-occurs with a modal verb in the corpus E is one where the modal verb is used in a non-subjective manner. Witness the sentence: (17) Als Marxentius Communus erkennen musste, dass nun trotz allem eigentlich alles beim alten geblieben war, hüllte er sein Haupt in die Toga und ging davon. Wohin, hat man niemals erfahren. ‘When Marxentius Communus had to recognise that, despite everything, everything actually remained the old way, he covered up his head with a toga and walked away. Where he went, we have never found out.’ (p.48) First of all, this sentence as part of the narration reveals the viewpoint taken by the third-person narrator who is not involved in the story. The deliberately detached viewpoint of the writer can be seen clearly in both the propositional content of the last sentence above and the use of the indefinite personal pronoun man in it. With this distancing statement, the incidence introduced by the als-sentence is complete with a closure. Given the detached viewpoint, we can tell that the modal verb musste is used in a very much objective manner accordingly. That is, it signals that a high degree of necessity is imposed upon the subject by the external force of reality, as stated in the following complement clause introduced by dass. By virtue of objectivity, the situation can be taken for granted as the setting in which the accurate main line event unfolds. This example, rather than undermining the meaning we have postulated for als, provides a further proof of its meaning BACKGROUND GIVEN. A further observation that is of analytic interest with regard to the strategy of reinforcing is that, more often than not, wenn co-occurs with both the subjunctive and a modal verb, as in the following example: (18) Wenn ich das richtige Leben führen könnte, dann wäre ich ein ganz andere Mensch! ‘If I could live the right kind of life, I would be a completely different person!’ (p. 58)

95

96

Zhuo Jing

This sentence is a clear illustration of the speaker’s effort to reinforce the low degree of givenness, or high hypotheticalness in this context, by combining mood and modality. Thus far, the prediction of reinforcement as a strategy is confirmed on account of the frequency counts of wenn versus als in subjunctive and with modal verbs. As far as lexical devices for reinforcing are concerned, we found that als is more often than not accompanied by the following adverbs denoting the speaker’s perception of time: schließlich ‘finally’, nun ‘now’, noch ‘still’, wieder ‘again’, and einmal ‘once’. These adverbs serve to give an accurate measure of the speaker’s perception and experience of the situation in terms of time and thereby indicate that the hearer/ reader can depend on and take for granted the speaker/writer’s knowledge about the background information he provides. As for wenn, we found that it is frequently, that is, on 32 occasions out of a total of 82, used with the adverb dann, as in (18) above. We argue that the adverb dann ‘then’, which is not grammatically obligatory, serves to strengthen the contingency between the hypothetical background and the main line situation. Some modal particles such as nur ‘only’, selbst ‘even’, meistens ‘mostly’, and erst ‘not . . . until’ are also observed to highlight the high contingency of the situation. Apparently, the choice of these lexical items is not arbitrary, but rather coherent with the respective meanings of the two conjunctions. Thus far, all three statistical skewings predicted by our hypothesis are confirmed. In the following part of the chapter, I will re-examine some of the examples given in the Kontakte textbook in the light of the analysis of wenn and als, which I have presented here. Sentence (7a), repeated below for the reader’s convenience, is meant to illustrate what is called “conditional” by the textbook: (19) Wenn man auf diesen Knopf drückt, öffnet sich die Tür. ‘If you press this button, the door will open.’ This is a good example to serve the purpose because this sentence is about a theoretical consequence triggered by a certain event. However, the dependency between ‘the door will open’ and ‘you press the button’ is only theoretically certain. In other words, it is possible in reality that the door may not open even if you press the button due to whatever factors beyond the expectation based on the condition. Since ‘you press the button’ is something that has not taken place, the predicted consequence ‘the door will open’ cannot be taken for granted on that hypothetical background. Thus it is safe to replace — as an explanation for the use of wenn — the truth-conditional notion of conditionality with the concept of an instruction by the speaker to the hearer to understand the presented background as ungiven rather than given. This instructional meaning rather than a descriptive truth-conditional notion of conditionality better motivates the perceived message of hypotheticalness. Note that ambiguous cases abound in the conditional usage of wenn. Whereas a cautious English speaker struggles to decide between Give me a call if you have time and Give me a call when you have time in her discourse strategy, a German speaker

The givenness of background

need not worry about such a choice. By not offering a given background that can be taken for granted, wenn leaves enough room for the face-related issue of pushiness to be diluted and the hearer interprets the message based on other cues in the verbal or non-verbal exchange of information. In short, wenn is not necessarily or precisely what either conditionality or temporality is able to account for. In this respect, even some traditional grammars admit the dilemma the user has in differentiating conditionality and temporality. Hentschel and Weydt (1990) remark: Bei wenn ist der Bezug (temporal oder konditional) oft nicht eindeutig entscheidbar und kann nur anhand des Kontextes bestimmt werden, da die Konjunktion nicht zwischen Konditionalität und Temporalität unterscheidet. (p. 271) In the case of wenn it is often not unambiguous as to which meaning (temporal or conditional) is being referred to. The reference can be determined only with the aid of the context, because the conjunction does not differentiate between conditionality and temporality. [Translation mine, Z. J.]

The reason why the conjunction wenn does not distinguish between conditionality and temporality is that it means neither one nor the other, but assumes a higher degree of semantic abstraction. Let us now turn to temporality in relation to the contrast between wenn and als. Given the compatibility in terms of tense which wenn and als seem to share, at least partially, as traditional grammars describe, wenn still cannot be used in any even remotely familiar way als is used in any of the tenses with which both als and wenn show sufficient compatibility. In the past tense wenn is traditionally perceived as a word that refers to a repeated or habitual act or event, as the textbook describes it. The English translation ‘whenever’, which well renders the message of the German wenn, gives the illusion that the traditional description does sufficient justice to the semantic aspect of wenn. In fact, that wenn refers to a repeated or customary act or event is just a vague impression of one message conveyed by wenn in the given context, no matter how close the English translation seems to be to the German original form. Sentence (7b), repeated below as (20), does not emphasize an action Mr. Wagner repeatedly or habitually performed in the past: (20) Wenn Herr Wagner nach Hause kam, freuten sich die Kinder. ‘When(ever) Mr. Wagner came home, the children would be happy.’ Based on the information given in the sentence, we cannot precisely decide whether Mr. Wagner did actually come home or not. And this is the point of this sentence: What the wenn-clause focuses on is a hypothetical background that offers a circumstance in the past under which the content of the main clause can be expected to be true. Thus the sentence instructs the hearer that given the hypothetical background of Mr. Wagner’s coming home in the past, the children’s being happy would be true. The hearer is given the hint not to mind whether the event is real but to picture it in imagination on the hypothetical background. Precisely because wenn signals BACKGROUND NOT GIVEN, the main clause freuten sich die Kinder is translated into

97

98

Zhuo Jing

English as ‘the children would be happy’ rather than ‘the children were happy’ — a difference between contingency and fact. That the modal verb would substitutes the simple past were in the English translation reveals the nature of the original sentence in German. It is about something hypothetical, not actual. Logically, only something hypothetical can be treated as a variable that has the potential of being unlimitedly repeated. This is probably why the authors of the textbook decided that they were justified in describing the meaning of wenn in terms of repetition or habituality. Als used in the past tense forms a contrast to wenn, which is explained by the textbook in terms of once-versus-more-than-once. The textbook is limited to a simplistic solution to the semantic contrast: compared to wenn, als refers to non-recurring events. It is simplistic because it ignores the context in which each subordinating conjunction occurs by looking solely at the subordinate clause introduced by the conjunction in question and treating it as a self-sufficient entity. Contemplating the whole sentence (6a), repeated below as (21), one comes to realize that the weight of the sentence lies in the main clause, or the main line situation, hence the background-foreground distinction: (21) Als ich 15 Jahre alt war, sind meine Eltern nach Texas gezogen. ‘When I was 15 years old, my parents moved to Texas.’ It does not matter whether ‘I was 15 years old’ is recurring or non-recurring, what concerns the hearer is that ‘I was 15 years old’ is a fact to be taken for granted. This fact offers a fixed reference point or background for the hearer to approach what is to be said next: ‘my parents moved to Texas’, which is the occurrence intended to be the core of the information. It is not really important in which tense something is talked about. However, as I already suggest in the previous part of the study, something that has the nature of givenness has the logical tendency to be a fait accompli. This is why the textual frequency of als markedly skews towards the past tense while departing from the non-past tense. All of the 24 instances are observed in the past tense and none in the present, thus needless so say the future. The possibility of future tense that indicates any possible distance from the actual now is definitely ruled out wherever als is used as a subordinating conjunction to set the given background. Since the primary concern of a sentence with an als-clause is to introduce a given background, there may or may not be any necessary causality between the two occurrences distributed in the dependent clause and the main clause, respectively. Causality or non-causality is not of significance; what matters is the introduction of events according to the viewing arrangement made by the speaker. The two events may occur simultaneously or the event presented in the subordinate clause may precede or follow the event introduced by the main clause. This is what the German grammars such as Duden, Buscha, Hentschel & Weydt, and Genzmer, inter alia, call the simultaneity, the anteriority and the subsequency. In any case, however, the meaning of als remains its consistency in the sense of BACKGROUND GIVEN. It

The givenness of background

does not undergo any change because of the different time order the als-clause possesses each time. Now I would like to discuss in a more detailed fashion the descriptive account of wenn offered by Duden Grammatik regarding the hypothetical versus the conditional meaning, which has already been touched upon in the review of traditional grammars. Look at (9) again, reintroduced here as (22) below: Es würde mich sehr freuen, wenn ich ihn sehen würde. ‘I would be very much pleased, if I would see him.’ a′. Dass ich ihn sehen werde, wird mich sehr freuen. ‘That I would see him would please me very much.’ b. Wenn das wahr ist, dann müssen wir uns beeilen. ‘If this is true, then we have to rush.’

(22) a.

Contemplating the two examples given by Duden, one notices that the same sign wenn is used in two settings that differ from each other in the following ways: First, sentence (22a) is in subjunctive mood while (22b) is in indicative. Second, the wenn-clause in (22a) is placed behind the main clause whereas the wenn-clause in (22b) is placed in the first position. It is argued in this study, once again, that a syntactic paraphrase relationship is not a reliable criterion in semantic analyses. Sentence (22a) is not and cannot be semantically identical to (22a′) even though the transformed structure may sound close enough to the original. In fact, sentence (22a) conveys a very different message than sentence (22a′) does. Sentence (22a) suggests a high degree of hypotheticality that is unmistakably connected with wenn and the subjunctive as well as the sentential position of the wenn-clause. By contrast, (22a′) is almost on route to actuality because of the substitution of wenn with dass and because of the replacement of the subjunctive with the indicative. The relative pronoun dass is one subordinating conjunction that denotes objective reality, and the indicative mood in which objective assertions are made. Apparently, what is lost in the paraphrase relationship from (22a) to (22a′) is of great semantic and pragmatic significance. To claim that (22a) and (22a′) are the same is mistaken. It is obvious that the value of the syntactic paraphrase relationship is too questionable to be trusted in setting any norms for a semantic analysis. The differences between (22a) and (22b) cannot be attributed to whether or not they are subject to a syntactic paraphrase relationship. As a matter of fact, the differences between these two sentences do not invalidate the postulation of the invariant meaning for wenn, but rather support the idea that the same sign can be used differently in varying contexts. Sentence (22a) is different from (22b) not because wenn has a different meaning in one sentence than in the other. The meaning of the conjunction itself is not so powerful as to dictate the whole set of contrasts. The sign remains rather stable by maintaining the same meaning BACKGROUND NOT GIVEN while being in constant interaction with other grammatical elements on a number of linguistic levels. The

99

200 Zhuo Jing

subjunctive form in (22a), compared to indicative formulations, denotes a contrafact that is of lower probability. The secondary position of the wenn-clause in (22a) is a signal that indicates the fact that the wenn-clause is less salient than the main clause in the first position of the sequence. The two factors, the subjunctive plus the backgrounding of the wenn-clause, are coupled with each other in (22a). This combination reinforces the message of unlikelihood, one that is in tune with the meaning of wenn. In pragmatic terms, it is natural that the speaker tends to adopt a relatively humble tone to express her wish (isn’t any wish an emotionally triggered hypothesis?) of ‘I would be pleased’ when she evaluates the probability of ‘I would see him’ as low; or when she intends to show politeness by being pessimistic (cf. Brown & Levinson 1978 on strategies of politeness). It is true that the entire consideration of the linguistic texture of such sentences as (22a) is beyond native speakers’ awareness. German speakers, like speakers of many other languages, conventionally use them to show politeness in communication. However, the fact that such forms are associated with a certain degree of social finesse and dignity reveals the direct link between semantic structures and symbolic interaction. In contrast to (22a), sentence (22b) is in the indicative mood and the wenn-clause is pushed forward to the first place of the sequence and thus is more salient. That is to say, the wenn-clause indicates a hypothetical state of relatively high probability in comparison to its subjunctive counterpart. But in any case, by virtue of wenn, which indicates that a background is not given and that the hypothetical background cannot be taken for granted, it is not assumed that the hypothetical world is close to reality. The adverb dann leading the main clause refers back to the preposed wenn-clause and in so doing adds extra emphasis on the contingency it represents. The hypothetical world, though offering sufficient motivation, does not necessarily offer any condition for what is said in the main clause. Thus one can see that the validity of the meaning BACKGROUND NOT GIVEN postulated for wenn remains intact in this context as well. From my previous discussion of als, it is clear that this subordinating conjunction is always used in the sense of givenness, or actuality, no matter in which tense and whether at a certain point or in a certain period of time. Thus when we think of the separation of temporality from factuality made by Duden, we can dismiss it as unnecessary because the factual quality of als is present in all tenses where als can be used. Yet, the notion of factuality alone neither claims nor explains the syntactic function of subordination served by als. In other words, facts are asserted everywhere in verbal communications, but als is not used wherever facts are claimed. The key point is the relationship between the background and the foreground events. It is by means of the linking agent als that the telling of facts receives a perspective and stops being arbitrary and becomes meaningful and communicative. The same thing applies to wenn. For this reason, a linguistic analysis must reveal the relational meaning of the two items and not just the isolated sense of factuality and hypotheticality. This consideration justifies the concentration on the relational notion of background as the central aspect of the invariant meanings postulated for wenn and als.

The givenness of background

4. Conclusion I have argued in this study that the two modern german subordinating conjunctions wenn and als show differing distributions and uses across discourse frame, tense and mood and that they exhibit a striking difference in the ability to tolerate an indefinite subject. Wenn is favored by subjective discourses and analytic discourses whereas als is not utilized in dramatic dialogues nor in philosophical discourses, but in narratives for introducing episodes. While wenn can be used in just about any tense, it is more frequently used in the non-past tense and it has a high compatibility with the subjunctive, modal verbs, modal particles and indefinite subject NPs. Not all uses of wenn can be explained either in terms of conditionality or in terms of temporality. Als occurs only in the past tense, although it theoretically can be used in the present tense as well. Als does not tolerate the future tense and the subjunctive. Even though both wenn and als can be used in the past tense, they are not used in the same way and are never interchangeable without introducing significant differences in the massage being communicated. All these contrasting aspects are both formally real and statistically evidenced. They are not random and they demand explanation. In the present study, one invariant instructional meaning has been postulated for each individual conjunction within “the system of speaker’s instruction to the givenness of the background”. The advantage of this postulation is that the meaning is relatively abstract and thus allows a wider range of semantic interpretations in varying contexts. The meaning BACKGROUND NOT GIVEN of wenn and the meaning BACKGROUND GIVEN of als, from which concrete messages in specific situations can be derived, are shown to be successful in explaining all uses of the conjunctions. The postulation also makes it possible to explain why a certain meaning inhibits some grammatical representations and favors others in terms of the definiteness of the subject NP and in terms of tense, mood and modality.

Notes * This chapter developed from a term paper I had written for Professor Robert Kirsner’ s German 238 course at UCLA, “Linguistic theory and grammatical description.” I am grateful for Professor Kirsner’ s valuable effort to improve the earlier version of this work. However, errors in this chapter are exclusively mine. For further discussion of the Columbia School, see Contini-Morava 1995. 1. Duden does not explicitly term the semantic aspect as “factual”; but it clearly claims that this aspect is non-temporal. Duden decides that sentence 5d can be transformed into a sentence with a factual dass-clause: Dass er plötzlich auftauchte, (das) freute mich ungemein. Duden points out that both 5d and its paraphrase relationship are to be viewed as factual: “In beiden Fällen wird der Inhalt des Inhaltsatzes als tatsächlich gegeben gesetzt.” (p. 745) 2. See Langacker (1990) on speaker’s viewing arrangement and subjectivity. 3. The notion of a “frame” is born out of the sociolinguistic research known as symbolic

20

202 Zhuo Jing interaction. Bateson, in his celebrated essay A theory of play and fantasy (1972), showed that even monkeys engaged interactive actions that as a whole sequence suggests “this is play” versus “this is combat”. The message that “this is play” is a frame, contrasting to other potential frames such as “I am fighting you” or “I am threatening you”. Goffman (1974) adopted Bateson’ s concept and used in his analyses of human social interaction. The “frame” in his sense is the way individuals define the situation within which the interaction occurs. 4. The present perfect is concerned more with aspect than with tense. It has been suggested that present perfect marks the boundedness of events and therefore signals the speaker’s perspective. See Sasse (2001) on aspectology and Löbner (1988) on tense, aspect and aktionsart. 5. German does not have overt future marking. Given appropriate contexts, the unmarked present tense can give rise to an inference of the future. For example, the German sentence Er kommt morgen früh an is literally ‘He arrives tomorrow morning’, understood as ‘He will arrive tomorrow morning’ in English. The so-called werden-future, rather than marking temporality, expresses the speaker’s certainty or determination with regard to the situation being communicated. 6. See Kirsner (1972) on the meaning of the Dutch definite article. 7. See Bouma (1973) on modal verbs in German.

References Bateson, G. 1972. Steps to an Ecology of Mind. Chicago & London: University of Chicago Press. Bolinger, D. 1972. Degree Words. The Hague/Paris: Mouton. Bouma, L. 1973. The Semantics of the Modal Auxiliaries in Contemporary German. The Hague, Paris: Mouton. Brown, P. & S. Levinson 1978. Politeness. Studies in Interactional Sociolinguistics 4. Cambridge: Cambridge University Press. Buscha, J. 1989. Lexikon Deutscher Konjunktionen. Leipzig: VEB Verlag Enzyklopädie. Chafe, W. 1976. “Givenness, contrastiveness, definiteness, subjects, topics, and point of view.” In Subject and Topic, C. Li (ed), 25–55. New York: Academic Press. Contini-Morava, E. 1995. “Introduction: On linguistic sign theory.” In Contini-Morava and Goldberg (eds.), 1–39. Contini-Morava, E. and Goldberg, B. S. (eds). 1995. Meaning as Explanation: Advances in Linguistic Sign Theory. Berlin: Mouton de Gruyter. Croft, William.1990. Typology and Universals. Cambridge Textbooks in Linguistics. Cambridge: Cambridge University Press. Diver, W. 1969. “The system of relevance of the Homeric verb.” Acta Linguistica Hafniensia 12: 45–68. Diver, W. 1981. “On defining the discipline.” Columbia University Working Papers in Linguistics 6, 59–117. Diver, W. 1995. “Theory.” In Contini-Morava and Goldberg (eds.), 43–114. Duden 4 Die Grammatik. 1995. 5. Auflage. Mannheim/Leipzig/Wien/Zürich: Dudenverlag. Duden Deutsches Universal Wörterbuch. 1998. Mannheim, Leipzig, Wien, Zürich: Dudenverlag. Ende, M. 1973. Momo. Stuttgart: K. Thienemanns Verlag.

The givenness of background 203 García, E. 1975. The role of Theory in Linguistic Analysis: The Spanish Pronoun System. Amsterdam: North Holland. Genzmer, H. 1995. Deutsche Grammatik. Frankfurt am Main & Leipzig: Insel. Givón, T. and C. Li (eds.) 1979. Discourse and Syntax. New York: Academic Press. Goffman, E. 1974. Frame Analysis. Boston: Northeastern University Press. Halliday, M. A. K. 1970. “Functional diversity in language as seen from a consideration of modality and mood in English.” Foundations of Language 6: 322–361. Heidegger, M. 1959. Unterwegs zur Sprache. Stuttgart: Verlag Günter Neske. Hentschel, E. and Weydt, H., 1990. Handbuch der deutschen Grammatik. New York, Berlin: de Gruyter. Hopper, P. 1979. “Aspect and foregrounding in discourse. In Givón and Li (eds.), 213–241. Joos, M. 1968. The English Verb: Form and Meanings. Madison: University of Wisconsin Press. Kirsner, R. S. 1972. On Deixis and Degree of Differentiation in Modern Standard Dutch, unpublished doctoral dissertation. Columbia University, New York. Kirsner, R. S. 1979a. “Deixis in discourse: an exploratory quantitative study of the modern Dutch demonstrative adjectives.” In Givón and Li (eds.), 355–375. Kirsner, R. S. 1979b. The problem of Presentative Sentences in Modern Dutch. Amsterdam: North Holland. Lambrecht, K. 1994. Information Structure and Sentence Form. Cambridge: Cambridge University Press. Langacker, R. 1990. “Subjectivity.” Cognitive Linguistics 1 (1): 5–38. Löbner, S. 1988. “Ansätze zu einer integralen semantischen Theorie von Tempus, Aspekt und Aktionsarten.” In Temporalsemantik, V. Ehrich and H. Vater (eds.), 163–191. Beiträge zur Linguistik der Zeitreferenz. Tübingen: Max Niemeyer Verlag. Lyons, J. 1995. Linguistic Semantics. New York/Melbourne: Cambridge University Press. Webster’s Third New International Dictionary.1986. Springfield, MA: Merriam Webster, Inc. Reid, Wallis. 1976. “The quantitative validation of a grammatical hypothesis: the passé simple and the imparfait” In Proceedings of the 7th Meeting of the North Eastern Linguistic Society, J. A. Kegal, D. Nash and A. Zaenen (eds), 315–333. Sasse, H-J. 2001. Recent Activity in the Theory of Aspect. Arbeitspapier 40 (Neue Folge). Institut für Sprachwissenschaft. Universität zu Köln. Searle, J.1969. Speech Acts: An Essay in the Philosophy of Language. Cambridge: Cambridge University Press. Tannen, D. 1988. “Hearing voices in conversation, fiction, and mixed genres. In Linguistics in Context: Connecting Observation and Understanding, D. Tannen (ed.), 89–113. Vol. XXIX. in Advances in Discourse Processes, Norwood, NJ: Ablex. Terrell, T. D., Tschirner, E., Nikolai, B. and Genzmer, H., 1996. Kontakte. New York: McGrawHill. Traugott, E.1989. “On the rise of epistemic meanings in English: An example of subjectification in semantic change.” Language 65 (1): 31–55. Whaley, L. 1997. Introduction to Typology. Thousand Oaks/London/New Delhi: SAGE Publications.

CHAPTER 7

The relevance of relevance in linguistic analysis Spanish subjunctive mood Bob de Jonge Groningen University

1. Introduction In this chapter I will argue, as others have done previously (cf. for instance Klein 1975), that the various phenomena observed in the distribution of the subjunctive and the indicative moods in Spanish may be explained by means of a single, unitary meaning for, respectively, both indicative and subjunctive.¹ The proposed hypothesis has the advantage over others in the fact that it can offer a general explanation for the distribution of both forms and does not need to rely on various descriptionlike “meanings” for the occurrences of the subjunctive in different distributional environments. The meaning presented in this study is non-arbitrary in the sense that it is not only compatible with the distribution of the forms but also justifies or motivates that distribution. If a non-arbitrary meaning is capable of justifying and predicting the distribution of linguistic forms, then all meanings of all linguistic forms could, and in my view should, be taken as responsible for linguistic structure as a whole.², ³ Before presenting the hypothesis to be examined here, some of the most important previously hypothesized meanings for the subjunctive mood will be briefly reviewed.⁴

2. Some previous studies Klein (1975), who bases herself on Hooper (1975) and Terrell and Hooper (1974), tries to demonstrate that the indicative mood indicates ‘assertion’ of the occurrence expressed by the verb, and that the subjunctive indicates ‘non-assertion’ in complement clauses (Klein 1975: 353). However, there are some problematic cases, as shown by (1), taken from Klein (1975: 355): (1) Lamento que aprenda (S) ‘I regret that he should learn’ [translation Klein’ s]

206 Bob de Jonge

In Klein’ s view, with “main verbs of comment, such as lamentar ‘regret’ [. . .] the nonassertion of the complement can only be taken to indicate that it is not the purpose of the sentence to state this fact, but only to comment on it [. . .]”. This is essentially circular: reasoning that the meaning of the subjunctive mood is non-assertion, and that then aprenda cannot be taken as an assertion (in spite of its factuality in the real-world-situation) is only valid because of the assumption that the subjunctive mood means non-assertion. Moreover, if the subjunctive mood is caused by the verb lamentar, whose purpose is to comment on aprender and not to assert it, then other constructions which also comment on the subordinate clause should behave in precisely the same way. That this is not the case is shown by (2): (2) Es evidente que aprende zz(I) ‘It is obvious that he is learning’ There are, at least, two observations to be made. In the first place, it might be objected that que aprende here constitutes a subject clause that may therefore behave differently. However, this would allow arbitrariness in the analysis, whereas we hypothesize that the meaning of the subjunctive should determine its distribution. If meaning is the basic ingredient of a linguistic form, then that meaning (and that form) should occur only in combinations with those other meanings (and the linguistic forms signaling them) that are congruent with it. Of course, this point of view does not allow (meaningless) syntactic categories to determine the distribution of the other linguistic forms in their contexts.⁵ The second observation to be made is that es evidente indicates a comment on a factual occurrence. This observation contradicts the explanation of (1): if a main clause that indicates a comment — any comment — on the subordinate clause causes subjunctive mood, then it should always do so, and not only with certain kinds of comments in certain constructions. This certainly must be the case if we are to take meaning seriously, and as I said in the beginning, it is my intention to do exactly that. Other important studies that focus on meaning related to subjunctive mood, such as Haverkate 1989, contain essentially the same sort of argument found in Klein 1975. Haverkate assumes that one of the fundamental functions of the subjunctive mood is to affirm that the sentence with the verb in subjunctive mood does not convey information (1989: 98), but should be taken as already conceived, explicit or not, in the previous context. The occurrence, rendered in the subjunctive would then indicate background information. But here again, cases like (2) are problematic for this theory, since the preconception of the information is as valid for this kind of sentence as for the other verbs of comment, as exemplified in (1). Moreover, the postulated negative meaning as stated by Haverkate raises another question: if the subjunctive expresses that the sentence does not convey (new) information, what is it that the subordinate part that shows use of the subjunctive does convey? It is not the comment in itself, for that part is rendered in the main clause

Spanish subjunctive mood 207

of the sentence. We should, then, look for a meaning of the subjunctive that can be stated in positive terms in order to be able to explain (and not merely describe) all its uses as opposed to the indicative.

3. Hypothesis It should be stressed that neither of the two moods is intended to convey the truth as related to the real world and that neither has more truth-value than the other. Speakers by definition express a vision on reality that is subjective and can only be taken as the version of reality they want to get across to the hearer. The best proof of this statement is lying: all speakers are capable of lying, and clearly, lying can be done in the indicative mood.⁶ Thus, an occurrence expressed in indicative mood is not necessarily factual, but its effect on the hearer is intended to make him or her believe it is. The meaning assertion for indicative mood is therefore not badly chosen; it is non-assertion that needs refinement. What examples like (1) and (3) below have in common is that, in both sentences, there is another situation relevant apart from the one expressed by the subjunctive mood: (3) Quiero que aprenda (S) ‘I want that he/she learns’ In (1), the speaker has in mind a situation different from the actually presented one, namely a situation in which the person he or she is talking about had not learnt what he or she is actually doing. In (3), it is just the other way round: the person in question is not learning, and the speaker is expressing his or her wish that the other should do so. If this correspondence is the core of the meaning of subjunctive mood, then we could hypothesize the following meanings for the two moods: (4) Indicative mood (I) = assertion of the occurrence expressed by the verb (5) Subjunctive mood (S) = contextual relevance of an alternative for the occurrence expressed by the verb An important difference between this hypothesis and most others is that it considers the communicative situation in which the utterances containing the subjunctive mood occur. In order to test a hypothesized meaning, it is necessary in my view to take into account its communicative effects on the message. Therefore, use of terms like contextual and relevance are indispensable in this kind of hypothesis. Below, I will briefly outline the role and importance of these terms in the analysis. The proposed hypothesis should, and can, as shown in De Jonge (1998), account for so-called syntactic categories as in (3), in which use of the subjunctive is said to be ‘governed’ by the verb querer. In a meaning-based hypothesis, these cases should

208 Bob de Jonge

also be dealt with in a non-arbitrary way, rather than by writing mechanical “rules” that describe the co-occurrence of, say, certain types of verbs with the subjunctive. These verbs should be shown to have a meaningful characteristic in common in order to explain the use of the subjunctive with them; it does not suffice to merely say that it is always there. This characteristic could very well be the fact that in all communicative contexts in which querer que ‘to want to’ occurs, a desired situation is in fact not taking place. It would then follow that in situations in which a verb of desire is used, use of the subjunctive should indicate the current relevance of the opposite situation, the most logical alternative, in the given context. Therefore, since in all these situations an alternative is in fact relevant, categorical use of the subjunctive should not surprise us. De Jonge (1998) demonstrates qualitatively that this is so for verbs of desire, verbs and other constructions of emotion, hacer que ’to make that’ in the matrix clause and with conjunctions like aunque ‘although’ and el hecho de que ‘the fact that’.⁷ This way of relating meaning to distribution is not common in traditional grammars, since normally only the syntactic environment is taken into account in formulating rules. In the meaning-based hypothesis of the present study, on the other hand, one must consider the prototypical communicative situation in which the construction with subjunctive mood is used. Before beginning the actual analysis, it is important to define the terms of our analysis. In the first place, by alternative I mean an alternative to the verb, indicated in the subjunctive mood. The most common alternative to a verb is its negated counterpart. Therefore, while considering possible alternatives to the verb under analysis, the verb-negation is always taken into consideration. However, negations implied in subordinate conjunctions as in sin que ‘without that’ are not, since it is precisely the meaning of this conjunction that may explain the relevance of an alternative, expressed by the subjunctive.⁸ It should also be stressed that the word relevance is crucial in the analysis of the proposed hypothesis. It is possible to think of an alternative for any given situation, but this alternative is not necessarily relevant in the given context. In analyses of categorical uses of the subjunctive, it is possible to test the proposed hypothesis on the basis of the meaning of the relevant syntactic element, as shown above and in De Jonge (1998). However, it is impossible to do so in those cases where variation between subjunctive and indicative is observed, for only analysis of the wider context can demonstrate the relevance of an alternative. In this chapter, the hypothesis will be tested on a corpus of independent examples in syntactic contexts where variation between subjunctive and indicative is observed. All were taken from a collection of short stories by García Márquez (GGM 1994). In order to demonstrate the correctness of the proposed hypothesis in those cases where subjunctive is used, the relevance of the alternative will be studied in the communicative setting.

Spanish subjunctive mood 209

4. The data In order to investigate this hypothesis systematically, a corpus of examples of both indicative and subjunctive mood in similar contexts was collected. For this investigation, the largest group of contexts where both moods occur was chosen, namely all clauses introduced by some form of the conjunction que ‘that’. All instances of these clauses were taken from GGM (1994), yielding 728 instances of clauses introduced by some form of que, of which 613 (84%) were in indicative mood, and 115 (16%) in subjunctive mood.⁹ Table 1 presents the distribution of both moods in different kinds of clauses introduced by que. One thing is immediately clear from Table 1: practically all clause types show both indicative and subjunctive mood, so it is safe to say that there is no relation between syntactic categories and use of mood in Spanish. There is, therefore, no reason to carry out the analysis on the basis of such categories, so they play no further role in this chapter. In what follows, we focus solely on the meaning-based hypothesis proposed above.

5. The meaning-based analysis The analysis to be discussed below shows the contextual presence of an alternative with both moods. It should be noted, however, that the mere indication of an alterTable 1. Distribution of indicative vs. subjunctive mood in subordinate clauses introduced by (x)que, in García Márquez (1994)

Total complement clauses Total (X) prep.que Subject clauses Relative clauses Total (X) adj. que Adv.que Total prep.N que Imperative que sino que #Que Total

Indicative mood

Subjunctive mood

Rate %

Tokens

Rate %

81 67 67 95 95 89 100 50 100 78

140 93 32 284 19 16 17 1 2 7

19 33 33 5 5 11 — 50 — 22

33 46 16 15 1 2 — 1 — 2

84

611

16

116

Tokens

Note : Different types of subordinate clauses are given. The abbreviations used are: prep. ‘preposition’, as in para que ‘so that, in order to’; adj. ‘adjective’, as in lo más rápido que pueda ‘as fast as he can (S)’; adv. ‘adverb’, as in seguro que ‘for sure that’; prep.N ‘preposition noun’, as in de manera que ‘in a way that, so that’; sino que ‘but that (after a negated main verb)’. However, these distinctions in types of subordinate clauses will turn out to be irrelevant to the analysis, since both moods occur in practically all of them.

20

Bob de Jonge

native in Table 2 (to be presented shortly) does not automatically mean that this alternative is also relevant in the given context. Examples (6) and (7) are cases in which an alternative is present in the given context, but, while an indicative is used in (6), a subjunctive is used in (7): (6) Cuando murió don José Montiel, todo el mundo se sintió vengado, menos su viuda; pero se necesitaron varias horas para que todo el mundo creyera que en verdad había muerto (I). ‘When José Montiel died, everybody felt revenged, except his widow; but it took various hours before people believed that he had really died (I).’ (GGM 1994: 83) [all translations mine, BdJ] (7) Fue preciso que atornillaran (S) la tapa del ataúd y que lo emparedaran (S) en el aparatoso mausoleo familiar, para que el pueblo entero se convenciera (S) de que no se estaba haciendo el muerto. ‘It was necessary to screw (S) the lid onto the coffin and to put (S) him behind the wall of the pompous family grave in order to convince (S) the whole village that he was not just pretending to be dead.’ (GGM 1994 : 83–4) In (6), the alternative is that Montiel did not die. This alternative is present, because obviously many people at first did not believe he died. However, an indicative is used, because after some time this alternative had ceased to be relevant, and if people finally really believed he did die, there is no other option than to assert this fact in indicative mood. In (7), on the other hand, three cases of subjunctive mood are observed. The alternatives are clearly present in the given context: as long as the lid had not been screwed onto the coffin, and he had not been put behind the wall of his grave, people were not convinced he was dead. The relevance of these alternatives lies in the fact that an utterance of this kind can only be conceived of from a perspective where the lid has not yet been screwed onto the coffin and the coffin has not yet been put into the grave. In (8), an example is given in which no alternative is present in the context: (8) José Montiel murió en su hamaca, un miércoles a las dos de la tarde, a consecuencia de la rabieta que el médico le había prohibido (I). ‘José Montiel died in his hammock, on a Wednesday at two o’clock in the afternoon, as a result of an attack of rage that his doctor had forbidden (I) him to have.’ GGM 1994: 84) In (8), there is no alternative present to the relevant verb había prohibido ‘had forbidden’ in the given context: his doctor had forbidden him to get very angry and there is no way of expecting a presence of not forbidding this in this context, let alone its relevance. Table 2 gives the total distribution of subjunctive vs. indicative mood in contexts where there is an alternative present, vs. those where there is not. All contexts

Spanish subjunctive mood Table 2. Distribution of indicative vs. subjunctive mood in subordinate clauses introduced by (x)que, in García Márquez (1994) with contextual alternatives χ2=283.6; p < 0.001 + alternative − alternative Total

Indicative mood

Subjunctive mood

Rate %

Tokens

Rate %

Tokens

51 100

122 489

49 0

116 —

84

611

16

116

were classified according to the contextual presence of an alternative, independently of its contextual relevance, and independently of the mood being used. What we would expect to see, then, is a relative preference of alternatives for the subjunctive mood, whereas the absence of an alternative should never lead to use of the subjunctive, in view of the meaning of the latter. The results are set out in Table 2. It should be stressed that the results in Table 2 were obtained without taking into consideration the contextual relevance of the alternative. Table 2 shows clearly that the mere presence of a contextual alternative greatly favors the use of the subjunctive mood, whereas its absence leads to exclusive use of the indicative mood.¹⁰ The next step in the analysis is to show whether the difference between the use of the indicative in some cases and the subjunctive in others depends on the contextual relevance of the given alternative. According to our hypothesis, use of the subjunctive should occur ONLY when the alternative is indeed relevant to the message conveyed.

6. Demonstrating relevance In principle, there should be two ways to demonstrate the relevance of an alternative in given contexts with use of the subjunctive mood: qualitatively in each individual context, and quantitatively, showing the presence of a contextual factor that correlates with the subjunctive mood. But there is one methodological problem to be solved: there are also observed alternatives in contexts with the indicative mood, as the next pair of examples shows, but it is impossible to prove a negative, or, the irrelevance of the alternative in these contexts. In (9), a subjunctive is used, and therefore an alternative should be relevant in the given context; in (10) an alternative appears to be present, but should not be relevant, since an indicative is used:¹¹ (9) – Baltazar – dijo Montiel, suavemente. Ya te dije que te la lleves (S). ‘Baltazar, Montiel said, gently. I already told you that you should take (S) it away.’ (GGMs 1994: 78) (10) – ¿Entonces? Ana se arrodilló frente a la cama.

2

22

Bob de Jonge

– Que además de ladrón eres embustero – dijo. –¿Por qué? –Porque me dijiste que no había (I) nada en la gaveta. ‘And so? Ana kneeled down in front of the bed. – That apart from a thief, you are also a cheater -she said. – Why? – Because you said to me that there was (I) nothing in the drawer.’ (GGM 1994: 35–6) In (9), Baltazar, a carpenter who made a birdcage for the son of Montiel without Montiel knowing it, refuses to take the cage away and finally gives it to the boy. Not taking it away is therefore an alternative and Baltazar’s refusal to do so indicates its relevance in the given context. This corresponds, as expected, to the subjunctive form. In (10), on the other hand, Ana claims that Dámaso, who has stolen a set of French snooker balls from a bar, has lied to her about the amount of money he is supposed to have found in the drawer of the bar. She is commenting on his uttered words (that in the end turn out to be true) and, although obviously there is a possible alternative, Ana just wants to assert his alleged lies and thus has to use an indicative. However, it can be argued that Ana firmly believes there was something in the drawer, making this alternative relevant for the given context, and then the reasoning as given immediately above, would be circular, for an indicative is observed. Moreover, it could be objected that these cases are special, since they represent reported speech, and that (9) is a reported imperative. However, there are historical relations between the imperative and the subjunctive mood; in Old Spanish both indicative and subjunctive forms could communicate “imperative-type” messages, with the latter being a form with a weaker meaning (Lapesa 1981: 216, Menéndez Pidal 1976: 345–6). It is not surprising, therefore, that the indicative forms specialized for the non-formal relations between speaker and 2nd person tú and vosotros in affirmative contexts, and that the subjunctive specialized for formal relations between speaker and 2nd person usted(es), and for all negated forms, creating together a new imperative paradigm in the Spanish verb system. Moreover, it is not difficult to link the function of the imperative to the proposed meaning of the subjunctive mood: in cases where an imperative is used, the actual communicative situation presents a true alternative situation. For example, the iconic situation for saying ‘eat your supper’ to a child, is that of the child refusing to do so. Therefore, neither from a formal point of view nor from a historical or communicative one is it necessary to distinguish examples like (9) from other cases of the subjunctive. It seems better to look for confirmation of the proposed hypothesis in quantitative tendencies. As shown above in the qualitative analyses, an alternative for a given eventuality is usually provided by its negative counterpart. Therefore, negation might be a useful variable to study. Negated sentences generally refer to things

Spanish subjunctive mood

that do not take place, which in itself suggests an alternative to things that do. This is generally not the case with affirmative sentences. Accordingly, there are two correlations that must be examined: that between negation in the subordinate clause with subjunctive mood, and that between negation in the main (or superordinate) clause with subjunctive in the subordinate clause. One might expect that the second correlation would be stronger, because a negative main clause is more likely than an affirmative one to create a relevant alternative in the subordinate; cf. ‘he believes that John was not the thief ’ vs. ‘he does not believe that John was the thief ’. In the both situations, John is not believed to be the thief, but in the second one, the starting point is a subordinate clause saying he is, namely ‘- that John was the thief ’, thus suggesting an alternative relevant in the context. It is in the latter kind of situations where we expect to observe a relative preference for use of the subjunctive in Spanish. In the following two sections, we will see whether this is indeed the case.

6.1. Negation in the subordinate clause When taking into consideration the effect negation in the subordinate clause might have on the use of subjunctive mood, it is obvious that for the (relevant) alternative it makes no difference whether the verb in question is negated or not: when negated, its alternative is the affirmative form, and when affirmative, the negated one is. For example, when taking (9) and (10), both examples would not change as far as the use of moods is concerned if we were to replace the clauses by their negated counterparts (cf. (11) and (12)): (11) te dije que no te la lleves (S) ‘I told you that you should not take (S) it away’ (12) me dijiste que había (I) algo en la gaveta ‘You told me there was (I) something in the drawer’ Therefore, there seems to be no reason to expect a significant correlation of negation in the subordinate clause with the use of the subjunctive. Table 3 shows the results of this test. As Table 3 indeed shows, there is no difference in distribution of indicative vs. subjunctive mood with negated or not negated verbs (84% for indicatives Table 3. Distribution of indicative vs. subjunctive mood with presence or absence of a negation in the subordinate clause χ2=0.02; p < 0.9 − negation subordinate + negation subordinate

indicative mood

subjunctive mood

Rate %

Tokens

Rate %

Tokens

84 84

530 81

16 16

100 16

23

24

Bob de Jonge Table 4. Distribution of indicative vs. subjunctive in subordinate clauses with presence or absence of a negation in the main clause χ2=7.9; p < 0.01 − negation main + negation main

indicative mood

subjunctive mood

Rate %

Tokens

Rate %

Tokens

85 70

576 35

15 30

101 15

both without and with negations respectively, and corresponding values of 16% for the subjunctive).

6.2. Negation in the main clause Things could change when we take into account the presence of a negation in the main clause. As stated above, just before the end of the first part of Section 5, a negation in a main clause could suggest a contrast between the content of the subordinate clause and its (implicitly present) negative counterpart, thereby making that negative potentiality relevant in the general context. Accordingly, one expects to observe a correlation between a negated main clause and the presence of subjunctive mood in the subordinate clause. Table 4 presents the data and shows that our prediction is confirmed. Of all main clauses without a negation, 15% subjunctive (vs. 85% indicative) is observed in the subordinate clause, whereas this percentage is twice as high (30%) in subordinate clauses with a negation with the main verb.¹² That our hypothesis explains facts which otherwise would seem mysterious is shown by the following: in (13) and (14), a minimal pair of examples is given where one of them, (13), shows a negation in the main clause and the other, (14), does not. In both cases, the subordinate clause is a relative clause. Now traditional grammars, such as Ramsey (1956), claim that “after a relative pronoun referring to a person, thing or idea which is either unknown or not definitively known, the verb of the dependent clause is subjunctive” (1956: 420). That is, there is supposed to be a correlation between the degree of familiarity or givenness of the antecedent of the relative clause and mood. Furthermore, according to some grammars, the use of the definite article can be taken as an independent indication of the relatively greater familiarity of the antecedent, whether implicit or explicit, and lead to the use of the indicative. An indefinite article or its absence would, by contrast, indicate an unknown antecedent, which, in its turn, should lead to the subjunctive. However, (13) and (14) both have unknown antecedents, according to the use of the indefinite article, and nonetheless both moods are observed. But if we scrutinize the main clauses, we observe a negation in (13), and its absence in (14), corresponding exactly to the use of the moods that we have predicted:

Spanish subjunctive mood

(13) A pocas cuadras de allí, en una casa atiborrada de arneses donde nunca se había sentido un olor que no se pudiera (S) vender, permanecía indiferente a la novedad de la jaula. ‘Only a few blocks away, in a house filled with harnesses, where no odor had ever been smelled that could (S) not be sold, he [Montiel] remained indifferent to the novelty of the cage.’ (GGM 1994: 74–5) (14) Después de que el alcalde les perforaba las puertas a tiros y les ponía el plazo para abandonar el pueblo, José Montiel les compraba sus tierras y ganados por un precio que él mismo se encargaba (I) de fijar. ‘After the mayor perforated their doors with gunshots and gave them a deadline to leave the village, José Montiel bought their land and cattle for a price he himself took (I) the trouble to establish.’ (GGM 1994: 89–90) In both cases, the antecedent is inanimate and unknown in the previous context, as indicated by the indefinite article. Nevertheless, in one case the subjunctive is used and in the other, the indicative. Traditional grammars like the one mentioned above would have to conclude that (14) should be taken as an exception, since no adequate explanation can be given as to why this example does not follow the proposed distribution of subjunctive and indicative mood with (un)known antecedents. Our proposed hypothesis, however, serves perfectly to explain these two cases: in (14) a description is given of the situation of how people were expelled from the village and how Montiel bought their properties and how prices were so arbitrarily fixed. There is no relevant alternative present in the context, and in fact, the lack of an alternative is an important feature of the message: these people had no choice. Thus, an indicative form is used. In (13), on the contrary, the context implies that anything that had a smell in that house was a commodity that could be sold; that this is a possibility is a consequence of negation in the main clause: the existence of no odor that could not be sold implies a wide variety of odors that could.¹³ Multiple alternatives are implicit in this example, and therefore, the subjunctive mood is appropriate here.

7. Conclusion What I have tried to demonstrate in this chapter is that a theory based on the meaning of the linguistic sign is capable not only of describing but also of explaining the distribution of linguistic forms. Although this is a brief chapter which has not discussed the problem of subjunctive versus indicative in all possible contexts, the first results of the semantically oriented investigation outlined here are, in my view, quite promising. We therefore have reason to be optimistic about our further research on this topic.

25

26

Bob de Jonge

Notes 1. This is a further developed version of De Jonge (2001) and of a paper presented at the 6th Columbia School Linguistics Conference held at Rutgers University (State University of New Jersey), New Brunswick, NJ, October 1999. I am indebted to Reineke Bok-Bennema, Concepción Company Company, Brigitte Kampers-Manhe, Arie Molendijk, Dorien Nieuwenhuijsen, Co Vet and the editors of this volume for comments on earlier versions of this chapter. All remaining flaws and errors are my full responsibility. 2. ‘Linguistic forms’ should be taken in the broadest possible sense: also syntactic phenomena may be taken, in my view, as linguistic forms and thus as meaningful forms; cf. for instance Bolinger (1991) for the meaning of word order in Spanish. 3. For a more detailed discussion of this theoretical point of view, see García (1975, especially chapter X), Diver (1975 and 1995), Reid (1991) and Contini-Morava (1995). 4. I am well aware of the fact that the literature on mood in Romance linguistics is abundant. However, due to space limitations I will limit myself to two studies that fall within the scope of the hypothesis presented here. 5. It may be clear that this point of view invalidates any traditional analysis of subjunctive vs. indicative mood based on the distinction between main and subordinate clauses as related to mood. Moreover, both moods occur in both clause types, although the subjunctive has a preference for subordinate clauses. And again, as we will see below, in different types of subordinate clauses, both moods occur. 6. I would even say more: I assume that all lying takes place in the indicative mood. This is the most logical thing to do: the speaker wants the hearer to take the lie as a fact, and the best way to make him believe it is, is to assert its authenticity by use of the indicative mood. 7. Aunque may take both subjunctive and indicative in the subordinate clause. Here, too, the proposed hypothesis can explain this distribution. Consider a sentence such as aunque hace/haga mal tiempo, vamos a la playa ‘although/even if the weather is bad, we will go to the beach’. The second option takes into account the possibility of bad weather, presupposing the alternative, namely good weather, the normal situation for beach visits. The first option, on the other hand, merely takes bad weather as an observed factuality. 8. This reasoning may seem circular, but in fact it is not. In the search for an alternative, only the verb in subjunctive is taken into consideration, including any occurring negation. With regard to negation, note that it has a fixed position in relation to the verb: between the negation no ‘not’ and the inflected verb no other element than an oblique pronoun may be inserted, as in no (lo) sé ‘I do not know (it)’. The negation thus forms an inseparable unit with the inflected verb and accordingly has to be taken into account when looking for its alternative. The relevance of the alternative, then, is sought in the rest of the context, which (of course) is not fixed. Some subordinate conjunctions, such as para que ‘so that’ and sin que ‘without (that)’ trigger this contextual relevance through their general meaning, and so no further context needs to be taken in consideration. 9. GGM (1994) consists of six short stories, all first published in 1963 adding to a total of almost 20.000 words (19.934, to be precise). Although García Márquez is of Colombian origin, the Spanish used in his books may be considered to represent Latin American Spanish in general and not Colombian Spanish in particular. 10. The results obtained in this table may seem to have been gathered circularly, but in fact

Spanish subjunctive mood they are not. The classification was made on the basis of the possible presence of an alternative, and not of its absence. In the worst case, we might have missed an instance of an alternative present, with the consequence that the number of indicatives with an alternative would rise (with a slight change of percentages). But this would not alter the generally observed preferences. 11. It could also be that the indicative is indifferent to the contextual relevance of an alternative; its use would then indicate that the assertion of the occurrence is more relevant than the relevance of its alternative. I am indebted to the editors of this volume for drawing my attention towards this possibility. However, it does not really change my analysis, for it does not eliminate the danger of circularity, which I will discuss below. 12. Traditional grammars state that with verbs of thought, such as creer ‘to believe’ and pensar ‘to think’, their negative counterparts should show use of the subjunctive. In this table, however, only four cases of these negated verbs were observed out of a total of 50 negated verbs in main clauses. As the following example shows, these four did not show exclusive use of the subjunctive. (i)

No pensó -como no lo había pensado la primera vez- que Dámaso estaba (I) aún frente al cuarto, diciéndose que el plan había fracasado, y en espera de que ella saliera dando gritos. ‘She did not think -just as she had not thought the first time- that Dámaso was (I) still at the other side of the door, telling himself that the plan had failed, and waiting for her to come out of the room screaming.’ (GGM 1974: 63)

Here, an assertion is being made of a possible thought Ana did not have -Dámaso is at the other side of the door, etc.-, it is not a comment on a situation she was actually thinking about. Only in the latter case, use of the subjunctive would be appropriate, since Dámaso had indeed left. 13. In syntactic and traditional analyses, the negation nunca ‘never’ in the matrix is said to trigger use of the subjunctive in the subordinate clause (cf. for instance Kampers-Manhe (1991: 49) for French and Quer (1998: 37) for Catalan). However, as stated above in general terms, this observation does not explain its appearance in these contexts. The hypothesis proposed here, on the other hand, reveals why subjunctive is the most coherent choice: a negation of this kind of situation automatically implies the relevance of an alternative.

References Bolinger, D. 1991. “Meaningful word order in Spanish.” In Essays on Spanish: Words and Grammar, D. Bolinger, 218–230, Newark, Delaware: Juan de la Cuesta. Contini-Morava, E. 1995. “Introduction: On linguistic sign theory.” In Meaning as Explanation. Advances in Linguistic Sign Theory [Trends in Linguistics, Studies and Monographs 84, E. Contini-Morava and B. Sussman Goldberg (eds), 1–39. Berlin/New York: Mouton de Gruyter. Diver, W. 1975. “Introduction.” Columbia University Working Papers in Linguistics, 2 (Fall), 1–26. Diver, W. 1995. “Theory.” In Meaning as Explanation. Advances in Linguistic Sign Theory [Trends in Linguistics, Studies and Monographs 84], E. Contini-Morava and B. Sussman Goldberg (eds), 43–114. Berlin/New York: Mouton de Gruyter.

27

28

Bob de Jonge García, É. 1975. The role of theory in linguistic analysis: the Spanish pronoun system. Amsterdam: North Holland. GGM = García Márquez, G. 1994. Los funerales de la mamá grande. Barcelona: Plaza & Janes. Haverkate, H. 1989. Modale vormen van het Spaanse werkwoord. Dordrecht: Foris. Hooper, J. 1975. “On assertive predicates.” In Syntax and Semantics IV, J. P. Kimball (ed.), 91– 124. New York: Academic Press. Jonge, R. de 1998. “El uso del subjuntivo: ¿un problema para los hablantes de lenguas germánicas?” In Las lenguas en la Europa comunitaria [Proceedings from the 3rd International Colloquium, Amsterdam 1997], F. Sierra Martínez et al. (eds), 75–84. Amsterdam: Rodopi. Jonge, R. de. 2001. “Spanish subjunctive mood: One form, more than one meaning?” In Adverbial Modification [Faux Titres 203], R. Bok-Bennema et al. (eds), 79–93. Amsterdam: Rodopi. Kampers-Manhe, B. 1991. L’opposition subjonctif/indicatif dans les relatives, PhD thesis, University of Groningen. Klein, F. 1975. “Pragmatic constraints on distribution: the Spanish Subjunctive.” Papers from the Regional Meeting of the Chicago Linguistic Society 11: 353–365. Lapesa, R. 1981. Historia de la lengua española. Madrid: Gredos. Menéndez Pidal, R. 1976. Cantar de mío Cid.Texto, gramática y vocabulario. Vol. I. Madrid: Espasa Calpe. Quer, J. 1998. Mood at the interface. The Hague: Holland Academic Graphics. Ramsey, M. M. 1956. A Textbook of Modern Spanish. Revised by R. Spaulding. New York: Holt, Rinehart and Winston. Reid, W. 1991. Verb & Noun Number in English: A Functional Explanation. London and New York: Longman. Terrell, T. and J. Hooper. 1974. “A Semantically Based Analysis of Mood in Spanish.” Hispania 57: 484–494. Vet, J. P. 1996. “Analyse syntaxique de quelques emplois du subjonctif dans les complétives.” Cahiers de Grammaire 21: 135–152.

CHAPTER 8

A sign-based analysis of English pronouns in conjoined expressions Nancy Stern The City College of New York

1. Introduction The use of pronouns in conjoined expressions vexes writing teachers and puzzles linguists: (1) a. Me and him went to the movies. b. That’s between he and I. c. According to John, the article was written by Ann and himself. (Kuno 1987: 121) d. As for Harry and myself, we won’t be able to attend. While there remain many unanswered questions, this chapter offers a sign-based account of one part of this corner of English grammar, specifically, the choice between -self and simple (object) pronouns in conjoined expressions.¹ The use of -self forms in (1c) and (d) is unusual because, as is well known, most -self pronouns find their antecedent within their clause, as in the much studied reflexive and emphatic uses, shown in (2) and (3): (2) Mick saw himself in the mirror. (3) Mick himself was on stage. Structural syntactic accounts treat cases like (1c) and (d) as exceptions to the rule that -self forms are used either reflexively or emphatically. In fact, few accounts of pronouns, functional or formal, provide an explanation of why -self forms occur in conjoined expressions, and most theorists just exclude them from the analysis.² I have proposed that -self and simple pronouns are signals of invariant meaning (Stern 2001). The aim of this chapter is to demonstrate that the meaning of -self forms makes them particularly felicitous in conjoined expressions. While speakers’ concerns about prescriptive rules probably account for some of the distribution of pronouns in conjoined expressions, this explanation is, at best, incomplete. This chapter will show that -self pronouns are often deployed effectively in conjoined contexts when the conjuncts are to be distinguished from each other, and understood as differentiated and separated entities.

220 Nancy Stern

2. Confusion over prescriptive rules Before turning to a linguistic meaning hypothesis, let us consider the role of prescriptivism as a possible explanation. There appears to be a sense among linguists, teachers, and others that -self forms appear in conjoined expressions as the result of the misapplication of prescriptive rules. The thinking is that speakers are sometimes unsure whether it is correct to use I or me, and they may use myself to avoid the conundrum altogether. Following is a passage from a popular usage guide that tries to set the record straight: In the contest between I and me, the booby prize often goes to myself. That’s why we see sentences like Jack and myself were married yesterday. (It’s Jack and I.) Or like this more common self-promotion: The project made money for Reynaldo and myself. The speaker isn’t sure whether it’s Reynaldo and me or Reynaldo and I, so she goes with Reynaldo and myself. Tsk, tsk. (It’s for Reynaldo and me.) (O’Conner 1996: 13)

We will return to the issue of prescriptive rules in Section 6, but this factor can only be part of a successful explanation. Skilled writers use elegantly constructed sentences such as those in (4), and (5), and this does not seem to be related to a misunderstanding of prescriptive rules: (4) the principal hazards were to himself and not the neighbors in whose children he was deeply interested . . . (MBW 42) (5) As Hedda Nussbaum, in a red coat with faux-fur trim, met the press on the courthouse steps in Manhattan on March 10 to announce a legal victory giving herself and other battered women more time to sue their abusers, she dabbed her left eye with a tissue. (RHN) Examples (6) and (7), from best-selling novels, show two near minimal pairs, in which the choice of pronoun is not contingent on whether the referent has been mentioned elsewhere in the clause. In (6a) and (b), both from the same novel, both types of pronoun are coreferent with the understood subjects: (6) a.

Behind him emerged a woman’s face and graceful hands with redlacquered nails, locked on the boy’s shoulders as if in a death grip, holding him between herself and Trudi. (HSR 321) b. Suspended in the beams of pearl-gray light, he kept Trudi there though his arms began to quiver with her weight, held her there between him and his wife for what seemed the span of an entire lifetime . . . enveloping the three of them with the musty smell of earth. (HSR 20)

In (6a), the understood subject of holding is the woman, referred to in the between-phrase by herself. In (6b), the referent of he is the subject of held and is also the referent of the simple pronoun in the between-phrase. By contrast, in (7), the ref-

English pronouns in conjoined expressions

erents of the pronouns have not been mentioned elsewhere within the clause. That is, there is no clause-internal antecedent for either the -self pronoun in (a) or the simple pronoun in (b): (7) a.

Between herself and the door was a pungent memory, a reprise of another time she’d walked the length of the hall and opened the door to him, a moment when all her life had changed, had altered its course for good. (SPW 286) b. Trudi loved quickly, rashly — Sister Mathilde, whose voice would tremble with emotion when she spoke of the martyrs; Eva Rosen . . . ; Herr Pastor Schuler, who would hear Trudi’s first confession and tell her not to forget that she was God’s child — loved quickly, rashly, as she had once loved George, as though there were no air between her and the other person. (HSR 91)

Neither dialect nor register appears to account for this variation. Hypothesizing that -self forms are simply reflexes of underlying syntactic structure does not account for this variation either. However, the distribution of these forms can be explained by postulating that both the -self and simple pronouns are signals of (distinct) invariant meaning, used by speakers to meet communicative goals.

3. The meaning of -self pronouns I have proposed that the grammatical difference between -self and simple pronouns is a semantic one: while all pronouns signal some combination of Person, Number and Sex meanings, -self forms are hypothesized to signal an additional meaning, INSISTENCE ON AN ENTITY(S) (Stern 2001).³ This analysis therefore rejects the traditional characterization of -self as a reflexive pronoun, but it does borrow from the tradition by analyzing -self as a kind of emphatic pronoun. INSISTENCE can be understood as a vigorous reference that draws additional attention to an entity or entities. INSISTENCE is a strong pointing to an entity, a way of saying to the hearer, “Yes, this entity. I do mean this one(s)!” There are many different reasons why a speaker might choose to INSIST on a referent. Hearers must infer a speaker’s communicative intent, and her motivation for choosing to INSIST.

3.1. Reflexive uses Reflexive messages are the most frequent ones associated with -self forms, so it will be instructive to see how the meaning INSISTENCE contributes to reflexive interpretations. While it is beyond the scope of this chapter to explicate the use of INSISTENCE in all the contexts in which -self forms occur, we should address reflexive uses at least briefly to assure the reader that the hypothesized meaning can, in fact, account for this use of -self forms.⁴ In reflexives, the referent of the -self pronoun un-

22

222

Nancy Stern

expectedly plays more than a single role in one event (Levinson 1991; Givón 1993; Kemmer 1995). INSISTENCE serves as an alert to the hearer that there is something unusual about the referent, specifically, that the referent is playing more than a single role at just one time (Stern 2001, 2003). This analysis, dubbed the Role Conflict Hypothesis, accounts for the appearance of -self pronouns in reflexive environments, as well as in some environments that are not syntactically reflexive. It also explains the occurrence of simple pronouns in grammatically reflexive environments. The next three attested utterances will illustrate: (8) a.

Edward Sapir continually berated himself for idleness, when, in fact, he was continually turning out grammars for complex and difficult languages. (MBW 287) b. “Let Me Count the Times” is about a man who compiles statistics on his love life, which includes an affair with himself. (SHE) c. Even as a child I could tell that Mr. Tanaka saw the world around him as it really was. (GMG 14)

In (8a), a traditional reflexive context in which a -self pronoun appears, Sapir has two roles, as the berater and the one who was berated. Generally, when someone is berated, he berates someone else, so the message that the same entity plays both roles at the same time is unexpected. In (b), a -self pronoun occurs even though its clause is not syntactically reflexive. Nevertheless, there is a role conflict because the man has two roles: the one having the affair, and the person the affair is with. Under normal circumstances, these two roles are played by different people. But here, one individual plays both roles, so the -self pronoun appears. By contrast, in (c) a simple pronoun appears even though its clause is syntactically reflexive. This is because there is nothing unexpected, as there is no dual role for Mr. Tanaka. When a man sees the world, he typically sees the world around himself, not around someone else. There is no reason to INSIST on this referent, because there is nothing unexpected about him.

3.2. He behaved himself The brief explanation offered above does not explain why -self pronouns occur with certain verbs in which simple pronouns never seem to occur, as in: (9) He behaved himself. Clearly, the meaning of behave leads to the expectation that the second participant will be the same as the first, since one cannot behave someone else. Nevertheless, -self pronouns appear here, and it is difficult to imagine a simple pronoun in this context. An alert reader may wonder, then, why a simple pronoun does not occur in (9), if the referent’s role is expected. There is a role conflict in (9), but its source is grammatical: the meanings of the English Control System.⁵ (This grammatical source of

English pronouns in conjoined expressions

The girl kissed the boy. C1 C2 C1 = Higher Controller (in position before the verb) = the girl C2 = Lower Controller (in position after the verb) = the dog Figure 1. The English Control System

role conflict is found in (8a) as well.) The word order meanings of the Control System, which signal that the participants exert different degrees of Control over an event, are schematized in Figure 1. In the girl kissed the boy, the phrase the girl, by virtue of its position prior to the word kissed, is a signal of Higher Control, and the boy, by virtue of its position after kissed, is a signal of Lower Control. There is an inherent conflict when one individual is signaled to have both a Higher Control and a Lower Control role simultaneously. It is this conflict that motivates the use of the -self form (as opposed to a simple form) in he behaved himself. While there is no apparent emphasis on the -self form, INSISTENCE serves to warn the hearer that there is something unexpected about this referent, and in this case, it is that he is playing more than one role in the single event of behaving. Why then is the second participant (himself) mentioned at all? Why not simply, he behaved? The -self form occurs in such contexts in order to invoke the Control System. In one-participant events such as he behaved, there is no Control meaning signaled. The appearance of the second participant, himself, in (9) results in the invocation of the Control System, so that he is a signal of Higher Control, adding to the agentivity of the message. Turning to authentic data in which the appearance of -self as a second participant is, in a sense, optional, we see that -self forms may occur specifically to invoke the Control System, to suggest that their referents are in control of their situations: (10) a.

My big mistake here was that I didn’t realize how very, very angry I was . . . . And I needed support. Certainly, deathbed wills became exceedingly difficult and depressing, as if they weren’t in the first place. Sometimes I found myself angry that the client had gotten himself sick in the first place, a notion only one step away from the anger I was really feeling — anger at Joe for getting sick and for dying and of course for leaving me. (SFC 212–213) b. We are brought up to control and manage our bodies’ functions, holding back coughs and yawns, fending off sleep in public . . . . But childbirth cannot be controlled in this sense. Once labor begins, your baby is going to get himself born with or without your conscious cooperation. (LYB 27)

In (10a), when the writer says I found myself angry that the client had gotten himself sick, he wants to blame the client’s sickness on the client himself, to impute control to the person who had died, in order to have a target for his own anger. Later in the

223

224 Nancy Stern

passage, he writes anger at Joe for getting sick and for dying . . . . Here, the writer is speaking more rationally about what he is angry about — his friend’s death — and, being realistic, does not blame his friend, and does not invoke Control. In (10b), the writer is trying to express the agency of the baby, and his control over the birth situation. His birth will not just happen; rather, he will be an active, controlling participant.

4. Differentiation Another communicative effect that may result from the use of -self pronouns is the differentiation of their referents from other, closely related entities. As we will see, this characteristic of -self forms makes them particularly suitable for conjoined expressions. For instance, among emphatic appositive uses, we find -self pronouns used to differentiate their referents from other entities. That is, if we go beyond the overly general rubric “emphatic” and look more closely at the message effects among -self forms used as appositives, we find that in some cases, the pronoun steers the hearer away from unintended connections between the actual referent and other possible referents, as in (11): (11) It is wrong, almost always, to review the blurbs, cover, publicity, contract or circumstances surrounding a book instead of the book itself (PSM) The writer of this passage wants to differentiate the content of a book from other elements that a reader might consider when thinking of that book. A similar effect is found in (12), in which a speaker wants to distinguish the actual dinner from other details of an event: (12) – How was the banquet? – The dinner itself was fine, but after that it was all downhill. I was bored to death. (Kemmer 1995: 60) Kemmer notes that a banquet consists of many parts, including possibly dinner, dessert, drinks, décor, other guests, etc. The speaker wants to be sure to differentiate the dinner from these closely related entities. Even in non-appositive argument uses, -self forms can be used to differentiate referents.⁶ From a structural point of view, the next two examples are unexpected, since there are no clause-internal antecedents for the pronouns. But they share the semantic feature we have just seen among appositives, of excluding others and differentiating closely related entities: (13) But when [young men during the Vietnam War] burned their draft cards, no one died. Their protest affected themselves alone as sovereign individuals. (CKD)

English pronouns in conjoined expressions

While there is no apparent structural motivation for the use of a -self pronoun here, there appears to be a semantic motivation: the referents are to be sharply differentiated from others. In (13), the writer has also used the words alone and sovereign to emphasize that the actions of the Vietnam protesters affected only them, and not anyone else. The same message effect is seen in (14), in which the speaker is answering a question about an irreverent book he has written about celebrities called Naked Pictures of Famous People. (14) I am writing about their iconic basis more than I’m writing about themselves. (SNP) In this example, the speaker wants to differentiate the people themselves from their iconic value, what he terms their “iconic basis”. These are obviously two closely related aspects of being a celebrity, but the speaker wants to differentiate them. In (15), the speaker wants the referent of the -self pronoun to be understood to the exclusion of others, and also uses the word solely to contribute to this message: (15) a.

“We created an ad campaign solely directed at women for this watch.” The thinking, she said, was that Patek Philippe needed “to be speaking to women in a language solely for themselves.” (CAD) b. “We created an ad campaign solely directed at women for this watch.” The thinking, she said, was that Patek Philippe needed “to be speaking to women in a language solely for them.” (Unattested)

The speaker in (15a) could have used a simple pronoun, as in (b), and the scene described would have remained the same. The attested choice however, is more effective in expressing the idea that the ad campaign is targeted very specifically — exclusively — to women. The -self pronoun, with its meaning INSISTENCE, serves to differentiate its referent from other possible referents.

4.1. Differentiation in conjoined expressions We have just seen that -self forms may be used to differentiate their referents from other, closely related entities. And this is just what they may do in conjoined expressions: -self forms suggest that their referents are bounded and distinct from other, related entities in the group created by the conjuncts. Conjoined expressions contain the word and. And draws things together, creating the notion of a group that consists of the entities named by the conjuncts, as in bacon and eggs; salt and pepper; Donny and Marie. The -self form may be used to differentiate or individuate its referent, to, in a sense, extract its referent from the relationship suggested by the word and. To put it another way, the word and puts things together, but the meaning of the -self form, INSISTENCE, can serve to pull them apart. For this reason, -self forms can be quite felicitous in conjoined expressions.

225

226 Nancy Stern

4.2. Between-phrases Evidence for this hypothesis can be found by examining a collection of between-phrases, where it will be shown that simple pronouns occur when there is a greater closeness or intimacy between the referent and other entities, and -self pronouns appear when there is more of a distance between the conjuncts. Let us return to the near minimal pairs seen earlier in (6) and (7). Syntactic conditions do not predict which form will appear, but a semantic account fares quite well. (6a) is from the novel Stones from the River. A woman who is fleeing the Nazis with her young son unexpectedly encounters a stranger, Trudi. Naturally, the woman is terrified and mistrustful, fearing for her son’s life and for her own. There is no intimacy or closeness whatsoever between the woman and Trudi. The -self pronoun helps express this great gulf between them. In (6b), we see the use of a simple pronoun, from the beginning of the same novel, when Trudi is an infant. Here, the three people on the scene are a family: a father, mother and their new baby. The mother has rejected her child following her birth. This passage describes a scene in which the father, who loves both his baby and his disturbed wife, hopes his wife will embrace them both and that they can live the close, happy family life he dreams of. The use of the simple pronoun suggests the closeness the father longs for, the closeness which is the goal of him holding the baby there between them. A -self pronoun would not be effective in this passage, as it would suggest a distance between the husband and wife which was not part of the intended message. Likewise in (7), a syntactic account is unsuccessful, but our signal-meaning hypothesis offers an explanation. (7a) is from the best-selling novel The Pilot’s Wife, in which the main character, Kathleen, learns of her husband’s death when a man comes to her front door to deliver the news of a plane crash. Her entire world disintegrates as she discovers that her husband had been living a double life, married to another woman, with children in another city — all at the same time he was married to her. In the scene described in (7a), near the end of the novel, Kathleen is approaching the door again, remembering that other time she answered the doorbell and first learned of her husband’s death and then of his duplicitous other life. In that memory, the distance between her life before she answered the door and the utter chaos that ensued afterward is immense. The -self pronoun occurs in this between-phrase to help express the conceptual distance between the referent of -self, Kathleen, and the door, an entire world away. By contrast, the between-phrase in (7b) contains a simple pronoun. This is from the same novel in which (6a) and (b) appear. One of the goals of this passage is to express the closeness Trudi felt toward those she loved. The conceptual distance between the referent of her (Trudi) and the other conjunct (the other person) is quite small, as though there were no air between [them]. This conceptual closeness appears to be one of the reasons for the writer’s choice of the simple pronoun. In the passages in (16), a simple pronoun occurs in the between-phrase. In each example, there is a

English pronouns in conjoined expressions

closeness, a bond, an “intimate connection” as in (g), between the conjuncts: (16) a. b. c.

d.

e. f. g.

h.

Nothing comes between me and my Calvins! (Former ad campaign for Calvin Klein jeans) Whatever the product, the goal is always emotional identification between it and the consumer. (MAL) “It’s much easier to play someone else than to be yourself onscreen,” says Michael Douglas, who transforms himself into a bumbling, outof-shape English professor in the drama Wonder Boys. “And when it’s going good, there’s an element of feeling that you’re living that character. On Wall Street I felt that. There was no separation between me and Gordon Gekko until the end of the day.” (ACH) About a year ago, Ms. Whitman [CEO of eBay] set up a consumer insight and analysis group to find out. In contrast to the situation at traditional retailers, she said, “there is no one between us and the consumer.” (HDO) Some mothers, like Dot Hewitt of Paramus — can no longer find the gift that best captured the bond between her and her children, but it doesn’t matter. (AHM) The Best Medicine Can Help You Improve One of the Most Important Relationships in Your Life: The One Between You and Your Doctor. (MAB) And because she missed the written words that had formed the intimate connection between her and Stefan for seventeen years — so much more satisfying to be with him on the page than in person — she began to write to him again, fervent notes in German that she would hide and never show him, letters in which she didn’t have to hold back with her love for him or with her anger . . . (HVE 105) Between you and me, she’s an awful singer.

The conspiratorial between you and me in (16h) fits the pattern of (a) through (g): there’s no need to differentiate us; instead, we’re bonded together by this secret we’re sharing. By contrast, in the examples in (17), a -self form appears in the between-phrases, and there are considerable distances between the conjuncts: (17) a.

Despite the revelations of the last several minutes, despite the unnatural relationship between herself and the baby (despite the very fact of the child’s existence at all), Kathryn felt an urge, akin to sexual, to hold the infant to her breast, to that hollow space that wants always to embrace a small child (SPW 215) b. There was too much of everything in the banker’s office, the banker’s house. It all only emphasized the gap between Elizabeth and himself. [Elizabeth is the wife of the referent of himself, and the banker is her wealthy father.] (HVE 31)

227

228

Nancy Stern

c. d.

e.

f.

g.

“Gore is on the offensive because he has to convince people that there are differences of substance between himself and Bradley,” says Tumulty. (GVB) Unsatisfied by ham-fisted stereotypes, I went in search of political scientists, historians, legal scholars, biologists and psychologists who might cast more light on why most women make their first public act as mothers an etymological suicide, obliterating the most visible link between their children and themselves. (LWS) An ad for a computer, headline: Christian Murphy was non-verbal. He lived with a wall between himself and the rest of the world. . . . Caption: This is Christian. He can see, hear, feel touch, but he cannot speak. He was born with autistic tendencies that cut him off from the world around him. (ADA) And I puzzled about the contrasts between Ruth and myself, especially when she made what seemed to me contradictory remarks, such as that it was quite impossible to imagine me as a man and that “you’d make a better father than a mother.” (MBW 196) Gradually, she recognized more words. During the day she spoke German to the children: it was easier than taking every thought, translating it into English inside her head, and then saying it aloud, a process that always meant one extra step between herself and the meaning. (HVE 106–7)

Figure 2 provides a summary. The phrases containing the -self pronoun suggest either a conceptual, physical or emotional distance, while the phrases containing the simple pronoun suggest closeness and connection. The qualitative data examined here provide support for the INSISTENCE hypothesis. INSISTING on an entity can differentiate that referent, thereby suggesting a conceptual distance between that referent and others. between + -self

between + simple

(conceptual distance) a pungent memory the unnatural relationship emphasized the gap differences of substance obliterating the most visible link he lived with a wall

(conceptual closeness) as though there were no air nothing comes between . . . emotional identification no separation there is no one the bond between [a mother] and her children One of the most important relationships in your life the intimate connection

I puzzled about the contrasts one extra step

Figure 2. Summary of between-phrases

English pronouns in conjoined expressions 229

The pattern shown here results from an examination of a large group of examples, collected at random. While this type of collection is unstructured and cannot be used to quantitatively measure the frequency of forms or of specific contexts, it provides a relatively large number of relevant tokens. It is left to future research to validate the hypothesis quantitatively.

5. Use of forms is multi-determined It is important to note that the use of -self and other forms is multi-determined. -Self pronouns contribute to many types of interpretations in both argument and appositive uses, and these interpretations are found in conjoined expressions as well. For example, in (4), in addition to differentiation, the -self pronoun appears to contribute to a message of contrast, and it also suggests the importance of the referent, both of which are frequent interpretations resulting from the use of -self forms. In (4), the word not demonstrates that the writer wishes to distinguish the referent of himself from the other entity in the phrase, the neighbors. The writer notes that these two entities do not share the same relationship to the hazards, which were to her father, but not to the others. In (5), the meaning INSISTENCE is deployed to separate the form’s referent from the rest of the group. The same article describes the referent, Hedda Nussbaum, as “one of the most famous battered women in America,” and in addition, this passage tells us that she is a newly empowered crusader. She is, therefore, not an ordinary member of the group. There is another motivation for the use of -self in this passage. As we saw in the discussion of reflexives, -self pronouns are often used when there is something unexpected about the referent. Nussbaum’s role here as a strong, healthy spokesperson is rather surprising. The last time she had been seen in public was after her daughter had been murdered, and she herself had been brutally beaten over a period of ten years. So, the meaning INSISTENCE is useful in this example on at least two counts: to differentiate Nussbaum as a special member of this group, and because her role is unexpected. In contrast to (5), the following example shows a simple pronoun in a conjoined expression: (18) What the arrest has also done, though, is thrust into the spotlight, once again, a man [Michael Skakel] who has spent much of his life trying to escape the rumors and accusations dogging him, his older brother Thomas and by extension, his entire family. (CGO) Skakel, a 40-year-old member of the Kennedy family, had just been arrested on charges of committing a murder when he was a teenager. While no arrest had been made at the time of the crime, 25 years earlier, many residents and investigators had suspected Skakel since that time. The meaning INSISTENCE is not used here because there appears to be nothing unexpected about this referent, and no need to differ-

230 Nancy Stern

entiate him from the other conjuncts. It is not surprising that Skakel and his brother had been dogged by rumors. Skakel’s brother had also been a suspect, and because of a widespread belief in a cover-up by their wealthy family, the whole family had been under a cloud of suspicion. There is no need to differentiate Skakel from the others in the group formed by the conjoined expression. Example (19) features a -self pronoun in a conjoined expression, where the use of the meaning INSISTENCE is motivated not only by the need to differentiate the referent from the other conjuncts, but also because the role of the referent is unexpected: (19) The day before the gunman-arsonist turned a Harlem clothing store into an inferno that killed seven employees and himself, the store’s owner, who is Jewish, told a court that . . . arson and physical attacks seemed imminent. (MCF) The meaning INSISTENCE is deployed to advise the hearer that there is something noteworthy, something worth INSISTING about this referent. One motivation for the use of INSISTENCE here is to differentiate the arsonist from the group formed by the conjuncts (all those who were killed by the fire). The arsonist is a unique member of this collectivity, as he is the one responsible for all these deaths. Another motivation for the use of INSISTENCE in this example is that the role of the arsonist is unexpected, since arsonists generally set fires that kill other people, not themselves.⁷ To reiterate, the use of -self pronouns is multi-determined. -Self pronouns contribute to the same types of interpretation in conjoined expressions as they do in other environments. What makes them particularly apt in conjoined contexts is the effect of differentiating their referent from others. It is not the case that every time a -self pronoun is used in a between-phrase, it is to express distance, or that every time a simple pronoun occurs it is to express closeness. Nevertheless, there appears to be a strong skewing, and this discovery accounts for much of the distribution of pronouns in these contexts.

6. The human factor: prescriptive pressures As we discussed earlier, one of the factors that does seem to influence the choice of pronoun in conjoined expressions is speakers’ efforts to follow prescriptive rules. Here (as in other instances) language users are quite different from linguists. Linguists strive to ignore prescriptivist tenets in their work as analysts, but most speakers and writers are quite interested in using language “correctly” and in following the guidelines of stylebooks and other authorities. Thus, one of the factors that affects speakers’ choices of pronouns appears to be the desire to conform to social norms and prescriptive pressures. The role of prescriptive pressures is part of what Diver (1975) calls the Human Factor. As Cameron (1995: 6) describes it, “language-using is a social practice: the

English pronouns in conjoined expressions

human capacity for acquiring and using language is necessarily actualized within social relationships . . . .” Norms and evaluative views are the essence of such practices. While misapplication of prescriptive rules is probably one factor in the use of pronouns, another factor is the opposite — application of the prescriptive rules. Many speakers may deliberately use the forms according to the normative guidelines they know well. Therefore, in conjoined expressions as objects, some speakers might avoid -self pronouns if they are aware of the standard that prescribes simple (object) pronouns to be used. The next example demonstrates the difficulty of determining with certainty the reasons for a speaker’s choice in any given instance. We do find examples where we cannot settle on any single motivation or set of motivations. Nevertheless, the indeterminacy of individual examples like (20) should not deter us from pursuing functional, communicative explanations. In this example, the writer describes a significant age difference between herself and her students, yet she uses the simple pronoun, me. (20) As a teacher, I am at times most disconcerted by the style that reminds me of the 25-year gap in age between my students and me — the one my son calls the Hootchie-With-a-B. A.: smart girls in tight jeans, midriff-baring spandex shirts or body-hugging knit minidresses. (TWC) At first glance, this writer is discussing a large conceptual gap (a 25-year age gap), so a -self pronoun may have been more effective in communicating her intended message. However, it is also possible that this writer used the simple pronoun me precisely to achieve the conceptual closeness effect we have been discussing, that is, to ally herself with her students. After all, she tells us that she is disconcerted by this age gap (more precisely, by the style that reminds her of it). She may have used me to minimize the gap, to suggest a closeness and affinity with her students. But the possible influence of prescriptive pressures in this example should not be ignored. It is likely that the writer, a high school teacher, is well aware of the rules laid down in grammar books, and she may be devotedly following them. It may be quite difficult to determine to what extent prescriptivism is a factor in a writer’s choice of pronoun in any individual example, but it would be naïve to believe that it plays no role at all. A more rigorous understanding of this aspect of the human factor — that is, the ways in which prescriptive pressures influence the distribution of forms — is clearly needed.

7. Pronouns as signals of meaning While most research on -self pronouns has focused on the structural conditions for their use, the findings presented here suggest that treating pronouns as meaningful signals can successfully account for their distribution. In conjoined expressions, -self pronouns contribute to the same types of message effects that motivate their

23

232

Nancy Stern

use in other environments. They are especially felicitous in conjoined expressions, because they can differentiate their referents from the other conjuncts. While this observation does not account for all the choices speakers and writers make between -self and simple pronouns in conjoined expressions, it offers the beginnings of an understanding of this complex and little studied phenomenon.

Data Sources Chuck Arnold. “Chatter.” People Weekly. March 6, 2000 (v53 i9 p168). [InfoTrac] ADA Ad for Apple Computers AHM Mary Amoroso. “Whether hand-made or bought, some Mother’s Day presents stand out as extra-special.” The Bergen Record. May 9, 1999. CAD Dana Canedy. Advertising Column, The New York Times, September 7, 1999. CGO David W. Chen. “Glare of an Old Horror Intrudes on a Priviliged Life,” The New York Times, January 22, 2000, p. B1. CKD Polyxane S. Cobb. “Kevorkian’s Disobedience” (letter to the editor), The New York Times, April 22, 1999. GMG Arthur Golden. 1997. Memoirs of a Geisha. New York: Knopf. GVB Tony Karon. “Gore vs. Bradley: Get Ready for a Wonkfest,” Web-only news October 27, 1999. HDO Laura M. Holson. “Defining the On-Line Chief; Ebay’s Meg Whitman Explores Management, Web Style,” The New York Times, May 10, 1999, p. C8. HSR Ursula Hegi. 1994. Stones from the River. New York: Scribner Paperback Fiction. HVE Ursula Hegi. 2000. The Vision of Emma Blau. New York: Simon and Schuster. LWS Carol Lloyd. “Why should a baby get the father’s last name?” www.Salon. com, January 20, 2000. LYB Penelope Leach. 1987. Your Baby and Child. New York: Alfred A. Knopf. MAB Ad appearing in The New York Times for a book by Mike Magee, M. D. and Michael D’Antonio, The Best Medicine. Saint Martin’s Press. MAL Douglas Martin. “What’s in A Name: The Allure of Labels,” The New York Times (Week in Review), January 19, 2000. MBW Margaret Mead. Blackberry Winter. 1972. New York: Touchstone Books. MCF Robert D. McFadden. “Harlem Store Owner Had Cited Anti-Semitism Before Fatal Fire,” The New York Times, December 12, 1995, p. A1. PSM Michael Pye. “Serial Mom” (Review of Joyce Carol Oates writing as Rosamond Smith, Starr Bright Will Be With You), The New York Times Book Review, April 11, 1999. RHN Francine Russo. “The Faces of Hedda Nussbaum,” The New York Times Magazine, March 30, 1997, p. 26. ACH

English pronouns in conjoined expressions

SFC SHE SNP SPW TWC InfoTrac

Mark Senak. 1998. A Fragile Circle. Los Angeles: Alyson Books. R. Z. Sheppard. “Bitter Sweets,” Time Magazine, February 9, 1999, p. 70. Jon Stewart, on The Today Show, NBC, October 4, 1999. Anita Shreve. 1998. The Pilot’s Wife. Boston, New York and London: Little, Brown and Company. Sallie Tisdale. “What Color Is Your Uniform,” The New York Times, May 24, 1998, Sunday Styles/Section 9, p. 5 The InfoTrac Reference Center, Magazine Index from the General Reference Center. 2001. The Gale Group Inc. This electronic database consists of articles from magazines, reference books, newspapers and news services on current events, popular culture, the arts and sciences, sports, business etc. that have appeared from 1980 to the present.

Notes 1. -Self pronouns include myself, yourself, himself, herself, itself, ourselves, etc., while simple pronouns include me, you, him, her, it, us, etc. 2. The most well-known syntactic account, generative grammar’s Binding Theory, is not intended to cover conjoined expressions (Chomsky 1981, 1982, 1995; Reinhart and Reuland 1993). Van Hoek (1997) and Kuno (1987) discuss data that include -self pronouns in conjoined expressions. 3. By hypothesis, meanings are present only when they are encoded linguistically, that is, when they are part of the invariant semantic contribution of the form. So, for instance, you signals Person, but does not signal a Number meaning or a Sex meaning, although of course the number and sex of referents are often part of the hearer’s interpretation of an utterance. 4. A more thorough description of the INSISTENCE hypothesis and the use of -self pronouns in reflexive and other contexts can be found in Stern (2001). 5. A more thorough description of the English Control System can be found in Reid (1991) and Huffman (1993). 6. Following Kemmer and Barlow (1996) and Cresswell (1997), the term “argument” is applied here not only to participants selected by the verb (e.g., Diana and Ben in Diana loves Ben), but to any entity with its own role, including objects of prepositions (e.g., Diana was looking at a picture of Ben in the newspaper), where Diana, picture, Ben and newspaper are all here described as arguments. 7. The order of the conjuncts may also be relevant, since a simple pronoun would seem like a more plausible option if the arsonist had been mentioned first. We can only speculate about non-occurring examples, but an inferno that killed him and seven employees would seem more likely than an inferno that killed seven employees and him.

References Cameron, Deborah. 1995. Verbal Hygiene. London and New York: Routledge. Chomsky, Noam. 1981. Lectures on Government and Binding. Dordrecht: Foris Publications.

233

234

Nancy Stern Chomsky, Noam. 1982. Some Concepts and Consequences of the Theory of Government and Binding. Cambridge, MA: MIT Press. Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press. Cresswell, Cassandra. 1997. Non-argument Reflexives: A Pragmatic Analysis. Senior honors thesis, University of Illinois at Urbana-Champaign. Diver, William. 1975. “Introduction.” Columbia University Working Papers in Linguistics 2: 1–20 [corrected reprint 1980]. Givón, Talmy. 1993. English Grammar: A Function-Based Introduction, Vols. 1 and 2. Philadelphia and Amsterdam: Benjamins. Huffman, Alan. 1993. “The Control-Focus interlock.” Part I, Events and Participants. Modern English: A Columbia School Grammar. Ms. Kemmer, Suzanne. 1995. “Emphatic and reflexive -self : Expectations, viewpoint and subjectivity.” In Subjectivity and Subjectivisation, Dieter Stein and Susan Wright (eds), 55–82. Cambridge University Press. Kemmer, Suzanne and Michael Barlow. 1996. “Emphatic -self in discourse.” In Conceptual Structure, Discourse and Language, A. E. Goldberg (ed), 231–248. Stanford, CA.: Center for the Study of Language and Information. Kuno, Susumu. 1987. Functional Syntax: Anaphora, Discourse and Empathy. Chicago: The University of Chicago Press. Levinson, Stephen C. 1991. “Pragmatic reduction of the Binding Conditions revisited.” Journal of Linguistics 27: 107–161. O’Conner, Patricia T. 1996. Woe Is I: The Grammarphobe’s Guide to Better English in Plain English. New York: G. P. Putnam’s Sons. Reid, Wallis. 1991. Verb and Noun Number in English: A Functional Explanation. London and New York: Longman. Reinhart, Tanya and Eric Reuland. 1993. “Reflexivity.” Linguistic Inquiry 24(4): 657–720. Stern, Nancy. 2001. The Meaning and Use of English -Self. Ph.D. dissertation, CUNY Graduate Center. Stern, Nancy. 2003.“The grammar of English reflexives: A new view.” In LACUS Forum XXIX: Linguistics and the Real World, Douglas W. Coleman, William J. Sullivan and Arle Lommel (eds), 327–337. Houston: LACUS. Van Hoek, Karen. 1997. Anaphora and Conceptual Structure. Chicago and London: The University of Chicago Press.

CHAPTER 9

Semantic oppositions in the Hebrew verb system A sign-oriented approach Noah Oron and Yishai Tobin Ben-Gurion University of the Negev

1. Introduction In this chapter we will explore the issue of semantic oppositions in the Hebrew verb system. Traditional analyses of the Hebrew verb system have typically been limited to listing and discussing the various syntactic, semantic and pragmatic functions of each of the seven major patterns that comprise the verb system. We will outline a sign-oriented analysis of the Hebrew verb system as presented in tobin (1995), that adopts the tenets that the distribution of linguistic signs and their relationships are motivated by invariant meaning and that the connection between signal and meaning in sign systems may be iconically reflected in the language code. Then we will go on to show how two of the patterns of the Hebrew verb system can be systematically opposed based on the notion of invariant meaning. Traditional analyses have attempted to attribute specific functions to each pattern in the verb system, but none has validated the entire system with all the possible roots that appear in each pattern. In the present analysis, the connection between two patterns in the verb system will be exhaustively tested from a semiotic linguistic approach, with the intention of contributing to a more general attempt towards a unified calculus of meaning (see van Schooneveld 1978) in language.

2. The Hebrew verb system Hebrew, like other Semitic languages, has a fundamentally different structure than the Indo-European languages. In the Semitic languages, the isomorphic connections between phonology, morphology, syntax and semantics are much more overt. The vast majority of the words of the language can be analyzed into consonantal roots signaling broad semantic fields which are combined with fixed morphophonemic patterns for what are traditionally called nominal adjectival (mishkalim ‘weights/scales’), and verbal (binyanim ‘building blocks’) forms. More often than

236

Noah Oron and Yishai Tobin

not, the connections and relationships between the root and these fixed morphophonemic patterns are rather transparent. The relatively transparent connection between lexical roots and the fixed morphophonemic nominal, verbal, or adjectival patterns may be briefly illustrated for the triliteral /C–C–C/ consonantal root K-T-V ‘write’ as follows: KoTeV ‘writes, writer’; KaTaV ‘wrote, reporter, correspondent’; niKTaV ‘was written’; KiTeV ‘inscribed’; KuTaV ‘was inscribed’; hiKTiV ‘dictated’, huKTaV ‘was dictated’; hitKaTeV ‘corresponded, wrote letters’; nitKaTeV ‘was written in a correspondence, letter’; KaTaVa ‘article’; KToVet ‘address’; KiTuVit ‘title, subtitle’; KTuVa ‘contract’; KtaV ‘handwriting’; miKTaV ‘letter’; miKTAVa ‘writing desk’; haKTaVa ‘dictation’; hitKaTVut ‘correspondence’; KTiV ‘spelling, orthography’; KaTVan ‘typist, scribe’; KaTVanut ‘typing’; KaTuV ‘was written, biblical verse, passage, text’; KTaVim ‘works, writings’; SHiKTuV ‘rewrite’. It should be clear from the above examples, that on the most fundamental level, the Hebrew lexicon is primarily based on (usually triliteral) /C-C-C/ consonantal roots. These lexical roots pass through elaborate conjugational, inflectional and derivational patterns and systems in order to form words. Some of these grammatical conjugational and inflectional patterns (mishkalim/binyanin) are signaled by alternating vocalic and consonantal — vocalic affixes (prefixes, infixes, suffixes). Different synthetic, semantic and syntactic forms are systematically created through these mishkalim/binyanim .

2.1. Traditional and neotraditional analyses The Hebrew verb system consists of seven or eight conjugational and inflectional patterns¹ termed binyanim, which are signaled by alternating vocalic and consonantal-vocalic affixes (prefixes, infixes and suffixes). The names of the traditional seven major binyanim (PAAL, NIFAL, PIEL, PUAL, HIFIL, HUFAL, HITPAEL) are derived from the triconsonantal (CCC) root P-‘-L / ‘action’ and the fixed morphophonemic pattern of each binyan. The traditional seven major binyanim usually have been defined according to their most frequent or prototypical function: simple, intensive, causative, reflexive and/or voice (active, passive). These terms, primary functions

Table 1. The binyan system: traditional view (1) (2) (3) (4) (5) (6) (7)

PAAL — ‘simple’ conjugation (also known as QAL / ‘basic/ simple’) NIFAL — ‘passive’ of ‘basic-simple’ conjugation (PAAL/QAL) PIEL — ‘intensive’ conjugation PUAL — ‘passive’ of ‘intensive’ conjugation (PIEL) HIFIL — ‘causative’ conjugation HUFAL — ‘passive’ of ‘causative’ conjugation (HIFIL) HITPAEL — ‘reflexive’ conjugation

Semantic oppositions in the Hebrew verb system Table 2. The binyan system: root–pattern–verb Root

Binyan

Pattern

Resulting verb

(1) (2) (3) (4) (5) (6) (7)

PAAL NIFAL PIEL PUAL HIFIL HUFAL HITPAEL

CaCaC niCCaC CiCeC CuCaC hiCCiC huCCaC hitCaCeC

KaTaV / ‘wrote’ niKTaV / ‘was written’ KiTeV / ‘inscribed’ KuTaV / ‘was inscribed’ hiKTiv / ‘dictated’ huKTaV / ‘was dictated’ hitKaTeV / ‘corresponded’

K-T-V K-T-V K-T-V K-T-V K-T-V K-T-V K-T-V

and designations of active and passive voice, have served, over the centuries, as the traditional labels for the binyanim as shown in Table 1. Most, if not all of the usually triconsonantal (CCC) roots in the language are combined with the morphophonemic patterns of one or more binyan to create a lexical verb. This “root + binyan = verb” process is illustrated in Table 2 (for the third person, masculine, singular past forms from which the traditional labels are taken) using the triconsonantal root K-T-V ‘write’.

2.2. A sign-oriented analysis of the binyan system There have been many attempts at characterizing the binyan system. According to Tobin (1995), however, all of these, whether syntactic or lexicalist, have not gone beyond the listing of a variety of semantic, syntactic and pragmatic functions of each binyan. Such listing and discussion of the various syntactic, semantic and pragmatic functions have failed to assign a single function or a set of functions consistently and systematically to each pattern. The result is that the binyan system appears to be asymmetrical and random. Tobin proposes that a set of invariant meanings may be postulated for each binyan, and that these comprise a larger semantically motivated grammatical system. He argues for an eight-member binyan system, with the eighth pattern being the passive reflexive NITPAEL. The binyan system is conceived as composed of three interlocked deictic systems, defined by the following semantic domains: 1. Domain 1: the Objective versus Subjective conceptualization of actions, states, or events: i.e. an objective description or more of a subjective description: this means that the encoder may either be giving an objective description of an action, state, or event, or making a more subjective assessment or comment on an action, state, or event. In other words, merely “presenting the facts” or “appraising the facts”, i.e. revealing more of a personal point of view, an impression, or opinions of the scene. For example, KaTaV-‘wrote’ as compared with KiTeV-‘inscribed’ (or in other words, ‘pressed down very hard on the paper’/ ‘wrote roughly’). The notion of Objective versus Subjective may best be viewed as a continuum or cline across the lines of denotation — connotation respectively.

237

238

Noah Oron and Yishai Tobin

2. Domain 2: the Single, versus Multiple perceptions of these actions, states, or events: this means that the action, state, or event may either be identified at a Single perception, or that a pluralty or more than one — a Multiple — perception is necessary to identify it. For example, KaTaV-‘wrote’ as opposed to hitKaTeV‘corresponded’. Within this domain, actions, states, or events requiring a plurality of Multiple perceptions may also be perceived within Integral space and time, i.e. occupying a single continuous space and time rather than discrete discontinuous space and time (in a way similar to Semantic Integrality (Tobin 1995: 71)). 3. Domain 3: the Automony of actions, states, or events: this means that the action, state, or event may be identified independently (Autonomous), or is dependent on other elements (either implicitly or explicitly) to be identified (Non-Autonomous). For example, hiKTiV-‘dictated’ as compared with huKTaV-‘was dictated’ (by whom?). Each binyan is defined by one value of each one of these systems. While the root presents the general semantic field of the action, state or event, the binyan conceptualizes and categorizes different ways of perceiving the action, state and event. This function of a binyan is termed its semantic domain. The different verb patterns are defined in the following manner with respect to the three semantic domains: (a) PAAL: objective, single perception, autonomous (KaTaV-‘wrote) (b) HITPAEL: objective, multiple perceptions, autonomous (hitKaTeV-‘corresponded’) (c) PIEL: subjective, single perception, autonomous (KiTeV-‘inscribed’) (d) HIFIL: subjective, multiple perceptions, autonomous (hiKTiV-‘dictated’) (e) NIFAL: objective, non-autonomous, multiple perceptions (of original single perception) (niKTaV-‘was written’) (f) NITPAEL: objective, non-autonomous, multiple perception (of original multiple perception) (nitKaTeV-‘was written in the correspondence’) (g) PUAL: subjective, non-autonomous, multiple perceptions (of original single perception) (KuTaV-‘was inscribed’) (h) HUFAL: subjective, non-autonomous, multiple perceptions (huKTaV-‘was dictated’) Tobin stresses that these interlocked invariant meanings are not absolute but rather relative in nature. When a root appears in all eight patterns, all of the opposed markedness values are defined relative to each other. If a root is only partially distributed within the system, some of these semantic features may be neutralized. However, if a root appears in more than one pattern, it will maintain the opposed markedness values for those patterns in which it appears. The fact that not all roots appear in all conjugations is related to semantic/pragmatic holes in meaning in the system which, in some cases, are filled by other roots. Junger (1988) lists which roots appear in which conjugation but no one, to the best of our knowledge, has attempted to explain the nature of this distribution.

Semantic oppositions in the Hebrew verb system

Since the binyan system falls somewhere between inflectional and derivational morphology, the meaning of a certain verb form is not entirely predictable based on the verb pattern. Rather, the meanings of verb forms are motivated by the meaning assigned to the prefix/infix of the different binyanim. The binyan system is not a totally closed one, in the sense that it is productive to a certain extent. Berman and Sagi (1981) list verbal neologisms from various studies on Hebrew child language. They assume that children do this either to fill lexical gaps in the system or in their own lexicons. Some of these have entered normative language such as kile’ax-‘washed (trans)’, a PIEL form of hitkale’ax (HITPAEL pattern)showered himself. Based on their examples, children coin new verbs in all the four autonomous verb patterns. The reason that children do not coin verbs in the non-autonomous verb patterns is probably that these are conceptually more complex. Another interesting observation made by Berman and Sagi (p. 41), that lends support to this assumption, is that children very often inflect a HITPAEL verb form in the NIFAL conjugation (in terms of Tobin’s system, HITPAEL and NIFAL differ only in one domain: NIFAL is non-autonomous and HITPAEL is autonomous). Hebrew-speaking adults very often create new verbs for the purpose of being ironic or facetious or making subjective remarks. An interesting example is the newly coined hitputara combination of the HITPAEL form hitpater (‘resigned’) and the PUAL form putar (‘dismissed’)- meaning that a person was ‘induced by his/her superiors to resign.’

3. PAAL–HITPAEL verb alternations In what follows, we will present two analyses of PAAL–HITPAEL verb alternations- a traditional sentence-oriented analysis (based on Oron 2001) and a sign-oriented analysis. It will be shown that a traditional approach, because of conceiving the sentence as the basic unit of analysis and assuming preconceived categories, is unable to make the same types of extensive generalizations as a sign-oriented approach. Signoriented theory differs from more traditional approaches to language in that it views language as a holistic code used by human beings for communicating. This code is comprised of linguistic signs which integrate signals and meanings and the relationship between them. The distribution of linguistic signs and their relationships are motivated by invariant meaning and the connection between signal and meaning may be iconically reflected. Moreover, the invariants of the language code are not preconceived categories, but rather must be discovered. Another important point is that the invariant meanings that underlie the code level are not equivalent to sentence or dictionary or discourse meanings and messages nor to the syntactic, logical or pragmatic communicative discourse functions and exploitations of this code.

3.1. PAAL–HITPAEL verb alternations — a traditional analysis As previously stated, the binyanim have been traditionally defined according to

239

240 Noah Oron and Yishai Tobin

their most frequent meanings or functions and/or voice. Under this approach, PAAL represents the ‘simple’ or ‘basic’ conjugation, and HITPAEL represents reflexive conjugation (e.g. Berman 1975a). There are roughly 150 roots that can appear in both binyanim. However, only a relatively small amount of these (24) fit the ‘basic-reflexive’ dichotomy. Another possibility of exploring the difference in meaning between PAAL and HITPAEL is to group the verb pairs into different verb categories, and to see whether a consistent distinction can be found between the two binyanim within each group. We will attempt to do this with intransitive verbs of motion. There are ten such verb pairs, and they are listed in Table 3. A consistent distinction can be found between the PAAL and HITPAEL forms of the first six verb pairs: both forms denote activities with similar semantic content. However, only the PAAL form can be telicized by adding a goal phrase, whereas the HITPAEL form remains an activity even when the same type of goal phrase is affixed to it. This point can be illustrated by the following example: (1) a.

Ha-ish halax me-ha-gan la-bait be-sha’a. the-man walk.PAAL.PAST from-the-park to-the.house in-hour ‘The man walked from the park to the house in an hour.’ b. Ha-ish hithalex me-ha-gan la-bait bemeshex the-man walk.HITPAEL.PAST from-the-park to-the.house during sha’a. hour ‘The man walked back and forth from the park to the house for an hour.’

Table 3. PAAL-HITPAEL intransitive verbs of motion Verb root

PAAL form

English gloss

HITPAEL form

English gloss

1. h.l.x. 2. r.ts. 3. n’.a.

halax rats n’a

walked ran moved

hithalex hitrotsets hitnoe’a

4. z.z.

zaz

moved

hitazez

5. n.d. 6. n.o.d.

nad (berosho) bowed (his head) nadad wandered, migrated (birds) ‘af flew savav encircled, spun, turned around, wandered galash surfed, skied, slid zaxal crawled

walked back and forth ran back and forth moved from side to side, back and forth ran back and forth from one place to another swung back and forth swayed back and forth

7. .‘.f. 8. s.o.v. 9. g.l.sh. 10. z.x.l.

hitnadned hitnoded hit’ofe histovev

flew away, took off spun, rotated, turned around

hitgalesh hizdaxel

slid down wriggled, crawled into

Semantic oppositions in the Hebrew verb system

This example can be said to illustrate that the two verbs with similar semantic content, in identical syntactic structures, have different aspectual interpretations. In the second structure, it can be claimed, the verb form hithalex has taken on an iterative meaning, and this is the reason why it cannot be delimited by a frame adverbial such as in an hour. This claim can be extended to all of the HITPAEL forms under consideration. More specifically, it can be argued that the HITPAEL forms have an iterative component in their Lexical Conceptual Structure, which is only made explicit when an adjunct defines two or more points in space in between which the motion is carried out. These observations, however, fail to explain why these verbs, despite the fact that they retain their basic semantic content in both patterns, gain an iterative component when conjugated in the HITPAEL pattern. The solution would have to lie in a difference in meaning between the two patterns. This difference would have to be consistent since in all of the cases it was the HITPAEL form that carried an iterative component (and there are no examples of an alternation where the HITPAEL form carries the ‘basic’ meaning and the PAAL form gains an iterative component). Furthermore, ‘iteration’ would have to be only one aspect of this difference in meaning between the two patterns, since the distinction between the remaining four PAAL– HITPAEL verb alternations cannot be described in the same way. The problem at the core of this analysis is the preconception of the category intransitive verbs of motion. First, it prevents addressing the fact that other PAAL–HITPAEL alternations display the same distinction despite not being ‘verbs of motion’, for example, nasham‘to breathe’ vs. hitnashem- ‘to pant’. Second, it does not allow the discovery of other types of distinctions between PAAL–HITPAEL alternations which might lead to a generalization about the fundamental difference between the two patterns. In any case, however, such a generalization would be difficult to reach within a sentence-based approach, as former attempts in the literature have shown. An example of such an attempt is Berman (1975b). Berman lists a wide variety of uses/ meanings associated with HITPAEL which appears to be riddled with inconsistencies and contradictions. For example, the pattern HITPAEL is characterized by a mixture of syntactic and semantic features; it is said to carry the following possible meanings: 1. intransitivity; 2. reciprocity; 3. iteration. In addition, HITPAEL is said to function as the middle form of verbs appearing in PIEL, and by virtue of that, to be a derived pattern par excellence. However, verb forms appearing only in HITPAEL and carrying a “basic” meaning are also cited. An additional feature attributed to HITPAEL is that it conveys an intensified or more specific meaning in relation to its PAAL or PIEL counterparts, but other examples where HITPAEL behaves “idiosyncratically” vis-à-vis PAAL and PIEL are also given. Berman concludes (p. 37) that a systematic description of the binyan system of contemporary Hebrew is difficult to achieve because of the multitude of gaps and inconsistencies. However, she postulates that some of the consistencies outlined must have a basis, otherwise certain generalizations would go unexplained and it would imply that speakers of the language must acquire each lexical item separately and independently.

24

242 Noah Oron and Yishai Tobin

Mandelblit (2000) has taken a different approach in attempting to characterize the different patterns of the binyan system. She proposes that, while the verb root provides the semantic content about participants and relations in the conceived event, and the syntactic construction defines the generic structure of the event, the binyan marks the relative role the participants play in the more elaborate conception of the event structure. Mandelblit outlines a ‘blending schema’ that is typical of each pattern. This approach is different in positing that the grammatical meanings traditionally associated with the different patterns are not inherent to the patterns themselves, but rather produced by the different mapping schemes. However, none of these schemes is consistent for all of the forms that can appear in a particular pattern. For PAAL, for example, the blending characterization works only for transitive verbs.

3.2. PAAL–HITPAEL verb alternations within sign-oriented theory We will now explore the question of PAAL–HITPAEL verb alternations from a signoriented approach as outlined in Tobin (1995), extending the discussion to all 150 roots displaying this alternation, based on entries from RAV-MILIM: a comprehensive dictionary of Modern Hebrew by Choueka (1997). Most English glosses were taken from the Oxford English–Hebrew/Hebrew–English Dictionary (1995). In those cases where the verbs had newer or more common or colloquial messages than those that appeared in the dictionary, we provided our own glosses. This will be the first time that the PAAL–HITPAEL connection will be tested exhaustively according to Tobin’s conception of the Hebrew verb system. Under this approach, each binyan is characterized according to three semantic domains: (1)objective versus subjective conceptualization of actions, states or events; (2)single versus multiple perception of actions, states or events; (3)autonomous versus non-autonomous character of actions, states and events from the point of view of the decoder at the point of encoding. In this system, PAAL is defined by the following three features: objective, single perception and autonomous. HITPAEL is defined thus: objective, multiple perceptions, autonomous. As is apparent, PAAL and HITPAEL share the features objective and autonomous, but differ with respect to the domain of ‘perception’: whereas PAAL is marked for single perception, HITPAEL is marked for multiple perception. This means that an action, state or event denoted by the pattern PAAL is identified at a single perception (e.g. KaTav-‘wrote’), whereas an action state or event denoted by the pattern HITPAEL requires more than one perception (e.g. hitKaTeV‘corresponded’). Tobin points out (p. 255) that the simpler morphological forms are marked for the simpler single perception, whereas the more complex morphological forms, like HITPAEL and HIFIL that bear the prefixes HIT- and HI respectively, are marked for multiple perception reflecting language iconicity. Some of the PAAL forms may be arguably said to be non-uniform in their structure, therefore requiring more than a single perception. We propose here, then, that PAAL is identified by a singular perception (a single or an integral multiple percep-

Semantic oppositions in the Hebrew verb system

tion) whereas HITPAEL is identified by a plurality of perceptions. An alternative analysis to the one presented above, therefore, is that with respect to the domain of ‘perception’, PAAL is unmarked for number of perceptions and HITPAEL is marked for plurality of perceptions, always having more complex perceptions than the corresponding PAAL form.

3.2.1. The data

According to Stern (1994), the HITPAEL pattern cannot be characterized by a coherent semantic system. He recognizes seven message types that can be conveyed by the HITPAEL pattern, apart from iteration which is usually attributed to this pattern: 1. a one-place predicate with a basic meaning; 2. a two-place predicate with a basic meaning; 3. reciprocity; 4. a process or coming into being; 5. a state; 6. a sort of passive; 7. reflexiveness. Stern’s description of the different binyanim is among the most thorough and extensive, exemplifying the accepted categories created by sentence-oriented grammarians. For this reason the PAAL–HITPAEL alternations will be examined according to this categorization of HITPAEL. We will attempt to show that a distinction based on number of perceptions can achieve systematicity in explaining the difference in meaning between the two verb forms, and that it can attribute a consistent meaning component to each one of the verb patterns regardless of traditional verb type categorizations. Obviously some of the meaning types overlap, and some forms may be classified in more than one way. In these cases, the classification that most emphasizes the distinction between the PAAL and HITPAEL forms will be chosen. 3.2.1.1. Iteration. In some cases, both verb forms denote the same type of action (due to the invariant meaning of the root), however, the HITPAEL form carries an iteration meaning component. This is true for the six pairs of “intransitive verbs of motion” discussed earlier, but it is also true for four additional PAAL–HITPAEL alternations, as seen in Table 4. As for the first six verb pairs discussed earlier, the HITPAEL forms necessarily require more than one perception on the part of the decoder in order to perceive the action since the action is not uniform; it consists of repetitions of the action depicted by the PAAL form, which only requires a singular perception. The last four verb pairs can also be consistently distinguished in that the HITPAEL form of the verb conveys a repetitive action that is semantically similar to its PAAL counterpart: 7) PAAL- knock vs. HITPAEL-knock repeatedly; 8) PAAL- cry vs. HITPAEL-whine; 9) PAAL- exhale vs. HITPAEL-pant/gasp; 10) PAAL- inhale vs. HITPAEL pant/gasp. 3.2.1.2. Reciprocity. The following PAAL–HITPAEL verb pairs also depict actions with similar semantic content, with the difference that the HITPAEL forms also convey reciprocity. This means that the action is directed towards an entity, and that that same entity directs the action back without this being explicitly stated. Also, when the action is carried out by more than one entity, it is implicitly understood that the entities are directing the action towards one another and, therefore, more than a sin-

243

244 Noah Oron and Yishai Tobin Table 4. HITPAEL verbs of iteration Verb root

PAAL form

English gloss

HITPAEL form

English gloss

1. h.l.x. 2. r.ts. 3. n.a.

halax rats na

walked ran moved

hithalex hitrotsets hitnoe’a

4. z.z.

zaz

moved

hitazez

walked back and forth ran back and forth moved from side to side, back and forth ran back and forth from one place to another swung back and forth swayed

5. n.d. 6. n.o.d.

hitnadned hitnoded

7.

hitdaPek

8. 9. 10.

nad (berosho) bowed (his head) nadad wandered, migrated (birds) d.p.k.² daak knocked (on a door) b.x.a. baxa cried n.sh.f. nashaf blew (air), exhaled n.sh.m. nasham breathed, inhaled

hitbaxjen hitnashef hitnashem

knocked (on a door) repeatedly whined panted, gasped panted, gasped, breathed heavily

gular perception is required for decoding the action. In some cases, the similarity in semantic content between the PAAL form and the HITPAEL form is not as transparent as in the previous cases; however, it is still evident as may be seen in Table 5. The HITPAEL forms in examples (1), (2), (3), (4), (5), (8) and (9) all denote an action that is carried out with an additional participant, conveyed by the preposition with that is absent from the English glosses of their PAAL counterparts:1) PAAL-see vs. HITPAEL- meet with someone; 2) PAAL-compose rhymes vs. HITPAEL- rhyme with; 3) PAAL-write vs. HITPAEL- correspond with; 4) PAAL-crush vs. HITPAEL- fight with; 5) PAAL-whisper vs. HITPAEL- communicate with someone by whispering; 8) PAALrape vs. HITPAEL- have sexual intercourse with; 9) PAAL-unite vs. HITPAEL- become united with. In other words, the HITPAEL forms denote a more complex activity, requiring more than a singular perception in order to be decoded. Note that although in example 8) both shagal and hishtagel can mean had sexual intercourse with, only hishtagel implies that all of the participants in the act were active. Note also that only shagal can mean raped which is clearly not reciprocal. In examples 7) PAAL-kissed (someone) vs. HITPAEL-kissed each other, 11) PAAL-hugged vs. HITPAEL-hugged each other, 12) PAAL-gored vs. HITPAEL-gored each other, 13) PAALheard vs. HITPAEL-heard from each other, and 14) PAAL-came closer to vs. HITPAELcame closer to each other (fig.) , the PAAL forms describe an action that involves an agent and possibly a patient, but only the HITPAEL forms denote an action with multiple active participants, as conveyed by the pronoun each other. This holds for the HITPAEL forms in examples 6) PAAL-poured vs. HITPAEL-merged, mingled together and 10) PAAL-gathered dates vs. HITPAEL-gathered together, as witnessed by the adverb together.

Semantic oppositions in the Hebrew verb system Table 5. HITPAEL verbs of reciprocity

1. 2. 3. 4. 5.

Verb root

PAAL form

English gloss

HITPAEL form

English gloss

r. ʔ.a. x.r.z. k.t.v. k.t.sh. l.x.sh.

ra ʔ a xaraz katav katash laxash

saw (something) composed rhymes wrote pounded, crushed whispered

hitraʔa hitxarez hitkatev hitkatesh hitlaxesh

poured kissed (someone) had sexual intercourse with, raped united, bound, brought together

hitmazeg hitnashek hishtagel

picked or gathered dates embraced (someone), encircled gored, butted

hitgoded

met with someone rhymed with corresponded with fought with communicated with someone by whispering merged, mingled together kissed (each other) had sexual intercourse with someone became united with, formed a cooperative with gathered together (intrans.)

hitxaBek

hugged (each other)

hitnageax

heard came closer (to something or someone)

hishtamea hitkarev

gored (each other), argued vehemently heard from each other came closer (to someone or something or, figuratively, to each other)

6. m.z.g. mazag 7. n.sh.k. nashak 8. sh.g.l. shagal 9. ʔ.g.d.

ʔagad

10. g.d.d.

gadad

11. x.B.k.

xaak

12. n.g.x.

nagax

13. sh.m.a. shama 14. k.r.v. karav

hitʔaged

All of the HITPAEL forms in the table above have an additional participant understood to play an active role. This is not true of their PAAL counterparts. The HITPAEL forms, therefore, clearly require more than a singular perception. 3.2.1.3. Process or coming into being. A large number of verbs in the HITPAEL form that have a PAAL counterpart denote a process or a coming into being. This component is missing from most of the PAAL counterparts. A process or coming into being implies that the action, state or event conveyed by the verb is not uniform and therefore requires more than a singular perception on the part of the decoder. In those cases where a PAAL form does denote a type of process or coming into being, the focus of the event is on the end-point of the process, whereas their HITPAEL counterparts denote both the process and result- clearly a more complex event. This is illustrated in Table 6. Nearly all of the HITPAEL forms in Table 6 explicitly denote a process or coming into being as denoted by the words became or grew in their English glosss. This is true for the following examples 1)PAAL-shone vs. HITPAEL-became shiny, 2)PAAL-reached old age vs. HITPAEL-grew older, 3)PAAL-raged vs. HITPAEL-became enraged, 4)PAAL-was furious vs. HITPAEL-became furious, 5)PAAL-rejoiced vs.

245

246 Noah Oron and Yishai Tobin

HITPAEL-became exultant, 6) PAAL-eroded vs. HITPAEL-became eroded, 7) PAAL-preserved vs. HITPAEL-became preserved, 8) PAAL-worked vs. HITPAEL-became enslaved, 9) PAAL-bristled vs. HITPAEL-came to be bristled, 10) PAAL-prevailed vs. HITPAELcame into existence, became perceived, 11) PAAL-remained silent vs. HITPAEL-became silent, 12) PAAL-lasted vs. HITPAEL-grew longer, 13) PAAL-was angry vs. HITPAELbecame angry, 14) PAAL-was agitated vs. HITPAEL-became agitated, 15) PAAL-ate, consumed vs. HITPAEL-became consumed, 16) PAAL-reached adulthood vs. HITPAELmatured, became an adult. Note that, although both forms can mean ‘to reach adulthood’, only the HITPAEL form can mean ‘mature intellectually/emotionally’, clearly denoting a more complex event. 17) PAAL-fabricated, invented vs. HITPAEL-became disillusioned, 19) PAAL-withered vs. HITPAEL-became withered, worn out, 20) PAALswallowed vs. HITPAEL-became confused (the meaning of hitbala is probably derived from the literary expression hitbal’a da’ato, literally, ‘his reason was swallowed’, and metaphorically, ‘he lost his sanity’) 21) PAAL-raged, seethed vs. HITPAEL-became turbulent and seething, 22) PAAL-grew, grew up vs. HITPAEL-became large or lofty, 23) Table 6. HITPAEL verbs of process or coming into being Verb root

PAAL form

English gloss

HITPAEL form

English gloss

1. z.h.r. 2. z.k.n. 3. z.’.f. 4. z.’.m. 5. ts.h.l. 6. sh.x.k. 7. sh.m.r. 8. ‘.B.d. 9. s.m.r. 10. s.r.r.

zahar zakan za’af za’am tsahal shaxak shamar ’aad samar sarar

shone, glowed reached old age raged, scowled was furious rejoiced wore down, eroded guarded, preserved worked, worshiped bristled prevailed

hizdaher hizdaken hizda’ef hizda’em hitztahel hishtaxek hishtamer hishta’aBed histamer histarer

11. sh.t.k.

shatak

hishtatek

12. ʔ.r.x.

ʔarax

hit ʔ arex

grew longer, dragged on

13. 14. 15. 16.

ragaz ragash ʔaal bagar

remained silent, refrained from or stopped speaking lasted (a certain period of time) was angry seethed, was agitated ate, consumed graduated, reached maturity

became shiny grew older became enraged became furious became exultant became worn down, eroded became preserved became enslaved, subjugated came to be bristled came into existence, became perceived became silent

hitragez hitragesh hit ʔaKel hitbager

became angry became agitated, excited became consumed matured (intellectually/ emotionally), became an adult became disillusioned burst open, split open

r.g.z. r.g.sh. ʔ.K.l. b.g.r.

17. b.d.a. 18. b.k.a.

bada baka

fabricated, invented hatched, emerged, burst

hitbada hitbakea

Semantic oppositions in the Hebrew verb system 247

PAAL-knew vs. HITPAEL-became familiar with, 24) PAAL-dried vs. HITPAEL-became dried, 25) PAAL-was weary of vs. HITPAEL-became weary of, 26) PAAL-sat vs. HITPAEL-became seated, 27) PAAL-darkened vs. HITPAEL-became darker, 28) PAAL-was angry vs. HITPAEL-became angry, 29) PAAL-burned, blazed vs. HITPAEL-grew hotter, became intense, 30) PAAL-sold vs. HITPAEL-became addicted to (literally, ‘sold himself ’ to a cause, a habit. Probably derived from the expression maxar et nafsho la-satan- ‘sold his soul to the devil’) , 32) PAAL-boiled, was angry vs. HITPAEL-became angry, 33) PAAL-tired of vs. HITPAEL-became tired of, 35) PAAL-beat (intrans.), throbbed vs. HITPAEL-became excited with, 38) PAAL-was dark, gloomy vs. HITPAELbecame dark, gloomy, 38) PAAL-was diminished vs. HITPAEL-became diminished, and 39) PAAL-thought, assumed vs. HITPAEL-became known. In example 31) a similar distinction exists between the meaning of the PAAL form- smelt/reeked of- and the meaning of the HITPAEL form- evaporated, dissipated, disappeared. The PAAL form describes a state; the HITPAEL form describes a change of state, clearly a more complex event, requiring more than a singular perception in order to be decoded. 19. b.l.’. 20. b.l. ʔ. 21. g.’.sh

bala bala ga’ash

withered swallowed raged, seethed

hitbala hitbale ʔa hitga’esh

22. 23. 24. 25.

gadal jada jaash jaga

hitgadel hitvada hitjabesh hitjaga

hitkaha hitka’es hitlahet hitmaKer hitnadef

g.d.l. j.d.a. j.B.sh. j.g.a.

26. j.sh.v.

jashav

grew, grew up knew dried labored, toiled, was weary of sat

27. 28. 29. 30. 31.

kaha ka’as lahat maar nadaf

darkened was angry burned, blazed sold smelled of, reeked of

32. r.t.x. 33. ‘.j.f. 34. p.g.

ratax ajaf pag

35. p.’.m.

pa’am

boiled, was angry hitrateax tired of hit’ajef expired, vanished, hitpogeg stopped beat (heart), throbbed hitpa’em

36. p.k.a.

paka

37. k.d.r. 38. k.t.n.

kadar katan

39. s.B.r.

saar

k.h.h. k.’.s. l.h.t. m.K.r. n.d.f.

burst, tore, broke off, cracked, expired was dark, gloomy, sad was diminished in size, number or amount thought, assumed

hitjashev

hitpakea hitkader hitkaten histaBer

became withered, worn out became confused became turbulent and seething became large or lofty became familiar with became dried became weary of, toiled, labored sat down, became seated, settled became darker became angry grew hotter, became intense became addicted evaporated, dissipated, disappeared (fig.) became angry became tired (of) faded away, gradually vanished became excited or filled with admiration burst, cracked as a result of excessive pressure became dark, gloomy, sad became diminished in size, number or amount became known

248 Noah Oron and Yishai Tobin

Some of the PAAL–HITPAEL verb pairs seem to carry an identical message: 18) baka (‘hatched, emerged, burst’)-hitbakea (burst open), 19) bala (‘withered’)hitbala(‘became withered’), 24)javash (‘dried’)-hitjabesh (‘became dry’), 27) kaha (‘darkened’)-hitkaha (‘became darker’), 34) pag (‘expired’, ‘vanished’)-hitpogeg (‘faded away’, ‘gradually vanished’), 36) paka (‘burst’, ‘cracked’, ‘expired’)-hitpakea (‘burst, cracked’), 37) kadar (‘was dark’, ‘gloomy’)-hitkader (‘became dark, gloomy’), 38) katan (‘was diminished’)-hitkaten (‘became diminished’). If the assumption that the two binyanim- PAAL and HITPAEL- differ in terms of number of perceptions is correct, then a difference in message is expected despite the fact that in both cases the same verb root, with the same semantic content, is employed. Subtle as it may be, a difference exists in that, although both forms describe a process, the PAAL form focuses on the result of the process, whereas the HITPAEL form seems to denote both the process and the result (Tobin 1993). This can be illustrated by the following expressions with the paka-hitpaka alternation, where the forms are in complementary distribution: (2) a.

Savlanut-o pak’a. patience-POSSESSIVE expire-PAAL-PAST ‘He ran out of patience.’ b. hu hitpake’a mi-tsxok/ka’as/kin’a. he burst.HITPAEL.PAST of-laughter/anger/jealousy ‘He burst out laughing.’

The HITPAEL form very clearly depicts a process of tension building up until the final result, namely, two perceptions. The PAAL form, on the other hand, depicts a singular perception by focusing on the final result. The pair pag–hitpogeg differ in the following manner: (3) a.

ta’arix pag tokef date expire.PAAL.PAST validity ‘expiry date’ b. ha-arafel hitpogeg. the-fog expire.HITPAEL.PAST ‘The fog dissipated.’

As in the previous examples, the PAAL form depicts an event involving only an endpoint, therefore requiring a singular perception for decoding. The HITPAEL form, by contrast, depicts a process with a result/endpoint that requires a plurality of perceptions in decoding. Another way of demonstrating the difference in meaning between the two verb forms is by relating to their corresponding mishkalim. As stated in Section 2, roots are combined with morphophonemic patterns to create nominal, verbal and adjectival forms- binyanim in the case of verbal forms and mishkalim in the case of nominal and adjectival forms. Berman (1994) lists five “action nominals” (p.83) that correspond to each one of the “active verb binyan patterns”. The corresponding action

Semantic oppositions in the Hebrew verb system 249 Table 7. HITPAEL verbs of action

1. 2. 3. 4.

Verb root

PAAL form

English gloss

HITPAEL form

English gloss

ts.v.x. ts.m.x. ts.’.k. z.’.k.

tsavax tsamax tsa’ak za’ak

screamed grew shouted cried out

hitstaveax hitstameax hitsta’ek hizda’ek

began to scream began to grow began to shout let out a cry

nominal pattern for PAAL is CCiCa and the corresponding action nominal pattern for HITPAEL is hitCaCCut. The following nominal actions correspond to the verb pairs in 18), 19), 24), 27), 34), 36), 37) and 38). There are some gaps in the system so not all of the PAAL forms have a corresponding action nominal. All of the HITPAEL forms, however, have a corresponding nominal action that in all cases depicts a process and result, involving more than a singular perception. In those cases where an action nominal exists for both verb forms, the one corresponding to the HITPAEL verb form will always require more perceptions for decoding: 18) bki’a- ‘the act of emerging’; hitbak’ut- ‘the process of becoming cracked’, 19) blaia- ‘the disintegration of rocks’; hitbalut- ‘the process of becoming worn out’, 24)hitjabshut- ‘the process of drying’, 27) hitkahut- ‘the process of darkening’, 34) hitpogegut- ‘the process of becoming dissipated’, 36) pki’a- ‘expiry’; hitpak’ut- ‘the process of bursting’, 37) hitkadrut- ‘the process of becoming dark or sad’, 38) hitkatnut- ‘the process of becoming diminished’. There is an additional group of verbs that denote the initiation of an action. They differ from verbs that describe a process in that they focus on the starting point, as opposed to the end point, of an event. In both cases, however, the event can be said to be non-uniform and to require more than one perception in order to be decoded: in one case, both the beginning point of the action and the ongoing action must be perceived; in the other case, a process and its ending point or result must be perceived. These are illustrated in Table 7. In these examples, the HITPAEL form of the verb depicts two points in time: one prior to the initiation of the action and one subsequent to it. Consequently, more than a singular perception is required on the part of the decoder. 3.2.1.4. Reflexiveness. Several PAAL–HITPAEL alternations denote very similar semantic contents. In these cases, the HITPAEL verb forms differ from the PAAL verb forms in that they encode reflexiveness. This means that the performer of the action is also the recipient of the action. Since the recipient of the action is different from what would ordinarily be expected given the lexical meaning of the verb, more than a singular perception is required for the conceptualization of the event. This may be seen in Table 8. Most of the HITPAEL forms in Table 8 are evident reflexives of their PAAL counterparts as the pronouns himself and oneself suggest. This is true for the following examples: 1) PAAL-straightened up (trans.) vs. HITPAEL-straightened himself up,

250

Noah Oron and Yishai Tobin Table 8. HITPAEL reflexive verbs Verb root

PAAL form

English gloss

HITPAEL form

English gloss

1. z.k.f. 2. z.k.r.

zakaf zakar

hizdakef hizdaker

straightened himself up stood out, protruded

3. 4. 5. 6.

sagar sarag tsaa tsaar

straightened something up straightened something up or thrust out closed knit painted accumulated, gathered (something) twined, interweaved spread something out rolled something, rolled something up flowed turned something over washed (something) collected or gathered (something) wore (something) handed over

histager histareg hitstaBea hitstaBer

shut himself up entwined himself painted himself accumulated, amassed

hishtazer hishtateax hitgolel

became interwoven stretched himself out rolled about, wallowed

hitkaleax hithaPex hitraxets hit ʔasef

showered (himself) turned (himself) over washed himself accumulated

hitlaBesh hitmaser

dressed himself devoted himself, gave himself (sexually) smeared himself with something called himself (by a name) wrapped himself up stretched himself spread (himself out), extended (himself) undressed himself resigned, rid oneself (of something) bowed, bent down laughed to himself

s.g.r. s.r.g. ts.B.a. ts.B.r.

7. sh.z.r. shazar 8. sh.t.x. shatax 9. g.l.l. galal 10. 11. 12. 13.

k.l.x. h.P.x. r.x.ts. ʔ.s.f.

kalax haax raxats ʔasaf

14. l.B.sh. laash 15. m.s.r. masar 16. m.r.x. marax

spread (something), smeared called (someone by name)

hitkara

18. ‘.t.f. ’ataf 19. m.t.x. matax 20. p.r.s. paras

wrapped something up stretched (something) spread out (something)

hit’atef mitmateax hitpares

21. p.sh.t. pashat 22. p.t.r. patar

took off (clothes) exempt, dismissed

hitpashet hitpater

23. sh.x. shax 24. ts.x.k. tsaxak

bowed (his head) laughed

hishtoxeax hitstaxek

17. k.r.a.

kara

hitmareax

3) PAAL-closed vs. HITPAEL-shut himself up, 4) PAAL-knit vs. HITPAEL-entwined himself, 5) PAAL-painted vs. HITPAEL-painted himself, 8) PAAL-spread something out vs. HITPAEL-stretched himself out, 10) PAAL-flowed vs. HITPAEL-showered (himself ), 11) PAAL-turned something over vs. HITPAEL-turned (himself) over, 12) PAAL-washed (trans.) vs. HITPAEL-washed (himself), 14) PAAL-wore vs. HITPAEL-dressed himself, 15) PAAL-handed over vs. HITPAEL-devoted himself, gave himself (sexually), 16) PAALspread, smeared vs. HITPAEL-smeared himself with something, 17) PAAL-called vs.

Semantic oppositions in the Hebrew verb system

HITPAEL-called himself (by a name), 18) PAAL-wrapped vs. HITPAEL-wrapped himself, 19) PAAL-stretched vs. HITPAEL-stretched himself 20) PAAL-spread out vs. HITPAEL-spread himself out, 21)PAAL-took off (clothes) vs. HITPAEL-undressed himself, 22) PAAL-exempt, dismissed vs. HITPAEL-resigned, rid oneself of, and 24) PAALlaughed vs. HITPAEL-laughed to himself. In examples 2) PAAL-straightened something up vs. HITPAEL-stood out, protruded, 6) PAAL-accumulated, gathered (trans.) vs. HITPAEL-accumulated, amassed, 7) PAAL-twined, interweaved vs. HITPAEL-became interwoven, 9) PAAL-rolled (trans.) vs. HITPAEL-rolled about, wallowed, 13) PAAL-collected or gathered (trans.) vs. HITPAEL-accumulated and 23) PAAL-bowed (his head) vs. HITPAEL-bowed, bent down, the PAAL and HITPAEL forms are very close in semantic content apart for the fact that for the HITPAEL forms the agent and patient are one and the same despite the fact that this is not made explicit in the gloss by means of a pronoun. In all of these cases, a multiple perception is inherent in the interpretation. 3.2.1.5. “Sort of passives”. According to Stern (1994), one of the message types carried by HITPAEL is that of “a sort of passive”. He gives the example of hitkabel which can be translated as ‘was accepted’. Stern treats this as a passive since it can be followed by a by-phrase. This particular form will not be analyzed since it has no PAAL alternation. However, the forms in Table 9, appearing to be the passive forms of their PAAL counterparts, were listed in Rav-Milim: a comprehensive dictionary of modern Hebrew by Choueka (1997). The distinction of singular perception versus a plurality of perceptions is still maintained since the HITPAEL forms all denote a process or change of state very similar in semantic content to the action depicted by their PAAL counterparts. Moreover, in all of the HITPAEL forms, the “subject” of the verb takes on expanded roles and is understood as playing the role of both agent and patient. The claim that the agent and patient are simultaneously implied is supported by the fact that these verbs cannot be followed by a by- phrase. Note that all of these HITPAEL forms have a NIFAL counterpart denoting the same type of event in terms of semantic content and process, but where the agent and the patient are distinct. This is in line with the stated assumption that HITPAEL and NIFAL differ in terms of the meaning componTable 9. HITPAEL ‘passive’ verbs Verb root

PAAL form

English gloss

HITPAEL form

English gloss

1. sh.k.x. 2. s.x.f. 3. s.t.m. 4. sh.B.r. 5. m.’.x. 6. ‘.k.r.

shaax saxaf satam shaar ma’ax ’akar

forgot swept away blocked, clogged broke (something) crushed, squashed extracted, uprooted

hishtaKeax histaxef histatem hishtaBer hitma’ex hit’aker

came to be forgotten came to be swept away came to be blocked, clogged came to be broken came to be crushed, squashed came to be extracted, uprooted

25

252

Noah Oron and Yishai Tobin Table 10. HITPAEL passive Verb root

PAAL form

English gloss

HITPAEL form

English gloss

sh.k.l.

shakal

weighed (trans.), considered

hishtaklel

had its relative part taken into account

ent referred to as autonomous/non-autonomous: the HITPAEL form depicts a process with a result, whereas the NIFAL form depicts a change-of-state with causation. An additional HITPAEL form that can be considered a type of passive and that has a PAAL alternation is illustrated in Table 10. The distinction between these two forms differs from the one described for the first six pairs. hishtaklel cannot be said to be a process of shakal but it clearly describes an event that is more complex and less uniform, and therefore requires a plurality of perceptions in order to be decoded. Note that the ‘real’ passive of shakal is derived in the NIFAL pattern, where an external agent is implied: nishkal- ‘was weighed’. 3.2.1.6. One-place predicates. Many HITPAEL forms that alternate with a PAAL form do not fit as neatly into one of the categories above. In terms of stern’s categorization, they can best be described as “one-place predicates with a basic meaning”. However, just as with the pairs already reviewed, the root will contribute similar semantic content to both forms, and the HITPAEL form will describe an event or action that is not uniform and that therefore requires more than one perception on the part of the decoder. This may be seen in Table 11. The difference in message for some of the PAAL–HITPAEL verb pairs listed above is quite subtle. For example, 9)PAAL-natsats and HITPAEL-hitnotsets both mean ‘to shine, sparkle’, however, there is a subtlety captured by Choueka’s dictionary meaning: natsats is said to mean ‘to glow as if emitting light’, whereas hitnotsets is defined as ‘emitting sparks of light’. This distinction clearly captures the singular perception versus plurality of perceptions dichotomy. Another pair of verbs that seem nearly identical in meaning are 13) PAAL-savav and HITPAEL-histovev. The difference between them can best be illustrated by a pair of colloquial expressions where only the HITPAEL form is acceptable, in the first expression, because an iterative event is being depicted, and in the second expression, because a process or change of state is being depicted. (4) a.

Histovev ben ha-raglajim. turn.HITPAEL.PAST between the-legs ‘Wandered around someone in an annoying and disruptive manner.’ b. histovev lo boreg. turn.HITPAEL.PAST to.him screw ‘Suddenly began acting in a strange manner.’

Semantic oppositions in the Hebrew verb system

In examples (1) and (10), the PAAL and HITPAEL forms also appear to convey very similar meanings, however, it is quite clear that the HITPAEL form is more complex than its PAAL counterpart: 1) PAAL-crawled as compared with HITPAEL-wriggled, crawled into. The HITPAEL form is more complex in that the first English glosswriggled contains an element of iteration and the second English gloss-crawled into depicts a process and result. Both these actions clearly require a plurality of perceptions in order to be decoded. A similar case can be made for example 10) PAALflew; HITPAEL-flew off, flew by battering wings. The first English gloss of the HITPAEL form depicts a process and result and the second English gloss contains an element of iteration. In 11) there is a greater semantic discrepancy between the two verb forms- PAAL-labored vs. HITPAEL-exercised. Nonetheless, a similarity in semantic content still exists but the HITPAEL form in this case also contains an element of iteration missing from the PAAL form. In examples (2), (3), (4) and (14), the HITPAEL forms denote a more complex activity in that they carry an element of reflexiveness wherein both agent and patient are simultaneously implied: 2)PAALwas right vs. HITPAEL-justified himself; 3) PAAL-got lost vs. HITPAEL-commited suicide, 4)PAAL-was depleted vs. HITPAEL-calibrated himself (both are derived from the root meaning ‘zero’, where depletion goes to zero and calibration is based on setting at zero), 14)PAAL-choked (trans.) vs. HITPAEL-choked (himself). Note that the Table 11. HITPAEL one-place predicates Verb root

PAAL form

1. z.x.l. zaxal 2. ts.d.k. tsadak 3. ‘.B.d. 4. ʔ.P.s.

’aad ʔaas

5. 6. 7. 8.

ran galash ratsa natsats

r.n. g.l.sh. r.ts.a. n.ts.ts.

9. n.m. 10. ‘.f. 11. ‘.m.l. 12. s.o.v.

nam ’af ’amal savav

13. sh.P.x. shaax 14. sh.n.k. shanak 15. sh.P.r. shafar

English gloss

HITPAEL form

English gloss

crawled was right (about something) got lost was depleted

hizdaxel hitstadek

wriggled, crawled into apologized, justified himself

hit’aBed hit ʔaPes hitronen hitgalesh hiratsa hitnotsets

committed suicide restarted, calibrated himself, came to his senses rejoiced slid down relented, was reconciled shone, sparkled, glittered

hitnamnem hit’ofef hit’amel histovev

drowsed off, drowsed flew off, flew by battering wings exercised spun, rotated, turned around

hishtaPex hishtanek hishtaper

‘poured his heart out’ choked (intrans.) improved

sang surfed, skied, slid wanted shone, sparkled, glittered slept, slumbered flew labored encircled, spun, turned around, wandered poured choked (trans.) had good qualities

253

254

Noah Oron and Yishai Tobin

form HIT’ABED-committed suicide is most probably derived from the more literary phrase (5) ‘ibed ‘atsmo la-da’at lose.PIEL.PAST himself to-reason ‘committed suicide’ The HITPAEL forms in examples (6), (7) and (9)-slid down; relented/was reconciled; drowsed off, respectively- are more complex than their PAAL counterparts-surfed/ skied/slid; wanted; slumbered, respectively- in that they depict a process as opposed to a uniform action. In examples 5) and 13), the HITPAEL forms are semantically more complex and depict actions that are less uniform than their PAAL counterparts, and therefore, more likely to require a plurality of perceptions in decoding: 5)PAAL- sang vs. HITPAEL-rejoiced; 13)PAAL-poured vs. HITPAEL-poured his heart out (which also includes an endpoint/result). In 15), there is a contrast between a state and a process with a result: PAAL-had good qualities vs. HITPAEL-improved(himself). 3.2.1.7. Two-place predicates. The motivation for this category is the same as for the previous one. Again it will be apparent that verbs in the HITPAEL pattern denote actions or events that are not uniform as compared with their PAAL counterparts, and that they, therefore, require a plurality of perceptions on the part of the decoder. This is shown in Table 12. In many of the examples above, the two verb forms denote a similar type of action or event, but whereas the PAAL form refers mainly to the result of the action or event, the HITPAEL form addresses both the process and the result. A similar distinction was found for the verb pairs in Table 6. As claimed before, a verb denoting both process and result requires a plurality of perceptions on the part of the decoder. This distinction is found in the following pairs: 5) PAAL-dwelled vs. HITPAEL-settled in, lodged, 7) PAAL-controlled vs. HITPAEL-gained control of, 9) PAAL-loved vs. HITPAELTable 12. HITPAEL two-place predicates Verb root

PAAL form

English gloss

HITPAEL form

English gloss

1. z.K.h.

zaa

won

hizdaKa

2. s.m.x.

samax

trusted, relied on (in general)

histamex

3. 4. 5. 6. 7. 8.

sa’ar shaha shaan saar shalat shamat

raged stayed dwelled hired, rented controlled dropped, let loose

hista’er hishtaha hishtaKen histaKer hishtalet hishtamet

was refunded for turning back equipment specifically relied on something/ someone to do something charge at, assault lingered, tarried settled in, lodged earned (wages) gained control of evaded, dodged

s.’.r. sh.h.h. sh.K.n. s.K.r. sh.l.t. sh.m.t.

Semantic oppositions in the Hebrew verb system 9. 10. 11. 12. 13. 14. 15. 16.

ʔ.h.v. k.m. g.r. g.n.v. x.B.t. x.m.k. x.P.r. x.sh.v.

ʔahav kam gar ganav xaat xamak xaar xashav

loved, was fond of rose lived somewhere stole beat, struck slipped away dug thought, believed

hit ʔ ahev hitkomem hitgorer hitganev hitxaBet hitxamek hitxaPer hitxashev

17. l.m.d. 18. n.g.d.

lamad nagad

studied conflicted with

hitlamed hitnaged

19. r.’.m.

ra’am

roared

hitra’em

20. n.P.l. 21. ‘.l.a.

naal ’ala

fell climbed, ascended

hitnaPel hit’ala

22. ‘.s.k.

’asak

hit’asek

24. j.’.ts 25. m.r.d.

did (for a living), was engaged in pakad commanded, counted, took a census of, visited, came upon ja’ats gave advice marad refused to abide by

26. n.x.l.

naxal

inherited, obtained, suffered (defeat), gained (victory)

hitnaxel

27. ʔ.z.r.

ʔazar

gathered (strength, courage)

hitʔazer

28. x.B.r.

xaar

united with, merged

hitxaBer

29. k.sh.r.

kashar tied up, brought together

hitkasher

30. b.l.t.

balat

hitbalet

23. p.k.d.

stood out, protruded, excelled in 31. g.B.r. gaar became stronger, bigger; defeated 32. n.k.m. nakam took revenge 33. p.r.ts. parats broke into; infiltrated; broke out 34. sh.l.x. shalax sent

hitpaked hitja’ets hitmared

hitgaBer hitnakem hitparets hishtaleax

fell in love with rose against lived somewhere temporarily stole into wrestled with a problem evaded dug himself into behaved in a thoughtful manner, took into account became apprenticed opposed expressed indignation at something pounced (on/at), attacked gave up a cause for the sake of a more noble one occupied himself with, fooled around with took part in a census, counted off attendance check consulted with someone opposed in an active or violent manner settled on land that was bequeathed to him, took over and settled on foreign land adopted (a behavior, characteristic), became stronger, greater became connected to, established a contact or connection phoned, established a connection (by modem, transmitter etc.), became emotionally attached to excelled in, stood out, protruded overcame; defeated; became stronger, bigger took revenge barged in, burst out, erupted, broke out lashed out at, assaulted verbally

255

256

Noah Oron and Yishai Tobin

fell in love with, 10) PAAL-rose vs.HITPAEL-rose against, 12) PAAL-stole vs. HITPAELstole into, 15) PAAL-dug vs. HITPAEL-dug himself into, 17) PAAL-studied vs. HITPAELbecame apprenticed, 20) PAAL-fell vs. HITPAEL-pounced at, attacked, 23) PAAL-visited, came upon vs. HITPAEL-counted off in an attendance check, and 27) PAAL-gathered (strength, courage) vs. HITPAEL-adopted (a behavior)/became stronger, greater. 34) PAAL- sent vs. HITPAEL- lashed out at. In other cases, the HITPAEL form of the verb describes a more complex event in that it is less uniform than that depicted by the PAAL form. Also, many of the HITPAEL forms take on a figurative meaning of their PAAL counterparts, which necessarily denotes a more complex event. The following pairs portray these distinctions: 1) PAAL-won vs. HITPAEL-was refunded for turning back equipment, 2) PAAL-trusted/ relied on vs. HITPAEL-relied on something/someone to specifically do something, 3) PAAL-stayed vs. HITPAEL-lingered/tarried, 8) PAAL-dropped/let loose vs. HITPAELevaded/dodged, 11) PAAL-lived somewhere vs. HITPAEL-lived somewhere temporarily, 13) PAAL-beat/struck vs. HITPAEL-wrestled with a problem, 14) PAAL-slipped away vs. HITPAEL-evaded, sneaked away from, 16) PAAL-thought/believed vs. HITPAEL-behaved in a thoughtful manner/took into account, 19) PAAL-roared vs. HITPAEL-expressed indignation at, 21) PAAL-climbed/ascended vs. HITPAEL-gave up a cause for the sake of a more noble one, 22) PAAL-did (for a living)/was engaged in vs. HITPAELoccupied himself with/fooled around with, 24) PAAL-gave advice vs. HITPAEL-consulted with someone, 25) PAAL-refused to abide by vs. HITPAEL-opposed in an active or violent manner, 26) PAAL-inherited/obtained/suffered (defeat)/gained (glory) vs. HITPAEL-settled on land that was bequeathed to him/took over and settled on foreign land, 28) PAAL-united with/merged vs. HITPAEL-became connected to/established a contact or connection and 29) PAAL-tied up/brought together vs. HITPAEL-phoned/established a connection (by modem, transmitter etc.)/became emotionally attached to. In two cases, the distinction between the two verb pairs lies in that the PAAL form depicts a state whereas the HITPAEL form depicts a process and result: 3) PAALraged vs. HITPAEL-charged at/assaulted and 18) PAAL-conflicted with vs. HITPAELopposed. In the following verb pairs, the HITPAEL form, as opposed to the PAAL form, depicts an event that occurs repeatedly over time, and therefore, requires more than one perception for decoding: 6) PAAL-hired/rented vs. HITPAEL-earned (wages). The last four pairs in Table 12 seem very similar, and in some cases, identical in meaning. However, their distribution differs in certain situations: this can best be exemplified through expressions and idioms. 31) gavar–hitgaber: (6) a.

gavar al-av jitsr-o. g.v.r..PAAL.PAST on-him desire-his ‘Was overcome by his desire.’ b. ka-ma’ajan ha-mitgaber. as-a.spring that-g.v.r.HITPAEL.PRESENT ‘That is constantly renewed and invigorated.’

Semantic oppositions in the Hebrew verb system

The verbs seem very different because of the other lexical items in the idiom but the semantic connection between them can be clarified. Both verbs are derived from the root meaning ‘strength’. In gavar alav jitsro, literally, ‘his desire defeated him’, the event depicted focuses on an endpoint. In kama’ajan ha-mitgaber, literally, ‘as the spring that intensifies’, the event depicted involves both a process and result (and possibly an element of iteration). 32) nakam–hitnakem: (7) a.

ha-shem ji-kom dam-o. the-name FUTURE-n.k.m.PAAL blood-his ‘God will avenge his death.’ b. ze jitnakem b-o. this FUTURE.n.k.m.HITPAEL in-him ‘It will backfire on him.’

The literal meaning of the expression with the HITPAEL verb form is ‘this will take revenge on him’, meaning that one will bear the consequences of one’s acts. Clearly this is a more complex event in terms of perception than the act of avenging depicted by the PAAL form. 33) Parats–hitparets: (8) a.

parats la-bajit p.r.ts.-PAAL.PAST to.the-house ‘Broke into the house.’ b. hitparets la-bajit. p.r.ts.HITPAEL.PAST to.the-house ‘Burst into the house.’

The PAAL form parats focuses on an endpoint- the breaking into the house, whereas the HITPAEL form hitparets requires the decoder to conceive not only the endpoint of the agent being in the house but also the process of entering the house. The difference in meaning between the pair 30) balat–hitbalet can best be explained by referring to their corresponding action nominals: blita- ‘protrusion’ and hitbaltut- ‘the act of standing out, excelling’. Clearly the HITPAEL action nominal is broader and more complex in meaning. These examples illustrate that the HITPAEL form depicts an event that is more complex and less uniform, and hence more likely to require a plurality of perceptions on the part of the decoder. 3.2.1.8. States. Only two forms seem to fit this category, although both can also fit into one of the previous categories (such as ‘one-place predicate’ in the case of the first verb and ‘iteration’ in the case of the second verb). In both cases shown in Table 13, it is clear that the state depicted by the HITPAEL form-was overcome by the desire in example (1) and knew his way around, was acquainted with a certain topic

257

258

Noah Oron and Yishai Tobin Table 13. HITPAEL statives Verb root 1. x.sh.k. 2. m.ts.

PAAL form

English gloss

xashak desired matsa found

HITPAEL form

English gloss

hitxashek (lo) was overcome by the desire hitmatsa knew his way around, was acquainted with a certain topic

in example (2) — is more complex then the event depicted by the PAAL form-desired in example (1) and found in example (2) — and that it requires more than a singular perception on the part of the decoder.

4. Summary and conclusions The issue of PAAL–HITPAEL pattern alternations within the Hebrew verb system was explored first from a traditional sentence-oriented approach and then from a sign-oriented approach. In the first analysis, only intransitive verbs of motion were analyzed since it was claimed that the traditional labeling of PAAL (‘basic’ conjugation) and of HITPAEL (‘reflexive’ conjugation) could not account for the distinction between most of the 150 PAAL–HITPAEL verb pairs. Ten such alternations were examined but a consistent distinction was found for only six of the pairs. For these six pairs it was claimed that both forms shared the same semantic content but that the verbs in the HITPAEL form took on an iterative meaning component, and consequently differed in their event structure, so that as opposed to the PAAL forms, they could not be telicized. This observation did not explain why it was the HITPAEL form that took on an iterative meaning (and that there were no examples of the reverse occurring- a PAAL form taking on an iterative meaning component). Also, it could not account for the additional four verb alternations which also shared similar semantic content but did not display this distinction. Nor for that matter, could it account for the numerous other PAAL–HITPAEL alternations with similar semantic content that did not display this distinction or other PAAL–HITPAEL alternations with different verb types that did display this distinction. A second, sign-oriented analysis was extended to all 150 PAAL–HITPAEL alternations, based on an analysis of the binyan system put forth by Tobin (1995). According to this analysis, each pattern within the system carries a set of invariant meanings, which together comprise a larger semantically motivated grammatical system. This means that the patterns are defined relative to each other, and therefore, a pattern should not be contemplated independently but rather juxtaposed to the other patterns in the system. In this framework, each pattern is defined by three semantic domains which determine how the action, state or event conveyed by the verb is perceived: (1) objective/subjective; (2) single/multiple perceptions; and (3) autonomous/non-autonomous. In terms of PAAL and HITPAEL, they are said to both be ob-

Semantic oppositions in the Hebrew verb system

jective and autonomous but to differ in terms of perceptions: while PAAL conveys an action, state or event that requires a single perception in order to be decoded, HITPAEL conveys actions, states and events that require multiple perceptions in order to be decoded. In this second analysis, the HITPAEL verb forms were grouped according to the different semantic features attributed to this pattern by Stern (1994). It was shown that regardless of this classification, a single and consistent difference in meaning could always be found between the HITPAEL form and its PAAL alternation. This difference, in terms of number of perceptions required for decoding, explains why, when one of the verb alternations has an iterative meaning component, it will always be the verb in the HITPAEL form. Even in those cases where the PAAL form denotes an event that appears to require more than a single perception, the HITPAEL counterpart always involves more perceptions. Indeed, we proposed an alternative analysis whereby the PAAL verb forms are unmarked for number of perceptions, but are always simpler in this respect when opposed to their HITPAEL forms. This issue warrants further attention in future analyses of this type. Since the binyan system overlaps with both grammar and the lexicon, we do not claim that the meanings of the HITPAEL forms are entirely predictable based on the corresponding PAAL forms. That is, the meanings of the HITPAEL verbs are not transparently predictable in a compositional fashion (i.e. verb stem + pre-/infix > resulting meaning). Rather, their meanings are motivated by the meaning assigned to the pre-/infix. This means that even when a verb goes on to develop its own idiomatic uses, which can no longer be accounted for by the source semantics of the root, it still retains the notion of perceptions. For example, the verb form hitapes, derived from the noun efes ‘zero’, and which originally meant ‘recalibrated itself ’, has taken on the colloquial meaning of ‘regained one’s senses’ (both share the common message of returning to the starting point, namely, zero). An analysis of this type truly captures generalizations that were missing from the initial analysis of PAAL–HITPAEL verb alternations and from other analyses of its type. Previous research (e.g. Sachs 1987 and Sachs 2001) and the present attempt, are lending further support for a sign-oriented analysis of the Hebrew verb system, as outlined in Tobin (1995). We would like to think that any such analysis juxtaposing any two patterns within the system would be equally successful. Once all of the oppositions stipulated in Tobin (1995) are validated, a unified and coherent description of the Hebrew verb system may be achieved.

Notes 1. The system is traditionally described as composed of seven patterns. Tobin (1995, ch.8), however, diverges from this description, and postulates an eight-member system, including the pattern NITPAEL, which is traditionally considered synonymous with HITPAEL. 2. Upper-case characters in the root represent consonants that undergo allophonic occlusant–spirant alternations.

259

260 Noah Oron and Yishai Tobin 3. We would like to thank the editors of the volume for their helpful suggestions and comments. We are responsible, of course, for all errors.

References Berman, Ruth A. 1975a. “The representation of the verb-roots and patterns in the lexicon.” In Sefer Rozen Jerusalrm, 25–36. Council on the Teaching of Hebrew. Berman, Ruth A. 1975b. “The morphological realization of syntactic processes in the binyan system.” Hebrew Computational Linguistics 9: 25–39. Berman, Ruth A. 1994. “Word-formation as evidence.” Proceedings of the Annual Boston University Conference on Language Development, 19, 1, 82–95. Berman, Ruth A. and Sagi, Yisrael.1981. “On word-formation and neologisms in early child language.” Hebrew Computational Linguistics 18: 31–61. Choueka, Yaacov. 1997. RAV-MILIM: a Comprehensive Dictionary of Modern Hebrew. Israel: C. E. T., Miskal, Steimatzky. Junger, Judith 1988. Predicate Formation in the Verbal System of Modern Hebrew. Dordrecht: Foris. Mandelblit, Nili. 2000. “The grammatical marking of conceptual integration: from syntax to morphology.” Cognitive Linguistics 11: 197–251. Oron, Noah. 2001. “Hebrew PAAL and HITPAEL verbs of motion and event structure.” Department of Foreign Languages and Linguistics, Ben-Gurion University of the Negev. Unpublished Ms. Sachs, Jacques. 1987. The Opposed Meanings of the kal/nifal in the Hebrew binyanim: A Signoriented Approach. Ben-Gurion University of the Negev M. A. Thesis. Sachs, Jacques. 2001. The Autonomous/Non-Autonomous Distinction in the Hebrew Verb Form: A Sign-Oriented Analysis. Ben-Gurion University of the Negev PhD. Dissertation. Stern, Naftali. 1994. The Verb Dictionary: Meaning and Distribution of Verbs in Modern Hebrew. Ramat-Gan: Bar-Ilan University. Tobin, Yishai. 1993. Aspect in the English Verb: Process and Result in Language. London and New York: Longman. Tobin, Yishai. 1995. Invariance, Markedness and Distinctive Feature Analysis: A Contrastive Study of Sign Systems in English and Hebrew. Amsterdam and Philadelphia: John Benjamins and Beer Sheva: Ben-Gurion University of the Negev Press. van Schooneveld, C. H. 1978. Semantic Transmutations: Prolegomena to a Calculus of Meaning. Bloomington, IN: Physsardt.

CHAPTER 10

Grammaticization of ‘to’ and ‘away’ A unified account of -k and -m in Hualapai* Kumiko Ichihashi-Nakayama University of California, Santa Barbara Tokyo University of Foreign Studies

Hualapai, an Upland Yuman language, has a pair of morphemes, -k and -m, which show contrasting functions in various linguzistic environments. The present study argues that these various usages of the morphemes can be explained in a unified way by postulating their basic function as indicating the direction or distance from the focal point. Particular functions of the morphemes are determined according to what the focal point is in a given environment, which may be a physical location or an abstract conceptual domain in the speaker’s mind. Using the framework of grammaticization, the possible developmental process of these functions is proposed from concrete lexical functions to more abstract pragmatic/discourse functions.

1. Introduction Hualapai is one of the Native American languages spoken in the northwest part of Arizona, and a member of the Upland Yuman language family. In Hualapai, as well as in other Yuman languages, we find a pair of suffixes, -k and -m, which show contrasting functions in several different parts of the grammar. The morphemes may attach to nominal elements and indicate cases, or they may attach to verbal elements and indicate direction, evidentiality or reference-switch. A number of studies dealing with other Yuman languages suggest that the functions of these morphemes, or at least some of them, may be considered as manifestations of the same basic meanings (e.g. Langdon 1970; Kendall 1975; Winter 1976; Munro 1976; Jacobsen 1983). Among those, Kendall 1975 is a specific attempt to provide a unified and coherent account of these morphemes in Yavapai, another Upland Yuman language closely related to Hualapai. Kendall (1975: 9) proposed two basic meanings for the Yavapai -k and -m. One of the meanings, which explains the case as well as the mood functions of the morphemes, concerns the point of view of the speaker: the morpheme -k indicates states or events toward or at the location of the speaker, or evaluations or thoughts

262 Kumiko Ichihashi-Nakayama

originating in the speaker, while the morpheme -m indicates things away from or conceptually separated from the speaker. The second function of the morphemes is identified in terms of the reference-switching operation: -k indicates sameness and -m indicates difference of the subject referents from clause to clause. Although she suggests a possible merger of these two meanings, she leaves it as a hypothesis to be examined in other Yuman languages. The present study argues that the various usages of the morphemes -k and -m in Hualapai can indeed be explained in a unified way by postulating only one basic meaning for each morpheme. I will argue that the morpheme -k has the general function of indicating ‘inside/toward the focal point’, while -m has the general function of indicating ‘outside/away from the focal point’. The focal point may be a physical location or an abstract conceptual domain in the speaker’s mind, and particular functions of the morphemes are determined according to what the ‘focal point’ is in a given context. The different functions of the morphemes are further analyzed in the framework of grammaticization. It will be proposed that these usages may have been developed or grammaticized from the original meanings of the morphemes in the history of the Hualapai language.

2. -k and -m in Hualapai 2.1. Locative/directional case markers One of the functions of the Hualapai -k and -m is marking the locative/directional case as nominal suffixes. When -k follows a noun naming some kind of place, the morpheme marks allative case and indicates “motion to or location at a point of reference; at; on; around; to and toward”. On the other hand, -m marks ablative case and indicates “motion or location away from a point of reference; away from; (go) by; over” (Watahomigie, Bender, and Yamamoto 1982: 57–9). In (1), for example, ‘my father’ came toward ‘the house’ as marked by -k, while, in (2),‘my mother’ moved away from ‘the house’ as marked by -m:¹ (1) nya đála-ch ’wa:-v-a-k va:-k-yu-ny. my father-SUB house-DEM-EV-LOC 3.come-SS-AUX-PAST ‘My father came to this house.’ (2) nya jíđa-ch ’wa:-v-m jibám-k-yu-ny. my mother-SUB house-DEM-LOC 3.go.out-SS-AUX-PAST ‘My mother went out of this house.’ (Watahomigie, Bender, Watahomigie and Yamamoto 2001: 46)

A unified account of -k and -m in Hualapai

2.2. Directional verbal suffixes When they appear with certain motion verbs, the morphemes -k and -m show a similar semantic contrast to the one we just observed in nominal case markers, as seen in (3) below: (3) e:-k e:-m gami:-k gami:-m jiba:-k jiba:-m ’u:-k ’u:-m vo-k vo-m

‘give (toward the speaker) / receive’ ‘give (away from the speaker) / send’ ‘bring (to the speaker)’ ‘take (away from the speaker)’ ‘come out’ ‘go out’ ‘come and see’ ‘go and see’ ‘come home’ ‘go home’

Like the nominal suffixes, -k here indicates the motion ‘toward the reference point’, while -m indicates the motion ‘away from the reference point’. The ‘reference point’ is generally where the speaker is located. For example, compare examples (4) and (2), repeated as (5), below: (4) nya misí’-h-ch ’wa:-h-m jibá’k-k-yu. my girl-DEM-SUB house-DEM-LOC 3.come.out-SS-AUX ‘My daughter came out of the house.’ (5) nya jíđa-ch ’wa:-v-m jibám-k-yu-ny. my mother-SUB house-DEM-LOC 3.go.out-SS-AUX-PAST ‘My mother went out of this house.’ In both (4) and (5), the movement of the subject referents is ‘away from the house’ as indicated by the ablative case -m attached to the ‘house’. In (4), however, the speaker is outside the ‘house’, and, in reference to the speaker’s location, the movement of ‘my daughter’ is expressed as the one ‘toward’ the speaker by using the verb jiba:-k with the verbal suffix -k. On the other hand, in (5), the speaker is inside the ‘house’, and the movement of ‘my mother’ is expressed as the one ‘away from’ the speaker by the verb jiba:-m with the verbal suffix -m.

2.3. Pragmatic functions of -k and -m The morphemes -k and -m may appear at the end of sentences and express different degrees of evidentiality as shown in the following examples (Yamamoto 1976: 149– 150):

263

264 Kumiko Ichihashi-Nakayama

(6) a.

nya:vlwi:v-m John-ch kwekiviyaam qowa:m-k. noon-at John-SUB automobile drive-K b. nya:vlwi:v-m John-ch kwekiviyaam qowa:m-m. noon-at John-SUB automobile drive-M ‘John drives a car at noon.’

(7) a.

John-ch quwa:q keya:-k. John-SUB deer shoot-K b. John-ch quwa:q keya:-m. John-SUB deer shoot-M ‘John shoots a deer.’

In (6) and (7), both the (a) and (b) sentences have the same propositional meanings. In (6a) and (7a) with the morpheme -k at the end, however, the speaker is asserting what is been said and is somehow committed to it. Here the speaker is making a statement about the event that s/he either witnessed or knows for certain. On the other hand, the (6b) and (7b) with the morpheme -m describe ‘matter-of-fact’ events from an objective point of view. The speaker is making a statement without any implication of his/her belief or commitment toward its truth (Yamamoto 1976: 150). This very nature of -k and -m seems to explain why certain sentences take only -k or -m depending on the factuality of the proposition and the speaker’s perception of it as seen in the following examples (Yamamoto 1976: 150–1): (8) yapa:-m kwekiviya:m qowa:m-m misie:v-k/*-m night-at car drive-DS dangerous-K ‘Driving a car at night is dangerous.’ (9) John-ch kula keya:-m ’-tie:-k/?-m John-SUB rabbit shoot-DS 1-happy-K ‘John’s shooting a rabbit made me happy.’ (10) svo:v-m wayo:v-m/*-k wait-DS boring-M ‘Waiting is boring.’ In (8) and (9), the speaker expresses the proposition as more or less his/her personal opinion. This seems to be the reason why only -k, the morpheme associated with the speaker’s personal commitment to the statement, is acceptable or more appropriate in these examples. On the other hand, (10) denotes a generally-known fact, at least according to this particular speaker, and only -m, the morpheme associated with a general statement, is considered acceptable.

2.4. ‘Switch-reference’ markers The morphemes -k and -m may also be used in clause combining and function as

A unified account of -k and -m in Hualapai

switch-reference markers. Switch-reference has been claimed to be a device which “indicates whether or not its subject is identical with the subject of some other verb” (Haiman and Munro 1983: ix). In Hualapai, at least in the elicited data, the morphemes -k and -m seem to show this canonical function: -k indicates that the clauses combined by the morpheme share the same subject, while -m indicates that the clauses combined by the morpheme have different subjects. This is illustrated in (11) and (12): (11) Rhiánnon-ch he’-h tuy-k đathgwi:l-k-wi-ny. Rhiannon-SUB dress-DEM 3/3.take.off-SS 3/3.wash-SS²-AUX-PAST ‘Rhiannon took off the dress and (Rhiannon) washed it.’ (12) Jóhn-a-ch Máry baeq-m mi:-k-i-ny. John-EV-SUB Mary 3/3.hit-DS 3.cry-SS-AUX-PAST ‘John hit Mary and she (=Mary) cried.’ In example (11), the subject of the first clause, ‘Rhiannon’, is the same as that of the second clause, and the clauses are combined by the ‘same subject marker’ -k attached to the main verb of the first clause, tuy ‘take off ’.³ In example (12), on the other hand, the subject of the first clause is ‘John’ but the subject of the second clause is ‘Mary’, and the clauses are combined by the ‘different subject marker’ -m affixed to the main verb of the first clause, baeq ‘hit’. The examination of the textual data, however, reveals that there are some cases in which the usage of the morphemes does not agree with the continuing or switchreference of the subjects. That is, the morpheme -k is used even when the subjects of the combined clauses are different, or -m is used even though the subjects of the combined clauses are the same. In these cases, the selection between -k and -m seems to be more sensitive to the event structure or the speaker’s conceptualization of the denoted event, rather than the identity of the grammatical subject. The following examples (13), (14), and (15) are from a short Hualapai text, Robber’s Roost: (13) wa-k-tə-sči-:və-k hayku-:və-č ha-k-wa-k house-REL-FACT-x-DEM-LOC white.man-DEM-SUB DEM-LOC-stay-SS kwe-va-k-yo:-və-č ha-k-wa-k . . . thing-DEM-REL-steal-DEM-SUB DEM-LOC-stay-SS ‘At the place X, a white man stayed there. A thief stayed there.’ (Winter and Butcher 1976: 61) (14) nyə-wi-k hayku’ nyə-hwa:k-ya’ wampor-e PERF-do-SS white.man his-be.two-DEM train-? tə-nyu:r-k-wa:m-s’am-ya-k akiwi:-θ-m . . . FACT-multicolored-REL-carry-close-DEM-LOC stand-COT-DS ‘He (=thief) did that (=robbed the train). A white man, his partner, stood in the door of the mail-carrier part of the train . . .’ (Winter and Butcher 1976: 62)

265

266 Kumiko Ichihashi-Nakayama

In (13), although the subject of the first clause, ‘a white man’, and the subject of the second clause, ‘a thief ’, are different individuals, the two clauses are connected by the ‘same subject’ marker -k affixed to wa ‘stay’ in the first clause. The same thing happens later in the text, as seen in (14). In (14), the clauses are connected by -k affixed to wi ‘do’ although they have different subjects, ‘the thief ’ and ‘the white man’. In the story, the ‘white man’ is an accomplice of the ‘thief ’. Although they are different persons, it seems that the speaker perceives them as a ‘unit’, and this motivates the use of -k in (13) and (14). The opposite case is found in example (15). Here, all of the clauses share the same subject ‘they (= authorities)’. Despite that, the last clause in the example follows the ‘different subject’ marker -m affixed to the second occurrence of wim ‘go’: (15) ha-wim-čə-k ke-k ha-wim-čə-θ-m nyə-v-war-čə-k . . . DEM-go-DJ-SS Q-LOC DEM-go-DJ-COT-DS PERF-EMPH-fail-DJ-SS ‘They (= authorities) were going in a certain direction, somewhere, when they failed (to catch him).’ (Winter and Butcher 1976: 63) In this part of the story, the authorities are in pursuit of the thief by tracking his (his horse’s) footprints. What is expected as a result of the story line would be the capture of the thief. However, something went wrong, and they failed to catch him. The morpheme -m here seems to indicate this unexpectedness in the course of events. This kind of ‘exceptional’ usage of the switch-reference markers has been reported in Yuman as well as in other languages. Langdon and Munro (1979: 332–3) conclude that the switch-reference mechanism in Yuman does not simply operate based on the identity of the grammatical subject; they propose several possible determining factors such as agency or control. Mithun (1993) and Watkins (1993) both have convincingly shown that the ‘switch-reference’ systems in Central Pomo (Pomoan) and Kiowa (Kiowa-Tanoan) mark the event or rhetorical structure rather than the identity of the subject in discourse. According to Mithun, the primary function of the switch-reference system in Central Pomo is to link clauses with the ‘same subject’ markers when actions are presented as components of a single event and with the ‘different subject’ markers when they are distinct events. She argues that referenttracking is only a secondary effect of this primary function, since the closely associated actions are more likely performed by the same subject referent, while distinct events are performed by the different subject referents. Similarly, Watkins claims that the ‘same subject’ markers in Kiowa are used to indicate the identity of events rather than that of the referents, while the ‘different subject’ markers may be used sentence-initially to link subunits in the discourse which are thematically distinct but parts of the larger episode. Although more observation needs to be done using extensive discourse data, what we have seen in (13) through (15) seems to show that switch-reference in Hualapai operates in a similar manner. The morpheme -k indicates that the speaker perceives certain participants or events as a continuous unit, rather than simply indicating that the clause shares the same grammatical subject with the following clause. On the other hand, the morpheme -m indicates discontinuity of the events rather than difference of the subject referents.

A unified account of -k and -m in Hualapai

3. Unified functions of -k and -m in Hualapai In the previous sections, the various usages of the Hualapai morphemes -k and -m have been observed. We have seen that the morpheme -k may indicate direction toward a certain location, motion toward the speaker, the speaker’s commitment toward a proposition, and the unity of events. We have seen that the morpheme -m, on the other hand, may indicate direction away from a certain location, motion away from the speaker, the speaker’s non-commitment to a proposition, and the discontinuity of events. All these usages can be explained in a unified way if we postulate one general function of each morpheme. I propose that the basic function of the morpheme -k is to indicate the notion of ‘inside/toward the “focal point”’, while the basic function of the morpheme -m is to indicate the notion of ‘outside/away from the “focal point”’. The focal point is the pointz of reference the direction or distance to which is perceived by the speaker and reflected in his/her usage of the morpheme -k or -m. The focal point is not necessarily a concrete physical location. It can also be an abstract conceptual domain in the speaker’s mind. In the usage as directional markers, both on nouns and verbs, as observed in Sections 2.1 and 2.2, the focal point is a physical location expressed by the noun to which -k or -m is affixed or the speaker’s physical position. Then -k expresses an event or motion inside/toward this focal point, while -m expresses an event or motion outside/away from this focal point. When the morphemes indicate evidentiality, as seen in examples (6)–(10), the focal point is the speaker’s psychological domain, and the choice between -k and -m reflects the psychological distance the speaker feels toward the described event. The speaker’s commitment toward the statement can be considered as proximity to his/ her psychological domain and expressed by the morpheme -k. On the other hand, statements describing general facts are marked by -m, indicating their detachment from the speaker’s psychological domain. When the morphemes are used to combine clauses, as discussed in Section 2.4, the choice between -k and -m reflects the speaker’s conceptualization of the relationship between the expressed participants or events. When two events are perceived as continuous, that is, in the same conceptual domain or focal point, they are combined by -k. When two events are perceived as discontinuous, that is, the second event is conceptualized as distinct from the focal point, the first event, they are combined by -m. While she concludes her survey of the Yavapai -k and -m with two basic functions, Kendall (1975: 9) suggests the possibility of relating the switch-reference function to the locative/directional function. She suggests that as long as the speaker stays on the same topic or talks about the same subject, s/he stays in one ‘conceptual location’, while when the speaker switches topics or subjects, s/he is metaphorically moving away from that ‘conceptual location’. Her ‘conceptual location’ may be compatible with the ‘focal point’ I am proposing here, except that she assumes that the ‘conceptual location’ in the switch-reference function of the morphemes is represented by

267

268 Kumiko Ichihashi-Nakayama

the grammatical subject. As discussed in Mithun (1993) and Watkins (1993), and shown in Section 2.4 with Hualapai data, however, the focal point does not necessarily coincide with the grammatical subject. When and only when we recognize the fact that the ‘switch reference’ system operates on the conceptualization of the denoted events rather than the identity of the grammatical subject per se, can we confidently relate the ‘switch-reference’ function of the morphemes to the other functions and postulate one basic function of each morpheme.

4. Additional functions of -m in Hualapai The morpheme -m has a couple of additional functions in Hualapai. Once we recognize the basic function of the morpheme that I have proposed in the previous section, these seemingly ‘unrelated’ usages can also be explained in a unified way.

4.1. Instrumental/comitative case marker In addition to the locative/directional case, the morpheme -m in Hualapai may mark instrumental and comitative cases. (16) below shows -m as an instrumental marker. (17) shows -m as a comitative marker: (16) Jóhn-a-ch kwàsadsa:d-m gwe ma:-k-wi. John-EV-SUB fork-INST something 3/3.eat-SS-AUX ‘John eats with a fork.’ (17) Cindy-ch Jorigine-m qwa:w-v-k-i. Cindy-SUB Jorigine-COM talk-RECIP-SS-AUX ‘Cindy is talking with Jorigine.’ In these functions, there is no contrastive usage of the morpheme -k. While Kendall (1975) could not explicitly relate these functions to the locative/directional meaning of the morpheme, I argue that the instrumental and comitative functions of -m can also be attributed to the general meaning of the morpheme, ‘outside/away from the focal point’. What is introduced with an instrumental or a comitative marker is an additional participant, such as kwasadsa:d ‘fork’ in (16) and Jorigine in (17), who, in addition to the subject referent, plays some part in the denoted event. These additional participants are separate or ‘detached’ entities from the subject referents. Assuming that, in these cases, the subject referent is the speaker’s focal point or ‘starting point’ in his/her conceptualization of the denoted event (cf. Chafe 1976, 1994), this ‘detachment’ notion can be considered to motivate this usage of -m here.

4.2. Time ‘adverbials’ The morpheme -m is also used in time expressions, corresponding to English

A unified account of -k and -m in Hualapai 269

time adverbials, such as makanya:-m ‘yesterday’, va:-m ‘now’, ye:ka-m ‘tomorrow’, nya’du:y-m ‘in summer’. As pointed out in Munro (1976) and Langdon and Munro (1979), these time expressions are in fact not nominal but verbal in Yuman languages. For example, (18) actually contains two predicates/clauses and is literally translated as ‘I wrote something and it was yesterday’: (18) nya-ch makanya:-m gwe ’-đinyu:d-’-wi-ny. I-SUB yesterday-DS something 1/3-write-1-AUX-PAST ‘I wrote (something) yesterday.’ Therefore, this usage of -m should be regarded as switch-reference rather than a case marking. Here, the second predicate, ‘being yesterday’, states an additional or parenthetical remark and is considered as being outside or detached from the main event, ‘my action of writing’, which motivates the use of -m.

4.3. Habitual expressions The morpheme -m may attach to repeated verbs and express habitual activities. Example (19) below illustrates this usage: (19) nya-ch tu he gwe ma-m ma-m wi-j-wi I-SUB very there something eat-DS eat-DS do-DISTR-AUX ‘I always eat there.’ (Redden 1980: 68) In habitual expressions, each activity is perceived as a distinct, discontinuous event which occurs in a different time and place.⁴ This conceptualization motivates the use of the morpheme -m in these expressions. On the other hand, the following example (20) shows a nice contrast to this usage. (20) nyi-tha-ch wi-k wi-k wi-wi. DEM-3-SUB do-SS do-SS do-AUX ‘He did it over and over and over.’ (Redden 1980: 69) In (20), the verbs are marked with the morpheme -k , not -m, and express repeated actions on a single occasion. The whole event, although more than one action has taken place in a short period of time, is perceived as one event, and, as expected, the usage of -k reflects this conceptualization of the event.

5. Grammaticization of -k and -m in hualapai In the previous sections, I have shown the various usages of the morphemes -k and -m in Hualapai, and argued that these usages can all be explained by one basic function of each morpheme. Next, I would like to turn to the question whether this wide distribution of the usage of the morphemes -k and -m in Hualapai is a synchronic phenomenon or a result of diachronic development. Jacobsen (1979: 575,

270 Kumiko Ichihashi-Nakayama

1983: 175) suggests a possible developmental process of the morphemes in Yuman starting from verbal directional suffixes -k/-m that originally meant ‘hither’/ ‘hence’ to case markers then to switch-reference markers. I propose a similar developmental process of the Hualapai -k and -m and support the argument using the framework of grammaticization (Heine, Claudi and Hünemeyer 1991; Hopper and Traugott 1993). Grammaticization is a dynamic process by which a lexical item assumes a grammatical function or a grammatical item assumes a more grammatical function. Based on observations of many instances of grammaticization in typologically diverse languages, a set of universal principles underlying the process have been identified. Based on these principles, multiple functions of a certain linguistic form can be arranged into a hypothesized order of development. One of the basic principles of grammaticization is the unidirectionality of the process. As summarized in (21) below, the changes which occur during the grammaticization process tend to proceed in only one direction within various aspects of grammar. First, the meaning tends to change from concrete and specific to more abstract and general (this process is often referred to as ‘semantic bleaching’ or ‘generalization’). While the semantic content of a form may be reduced, its pragmatic meaning tends to increase, in that a form with largely propositional meaning gains textual (cohesion-making) and/or expressive (modal, and other pragmatic) meanings (‘pragmatic strengthening’ or ‘subjectification’) (Traugott 1982, 1986, 1989). At the same time, a form which was used only in a certain grammatical context comes to appear in a wider set of contexts (Bybee, Perkins and Pagliuca 1994). (21) Unidirectionality in grammaticization a. semantic bleaching: concrete abstract textual b. pragmatic strengthening: propositional expressive c. range of syntactic applicability: narrow wider When considering the different functions of the Hualapai -k and -m in terms of these parameters, we are able to project possible grammaticization pathways in the development of the morphemes. Among the different functions of the morphemes, as suggested by Jacobsen (1979, 1983), the usage as directional suffixes on verbs is considered as representing the earliest/original stage of the development of the morphemes -k and -m in Hualapai. In this function, the morphemes express more concrete meaning, the one concerning the physical location of the speaker. The other parameters of grammaticization also suggest the earliest development of the function. The morphemes express propositional meanings rather than textual or expressive ones. In terms of ‘range of syntactic applicability’, the morphemes in this function have limited range and only attach to certain motion verbs. The fact that the morphemes are not used productively and have been rather lexicalized or ‘fossilized’ as components of those verbs indicates that they are in the terminal or ‘old age’ phase in life of linguistic forms (cf. Lehmann 1991).

A unified account of -k and -m in Hualapai

While the morphemes still express concrete and propositional meanings concerning the physical location of the entity expressed by the noun to which they are affixed, their function as locative/directional case markers is hypothesized to be a later development. The syntactic range of application is considered expanded from a limited number of verbs to potentially any number of nouns or noun phrases expressing some kind of location. In this function, the morphemes are in no way lexicalized but may be used productively. The evidential function of the morphemes is considered to show the next stage of development. It shows rather abstract and pragmatically expressive meaning by indicating the speaker’s judgment or commitment toward the denoted event. Its syntactic range is a whole statement rather than a lexical item or a phrase. Finally, the switch-reference usage of the morphemes is considered as the latest development. It shows abstract meaning and textual function in that the morphemes are used by the speaker as devices for event or discourse organization. The syntactic range in this usage is accordingly the widest: it now concerns a certain stretch of discourse rather than a single lexical item, phrase, or sentence. This hypothesized grammaticization process of the development of the morphemes -k and -m in Hualapai is summarized in (22) below: (22) Proposed grammaticization process of -k and -m in Hualapai Directional Locative/Directional Evidential Switch-reference Although we lack data for a comprehensive diachronic analysis of Hualapai as well as other Yuman languages, all the parameters in the unidirectionality of grammaticization suggest the developmental order of directional > locative/directional case > evidential > switch-reference functions of the morphemes in Hualapai.

6. Conclusions In this chapter I argued that the various usages of the morphemes -k and -m in Hualapai can be explained in a unified way in terms of their basic functions of indicating the direction or distance from the ‘focal point’. These morphemes give different readings in the language not because they have many distinct meanings. These readings simply represent different manifestations of their basic meanings depending on what is taken to be the ‘focal point’. When they appear with nouns or verbal stems, they express physical movement in(to) or away from the focal point, i.e. the location expressed by the nouns or the speaker’s location. When they appear at the end of sentences, they indicate a relationship between the focal point, i.e. the speaker’s psychological domain, and the statement. When they are used to combine clauses, they reflect the speaker’s conceptualization of the relationship between the expressed participants or events. Furthermore, it is argued that each of these functions represents different stages of the developmental process of the morphemes in Hualapai. On the basis of unidirectionality principles in the theory

27

272

Kumiko Ichihashi-Nakayama

of grammaticization, we can suggest a possible grammaticization pathway in the development of the morphemes -k and -m from concrete directional functions to more abstract pragmatic/discourse functions.

Notes * I would like to thank Marianne Mithun and Sandra A. Thompson for their invaluable comments on an earlier version of this chapter. This study was supported by the Japanese Ministry of Education, Science and Culture (grant nos. 04041011 and 07041013), American Philosophical Society (Phillips Fund for Native American Studies, 1994), and Whatcom Museum, Washington (Jacobs Funds, 1995). 1. Hualapai examples cited in this study are from Watahomigie, Bender, and Yamamoto (1982) unless otherwise noted. Orthographies used in each source have been preserved: ch=[čh] but [č] as a subject marker; ny=[ñ]; đ=[t̪]; ’=[ʔ]. Abbreviations: #/#: subject/object person indexing; 1: 1st person; 2: 2nd person; 3: 3rd person; AUX: auxiliary verb; CAUS: causative; COM: comitative; COT: cotemporal; DEF: definitizer; DEM: demonstrative; DISTR: distributive; DJ: disjunctive; DS: different subject marker; EV: epenthetic vowel; EMPH: emphatic; FACT: factitive; INST: instrumental; LOC: locative; PAST: past tense; PL: plural; PL.OBJ: plural object; PERF: perfective; PRET: pretemporal; Q: interrogative; RECIP: reciprocal; REL: relativizer; SS: same subject marker; SUB: subject marker. 2. The ‘same subject marker’ -k also appears between the main and auxiliary verbs when the subject is the third person. The auxiliary verbs in Hualapai still show the morphological characteristics of the full-fledged verbs from which they originated in that they follow the main verb after the ‘same subject marker’ (because the main verb and the auxiliary verb share the same subject) and they themselves take the personal prefix. While the third person prefix is ‘zero’ marked, in the cases of the first and second person subjects, the sequence of the ‘same subject marker’ and the first or second person prefix has undergone phonological changes and the ‘same subject marker’ -k is no longer apparent. 3. When more than one clause is combined in Hualapai, only the last predicate takes the finite form, that is, is followed by the auxiliary and other inflectional suffixes. 4. The occurrence of the distributive marker -j in (19) (but not in (20)) might also support this ‘discontinuous’ interpretation of the expression.

References Bybee, Joan, Perkins, Revere and Pagliuca, William. 1994. The Evolution of Grammar: Tense, Aspect, and Modality in the Languages of the World. Chicago: University of Chicago Press. Chafe, Wallace. 1976. “Givenness, contrastiveness, definiteness, subjects, topics, and point of view.” In Subject and Topic, Charles N. Li (ed.), 25–56. New York: Academic Press. Chafe, Wallace. 1994. Discourse, Consciousness, and Time. Chicago: University of Chicago Press. Haiman, John and Munro, Pamela (eds.) 1983. Switch Reference and Universal Grammar. Philadelphia: John Benjamins. Heine, Bernd, Claudi, Ulrike and Hünemeyer, Friederike. 1991. Grammaticalization: a Conceptual Framework. Chicago: University of Chicago Press.

A unified account of -k and -m in Hualapai Hopper, Paul J. and Traugott, Elizabeth Closs. 1993. Grammaticalization. Cambridge: Cambridge University Press. Jacobsen, William H., Jr. 1979. “Hokan inter-branch comparisons.” In The Languages of Native America, Lyle Campbell and Marianne Mithun (eds.), 545–591. Austin and London: University of Texas Press. Jacobsen, William H., Jr. 1983. “Typological and genetic notes on switch-reference systems in North American Indian languages.” In Haiman and Munro (eds.) 1983, 151–183. Kendall, Martha B. 1975. “The /-k/, /-m/ problem in Yavapai syntax.” International Journal of American Linguistics 41: 1–9. Langdon, Margaret. 1970. A Grammar of Diegueño: the Mesa Grande Dialect [University of California publications in linguistics 66]. Berkeley: University of California Press. Langdon, Margaret and Munro, Pamela. 1979. “Subject and (switch-) reference in Yuman.” Folia Linguistica 13: 321–344. Lehmann, Christian. 1991. “Grammaticalization and related changes in contemporary German.” In Approaches to Grammaticalization, volume II, Elizabeth Closs Traugott and Bernd Heine (eds.), 491–535. Amsterdam: John Benjamins. Mithun, Marianne. 1993. “ ‘Switch reference’: clause combining in Central Pomo.” International Journal of American Linguistics 59(2): 119–136. Munro, Pamela. 1976. Mojave Syntax. New York and London: Garland Publishing. Redden, James E. 1980. “On Walapai /-k/ and /-m/.” In Proceedings of the 1979 Hokan Languages Workshop [Occasional Papers in Linguistics 7], James E. Redden (ed.), 68–71. Carbondale: Southern Illinois University. Traugott, Elizabeth Closs. 1982. “From propositional to textual and expressive meanings: some semantic-pragmatic aspects of grammaticalization.” In Perspectives in Historical Linguistics, Winfred P. Lehmann and Yakov Malkiel (eds.), 245–271. Amsterdam: John Benjamins. Traugott, Elizabeth Closs. 1986. “From polysemy to internal semantic reconstruction.” Berkeley Linguistics Society 12: 539–550. Traugott, Elizabeth Closs. 1989. “On the rise of epistemic meanings in English: an example of subjectification in semantic change.” Language 65(1): 31–55. Watahomigie, Lucille J., Bender, Jorigine and Yamamoto, Akira Y. 1982. Hualapai Reference Grammar. Los Angeles: American Indian Studies Center, UCLA. Watahomigie, Lucille J., Bender, Jorigine, Watahomigie, Philbert, Sr. and Yamamoto, Akira Y. 2001. Hualapai Reference Grammar (Revised and Expanded Edition). Endangered Languages of the Pacific Rim (ELPR) Publications Series A2–003. Watkins, Laurel J. 1993. “The discourse functions of Kiowa switch-reference.” International Journal of American Linguistics 59(2).137–64. Winter, Werner and Butcher, Christal Jarr. 1976. “Robber’s roost — a Walapai tale.” In Yuman Texts [IJAL-NATS3], Margaret Langdon (ed.), 61–67. Chicago: University of Chicago Press. Winter, Werner. 1976. “Switch reference in Yuman languages.” In Hokan Studies [Janua Linguarum, Series practica, 181], Margaret Langdon and Shirley Silver (eds.), 165–175. The Hague: Mouton. Yamamoto, Akira. 1976. “Notes on the interpretation of /-m/ and /-k/ in Walapai.” In Proceedings of the First Yuman Languages Workshop [University museum studies 7], James E. Redden (ed.), 149–152. Carbondale: Southern Illinois University.

273

PART IV

Below and above the level of the sign

CHAPTER 11

Interaction of physiology and communication in the makeup and distribution of stops in Lucknow Urdu* Shabana Hameed Aligarh Muslim University

1. Introduction This chapter is a case study applying the theoretical framework of Columbia School phonology¹ to the analysis of the stop phonemes in Lucknow Urdu.² here we seek to explain that the paradigmatic makeup (the traditional “inventory”) of the stop phonemes, and their syntagmatic skewings (the traditional “phonotactic distribution in words”), are motivated by five independently known and verifiable orienting principles of phonology — communication, human behavior, physiology, acoustics, and vision.³ Our explanation of the makeup and distribution of the stop phonemes here is however limited in that it is based on the interaction of only two of these five phonological principles, namely, physiology and communication. It may also be noted that the phonological analysis of the stops presented here is primarily based on an exhaustive collection of monosyllabic words of Urdu as spoken by the artisans and craftsmen living in the various localities of the culturally rich city of Lucknow in Uttar Pradesh. A total of five informants (artisans and skilled laborers) were selected for the collection of the data. Out of the five informants, four were females and one was a male. They belonged to the lower strata of the Urdu speaking community, lacking any kind of formal education. For uniformity and homogeneity of the data, all informants were selected from the same socioeconomic class. The data collected for the study is based on face to face interviews running into hours.⁴ In Section 2, we introduce the stop consonants of Lucknow Urdu and categorize them according to the active articulators used to produce them as the relative aperture of the vocal tract, showing their relationship in terms of degree of complexity. In Section 3, we establish a hierarchy of adroitness of the lingual articulators to show that the varying degrees of adroitness have a significant impact on the makeup and distribution of the stop consonants in Lucknow Urdu. In Section 4, the stop consonants are compared in the initial and final positions of the word. The skewings encountered in the distribution of stops in the two word positions are attributed to the joint interaction of physiology and communication with vision.

Shabana Hameed

2. The paradigmatic makeup of stops in Lucknow Urdu There are 21 stop phonemes in Lucknow Urdu. The paradigmatic makeup of these phonemes is systematically presented in the excerpt of the phonological grid in Figure 1.⁵ Whereas the 21 stop phonemes of Urdu, as presented in Figure 1, are organized in terms of the physiologically-based articulators and apertures (to be taken up shortly), it is noteworthy that these phonemes are established by communicative contrast through minimal pairs. As communication is a fundamental orienting principle of linguistic analysis in the Columbia School theory, we formally resort to the use of meaning in the discovery of phonemes.⁶ They are in fact the elemental units of communication in language. The communicative contrast in the minimal pairs is first presented in terms of articulators, as seen in Word list 1.⁷ The four-way contrast among the Urdu stops in terms of the traditional features of “voicing” and “aspiration” is also established through minimal pairs as presented in Word list 2. As demonstrated through the two word lists, the 21 stop phonemes are established in terms of minimal distinctiveness justified by the principle of communication. However, it is the physiologico-acoustic substance of these phonemes that justifies their particular projection on the phonological grid of Lucknow Urdu. As seen in Figure 1, the phonological grid is basically organized in terms of the physiology of the vocal tract. There are two main physiological parameters, namely, the articulators and the apertures. It is noteworthy that both of them represent devices for producing speech sounds. The articulators are the adroit organs of the vocal tract that play a significant role in shaping and exciting the vocal cavity for the production of speech sounds. Whereas the glottal articulator, i.e., the larynx or the glottis, only excites the vocal cavity, the supraglottal articulators, such as the various parts of the tongue and the lips, can both shape and/or excite the cavity. Rather than following the traditional classification of consonants by passive “points of articulation”, our classification of

p b ph bh

t d th dh

  h h

c ɟ ch ɟh

k g kh gh

q □ □ □

Glottis

Aperture ϕ:Stops Voiceless unaspirated Voiced unaspirated Voiceless aspirated Voiced aspirated

Front dorsum Post dorsum

Supraglottal

Apertures Medium

Articulators Apex ↓ teeth Apex ↓ palate

Apertures

Labium

278

Glottal

□ ←V ←A ←h

Not relevant 1 2 1½

Figure 1. Excerpt of the phonological grid of Lucknow Urdu: the stops

Stops in Lucknow Urdu Apex ↓ teeth Apex ↓ palate

Labium pa:l(Vt) ‘bring up’

ta:l (Nm) ‘lake’

a:l (Nm;f ) ‘firewood store’

Medium

Front dorsum Post dorsum

ca:l (Nf ) ‘gait’

ka:l (Nm) ‘famine’

□

kUI (Adj) ‘total’

qUl (Nm) ‘first word of some suras’

ba:l (Nm) ‘hair’

da:l (Nf ) ‘pulse’

a:l (Nf ) ‘branch’

ɟa:l (Nm) ‘net’

ga:l (Nm) ‘cheek’

phAk (Nm) ‘momentariness’

thAk (Vi) ‘be tired’

h Ak (Nf ) ‘sound of knocking’

chAk (Vi) ‘be tricked’

□

chIl (Vi) ‘be peeled’

khIl (Vi) ‘bloom’

ɟhAk (Nf ) ‘foolish talk’

□

ɟho:l (Nm) ‘puckering’

gho:l (Nm) ‘solution’

bhAk (Nm) ‘blast’

dhAk (Nf ) ‘small louse’

hAk (Vt;i) ‘cover’

Figure 2. Word list 1: minimal pairs in terms of articulators

stop consonants is physiologically based on the sound-producing articulators. As shown in Figure 1, there are five supraglottal articulators — labium, apex, medium, front dorsum, post dorsum — that are used in the production of Urdu Voiceless unaspirated

Voiced unaspirated

Voiceless aspirated

Voiced aspirated

pAk (Vi) ‘be cooked’

bak (Vt;i) ‘babble’

phAk (Nm) ‘momentariness’

bhAk (Nm) ‘blast’

ta:l (Nm) ‘lake’

da:l (Nf ) ‘pulse’

tha:l (Nf ) ‘platter’

□

tha:r (Nm) ‘name of a desert’

dha:r (Nf ) ‘edge’

□

ha:l (Nf ) ‘shield’

hAk (Nf ) ‘sound of knocking’

hAk (Vt;i) ‘cover’

a:l (Nm;f ) ‘firewood store’

a:l (Nf ) ‘branch’

ca:l (Nf ) ‘gait’

ɟa:l (Nm) ‘net’

cha:l (Nf ) ‘bark’

ɟha:l (Nm) ‘pungency’

kUl (Adj) ‘total’

gUl (Nm) ‘flower’

khUl (Vi) ‘open’

ghUl (Vi) ‘be dissolved’

Figure 3. Word list 2: minimal pairs in terms of voicing and aspiration

279

280 Shabana Hameed

stop consonants, i.e., labial (“bilabial” in traditional terms), apico-dental (“dental”), apico-palatal (“retroflex”), medial (“palatal”), front dorsal (“velar”), and post dorsal (“uvular”). It is noteworthy that the apex, unlike other articulators, forms two orders of stops. This asymmetrical formation of apical stops is discussed in Section 3. As discussed below, the glottis as the glottal articulator is also intricately involved in the production of stop consonants in Lucknow Urdu. As seen in the figure, the four types of Urdu stops (voiceless unaspirated, voiced unaspirated, voiceless aspirated, voiced aspirated) are differentiated from one another in terms of both glottal apertures and supraglottal apertures. The supraglottal (oral) apertures are reference points for the vertical closings and openings of the vocal tract by the upward and downward movement of the lower jaw. The apertures range from total closure, i.e., aperture ϕ (zero), for the stops, to maximum openings, i.e., larger apertures, for the vowels. The articulators act as a controlling factor for the formation of apertures. As a matter of fact, the vocal cavity formed in the production of individual speech sounds is jointly determined by the articulators and the relevant apertures.⁸ Thus, the stops are produced by a complete closure (aperture ϕ) of the air passage by various articulators, build up of pressure in the vocal cavity, and excitation produced by explosion. In Urdu and all its dialects, three glottal apertures are formed at the larynx — 1, 2, and 1½. Aperture 1, a narrow slit at the glottis formed by the vocal folds, produces the sound traditionally known as “voice”, and is represented in the figure by V under glottis as the articulator. Aperture 2, a triangular configuration of the glottis by the vocal folds, produces “voiceless aspiration” indicated by the A on the figure. Aperture 1½, a narrow triangular configuration, about half the size of aperture 2, produces voiced h, indicated by h on the figure. Here it must be pointed out that neither V nor A is realized as an independent phoneme in Urdu. However, it is noteworthy that these two glottal sounds are accorded the status of phonological units. For, as highly abstract component entities,⁹ they both combine with simple supraglottal phonemes, such as p, t, k, (to be defined below), to produce complex phonemes, such as b, d, g, and ph, th, kh, respectively. The voiced h produced by the glottal articulator at aperture 1½ does occur as an independent phoneme in Urdu. Its inclusion in the phonological grid for the stops here is significant. For this glottal phoneme is also used in combination with simple supraglottal phonemes — p, t, k, etc. — to produce complex phonemes — bh, dh, gh, etc. As manifest in our discussion above, the four-way classification of stops in Urdu is primarily based on the glottal articulation. Of the four types of Urdu stops, the voiceless unaspirated (p, t, k, etc.) are the simplest. For, as shown in the figure, the glottal aperture is not relevant to the formation of these stops; they are produced at supraglottal aperture ϕ by the oral articulators alone. The other three stop types (voiced unaspirated, voiceless aspirated, voiced aspirated) are all more complex in that their production is brought about by the glottis as an additional articulator. Whereas the three latter types of stops are equally complex in terms of the number of articulators, they are differentiated from one another in that they are produced

Stops in Lucknow Urdu

by progressively more complex glottal apertures. Among these latter types, the voiced unaspirated (b, d, g, etc.) are produced in combination with V(oice) by the glottal articulator at aperture 1. The voiceless aspirated stops (ph, th, kh, etc.) are produced in combination with A(spiration) by the glottal articulator at aperture 2. But as the generation of A(spiration) requires a rush of air from the lungs, this extra effort in the production of the voiceless aspirated stops makes them more complex than the voiced unaspirated stops. Of the three complex stops types, the voiced aspirated stops (bh, dh, gh, etc.) are the most complex in that they are produced in combination with the voiced h, which is formed with an unusual and difficult configuration by the glottal articulator at aperture 1½. The rush of air coming from the lungs must now go through a narrowed triangular passage at the glottis. In fact, in view of their difficult glottal articulation, the voiced h and the voiced aspirates (bh, dh, gh, etc.) are rarely used in language. It appears that the use of these extraordinarily complex speech sounds as phonemes is confined to only Urdu and some other Indic languages.¹⁰

3. The impact of the hierarchy of adroitness of articulators: the makeup and distribution of stops As we have noted in the previous section, the articulators are the adroit organs of speech that produce speech sounds in contact with the places of articulation. But it is noteworthy that the articulators are not uniform in terms of their adroitness. And as we find, their relative adroitness has a direct bearing on the makeup and distribution of phonological units in Lucknow Urdu. Of all the supraglottal articulators, namely, the lower lip, the tongue, the velum, and the pharynx, the tongue is the most important in the production of speech sounds. For in terms of its musculature, the tongue can be divided into four parts, namely, the apex, the medium, the dorsum, and the post dorsum. The four parts of the tongue are clearly distinguishable from each other by their varying adroitness. As such, these four parts may be conveniently placed on the scale of adroitness of lingual articulators, as presented in Figure 4. We may say a few words to justify this scale of adroitness. The apex of the tongue is the most adroit because it is physiologically light in weight, it is triangular in shape, and it is placed at the free end of the tongue such that it can be moved freely along the length and breadth of the oral cavity. Apex Dorsum Medium Post dorsum

Most adroit More adroit Less adroit Least adroit

Figure 4. Scale of adroitness of the lingual articulators

28

282

Shabana Hameed

The post dorsum is the fixed portion of the tongue, with thick musculature, which is why it is the least adroit on the scale of adroitness. It acts as a hilt for the tongue in its entirety in general, and for the dorsum in particular. The medium-dorsum mass is rectangular in shape, and physiologically more massive than the apex. But the dorsum and the medium are not uniform in their adroitness. The physiology of the tongue is such that the musculature of the medium is more closely tied with that of both the apex and the dorsum. Further, the medium acts as a hilt to the apex. In view of the physiological factors outlined above, it is reasonable to place the post dorsum below the medium, the medium below the dorsum, and the dorsum below the apex in terms of the adroitness of lingual articulators. In addition to the tongue, the lower lip, being an adroit articulator, also plays an important role in the production of speech sounds. Given the musculature of the lower lip and the mobility of the lower jaw, it will be generally agreed that the labium occupies a place below the apex and above the medium in terms of degree of adroitness. Further, it can reasonably be argued that the labium occupies a place close to the dorsum in terms of the relative adroitness of articulators. Since both labium and dorsum appear to fall somewhere between apex (most adroit) and medium (less adroit) they are arbitrarily assigned to the same position on the scale of adroitness for the purposes of this study. In terms of the scale of adroitness of articulators, we may expect that in a language or a dialect, the most adroit apex will be most productively utilized in the formation of the consonantal units in the phonological paradigm and in the syntagmatic distribution of these units in the word. On the other hand, it is reasonable to expect that the least adroit post dorsum may only occasionally be utilized in a language to form a consonantal unit, such as the post dorsal q in Lucknow Urdu. At the same time, we may expect that such a post dorsal unit may be underutilized in the formation of morphemes and words in Lucknow Urdu, as in any other language or dialect. Further, in terms of the scale relationship, we may expect the labial-dorsal and the medial consonants to occupy a middle ground in the number of units and their frequency of usage in the word. With a view to gauging the impact of the scale of adroitness of articulators, we now present the actual distribution of the apical, labial, dorsal, medial, and post dorsal stops, both in terms of the number of units and the frequency of their combinatory usage in the monosyllabic words in Lucknow Urdu, in Table 1. Comments on Table 1: First, as shown in Column A of Table 1, the number of stops formed by the apex, labium, dorsum, medium, post dorsum, are 8, 4, 4, 4, 1, respectively. A clear manifestation of the extraordinary adroitness of the apex can be seen in the formation of two orders of stops, the apico-dental and the apico-palatal, in Urdu and in some other Indian languages. Further, as seen in this column, there are only two inequalities in the number of units. The most adroit apex forms 8 units, whereas the least adroit post dorsum articulates only 1 unit. This distribution fully conforms to our expectation in terms of the relative adroitness of articulators.

Stops in Lucknow Urdu Table 1. Makeup and distribution of stops in the monosyllabic words in terms of hierarchy of adroitness of articulators

Articulators

A

B

C

Units

Frequency of usage Number

Percentage

Apex → Teeth Apex → Palate

4 4

384 324

54.24 45.76

Apex Labium Dorsum Medium Post Dorsum

8 4 4 4 1

708 521 543 401 11

32.42 23.85 24.86 18.36 00.51

21

2,184

Total

100

Finally, as seen in the table, there is a parity in the number of units (4 each) for the labial, dorsal and medial stops. Whereas the parity between the labial and the dorsal stops is fully justified in terms of the parallel placement of the dorsal-labial articulators on the scale of the adroitness of articulators, the number of units formed by the less adroit medial articulator is contrary to our expectation. The parity for the medial stops observed here is brought about by the impact of communication for a fuller utilization of the intersections on the paradigm. There is a tendency in terms of the communicative factor to fill all intersections of the phonological grid. For the larger the number of phonemes, the greater will be their potential to build an extensive inventory of meaningful units, both grammatical and lexical. However, the principle of communication is here in conflict with the physiology of the vocal tract. For, in terms of the hierarchy of the adroitness of articulators, the medial stops should be less favored vis-à-vis the labial or the dorsal stops. The opposing forces of physiology and communication are balanced in conformity with our expectation when the medial stop phonemes are realized at all intersections of the grid, whereas these phonemes are markedly disfavored, vis-à-vis their labial or dorsal counterparts, in the frequency of their usage in the word in Lucknow Urdu. Thus, in Column B of Table 1, when we look at the frequency of occurrence of the five stop types among monosyllabic words, we find the skewings in the figures to be fully in conformity with our expectation. From the top down, in the order of decreasing adroitness of the articulators, the numbers are 708, 521, 543, 401, and 11. In fact, the most favored apical type (708) appears in a ratio of over 64 to 1 over the least favored post dorsal type (11). As expected, the labial stops (521) and the dorsal stops (543) compete well with each other. At the same time, the two stop types are far below in their frequency of occurrence in comparison with the apical stops.

283

284 Shabana Hameed

Further, as predicted, the less favored medial stops (401) appear less frequently than the more favored labial-dorsal stops, and occur more frequently than the least favored post dorsal stops. We may therefore conclude this section by observing that whereas a fuller utilization of the labial, the dorsal, and the medial stops in the phonological paradigm of Lucknow Urdu caters to the communicative need for a greater number of distinctive units, the marked skewings in the frequency of usage for the five stop types in the syntagmatic organisation of the word, responds to the degrees of preference based on the hierarchy of adroitness of articulators in the production of these stops.

4. Asymmetrical distribution of stops in word initial and final positions In the preceding section, we established a norm in terms of the hierarchy of adroitness of articulators for the relative preference in the syntagmatic distribution of the apical, labial, dorsal, medial, and post dorsal stops, irrespective of where they occur in monosyllabic words in Lucknow Urdu. But when we examine the distribution of the five stop types in initial and final positions of these words, we find that there is an additional bias in comparison with their normal, across-the-word distribution. This additional skewing is mainly brought about by the communicative factor. It will be readily agreed that the beginning of the word carries a greater communicative load than does the end of the word. As a matter of fact, there is a graded decrease in the communicative load of phonological units, as we move from the initial position to the final position of the word. Here it must be pointed out that the principle of decreased communicative load at the end of the word was articulated by G. K. Zipf in 1949. In consequence of this communicative rationale, we expect that there should be a competitive, more productive utilization of all relevant phonological units in word initial position. On the other hand, we expect the phonological units to be selectively underutilized in word final position. In Table 2, we present the frequency count for the five stop types in the initial and final positions of monosyllabic words. The total figures shown in Column C of Table 2 (apicals 708, labials 521, dorsals 543, medials 401, and post dorsals 11), represent the preferential norm for the five types of stops in terms of the hierarchy of adroitness of articulators, as already discussed in Section 3. But as seen at a glance in the table, the normal usage of the five stop types is significantly skewed in the initial and final positions of the word. As will be seen below, this skewing in the two positions of the word is brought about by the interaction of physiology with both communication and vision. As seen in Column A of the table, of the 1243 occurrences of stops in word initial position, there are 323 apicals, 384 labials, 265 dorsals, 260 medials, and 11 post

Stops in Lucknow Urdu

dorsals. Thus, in comparison with the norm, there is a substantial decrease in the frequency of usage for the physiologically favored apical stops (32.42 to 25.99), and dorsal stops (24.86 to 21.32), whereas there is a significant increase in the frequency of usage for the less favored medial stops (18.36 to 20.92) and the least favored post dorsal stop (0.51 to 0.88). This competitive usage of the four lingual stop types is motivated by the heavy communicative load carried by the first phonological unit in the word. Further, it is reasonable to expect that the frequency of occurrence of the labial stops should be on a par with that of the dorsal stops, in terms of physiology and communication. However, we find that there is an extraordinary bias in favor of the labial stops, vis-à-vis the other stop types, in the initial position of the word. In fact, as seen in the table, the frequency of usage for the labials (384) is even higher than that of the apicals (323) in this position of the word. The extraordinary preference for the labial stops in word initial position can only be explained by the visibility factor of the labial articulator. That is, there is an interaction of physiology and communication with vision as an orienting principle in this visually important position of the word. The process is reversed in the communicatively less important final position of the word. As seen in Column B of the table, there is an additional increase, in comparison with the norm, in the frequency of usage for the favored apical stops (32.42 to 40.92), and dorsal stops (24.86 to 29.54), whereas there is an additional decrease in the frequency of usage for the less favored medial stops (18.36 to 14.98), and a total skewing against the post dorsal stop (0.51 to 0.00). It is noteworthy that the physiologically disfavored post dorsal q, a rarely used phoneme in Lucknow Urdu, appears in only the communicatively most important word initial position. This selective utilization in favor of the apicals and dorsals and against the medials and Table 2. Frequency of stops in the initial and final positions of the monosyllabic words in terms of the hierarchy of adroitness of articulators Stop types in terms of articulators Apico-dental Apico-palatal Apical Labial Dorsal Medial Post dorsal Total

A

B

Initial Number 194 129 323 384 265 260 11 1,243

C

Final Percentage Number

Total Percentage Number

Percentage

60.06 39.94 25.99 30.89 21.32 20.92 00.88

190 195 385 137 278 141 0.00

49.35 50.65 40.92 14.56 29.54 14.98 00.00

54.24 45.76 32.42 23.85 24.86 18.36 00.51

100

941

100

384 324 708 521 543 401 11 2,184

100

285

286 Shabana Hameed

post dorsal is motivated by the low communicative load carried by the last phonological unit in the word. Finally, we expect that the frequency of occurrence of the labial stops should be on a par with that of the dorsal stops, in terms of physiology and communication. However, we find that there is an extraordinary reduction in the frequency of usage of the labial stops in word final position. In fact, as seen in the table, the frequency for the labials (137) is far below that of the apicals (385) and dorsals (278), and is even lower than that of the medials (141). This drastic tilt in the frequency of occurrence against the labial stops is brought about by the subordination of the role of vision in the communicatively less important final position of the word.

5. Concluding remarks In this chapter we have briefly provided a motivated rationale in terms of physiology and communication, for the paradigmatic relations and the syntagmatic skewings of the stop phonemes in Lucknow Urdu. In Section 2, we have presented oppositions of the Urdu stops as a paradigm of phonological units representing their interrelationships. The 21 stop phonemes are established in terms of communicative contrast through minimal pairs. At the same time, the opposing units are justified in the phonological grid in terms of their physiologico-acoustic substance. However, our partial justification of the grid is based on physiology alone. In Section 3, we set up a scale of adroitness of the supraglottal articulators. It is argued that in terms of the physiological musculature, the apex is the most adroit articulator, followed by the labium-dorsum, the medium, and the post dorsum, in that order. It is also argued that the adroit apex should be most productively utilized in the formation of consonantal units in the paradigm and in their frequency of usage in the word. It has been demonstrated through statistical support that the number of units for the apical, labial, dorsal, medial, and post-dorsal stops, and their across-the-word distribution, fully conform to our expectations in terms of the scale relationship. In Section 4, we compared the distribution of the five stop types in the initial and final positions of the word with the across-the-word norm . It has been shown that in consequence of the interaction with communication and vision, the normal usage of these stop types is significantly skewed in the initial and final positions of the word. It has been argued that the competitive usage of the five stop types in word initial position and the selective utilization of these stop types in word final position, are brought about by the interaction of physiology and communication. It is further argued that the drastic skewing in favor of the labial stops in word initial position and against these stops in word final position, is brought about by vision, in interaction with physiology and communication.

Stops in Lucknow Urdu

Notes * This is a revised and expanded version of the paper read at the Sixth International Columbia School Conference on Linguistics, Rutgers University, New Brunswick, NJ, October 9–11, 1999. I am grateful for the valuable comments and criticism of the earlier draft of this chapter by the editors of this volume, especially Prof. Ellen Contini-Morava and Prof. Betsy Rodriguez-Bachiller. Words cannot suffice to express my profound gratitude to my teacher Prof. Abdul Azim without whose moral support, personal attention, and invaluable academic guidance, this chapter would not have been possible. 1. The distinction of Phonology versus Grammar in Columbia School theory is based, with some modifications, on André Martinet’ s concept of the double articulation of language.For details, see Azim (1978: 18–20); Martinet (1965: 21–35). 2. For a classic account of Urdu consonants and vowels in the “phonematic” structure of the word in terms of the phonological concepts developed by the Prague and French Schools of Linguistics, see Husain (1955). 3. For a conclusive account of the Columbia School phonological theory, see Diver (1995: 43– 72); for classic phonological analyses in terms of the five principles, see Diver (1974, 1979, 1992), Davis (1987), Azim (1992, 2002), and Jabeen (1993). For a book-length treatment of Columbia School phonology, see Tobin (1997). 4. Data collection was done in two parts: non-selective collection of data was done where all kinds of words, phrases, and sentences were assembled without any reference to word list etc. Selective collection of data was done by keeping in view the potential slots in a graph based on combination of all the phonological units, plotted both vertically and horizontally, for actual realization of the words. 5. Whereas the traditional “phonemic inventory” is basically a statement of the total number of phonemes in a language, our phonological grid is a network of phonological units (phonemes) representing their interrelationships. 6. Even in traditional phonemic analysis, the phonemes are generally set up in terms of contrast through minimal pairs. However, they are formally established through distributionalsubstitutional criteria. 7. A note on transcription: /I/ is a high front unrounded short vowel; /U/ is a high back rounded short vowel; /A/ = schwa. 8. It may be pointed out that the apertures used for the production of consonants (stops, spirants, semi-vowels) are traditionally referred to as “manners of articulation”. 9. Of the two highly abstract units (V and A), A , that is, the voiceless h, does occur as an independent phoneme in many languages, as in Arabic and English. However, we know of no language that has V as an independent phoneme. Nevertheless, V(oicing), a necessary concomitant of all vowels and all voiced consonants in language, must be postulated as a glottal unit at a higher level of abstraction. 10. For a deeper understanding of the anatomy of the glottal articulation in Urdu, see Azim (2002: 278–283).

287

288 Shabana Hameed

References Aaronson, Doris and Robert W. Rieber (eds.). 1979. Psycholinguistic Research: Implications and Applications. Hillsdale, NJ: Lawrence Erlbaum Associates. Azim, Abdul 1978. Systems of the Verb in Classical Urdu. Ph.D. dissertation, Columbia University. Ann Arbor, Michigan, University Microfilms (Publication No. 79–8577). Azim, Abdul 1992.“Phonological and communicative asymmetry in modern standard Urdu”. (Paper presented at the International Seminar: The Role of Communication in Linguistic Theory, held at the Aligarh Muslim University, Aligarh, U. P., India) Azim, Abdul 2002. “Problems of aspiration in modern standard Urdu”. In Reid, Otheguy and Stern (eds.), 273–307. Contini-Morava, Ellen and Barbara Sussman Goldberg (eds.). 1995. Meaning as Explanation: Advances in Linguistic Sign Theory. Berlin and New York: Mouton de Gruyter. Davis, Joseph 1987. “A combinatory phonology of Italian”, Columbia University Working Papers in Linguistics 8: 1–99. Diver, William 1974. “Substance and value in linguistic analysis”, Semiotext (e) 1: 11–30. Diver, William 1979. “Phonology as human behavior”. In Aaronson and Rieber (eds.), 161– 186. Diver, William 1992. “The communicative motivation for the Centum-Satem split”. (Paper presented at the International Seminar: The Role of Communication in Linguistic Theory, held at the Aligarh Muslim University, Aligarh, U. P., India.) Diver, William 1995. “Theory”. In Contini-Morava and Goldberg (eds.), 43–114. Husain, Masud 1955. A Phonetic and Phonological Study of the Word in Urdu. Aligarh: Aligarh Muslim University, Department of Urdu, Studies 1. Jabeen, Shazi Shah 1993. Economy of Articulation in the Phonology of Bihar Urdu (as spoken in and around Gaya). Ph.D. dissertation, Aligarh Muslim University, Aligarh, U. P., India. Martinet, André 1965. La Linguistique Synchronique: Études et Recherches. Paris: Presses Universitaires de France. Reid, Wallis, Ricardo Otheguy and Nancy Stern (eds.). 2002. Signal, Meaning, and Message: Perspectives on Sign-Based Linguistics. Amsterdam/Philadelphia: John Benjamins. Tobin, Yishai 1997. Phonology as Human Behavior: Theoretical Implications and Clinical Applications. Durham: Duke University Press. Zipf, George Kingsley 1949. Human Behavior and the Principle of Least Effort: An Introduction to Human Ecology. Cambridge, MA: Addison-Wesley. 1965, Facsimile of 1949 edition. New York and London: Hafner.

CHAPTER 12

Between phonology and lexicon The Hebrew triconsonantal (CCC) root system revolving around /r/ (C-r-C)* Yishai Tobin Ben-Gurion University of the Negev

The notion of “semiological economy” has been one of the most fundamental and fruitful notions produced by functional semiology. . . . Given the well-founded assumption that, in general, inventories of items are more readily learned, digested and recalled when their items are sub-categorised . . . (Hervey 1985: 141, 155) Given that, formulated in a preliminarily intuitive way, “mnemonic economy” of a semiotic system is a memory-saving device — i.e. a way of cutting down the amount of effort required for memorising and remembering all the well-formed elements of a system . . . If, . . . the objects to which the figurae bear some similarity are closely connected with the semantic content of the signa in whose expression these figurae are used, the interrelatedness achieved by this similarity will also make the memorisation of the signa (and their appropriate form-meaning associations considerably more economical). (Hervey 1988: 27, 32)

1. Introduction In this chapter I will explore the connection between the triconsonantal (CCC) Hebrew roots containing the phoneme /r/ as C-II (C-r-C) and the various semantically related systems and sub-systems they may form in what appears to be a pattern of interconnected semantic fields revolving around these C-r-C roots (where the other consonants “flanking” /r/ (Diver 1995) belong to similar phonological classes): i.e. I will postulate a direct “classificational” connection between phonology and the lexicon as an example of “economy and motivation in semiotic systems” in the form of “mnemonic-classificatory economy” (Hervey 1985, 1988). I will first illustrate these functional, isomorphic, and synergetic semiotic concepts in examples (1–16) where a semantic field reflecting the general notion of “A Change In Structure” is postulated for a partial set of the C-r-C roots of the /p-r-C/ pattern:¹

290 Yishai Tobin

(1) /p-r-m/ ‘unstitch, undo stitches, untie, unbutton, rip; tear open, rend (clothes), rip, cut, chop’ (2) /p-r-f/ ‘fasten together (with a pin), pin up (hair), button’ (3) /p-r-t/ ‘expand, extend’ (4) /p-r-t’/ ‘divide into small parts; split, make change (‘break a dollar’); separate, specify, itemize, detail, divide’ (5) /p-r-d/ ‘separate, branch off, disintegrate, loosen, decompose, depart, divide, divorce’ (6) /p-r-s/ ‘slice (break, cut bread), spread, deploy, fan out, be divided, broken’ (7) /p-r-S/ ‘spread, cast, stretch out, expand, extend, unfurl’ (8) /p-r-z/ ‘exaggerate, overdo, overstate, be excessive, spread the fingers’ (9) /p-r-ts/ ‘break, break through, demolish; erupt, make a breach, crack, destroy; burst, rush upon; spread, increase, overflow, break out’ (10) /p-r-r/ ‘crumb, crumble, break into crumbs; shatter, undermine’ (11) /p-r-q/ ‘unload; deliver, set free, extricate; break, remove, dislodge, dismantle, disassemble, decompose, disjoint, demolish, liquidate, take apart, wind up, solve, resolve, crack’ (12) /p-r-x/ ‘crush, break, crumble; smash’ (13) /p-r-ħ ‘blossom, bloom, burst forth, burgeon, sprout; flourish, prosper, thrive, spread out’ (14) /p-r-ʔ/ ‘uncover (esp. hair), be dishevelled, disordered, unkempt; wildlooking’ (15) /p-r-ʕ/ ‘bring forth, bear, yield (fruit), bloom’ (16) /p-r-h/ ‘grow, produce, reproduce, be fruitful (fertile), thrive, flourish’ This general semantic field of “A Change in Structure” is reflected in this particular partial set of phonologically related roots in the following polaric processes: (a) the “diminishing/decreasing”, or “the closing of borders”, or “the division/separation of a whole into its component parts”: (1) /p-r-m/ ‘unstitch; . . . cut, chop’ (4) /pr-t’/ ‘divide; split, make change (‘money’) . . . ’, (5) /p-r-d/ ‘separate . . . disintegrate . . . decompose, depart, divide, divorce’, (10) /p-r-r/ ‘crumb, crumble; shatter . . . ’, (11) /pr-q/ ‘unload; . . . break, . . . dismantle, disassemble, decompose, disjoint, demolish, . . . take apart, crack’, (12) /p-r-x/ ‘crush, break, crumble; smash’; (b) the “enlarging/increasing”, or “the opening of borders”, or “the extension of the boundaries of a whole”: (3) /p-r-t/ ‘expand, extend’, (7) /p-r-S/ ‘spread, cast, stretch out, expand, extend, unfurl’, (p-r-z/ ‘exaggerate, overdo, overstate, be excessive, spread the fingers’, (13) /p-r-ħ/ ‘blossom, bloom, burst forth, burgeon, sprout; . . . spread out . . . ’, (15) /p-r-ʔ/ ‘bring forth, bear, yield (fruit), bloom’, (16) /p-r-h/ ‘grow, produce, reproduce, be fruitful (fertile), thrive, flourish’; or: (c) “A Change in Structure” that includes both the antonymous or antithetical meanings of these polaric processes in a single root (what Abel 1884 and Freud 1910, 1958, 1976 referred to as der Gegensinn der Urworte in general and Ben-Ezra

Hebrew triconsonantal roots revolving around /r/

1921, Blau 1928, Gordis 1936, Landau 1896, Morrison 1954 and Noeldoeke 1910 applied to Hebrew) that I have referred to elsewhere as “polaric semiology” (Aphek and Tobin 1988: ch. 5,7, 1989, 1991): (6) /p-r-s/ ‘slice (break, cut bread), spread, deploy, fan out, be divided, broken’, (9) /p-r-ts/ ‘break, break through, demolish; erupt, make a breach, crack, destroy; burst, rush upon; spread, increase, overflow, break out’; and/or (d) “a modification of the size, shape or proportions of a whole”: (2) /p-r-f/ ‘fasten together (with a pin), pin up (hair), button’, (14) /p-rʔʃ / ‘uncover (esp. hair), be dishevelled, disordered, unkempt; wild-looking’. I will further elaborate upon my hypothesis of the p-r-C pattern and present the other sets of C-r-C roots and the related semantic fields they may represent as part of a larger interrelated classification system from a sign-oriented perspective inspired by the work of linguists such as: Ferdinand de Saussure (1916/1959), Roman Jakobson (1941/1969, 1970), and William Diver (1974, 1975), and further exemplified in Andrews and Tobin (1996), Contini-Morava and Sussman Goldberg (1995), Contini-Morava and Tobin (2000), Jakobson and Waugh (1979), Klein-Andreu (1983), Reid, Otheguy and Stern (2002), Tobin (1988a, 1989a), and Waugh and Rudy (1991). I have also discussed similar isomorphic phonological-lexical classification systems elsewhere, including so-called irregular verbs in English, a phonologically exceptional infinitive class (‘defective’ I-NUN) in Hebrew, in the alternative infinitive endings of Spanish (-ar/-er/-ir), French (-er/-oir/-re/-ir) and Italian (-are/ere/-ire), and in a system of Hebrew particles revolving around the bilateral root ayin-dalet (Tobin (1989b, 1991a-b, 1993: ch. 12).

2. Defining language semiotically and synergetically Every linguistic theory is the direct result of a specific set of theoretical and methodological assumptions which are related to how the linguist: (i) defines language; (ii) defines a linguistic problem; (iii) determines the source, kind, and amount of data to be selected and analyzed; (iv) chooses a methodology to select and analyze the data; and (v) evaluates, compares, and contrasts analyses in light of all the above. These five criteria basically serve to describe how and what the particular linguist views as the goals of linguistic research. In other words, the way a linguist defines language is very often the first step of a linguistic analysis from which the other theoretical and methodological assumptions naturally follow. The semiotic or sign-oriented orientation to be applied in this study defines language in the following way: Language: a system of systems composed of various sub-systems (revolving around the notion of the linguistic sign) which are organized internally and systematically related to each other and used by human beings to communicate (Tobin 1990a: ch. 3, 1994/1995: ch. 1, 1997a: ch. 1). Theoretically, this definition of language implies the dichotomy between: (a) langue: an abstract code composed of signals and invariant meanings (i.e. lin-

29

292 Yishai Tobin

guistic signs) and their paradigmatic, associative or syntagmatic relationships: a complex abstract code shared by a community of speakers), versus (b) parole: the concrete and seemingly chaotic realization of this abstract code exploited by individual speakers to communicate specific discourse messages in different linguistic and situational contexts. Methodologically, this semiotic or sign-oriented theoretical outlook implies: (a) an explicit respect for and reliance on concrete language data taken from both spoken and written discourse in different linguistic and situational contexts (as well as introspective language data) and (b) at least an implicit commitment to deal with the “human factor” (i.e. the cognitive, perceptual, linguistic and non-linguistic behavior of human beings). This particular definition of language and its theoretical and methodological implications provide a holistic view of language based on following two fundamental orientations: the communication factor and the human factor. Both of these orientations can be further related to the larger isomorphic concept of language synergesis defined as: “the cooperative action of discrete agencies such that the total effect is greater than the sum of the discrete effects taken independently” (Tobin 1990a: 48, 1994/1995: 11, 1997a: 19). Put more simply, language (the linguistic whole) is greater than the sum of its individual parts (e.g. linguistic signs and systems such as the C–C–C and C-r-C root systems and sub-systems and their attached lexical meanings, morphemes, words, etc.). The human factor inherent in these definitions includes the following basic principles: (i) Human Intelligence - human beings can draw far-reaching conclusions from minimal cues, i.e. through the cognitive process of inference; (ii) Human Efficiency – Zipf ’s/Manzerath’s/Mandelbrot’s/Krylov’s/Bëothy’s Laws related to the investment of minimal effort to achieve maximal results in the semiotic communication process; (iii) Memory Limitations - human beings have large but limited memories which can be related to (i) and (ii) above. Zipf (1949) established the fundamental synergetic principle of “least effort” (or “economy”) in all domains of language, the principle which leads language users to unifications or diversifications of the properties of language units in order to alleviate their physical and mental effort. Some of the more obvious examples of Zipf ’s law regarding the lexicon include the relationship between the frequency and length of words and the number of possible “meanings” or “messages” (dictionary meanings or contextual messages) they convey. According to Zipf, the frequency of a use of a word results in a reduction in its size on the one hand and an increase in the number of messages it may convey on the other. Thus, from the point of view of the encoder, linguistic economy is best realized by a highly compressed lexicon composed exclusively of short words conveying a multiplicity of messages (dictionary meanings and contextual messages), or what Zipf calls The Force of Unification.

Hebrew triconsonantal roots revolving around /r/

From the point of view of the decoder, on the other hand, linguistic economy is best realized by a lexicon composed of words maximally distinct in form and restricted to one exclusive message, or what Zipf refers to as The Force of Diversification. Mandelbrot (1953) further refined Zipf ’s law by formulating that the encoder’s “economy of effort” is subject to the constraint imposed by the need to effectively communicate information to the decoder. The most obvious example of Mandelbrot’s law concerning the lexicon is related to the number of messages which can be assigned to a single word. The more messages a word conveys, the more “economical” it is. However, if over-efficient encoders limit themselves only to the exclusive repetition of the “most economical words conveying most messages”, they defeat the very essence of the purpose of communication. The least amount of information would be conveyed by the limited and predictable repetition of the same words. Consequently, the decoders would then be forced to augment their role in the cooperative effort needed for successful communication. Thus, Mandelbrot points out a “check and balance system” concerning the minimal average amount of information needed by an encoder to a decoder. In other words, there is a “balance of forces” between the ideal unification desired by the encoder and ideal diversification desired by the decoder. Menzerath (1954) discussed the synergetic principles regarding the quantitative relationships between a whole and its parts: “the greater the whole, the smaller its parts.” One of the more obvious examples of Manzerath’s law concerning the lexicon is the relationship between the length of words and syllables. The longer the word, the shorter its syllables. Menzerath’s law can also be extended to other realms within the lexicon: e.g. the synergetic relationship between the length of words and the number of morphemes they contain as well as the complexity of their meanings. The longer a word, the more morphemes it may contain and the more specific and complex its meaning. Zipf and Menzerath’s laws can be further applied to all the interrelated levels of language in general and to the concept of the reduplicated root patterns in Hebrew in particular. Krylov (1982) has shown that the distribution of meanings over the words in the lexicon follows a stochastic law restricting the process of polysemy — or the force of diversification — within the lexicon. Krylov’s Law of Polysemy or Polylexy further confirms Zipf ’s law linking an increase of the number of messages (dictionary meanings and contextual messages) conveyed by a word with an increase in its frequency. Krylov’s work also supports the basic hypothesis of Menzerath’s law by showing that the number of messages conveyed by a word depends directly on its length: the longer the word the more specific or complex its meaning and the fewer its potential contextual discourse messages. While Krylov’s law may be applied to the lexicon as a whole, similar findings concerning the heterogeneous frequencies of the different messages (and grammatical functions) which can be attributed to the same word were found by Beöthy (Beöthy and Altman 1984a, b). Like all the synergetic laws previously discussed, Beöthy’s law also represents the interplay of the opposed Zipfian forces of “encoder-oriented

293

294 Yishai Tobin

unification” versus “decoder-oriented diversification”. All of these synergetic laws of human efficiency related to the lexicon dealt with words and the potential number of their messages (dictionary meanings and contextual messages) and grammatical functions. The same implications and results can be extended to the Hebrew root system and the Hebrew lexicon concerning the postulation of a direct classificational connection between phonology and the lexicon in the Hebrew C–C–C root system in general and the C-r-C sub-system in particular as an example of economy and motivation in semiotic systems in the form of mnemonic–Classificatory economy.

3. Hebrew roots and structures Hebrew, like other Semitic languages, has a fundamentally different structure than the Indo-European languages. In the Semitic languages, the isomorphic connections between phonology, morphology, syntax and semantics are much more overt. The vast majority of the words of the language can be analyzed into consonantal roots signalling broad semantic fields which are combined with fixed morphophonemic patterns for what are traditionally called nominal (miʃkalim ‘weights/scales’), verbal and adjectival (binjanim ‘building blocks’) forms. More often than not, the connections and relationships between the root and these fixed morphophonemic patterns are rather transparent. The relatively transparent connection between lexical roots and the fixed morphophonemic nominal, verbal, or adjectival patterns may be briefly illustrated for the triliteral (CCC) consonantal root K-T-V ‘write’ as follows: KoTeV ‘writes, writer’; KaTaV ‘wrote, reporter, correspondent’; niKTaV ‘was written’; KiTeV ‘inscribed’; KuTaV ‘was inscribed’; hiKTiV ‘dictated’, huKTaV ‘was dictated’; hitKaTeV ‘corresponded, wrote letters’; nitKaTeV ‘was written in a correspondence, letter’; KaTaVa ‘article’; KToVet ‘address’; KiTuVit ‘title, subtitle’; KTuVa ‘contract’; KtaV ‘handwriting’; miKTaV ‘letter’; miKTAVa ‘writing desk’; haKTaVa ‘dictation’; hitKaTVut ‘correspondence’; KTiV ‘spelling, orthography’; KaTVan ‘typist, scribe’; KaTVanut ‘typing’; KaTuV ‘was written, biblical verse, passage, text’; KTaVim ‘works, writings’; ʃiKTuV ‘rewrite’; etc. It should be clear from the above examples, that on the most fundamental level, the Hebrew lexicon is primarily based on (usually triliteral (CCC)) consonantal roots.2 Studies of the triliteral (CCC) consonantal root system, particularly the combinatory phonotactic distribution of the consonants within this system based on the theory of Phonology as Human Behavior can be found in Tobin (1990b-c; 1997a:ch. 4).3 These lexical roots pass through elaborate conjugational, inflectional and derivational patterns and systems in order to form words. Some of these grammatical conjugational and inflectional patterns (mʃikalim/binjanin) are signaled by alternating vocalic and consonantal --vocalic affixes (prefixes, infixes, suffixes). Different synthetic semantic and syntactic forms are systematically created through

Hebrew triconsonantal roots revolving around /r/

these miʃkalim/binjanim some of which are analyzed in Tobin (1994/1995: ch. 8, 1996) from the points of view of isomorphism and iconicity. The underlying theoretical and methodological points underlying the interaction between consonantal roots and miʃkalim/binjanim from the sign-oriented approach of this chapter are the following: (i) The Hebrew root — the integral connection between a set of acoustic signals, a sound pattern and a concept of invariant meaning — represent a classic example of the linguistic sign. (ii) Each miʃkal/binjan may also be viewed as a linguistic sign: i.e. an integral connection between a sound pattern (infix, prefix + infix) and a concept (invariant meaning) or set of concepts (a set of invariant meanings belonging to interlocked systems). (iii) The interaction between a root and a miʃkal/binjan serves the communicative function of identifying, conceptualizing, and classifying different kinds of entities, qualities, actions, states, and events. (iv) The lexical root presents the general semantic field of the entity, quality, action, state, and event where the miʃkal/binjan deictically conceptualizes and categorizes different ways of perceiving these entities, qualities, actions, states, and events in a way similar to nominal, verbal and adjectival categories among other syntactic or grammatical and semantic or pragmatic categories. (vi) This categorization of entities, qualities, actions, states, and events is based on the orientation of the encoder at the here-and-now point of encoding. The relationship between roots and words, as well as the connection between the Cr-C roots and the associated semantic fields postulated here, support the concept of language synergy discussed above. The fundamental implications of isomorphic synergesis in language as a semiotic system are that the simpler the form, the more general the meaning and the greater its potential use and frequency while the more complex the form, the more specific its meaning, and the more limited its potential use and frequency.

4. The /C-r-C/ root system In the rest of this chapter I will present a fairly exhaustive list of all the /C-r-C/ roots and show how they can be attributed to the general semantic field of “A Change in Structure” as part of a functional, isomorphic and synergetic classificational system. These roots have literal, as well as possible figurative, and potential metaphoric connections to the concept of structural change which vary according to contexts ranging from the most transparent, and obvious physical or concrete connection to the less apparent and more abstract temporal, existential and associative connections. This spatio-temporal-existential gradation or cline has been illustrated elsewhere for “small words” such as prepositions in general and the preposition in in particu-

295

296 Yishai Tobin

lar (Tobin 1990a: 61): Small words (such as the preposition in) may often begin as locatives (in the room), are extended metaphorically from concrete spatial messages to the more abstract realm of temporal messages (in the morning), to the even more abstract realm of existential messages (in trouble, in pieces), to the point at which they may be nominalized (to be ‘in’) or made into adjectives (the ‘in-group’) [or even inny/outy in the context of navels]. Indeed, there may very well be a universal semantic development of linguistic forms on a universal spatio-temporal-existential cline going from the most concrete spatial messages to the more abstract temporal, to the most abstract existential kinds of messages. This spatio-temporal-existential cline has been considered a cognitive universal (e.g. Givón 1979, Traugott 1978, Wierzbicka 1972, 1980). It has been applied to the historical development of language in general and in the development of pidgins and creoles in particular (e.g., Traugott 1975); various word classes such as copulas, prepositions, and particles (Aphek and Tobin 1988, García, van Putte and Tobin 1987, Traugott 1975, 1978); first language acquisition (Traugott 1974, Clark 1973); the origin and development of idioms in general (Pike 1976) and the development of phrasal verb idioms in particular (Makkai 1972). These studies provide additional evidence for Kronasser’s Law (Kronasser 1952, Kovaćs 1961) which claims that semantic change tends overwhelmingly to move from the concrete to the abstract. It should also be remembered that the data presented in this chapter are prelexical /C–C–C/ roots and what is good for the ‘small word’ goose is probably even better for the triliteral Hebrew root gander.

4.1 Bilabial /C-r-C/ roots 4.1.1 The /p-r-C/ root system revisited

The largest set of /C-r-C/ roots is the /p-r-C/ system partially exemplified in examples (1)-(16) in the introduction. In addition to those sixteen /p-r-C/ roots previously introduced, I also have included here all the /p-r-C/ roots listed in standard dictionaries (e.g. Alcalay 1979), some of which share the same root and are listed separately as “homonyms” in dictionaries (that I will add to the original p-r-C1 with subscripts p-r-C2 , p-r-C3) . . . ⁴ (1) /p-r-m/ ‘unstitch, undo stitches, untie, unbutton, rip; tear open, rend (clothes), rip, cut, chop’ (2) /p-r-f/ ‘fasten together (with a pin), pin up (hair), button’ (3) /p-r-n/ ‘cut, divide, assign, provide, endow, pay a dowry’ (4) /p-r-t/ ‘expand, extend’ (5) /p-r-t’1/ ‘divide into small parts; split, make change (‘break a dollar’); separate, specify, itemize, detail, divide’ (6) /p-r-t’2/ ‘play (piano, stringed instrument), pluck, pinch’ (7) /p-r-d/ ‘separate, branch off, disintegrate, loosen, decompose, depart, divide, divorce’

Hebrew triconsonantal roots revolving around /r/

(8) (9) (10) (11) (12) (13) (14) (15) (16)

/p-r-s1/ /p-r-s2/ /p-r-s3/ /p-r-s4/ /p-r-S1/ /p-r-S2/ /p-r-z1/ /p-r-z2/ /p-r-ts1/

(17) (18) (19) (20)

/p-r-ts2/ /p-r-r/ /p-r-ʃ1/ /p-r-ʃ2/

(21) /p-r-g/ (22) /p-r-x/ (23) /p-r-q/ (24) /p-r-ħ1/ (25) (26) (27) (28)

/p-r-ħ2/ /p-r-ʕ1/ /p-r-ʕ2/ /p-r-ʕ3/

(29) /p-r-ʔ1/ (30) /p-r-ʔ2/ (31) /p-r-h/

‘slice (break, cut bread)’ ‘spread, deploy, fan out, be divided, broken’ ‘have (cloven) hoofs, tread with hoofs’ ‘declare in public, make public’ ‘spread, cast, stretch out, expand, extend, unfurl’ ‘slice (break, cut bread)’ ‘demilitarize, declare an open city’ ‘exaggerate, overdo, overstate, be excessive, spread the fingers’ ‘break, break through, demolish; erupt, make a breach, crack, destroy; burst, rush upon; spread, increase, overflow, break out’ ‘entreat, urge, press, plead, beg’ ‘crumb, crumble, break into crumbs; shatter, undermine’ ‘explain, clarify, distinguish, specify’ ‘retire, go into retreat, seclude (withdraw oneself), keep aloof from, withdraw; set out (to sea), sail’ ‘change for the worse, worsen, deteriorate’ ‘crush, break, crumble; smash’, ‘unload; deliver, set free, extricate; break, remove, dislodge, dismantle, disassemble, decompose, disjoint, demolish, liquidate, take apart, wind up, solve, resolve, crack’ ‘blossom, bloom, burst forth, burgeon, sprout; flourish, prosper, thrive, spread out’ ‘fly, disappear, vanish’ ‘riot, make a pogrom, cause disturbance; abandon, reject’ ‘pay, defray; reward, repay, punish; revenge, avenge’ ‘uncover (esp. hair), be dishevelled, disordered, unkempt, wildlooking’ ‘bring forth, bear, yield (fruit), bloom’ ‘become savage, grow (run) wild’ ‘grow, produce, reproduce, be fruitful (fertile), thrive, flourish’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /p-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (a) the “diminishing/decreasing”, or “the closing of borders”, or “the division/separation of a whole into its component parts”: (1) /p-r-m/ ‘unstitch; . . . cut, chop [cloth]’, (3) /p-r-n/ - ‘cut, divide, [a dowry]’, (7) /p-r-d/ ‘separate . . . disintegrate . . . decompose, depart, divide, divorce’, (18) /p-r-r/ ‘crumb, crumble, break into crumbs; shatter, undermine’, (22) /p-r-x/ ‘crush, break, crumble; smash’, (23) /p-r-q/ ‘unload; deliver, . . . extricate; break, remove, dislodge, dismantle, disassemble, decompose, disjoint, demolish, liquidate, take apart, wind up, solve, resolve, crack’; (b) the “enlarging/increasing”, or “the opening of borders”, or “the extension of the

297

298 Yishai Tobin

boundaries of a whole”: (4) /p-r-t/ ‘expand, extend’, (14) /p-r-z1/ ‘demilitarize, declare an open city’, (15) /p-r-z2/ ‘exaggerate, overdo, overstate, be excessive, spread the fingers’ [open or extend physical or discourse boundaries], (24) /p-r-ħ1/ ‘blossom, bloom, burst forth, burgeon, sprout; flourish, prosper, thrive, spread out’, (25) /p-r-ħ2/ ‘fly, disappear, vanish’ [removal or extension of borders or boundaries], (26) /p-r-ʕ1/ ‘riot, make a pogrom, cause disturbance; abandon, reject’, (27) /p-r-ʕ2/ ‘pay, defray; reward, repay, punish; revenge, avenge’ [extension or opening of physical or monetary boundaries], (31) /p-r-h/ ‘grow, produce, reproduce, be fruitful (fertile), thrive, flourish’; (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (5) /p-r-t’1/ ‘divide into small parts; split, make change (‘break a dollar’)’; separate, specify, itemize, detail, divide’, (6) /p-r-t’2/ ‘play (piano, stringed instrument), pluck, pinch’ [decreasing or change of shape in internal structure], (8) /p-r-s1/ ‘slice (break, cut bread)’ [decreasing], (9) /p-r-s2/ ‘spread, deploy, fan out, be divided, broken’ [increasing/decreasing], (10) /p-r-s3/ ‘have (cloven) hoofs, tread with hoofs’ [internal structure/extension of boundaries], (11) /p-r-s4/ ‘declare in public, make public’ [extension of boundaries], (12) /p-r-S1/ ‘spread, cast, stretch out, expand, extend, unfurl’ [increasing], (13) /p-r-S2/ ‘slice (break, cut bread)’ [decreasing], (16) /p-r-ts1/ ‘break, break through, demolish; erupt, make a breach, crack, destroy, burst, rush upon; spread, increase, overflow, break out’, (17) /p-r-ts2/ ‘entreat, urge, press, plead, beg’ [increasing and/or decreasing physical boundaries or discourse boundaries], (19) /p-r-ʃ1/ ‘explain, clarify, distinguish, specify’ [division or separation in discourse boundaries], (20) /p-r-ʃ2/ - retire, go into retreat, seclude (withdraw oneself), keep aloof from, withdraw; set out (to sea), sail’ [internal change and opening of boundaries], (21) /p-r-g/ ‘change for the worse, worsen, deteriorate’ [internal change],(28) /p-r-ʕ3/ ‘uncover (esp. hair), be dishevelled, disordered, unkempt, wild-looking’ [internal change and/or extension of boundary], (29) /p-r-ʔ1/ ‘bring forth, bear, yield (fruit), bloom’, (30) /p-r-ʔ2/ ‘become savage, grow (run) wild’ [extension of boundaries and/or internal change].

4.1.2 /b-r-C/ roots

There are seventeen single roots or so-called homonymous roots of the /b-r-C/ pattern: (1) (2) (3) (4) (5)

/b-r-d/ /b-r-s/ /b-r-z/ /b-r-ts/ /b-r-r/

‘hail, shower, storm’ ‘bray like a donkey’ ‘tap (faucet), bore; close the tap’ ‘overflow, run (brim) over, spill’ ‘select, choose, pick, prove, separate; examine, investigate; cleanse, purify, swift’

Hebrew triconsonantal roots revolving around /r/ 299

(6) (7) (8) (9) (10)

/b-r-ʃ / /b-r-g/ /b-r-x1/ /b-r-x2/ /b-r-q/

(11) /b-r-ħ1/ (12) (13) (14) (15) (16) (17)

/b-r-ħ2/ /b-r-ʔ1/ /b-r-ʔ2/ /b-r-ʔ3/ /b-r-h1/ /b-r-h2/

‘brush, be brushed’ ‘screw, fasten with screws’ ‘kneel, bend the knee, genuflect’ ‘bless, greet, congratulate, praise, thank; curse (euph.)’ ‘flash (lightning), send forth lightning, glitter, polish, shine, fulminate, brighten, become blind (dazzled)’ ‘run away, flee, take to one’s heels, escape, elope, abscond, make haste’ ‘go through, bolt, fasten with a bolt, make fast’ ‘create, form, shape, make, produce’ ‘cut down, fell (trees), deforest’ ‘recover (from illness), recuperate’ ‘eat, taste, feed, give food (to a mourner)’ ‘choose, select’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /b-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (a) the “diminishing/decreasing”, or “the closing of borders”, or “the division/separation of a whole into its component parts”: (3) /b-r-z/ ‘tap (faucet), bore; close the tap’; (b) the “enlarging/increasing”, or “the opening of borders”, or “the extension of the boundaries of a whole”: (4) /b-r-ts/ ‘overflow, run (brim) over, spill’, (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (1) /b-r-d/ ‘hail, shower, storm’ [change in weather], (2) /b-r-s/ ‘bray like a donkey’ [internal modification or change in discourse boundary], (5) /b-r-r/ ‘select, choose, pick, prove, separate; examine, investigate; cleanse, purify, swift’ [diminishing/separation and internal change], (6) /b-r-ʃ / ‘brush, be brushed’, (7) /b-rg/ ‘screw, fasten with screws’, (8) /b-r-x1/ ‘kneel, bend the knee, genuflect’ [modification of whole], (9) /b-r-x2/ ‘bless, greet, congratulate, praise, thank; curse (euph.)’ [extension of discourse boundary], (10) /b-r-q/ ‘flash (lightning), send forth lightning, glitter, polish, shine, fulminate, brighten, become blind (dazzled)’ [external change in weather or internal modification] (11) /b-r-ħ1/ ‘run away, flee, take to one’s heels, escape, elope, abscond, make haste’, (12) /b-r-ħ2/ ‘go through, bolt, fasten with a bolt, make fast’ [extension of boundary or internal modification], (13) /b-r-ʔ1/ ‘create, form, shape, make, produce’ [extend boundaries], (14) /b-r-ʔ2/ ‘cut down, fell (trees), deforest’ [diminish boundaries], (15) /b-r-ʔ3/ ‘recover (from illness), recuperate’ [internal modification], (16) /b-r-h1/ ‘eat, taste, feed, give food (to a mourner)’ [internal modification and/or extension of boundaries], (17) /b-r-h2/ ‘choose, select’ [decreasing/separation].

300 Yishai Tobin

4.1.3 /m-r-C/ roots There are fourteen single roots or so-called homonymous roots of the /m-r-C/ pattern: (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)

/m-r-t’/ /m-r-d1/ /m-r-d2/ /m-r-s1/ /m-r-s2/ /m-r-ts/

‘pluck, pull (feather, hair), polish, sharpen’ ‘rebel, revolt’ ‘ache, discharge matter’ ‘mix, stir, squeeze’ ‘fester, suppurate, discharge pus ’ ‘urge, stir, encourage, hearten, energize, force, goad, egg on, spur, stimulate’ /m-r-r/ ‘cause bitterness, embitter, grieve, distress’ /m-r-k/ ‘polish, rub, scrub, scour; cleanse’ /m-r-g/ ‘thresh’ /m-r-x/ ‘soften, intimidate’ /m-r-ħ/ ‘rub in, smear, plaster, anoint, spread’ /m-r-ʔ/ ‘fly, take off; soar’ /m-r-h1/ ‘rebel against, disobey’ /m-r-h2/ ‘fatten’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /m-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as: (b) the “enlarging/increasing”, or “the opening of borders”, or “the extension of the boundaries of a whole”: (12) /m-r-ʔ/ ‘fly, take off; soar’, and (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (i) physical modifications: (1) /m-r-t’/ ‘pluck, pull (feather, hair), polish, sharpen’, (3) /m-r-d2/ ‘ache, discharge matter’, (4) /m-r-s1/ ‘mix, stir, squeeze’, (5) /m-rs2/ ‘fester, suppurate, discharge pus’ (8) /m-r-k/ ‘polish, rub, scrub, scour; cleanse’, (9) /m-r-g/ ‘thresh’, (10) /m-r-x/ ‘soften, intimidate’, (11) /m-r-ħ/ ‘rub in, smear, plaster, anoint, spread’, or (ii) a change in social or emotional structure: (2) /m-r-d1/ ‘rebel, revolt’, (6) /m-r-ts/ ‘urge, stir, encourage, hearten, energize, force, goad, egg on, spur, stimulate’, (7) /m-r-r/ ‘cause bitterness, embitter, grieve, distress’, (13) /m-r-h1/ ‘rebel against, disobey’.

4.1.4 /w-r-C/ roots

(1) /w-r-d/ - ‘rose, make rose-colored, paint pink This root may indicate “a modification of the color of a whole”.5

Hebrew triconsonantal roots revolving around /r/

4.2 Apical /C-r-C/ roots 4.2.1 /t-r-C/ roots

There are ten single roots or so-called homonymous roots of the /t-r-C/ pattern: (1) /t-r-m/ (2) /t-r-f/ (3) /t-r-s1/ (4) /t-r-s2/ (5) /t-r-z/ (6) /t-r-ts/ (7) /t-r-n/ (8) /t-r-g/ (9) /t-r-ʕ/ (10) /t-r-h/

‘contribute, donate; lift, remove (ashes from the altar), separate (priestly dues) ‘weaken, soften’ ‘contradict, oppose, dispute, debate, protest against; (also) shield, protect’ ‘shutter’ (open and close) ‘have diarrhoea’ ‘explain a difficulty, give a pretext (an excuse), reply to a question’ ‘mast, hoist a mast’ ‘brighten, yellow, stain with citron-color, feed on citrons’ ‘sound the trumpet, blow the horn; warn, sound the alarm; cry bitterly, protest against’ ‘warn, forewarn’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /t-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (b) the “enlarging/increasing”, or “the opening of borders”, or “the extension of the boundaries of a whole”: (9) /t-r-ʕ/ ‘sound the trumpet, blow the horn; warn, sound the alarm; cry bitterly, protest against’, (10) /t-r-h/ ‘warn, forewarn’ [opening of discourse boundaries], (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (1) /t-r-m/ ‘contribute, donate; lift, remove (ashes from the altar), separate (priestly dues) [both opening and closing borders: giving/ taking increasing/decreasing], (2) /t-r-f/ ‘weaken, soften’ [internal change/modification], (3) /t-r-s1/ ‘contradict, oppose, dispute, debate, protest against; (also) shield, protect’ [opening/closing of boundaries discourse and physical], (4) /t-r-s2/ ‘shutter’ (open and close) [modification of structure of window], (5) /t-r-z/ ‘have diarrhoea’ [opening of boundary, internal change], (6) /t-r-ts/ ‘explain a difficulty, give a pretext (an excuse), reply to a question’ [separate or maneuver discourse boundaries], (7) /t-r-n/ ‘mast, hoist a mast’ [modification of structure of a boat], (8) /t-rg/ ‘brighten, yellow, stain with citron-color, feed on citrons’ [“a modification of the color of a whole”],

30

302

Yishai Tobin

4.2.2 /t’-r-C/ roots

There are seven single roots or so-called homonymous roots of the /t’-r-C/ pattern: (1) /t’-r-m/ (2) /t’-r-f1/ (3) /t’-r-f2/ (4) /t’-r-d/ (5) /t’-r-q/ (6) /t’-r-ħ/ (7) /t’-r-h/

‘do before, anticipate, precede, forestall, be anticipated, forestalled’ ‘prey upon, tear to pieces, rend’ (of wild best or bird of prey) seize (for prey); declare ritually unfit for food, knock, strike beat (eggs), wound, prevent, scramble’ ‘mix, confuse, shuffle (cards), beat up (whip) (eggs), stir, scramble’ ‘drive out, drive away, expel; push; trouble, perturb, disturb, distress, bother, annoy, harass, heckle’ ‘bang, slam ‘take pains, take trouble, trouble oneself ’ ‘(Aramaic) throw’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /t’-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (a) the “diminishing/decreasing”, or “the closing of borders”, or “the division/separation of a whole into its component parts”: (1) /t’-r-m/ ‘do before, anticipate, precede, forestall, be anticipated, forestalled’ [close the border before it begins]; (b) the “enlarging/increasing”, or “the opening of borders”, or “the extension of the boundaries of a whole”: (4) /t’-r-d/ - ‘drive out, drive away, expel; push; trouble, perturb, disturb, distress, bother, annoy, harass, heckle’ [open physical and discourse boundaries], (7) /t’-r-h/ ‘(Aramaic) throw’; (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (2) /t’-r-f1/ - ‘prey upon, tear to pieces, rend (of wild best or bird of prey), seize (for prey); declare ritually unfit for food, knock, strike beat (eggs), wound, prevent, scramble’; (3) /t’-r-f2/ ‘mix, confuse, shuffle (cards), beat up (whip) (eggs), stir, scramble’ [decrease, separate, and internal modification of whole], (5) /t’-r-q/ ‘bang, slam’, (6) /t’-r-ħ/ ‘take pains, take trouble, trouble oneself ’ [internal modification].

4.2.3 /d-r-C/ roots There are six single roots or so-called homonymous roots of the /d-r-C/ pattern: (1) /d-r-m/ ‘go south, face south’ (2) /d-r-s/ ‘trample, tread underfoot; prey upon, seize (prey), sink claws into (prey), tear, rend; press; slaughter (by pressing the knife on the throat of the animal)’

Hebrew triconsonantal roots revolving around /r/

(3) /d-r-r/ (4) /d-r-g/ (5) /d-r-ʃ / (6) /d-r-x/

‘liberate, free, release from servitude (slavery), emancipate, enfranchise, manumit’ ‘make steps, stairs, degrees; terrace (mountain slopes); grade, gradate, graduate’ ‘claim, ask, demand, investigate, require, request, enquire; lecture, preach; explain, expound, interpret, discourse’ ‘step, walk, go, march; press, tread down; trample on; bend (bow)’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /d-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (a) the “diminishing/decreasing”, or “the closing of borders”, or “the division/separation of a whole into its component parts”: (1) /d-r-m/ ‘go south, face south’, (2) /d-r-s/ ‘trample, tread underfoot; prey upon, seize (prey), sink claws into (prey), tear, rend; press; slaughter (by pressing the knife on the throat of the animal)’ [decreasing boundaries]; (b) the “enlarging/increasing”, or “the opening of borders”, or “the extension of the boundaries of a whole”: (3) /d-r-r/ ‘liberate, free, release from servitude (slavery), emancipate, enfranchise, manumit’ [opening of boundaries]; (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (4) /d-r-g/ ‘make steps, stairs, degrees; terrace (mountain slopes); grade, gradate, graduate’ [modification of internal structure], (5) /d-r-ʃ / ‘claim, ask, demand, investigate, require, request, enquire; lecture, preach; explain, expound, interpret, discourse’ [opening and closing of discourse borders], (6) /d-r-x/ ‘step, walk, go, march; press, tread down; trample on; bend (bow)’ [opening of boundaries and internal modification].6

4.2.4 /s-r-C/ roots

There are eleven single roots or so-called homonymous roots of the /s-r-C/ pattern: (1) (2) (3) (4) (5) (6)

/s-r-f/ /s-r-v/ /s-r-t’/ /s-r-s/ /s-r-r/ /s-r-g/

(7) /s-r-x/ (8) /s-r-q1/ (9) /s-r-q2/

‘cremate, cover with resin’ ‘refuse, decline, urge; (also) assume importance’ ‘make a film’ ‘castrate, emasculate, geld; distort’ ‘disobey, rebel; be stubborn, defiant, rebellious, mutinous’ ‘knit; make a net, lace, interlace, plait; Talmudical, also) gird, strap’ ‘be dragged, adhere, join’ ‘comb (also fig.); card; scratch, lacerate’ ‘paint red, rouge’

303

304 Yishai Tobin

(10) /s-r-ħ1/ (11) /s-r-ħ2/

‘stink; sin, corrupt’ ‘stretch, spread out, hang over, be dragged’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /s-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (1) /s-r-f/ ‘cremate, cover with resin’, (2) /s-r-v/ ‘refuse, decline, urge; (also) assume importance’ [diminish, close boundaries and modification of structure], (3) /s-r-t’/ ‘make a film’ [reflect or modify a structure], (4) /s-r-s/ ‘castrate, emasculate, geld; distort’ [modify a structure], (5) /s-r-r/ ‘disobey, rebel; be stubborn, defiant, rebellious, mutinous’ [open boundaries or internal modification], (6) /s-r-g/ ‘knit; make a net, lace, interlace, plait; Talmudical, also) gird, strap’ [internal modification of structure], (7) /s-r-x/ ‘be dragged, adhere, join’ [open borders and internal modification], (8) /s-r-q1/ - ‘comb (also fig.); card; scratch, lacerate’, (9) /s-r-q2/ ‘paint red, rouge’ [internal modification], (10) /s-r-ħ1/ ‘stink; sin, corrupt’, (11) /s-r-ħ2/ ‘stretch, spread out, hang over, be dragged’ [internal modification and opening of borders].

4.2.5 /S-r-C/ roots

There are eleven single roots or so-called homonymous roots of the /S-r-C/ pattern: (1) (2) (3) (4) (5) (6)

/S-r-f1/ /S-r-f2/ /S-r-f3/ /S-r-t’/ /S-r-d/ /S-r-g/

(7) /S-r-r/ (8) /S-r-x/ (9) /S-r-q/ (10) /S-r-ʕ/ (11) /S-r-h/

‘burn, destroy by fire’ ‘sip, quaff, suck (from Aramaic)’ ‘cover with resin’ ‘scratch, lacerate; make an incision’ ‘remain, escape, survive, be left’ ‘twine, interweave, interlace, be intertwined, interwoven, interlaced’ ‘rule, reign, prevail, dominate, domineer’ ‘twist, traverse; turn aside; deviate (from the right path); twine, entangle; lace’ ‘whistle; hiss’ ‘stretch oneself, extend’ ‘struggle, strive, contend; be a ruler’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /S-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows:

Hebrew triconsonantal roots revolving around /r/

(a) the “diminishing/decreasing”, or “the closing of borders”, or “the division/separation of a whole into its component parts”: (7) /S-r-r/ ‘rule, reign, prevail, dominate, domineer’ [close borders]; (b) the “enlarging/increasing”, or “the opening of borders”, or “the extension of the boundaries of a whole”: (9) /S-r-q/ ‘whistle; hiss’ [open discourse borders], (10) /S-rʕ/ ‘stretch oneself, extend’; (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (1) /S-r-f1/ ‘burn, destroy by fire’ [diminish, internal modification], (2) /S-r-f2/ ‘sip, quaff, suck (from Aramaic)’ [diminish], (3) /S-r-f3/ ‘cover with resin’ [modification of a whole], (4) /S-r-t’/ ‘scratch, lacerate; make an incision’ [modification], (5) /S-r-d/ ‘remain, escape, survive, be left’ [open borders/modification], (6) /S-r-g/ ‘twine, interweave, interlace, be intertwined, interwoven, interlaced’ [internal modification], (8) /S-r-x/ ‘twist, traverse; turn aside; deviate (from the right path); twine, entangle; lace’ [open borders/internal modification], (11) /S-r-h/ ‘struggle, strive, contend; be a ruler’ [internal modification, close borders].

4.2.6 /z-r-C/ roots

There are eleven single roots or so-called homonymous roots of the /z-r-C/ pattern: (1) (2) (3) (4) (5) (6)

/z-r-m/ /z-r-v/ /z-r-t/ /z-r-d1/ /z-r-d2/ /z-r-z/

(7) (8) (9) (10) (11)

/z-r-r/ /z-r-q/ /z-r-ħ/ /z-r-ʕ/ /z-r-h/

‘flow, pour down, stream; sweep away, flood’ ‘be made to flow; be heated’ ‘span’ ‘sprout, grow up, trim (young shoots)’ ‘howl, whine, roar’ ‘speed up, accelerate, urge, hurry, spur, hasten, expedite, stimulate, encourage, goad; catalyze (chem.)’ ‘sneeze’ ‘throw, toss, pitch, fling, hurl, dash; sprinkle’ ‘shine, rise (sun), glow, appear, bloom, phosphoresce’ ‘sow, seed, scatter’ ‘scatter, disperse, fab out, winnow, sprinkle’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /z-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (b) the “enlarging/increasing”, or “the opening of borders”, or “the extension of the boundaries of a whole”: (1) /z-r-m/ ‘flow, pour down, stream; sweep away, flood’, (3) /z-r-t/ ‘span’, [open physical borders], (6) /z-r-z/ ‘speed up, accelerate, urge, hurry, spur, hasten, expedite, stimulate, encourage, goad; catalyze (chem.)’ [open borders],

305

306 Yishai Tobin

(8) /z-r-q/ ‘throw, toss, pitch, fling, hurl, dash; sprinkle’, (10) /z-r-ʕ/ ‘sow, seed, scatter’, (11) /z-r-h/ ‘scatter, disperse, fan out, winnow, sprinkle’ [opening of borders]; (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (2) /z-r-v/ ‘be made to flow; be heated’ [open borders/internal modification], (4) /z-r-d1/ ‘sprout, grow up, trim (young shoots)’, (5) /z-r-d2/ ‘howl, whine, roar’ [open and close physical borders / open discourse boundaries], (7) /z-rr/ ‘sneeze’ [internal modification], (9) /z-r-ħ/ ‘shine, rise (sun), glow, appear, bloom, phosphoresce’ [open borders/internal modification].

4.2.7 /ts-r-C/ roots

There are eleven single roots or so-called homonymous roots of the /ts-r-C/ pattern: (1) (2) (3) (4)

/ts-r-m/ /ts-r-f1/ /ts-r-f2/ /ts-r-v/

(5) /ts-r-d/ (6) /ts-r-r1/ (7) /ts-r-r2/ (8) (9) (10) (11)

/ts-r-x/ /ts-r-ħ1/ /ts-r-ħ1/ /ts-r-ʕ/

‘grate (on the ears), injure, split’ ‘refine, purify; (fig.) test; burn’ ‘attach, annex, join, unite, combine; change (money)’ ‘burn, scorch, scald; be mordant; cauterize; corrode; stain (wood); cause heartburn’ ‘become hoarse, be husky, raucous, lose one’s voice’ ‘pack, tie, wrap, bind, parcel up’ ‘be hostile (inimical) towards; discord (mus.); be in distressed, pressed, in anguish, in dire straits’ ‘need, require; consume’ ‘scream, yell, cry, shriek, shout’ ‘castle (in chess)’ ‘be leprous, become leprous’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /ts-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (b) the “enlarging/increasing”, or “the opening of borders”, or “the extension of the boundaries of a whole”: (9) /ts-r-ħ1/ - ‘scream, yell, cry, shriek, shout’, (10) /ts-r-ħ1/ ‘castle (in chess)’ [opening borders discourse and physical]; (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (1) /ts-r-m/ ‘grate (on the ears), injure, split’, (2) /ts-r-f1/ ‘refine, purify; (fig.) test; burn’, (3) /ts-r-f2/ ‘attach, annex, join, unite, combine; change (money)’, (4) /ts-r-v/ ‘burn, scorch, scald; be mordant; cauterize; corrode; stain (wood); cause heartburn’, (5) /ts-r-d/ ‘become hoarse, be husky, raucous, lose one’s voice’ [internal modification/closure of borders], (6) /ts-r-r1/ ‘pack, tie, wrap, bind, parcel up’,

Hebrew triconsonantal roots revolving around /r/

(7) /ts-r-r2/ ‘be hostile (inimical) towards; discord (mus.); be in distressed, pressed, in anguish, in dire straits’, (8) /ts-r-x/ ‘need, require; consume’ [close borders/internal modification], (11) /ts-r-ʕ/ ‘be leprous, become leprous’ [internal modification].7

4.3 Anterodorsal /C-r-C/ roots 4.3.1 /ʃ-r-C/ roots

There are eleven single roots or so-called homonymous roots of the /ʃ-r-C/ pattern: (1) (2) (3) (4) (5) (6) (7)

/ʃ-r-v/ /ʃ-r-t/ /ʃ-r-r/ /ʃ-r-ts/ /ʃ-r-ʃ1/ /ʃ-r-ʃ2/ /ʃ-r-h1/

(8) (9) (10) (11)

/ʃ-r-h2/ /ʃ-r-h3/ /ʃ-r-h4/ /ʃ-r-h5/

‘be overcome by heat, get sunstroke’ ‘serve, officiate, act as, fill an office’ ‘validate’ ‘swarm, teem; abound, multiply, breed, pullulate’ ‘uproot, pluck, destroy, eradicate; extract a number’ ‘strike roots, implant, settle’ ‘immerse, soak, steep, saturate, drench, be immersed, soaked, saturated’ ‘prevail, rest, dwell’ ‘permit, allow’ ‘maintain, support’ ‘save, rescue, extricate’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /ʃ-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (b) the “enlarging/increasing”, or “the opening of borders”, or “the extension of the boundaries of a whole”: (4) /ʃ-r-ts/ ‘swarm, teem; abound, multiply, breed, pullulate’, (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (1) /ʃ-r-v/ ‘be overcome by heat, get sunstroke’, (2) /ʃ-r-t/ ‘serve, officiate, act as, fill an office’, (3) /ʃ-r-r/ ‘validate’ [internal modification/closure of borders], (5) /ʃ-r-ʃ1/ ‘uproot, pluck, destroy, eradicate; extract a number’, (6) /ʃ-r-ʃ2/ ‘strike roots, implant, settle’ [opening and closing of borders/internal modification], (7) /ʃ-r-h1/ ‘immerse, soak, steep, saturate, drench, be immersed, soaked, saturated’, (8) /ʃ-r-h2/ ‘prevail, rest, dwell’, (9) /ʃ-r-h3/ ‘permit, allow’, (10) /ʃ-r-h4/ ‘maintain, support’, (11) /ʃr-h5/ ‘save, rescue, extricate’ [internal modification/closing of borders].

4.3.2 /j-r-C/ roots

There are ten single roots or so-called homonymous roots of the /j-r-C/ pattern: (1) /j-r-t’/ (2) /j-r-d/

‘be precipitate, rush; intercept (an airplane), force down’ ‘come (go) down, descend, decrease, diminish’

307

308 Yishai Tobin

(3) (4) (5) (6) (7) (8) (9) (10)

/j-r-ʃ1/ /j-r-ʃ2/ /j-r-q1/ /j-r-q2/ /j-r-ʕ/ /j-r-ʔ/ /j-r-h1/ /j-r-h2/

‘inherit, take possession of, succeed, possess’ ‘be impoverished, become poor’ ‘spit, expectorate’ ‘become (turn) green; green’ ‘be afraid, tremble’ ‘fear, be afraid, apprehend; respect, stand in awe of, honor’ ‘shoot, fire; throw; pour (water)’ ‘teach, instruct, direct, show, decide, enjoin, command’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /j-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (a) the “diminishing/decreasing”, or “the closing of borders”, or “the division/separation of a whole into its component parts”: (2) /j-r-d/ ‘come (go) down, descend, decrease, diminish’ [close boundaries]; (b) the “enlarging/increasing”, or “the opening of borders”, or “the extension of the boundaries of a whole”: (1) /j-r-t’/ ‘be precipitate, rush; intercept (an airplane), force down’ [open boundaries], (9) /j-r-h1/ ‘shoot, fire; throw; pour (water)’, (10) /j-r-h2/ ‘teach, instruct, direct, show, decide, enjoin, command’ [opening of boundaries]; (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (3) /j-r-ʃ1/ ‘inherit, take possession of, succeed, possess’, (4) /j-r-ʃ2/ ‘be impoverished, become poor’, [close boundaries/internal change], (5) /j-r-q1/ ‘spit, expectorate’, (6) /j-r-q2/ ‘become (turn) green; green’ [open boundaries/internal change/modification of color], (7) /j-r-ʕ/ ‘be afraid, tremble’ [internal modification],8

4.4 Posterodorsal /C-r-C/ roots 4.4.1 /k-r-C/ roots

There are eleven single roots or so-called homonymous roots of the /k-r-C/ pattern: (1) (2) (3) (4) (5) (6)

/k-r-m1/ /k-r-m2/ /k-r-v/ /k-r-t/ /k-r-z/ /k-r-x/

(7) /k-r-ħ/

‘work in a vineyard’ ‘heap, pile up’ ‘fallow, plough’ ‘cut, cut off (down), fell, destroy, root out, lop, excise’ ‘declare, announce, pronounce, make public, advertise’ ‘bind (a book), wrap up, wind around, connect, tie, roll in, combine’ ‘compel, force’

Hebrew triconsonantal roots revolving around /r/ 309

(8) (9) (10) (11)

/k-r-ʕ/ /k-r-h1/ /k-r-h2/ /k-r-h3/

‘kneel, bow, bow down, crouch; collapse, sink, writhe in pain’ ‘dig’ ‘buy, trade, bargain’ ‘give a banquet, make a feast’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /k-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (a) the “diminishing/decreasing”, or “the closing of borders”, or “the division/separation of a whole into its component parts”: (4) /k-r-t/ ‘cut, cut off (down), fell, destroy, root out, lop, excise’ [closing of boundaries]; (b) the “enlarging/increasing”, or “the opening of borders”, or “the extension of the boundaries of a whole”: (5) /k-r-z/ ‘declare, announce, pronounce, make public, advertise’ [opening of discourse boundaries], (7) /k-r-ħ/ ‘compel, force’ [opening of boundaries], (9) /k-r-h1/ ‘dig’, (10) /k-r-h2/ ‘buy, trade, bargain’, (11) /k-r-h3/ ‘give a banquet, make a feast’; (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (1) /k-r-m1/ ‘work in a vineyard’, (2) /k-r-m2/ ‘heap, pile up’ [closure of boundaries/internal modification], (3) /k-r-v/ ‘fallow, plough’ [internal change], (6) /k-r-x/ ‘bind (a book), wrap up, wind around, connect, tie, roll in, combine’, (8) /kr-ʕ/ ‘kneel, bow, bow down, crouch; collapse, sink, writhe in pain’ [closing of boundaries/internal modification].

4.4.2 /q-r-C/ roots

There are twenty-one single roots or so-called homonymous roots of the /q-r-C/ pattern: (1) /q-r-m/ ‘crust, cover with crust, put on (overlay) with skin’ (2) /q-r-v/ ‘come (draw) near, approach, near; be sacrificed’ (3) /q-r-d/ ‘scrape, defur, remove the fur deposit (in kettle), be scraped, defurred’ (4) /q-r-s1/ ‘kneel, bend, bow (sink) down, writhe’ (5) /q-r-s2/ ‘become (turn) sour’ (6) /q-r-s3/ ‘be cracked’ (7) /q-r-ts1/ ‘wink’ (8) /q-r-ts2/ ‘cut, cut out shape (dough), pinch off; form, shape, mould, fashion’ (9) /q-r-n1/ ‘radiate, beam, shine’ (10) /q-r-n2/ ‘grow (have) horns’ (11) /q-r-r1/ ‘cool, chill, be cooled, chilled’

30

Yishai Tobin

(12) (13) (14) (15) (16)

/q-r-r2/ /q-r-ʃ / /q-r-ħ1/ /q-r-ħ2/ /q-r-ʕ/

(17) /q-r-ʔ1/ (18) /q-r-ʔ2/ (19) /q-r-h1/ (20) /q-r-h2/ (21) /q-r-h3/

‘well (gush) forth, flow, cause to well forth, make flow’ ‘freeze, congeal’ ‘make bald, shear, pluck, uproot’ ‘turn into ice’ ‘tear, tear to pieces (asunder), rend, rend a garment in mourning; split, divide; cancel abolish’ ‘read, call (cry) out; proclaim, pronounce, preach; name, call by name; study (the Bible), recite; summon, invite, assemble, convene; implore’ ‘happen, take place, occur’ ‘happen, chance, occur, come about, transpire, come to pass; encounter, meet unexpectedly; (also read /q-r-?/’ ‘roof, cover with rafters, be roofed, covered with rafters’ ‘have night pollution’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /q-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (a) the “diminishing/decreasing”, or “the closing of borders”, or “the division/separation of a whole into its component parts”: (1) /q-r-m/ ‘crust, cover with crust, put on (overlay) with skin’, (16) /q-r-ʕ/ ‘tear, tear to pieces (asunder), rend, rend a garment in mourning; split, divide; cancel abolish’, (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (2) /q-r-v/ ‘come (draw) near, approach, near; be sacrificed’ [closure of boundaries/internal modification], (3) /q-r-d/ ‘scrape, defur, remove the fur deposit (in kettle), be scraped, defurred’ [closure of boundaries/internal modification], (4) /q-r-s1/ ‘kneel, bend, bow (sink) down, writhe’, (5) /q-r-s2/ ‘become (turn) sour’, (6) /q-r-s3/ ‘be cracked’ [internal modification], (7) /q-r-ts1/ ‘wink’, (8) /q-r-ts2/ ‘cut, cut out shape (dough), pinch off; form, shape, mould, fashion’ [closure of boundaries/ internal modification], (9) /q-r-n1/ ‘radiate, beam, shine’, (10) /q-r-n2/ ‘grow (have) horns’, (11) /q-r-r1/ ‘cool, chill, be cooled, chilled’, (12) /q-r-r2/ ‘well (gush) forth, flow, cause to well forth, make flow’ [internal change/opening of borders], (13) /q-r-ʃ / ‘freeze, congeal’, (14) /q-r-ħ1/ ‘make bald, shear, pluck, uproot’, (15) /q-r-ħ2/ ‘turn into ice’ [closure of borders/internal modification], (17) /q-r-ʔ1/ ‘read, call (cry) out; proclaim, pronounce, preach; name, call by name; study (the Bible), recite; summon, invite, assemble, convene; implore’, (18) /q-r-ʔ2/ ‘happen, take place, occur’ [opening/closing of borders/internal change], (19) /q-r-h1/ ‘happen, chance, occur, come about, transpire, come to pass; encounter, meet unexpectedly; (also read /q-r-ʔ/)’, (20) /q-r-h2/ ‘roof, cover with rafters, be roofed, covered with rafters’ (21) /q-r-h3/ ‘have night pollution’ [opening/closing of boundaries/ internal modification].

Hebrew triconsonantal roots revolving around /r/

4.4.3 /g-r-C/ roots

There are seventeen single roots or so-called homonymous roots of the /g-r-C/ pattern: (1) (2) (3) (4) (5) (6) (7) (8) (9)

/g-r-m1/ /g-r-m2/ /g-r-m3/ /g-r-f/ /g-r-v/ /g-r-d1/ /g-r-d2/ /g-r-s1/ /g-r-s2/

‘cause, bring about’ ‘break bones, gnaw, bone, filet’ ‘kill, slaughter (an animal)’ ‘sweep, sweep away, clean, shovel, blow (the nose), collect’ ‘wear a stocking, rob’ ‘scratch, scrabble, scrape’ ‘plait, weave, braid’ ‘crush, pound, mill, grind, make grits; be crushed, pounded’ ‘learn, study, read (the meaning of), determine, maintain, lay down’ (10) /g-r-n1/ ‘laryngize’ (11) /g-r-n2/ ‘make a threshing floor, gather’ (12) /g-r-n3/ ‘be sent to the threshing floor, gathered’ (13) /g-r-r/ ‘drag, tow, tug, draw, pull along; involve, bring about; chew the cud; grate, saw’ (14) /g-r-l/ ‘cast (draw) lots, raffle’ (15) /g-r-ʃ / ‘drive out, expel, evict, deport’ (16) /g-r-ʕ/ ‘lessen, reduce, subtract, deduct, diminish, detract (derogate from), trim (beard), shear’ (17) /g-r-h/ ‘excite, provoke, irritate, stimulate, incite, tease, stir up, egg on’ The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /g-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (a) the “diminishing/decreasing”, or “the closing of borders”, or “the division/separation of a whole into its component parts”: (5) /g-r-v/ ‘wear a stocking, rob’, (8) /g-r-s1/ ‘crush, pound, mill, grind, make grits; be crushed, pounded’, (9) /g-r-s2/ ‘learn, study, read (the meaning of), determine, maintain, lay down’, (16) /g-r-ʕ/ ‘lessen, reduce, subtract, deduct, diminish, detract (derogate from), trim (beard), shear’; (b) the “enlarging/increasing”, or “the opening of borders”, or “the extension of the boundaries of a whole”: (14) /g-r-l/ ‘cast (draw) lots, raffle’, (15) /g-r-ʃ / ‘drive out, expel, evict, deport’, (17) /g-r-h/ ‘excite, provoke, irritate, stimulate, incite, tease, stir up, egg on’; (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (1) /g-r-m1/ ‘cause, bring about’, (2) /g-r-m2/ ‘break bones, gnaw, bone, filet’, (3) /g-r-m3/ ‘kill, slaughter (an animal)’ [opening/closing of boundaries], (4)

3

32

Yishai Tobin

/g-r-f/ ‘sweep, sweep away, clean, shovel, blow (the nose), collect’ [open boundaries/ internal change], (6) /g-r-d1/ ‘scratch, scrabble, scrape’, (7) /g-r-d2/ ‘plait, weave, braid’ [internal modification of structure], (10) /g-r-n1/ ‘laryngize’, (11) /g-r-n2/ ‘make a threshing floor, gather’, (12) /g-r-n3/ ‘be sent to the threshing floor, gathered’ [internal change/closing of boundaries], (13) /g-r-r/ ‘drag, tow, tug, draw, pull along; involve, bring about; chew the cud; grate, saw’ [opening of boundaries/internal change].9

4.5 Pharyngeal /C-r-C/ roots 4.5.1 /ħ-r-C/ roots

There are twenty-five single roots or so-called homonymous roots of the /ħ-r-C/ pattern: (1) (2) (3) (4) (5)

/ħ-r-m1/ /ħ-r-m2/ /ħ-r-f1/ /ħ-r-f2/ /ħ-r-f3/

(6) /ħ-r-f4/ (7) /ħ-r-v/ (8) /ħ-r-t/ (9) /ħ-r-t’1/ (10) /ħ-r-t’2/ (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25)

/ħ-r-d/ /ħ-r-z1/ /ħ-r-z2/ /ħ-r-ts1/ /ħ-r-ts2/ /ħ-r-r1/ /ħ-r-r2/ /ħ-r-r3/ /ħ-r-l/ /ħ-r-ʃ1/ /ħ-r-ʃ2/ /ħ-r-q/ /ħ-r-g/ /ħ-r-x/ /ħ-r-h/

‘swear, vow, confiscate, forfeit, ban, ostracize’ ‘spread a (fishing) net’ ‘winter, spend the winter, hibernate’ ‘insult, abuse, reproach’ ‘become more serious, become aggravated, be made worse, grow more acute; aggravate, worsen, exacerbate’ ‘be betrothed, affianced’ ‘be destroyed, ravaged, desolate, waste; dry up; destroy’ ‘engrave, inscribe, carve, incise, print, imprint, be engraved, inscribed, carved, incised’ ‘chisel, engrave, carve, etch, turn’ ‘repent, regret, feel remorse, change one’s mind (originally scratch oneself in regret’ ‘be afraid, tremble, be anxious, worried, feel uneasy, hurry’ ‘string (together), thread, arrange, array’ ‘rhyme, versify’ ‘cut, notch, nick; decide, resolve, determine’ ‘be diligent, alert, industrious’ ‘make a hole, bore through’ ‘be hot, be dry, burn, glow, overbake’ ‘set at liberty, free, set free’ ‘be covered with thorns’ ‘plough, devise’ ‘be silent, keep silence; be deaf ’ ‘grate, rattle, scratch, squeak, crackle, grind, gnash, cut’ ‘leap, spring forth; deviate, digress’ ‘singe, scorch, burn, char’ ‘be angry, grieved, vexed; burn, be kindled (of anger), glow (with anger)’

Hebrew triconsonantal roots revolving around /r/

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /ħ-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (a) the “diminishing/decreasing”, or “the closing of borders”, or “the division/separation of a whole into its component parts”: (7) /ħ-r-v/ ‘be destroyed, ravaged, desolate, waste; dry up; destroy’, (8) /ħ-r-t/ ‘engrave, inscribe, carve, incise, print, imprint, be engraved, inscribed, carved, incised’, (22) /ħ-r-q/ ‘grate, rattle, scratch, squeak, crackle, grind, gnash, cut’; (b) the “enlarging/increasing”, or “the opening of borders”, or “the extension of the boundaries of a whole”: (23) /ħ-r-g/ ‘leap, spring forth; deviate, digress’; (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (1) /ħ-r-m1/ ‘swear, vow, confiscate, forfeit, ban, ostracize’, (2) /ħr-m2/ ‘spread a (fishing) net’ [opening/closing of boundaries], (3) /ħ-r-f1/ ‘winter, spend the winter, hibernate’, [closing borders/internal change], (4) /ħ-r-f2/ ‘insult, abuse, reproach’, [opening discourse borders], (5) /ħ-r-f3/ ‘become more serious, become aggravated, be made worse, grow more acute; aggravate, worsen, exacerbate’, (6) /ħ-r-f4/ ‘be betrothed, affianced’ [internal modification], (9) /ħ-r-t’1/ ‘chisel, engrave, carve, etch, turn’, (10) /ħ-r-t’2/ ‘repent, regret, feel remorse, change one’s mind (originally scratch oneself in regret’ [closure of borders/internal change], (11) /ħr-d/ ‘be afraid, tremble, be anxious, worried, feel uneasy, hurry’, (12) /ħ-r-z1/ ‘string (together), thread, arrange, array’, (13) /ħ-r-z2/ ‘rhyme, versify’ [change in internal structure], (14) /ħ-r-ts1/ ‘cut, notch, nick; decide, resolve, determine’, (15) /ħ-r-ts2/ ‘be diligent, alert, industrious’ [closure of boundaries/internal modification], (16) / ħ-r-r1/ ‘make a hole, bore through’, (17) /ħ-r-r2/ ‘be hot, be dry, burn, glow, overbake’ [internal modification], (18) /ħ-r-r3/ ‘set at liberty, free, set free’ [opening of boundaries’], (19) /ħ-r-l/ ‘be covered with thorns’, (20) /ħ-r-ʃ1/ ‘plough, devise’, (21) /ħ-r-ʃ2/ ‘be silent, keep silence; be deaf ’ [change in internal structure], (24) /ħ-r-x/ ‘singe, scorch, burn, char’, (25) /ħ-r-h/ ‘be angry, grieved, vexed; burn, be kindled (of anger), glow (with anger)’ [internal modification].

4.5.2 /ʕ-r-C/ roots

There are twenty-three single roots or so-called homonymous roots of the /ʕ-r-C/ pattern: (1) /ʕ-r-m1/ ‘heap, pile, amass, stack, accumulate’ (2) /ʕ-r-m2/ ‘be cunning, crafty, shrewd; become wise, acquire knowledge, act prudently; fool, act cunningly, shrewdly, astutely, deceitfully; intrigue, make wise, impart knowledge’ (3) /ʕ-r-m3/ ‘denude, uncover lay bare’ (4) /ʕ-r-f1/ ‘break the neck, decapitate, behead, guillotine; demolish, destroy’

33

34

Yishai Tobin

(5) /ʕ-r-f2/ (6) /ʕ-r-v1/ (7) /ʕ-r-v2/ (8) (9) (10) (11) (12)

/ʕ-r-v3/ /ʕ-r-v4/ /ʕ-r-d1/ /ʕ-r-d2/ /ʕ-r-s1/

(13) /ʕ-r-s2/ (14) /ʕ-r-ts/ (15) /ʕ-r-r/ (16) /ʕ-r-l1/ (17) /ʕ-r-l2/ (18) /ʕ-r-q/ (19) /ʕ-r-x/ (20) /ʕ-r-g1/ (21) /ʕ-r-g2/ (22) /ʕ-r-h1/ (23) /ʕ-r-h2/

‘drip, shower, drop down’ ‘be dark, grow dark, become evening, be obscured (dusky), become gloomy’ ‘guarantee, stand surety, go bail for, be responsible, vouch for; pawn, pledge’ ‘mix, confuse, intermingle, assimilate, interfere, meddle’ ‘be agreeable, pleasant, sweet’ ‘expel, drive out, stir up’ ‘crumble’ ‘trail (branches of vine), intertwine, braid, form an arbour ((trellis; pergola)’ ‘roll (dough), mix dough’ ‘terrify, frighten, intimidate; destroy; fear, be afraid, adore, worship’ ‘appeal, contest’ ‘treat as unclean, leave uncircumcised, cast away as profane, count as forbidden’ ‘be poisoned’ ‘flee, escape, desert (esp. from army)’ ‘arrange, array, draw up, put (set) in order, lay (a table), prepare; organize, hold; muster, marshal (troops); edit; roll (dough); compare; be equal (comparable) to’ ‘long (pant) for, yearn, languish’ ‘make beds, divide into (flower) beds’ ‘lay bare, uncover, expose, strip, denude; empty, infuse, transfuse; pour out, decant; raze, destroy, demolish, spill’ ‘stick, attach, join, be stuck, attached, joined; be deeply rooted in’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /ʕ-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (a) the “diminishing/decreasing”, or “the closing of borders”, or “the division/separation of a whole into its component parts”: (14) /ʕ-r-ts/ ‘terrify, frighten, intimidate; destroy; fear, be afraid, adore, worship’, (15) /ʕ-r-r/ ‘appeal, contest’ [close borders] (b) the “enlarging/increasing”, or “the opening of borders”, or “the extension of the boundaries of a whole”: (18) /ʕ-r-q/ ‘flee, escape, desert (esp. from army)’ (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (1) /ʕ-r-m1/ ‘heap, pile, amass, stack, accumulate’ [modification of internal structure], (2) /ʕ-r-m2/ ‘be cunning, crafty, shrewd; become wise, acquire knowledge, act prudently; fool, act cunningly, shrewdly, astutely, deceit-

Hebrew triconsonantal roots revolving around /r/

fully; intrigue, make wise, impart knowledge’ [internal modification/opening of boundaries], (3) /ʕ-r-m3/ ‘denude, uncover lay bare’ [internal modification], (4) /ʕ-r-f1/ ‘break the neck, decapitate, behead, guillotine; demolish, destroy’ [closing of boundaries], (5) /ʕ-r-f2/ ‘drip, shower, drop down’ [opening of boundaries], (6) /ʕ-rv1/ ‘be dark, grow dark, become evening, be obscured (dusky), become gloomy’, (7) /ʕ-r-v2/ ‘guarantee, stand surety, go bail for, be responsible, vouch for; pawn, pledge’, (8) /ʕ-r-v3/ ‘mix, confuse, intermingle, assimilate, interfere, meddle’, (9) /ʕ-r-v4/ ‘be agreeable, pleasant, sweet’ [internal modification/closure of borders], (10) /ʕ-r-d1/ ‘expel, drive out, stir up’ [opening of borders], (11) /ʕ-r-d2/ ‘crumble’ [closing of borders], (12) /ʕ-r-s1/ ‘trail (branches of vine), intertwine, braid, form an arbour (trellis; pergola)’, (13) /ʕ-r-s2/ ‘roll (dough), mix dough’ [internal modification], (16) /ʕ-r-l1/ ‘treat as unclean, leave uncircumcised, cast away as profane, count as forbidden’, (17) /ʕ-r-l2/ ‘be poisoned’ [internal modification/ opening/closing of borders], (19) /ʕ-rx/ ‘arrange, array, draw up, put (set) in order, lay (a table), prepare; organize, hold; muster, marshal (troops); edit; roll (dough); compare; be equal (comparable) to’ [internal modification/closure of borders], (20) /ʕ-r-g1/ ‘long (pant) for, yearn, languish’ [internal modification], (21) /ʕ-r-g2/ ‘make beds, divide into (flower) beds’ [internal modification/closing of borders], (22) /ʕ-r-h1/ ‘lay bare, uncover, expose, strip, denude; empty, infuse, transfuse; pour out, decant; raze, destroy, demolish, spill’ [closing/opening of borders], (23) /ʕ-r-h2/ ‘stick, attach, join, be stuck, attached, joined; be deeply rooted in’ [internal modification of structure].10

4.6 Laryngeal /C-r-C/ roots 4.6.1 /ʔ-r-C/ roots

There are twelve single roots or so-called homonymous roots of the /ʔ-r-C/ pattern: (1) (2) (3) (4) (5) (6) (7) (8) (9)

/ʔ-r-v/ /ʔ-r-d/ /ʔ-r-s1/ /ʔ-r-s2/ /ʔ-r-z/ /ʔ-r-r/ /ʔ-r-q/ /ʔ-r-g/ /ʔ-r-x/

(10) /ʔ-r-ħ/ (11) /ʔ-r-ʕ/ (12) /ʔ-r-h/

‘lie in wait, lurk, ambush, waylay’ ‘bronze’ ‘be a tenant; lease’ ‘be betrothed, be engaged (affianced) to’ ‘pack, tie’ ‘curse’ ‘earth, ground, be earthed, grounded’ ‘weave, be woven’ ‘be long, continue, last, endure, be protracted, extend, stretch, prolong’ ‘travel, wander, join the group, accommodate, entertain’ ‘happen, occur, take place, befall’ ‘pick, pluck, gather’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /ʔ-r-C/ roots according to the

35

36

Yishai Tobin

previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (1) /ʔ-r-v/ ‘lie in wait, lurk, ambush, waylay’ [internal modification/closure of boundaries], (2) /ʔ-r-d/ ‘bronze’, (3) /ʔ-r-s1/ ‘be a tenant; lease’, (4) /ʔ-r-s2/ ‘be betrothed, be engaged (affianced) to’, (5) /ʔ-r-z/ ‘pack, tie’ [internal change in structure/ closure of borders], (6) /ʔ-r-r/ ‘curse’, (7) /ʔ-r-q/ ‘earth, ground, be earthed, grounded’, (8) /ʔ-r-g/ ‘weave, be woven’ [internal change in structure/closure of borders], (10) /ʔ-r-ħ/ ‘travel, wander, join the group, accommodate, entertain’ [opening of borders/ internal modification], (11) /ħ-r-ʕ/ ‘happen, occur, take place, befall’ [opening/closing of borders/internal modification].

4.6.2 /h-r-C/ roots

There are four single roots of the /h-r-C/ pattern: (1) (2) (3) (4)

/h-r-g/ /h-r-s/ /h-r-r/ /h-r-h/

‘kill, destroy, slay, murder’ ‘destroy, ruin, demolish, pull down; dare to be destroyed, ruined’ ‘rise like a mountain’ ‘conceive, be pregnant’

The semantic field of “A Change In Structure” is reflected directly or indirectly, (i.e. literally, or figuratively, or metaphorically) in all the /h-r-C/ roots according to the previously mentioned polaric processes as they function in specific contexts found in the universal spatio-temporal-existential as follows: (c) “A Change in Structure” that includes either or both the antonymous or antithetical meanings of these polaric processes and/or (d) “a modification of the size, shape or proportions of a whole” either in a single root or in a set of so-called homonymous roots: (1) /h-r-g/ ‘kill, destroy, slay, murder’, (2) /h-r-s/ ‘destroy, ruin, demolish, pull down; dare to be destroyed, ruined’ [closing of borders/internal modification], (3) /h-r-r/ ‘rise like a mountain’, (4) /h-r-h/ ‘conceive, be pregnant’ [internal change in structure].11

5. Summary and conclusions The general semantic field of “A Change in Structure” can be further divided into various related contextual sub-fields in the following descending order of the approximate number of /C-r-C/ roots they contain: Physical Changes (40), Extensions (general) (40), Extensions over Territory (20), Divisions (30), Border Changes (18), Heat Changes (12), Surface Changes (12), Mental State Changes (12), Connections (12), Control (10), Food/Health (8) Covering (7), Cloth (7) Collecting (3).

Hebrew triconsonantal roots revolving around /r/

This study represents only a very first tenuous step in exploring the hypothesis that there may be a functional, isomorphic and synergetic connection between the phonology of the /C–C–C/ root system and the semantic fields to which they may be interrelated and connected in the formation of the basic Hebrew lexicon. There may very well be additional generalizations connecting certain initial C-I and final C-III root consonants (pharyngeals, laryngeals, or other active articulators and/or stops, fricatives, affricates or other manners of articulation) with certain semantic fields for the present data. There may also be an iconic connection between the trilled/vibrant/mobile /r/ phoneme and the semantic field of “A Change in Structure”. Other phonological classes of C–C–C roots such as /C-l-C/ which is a continuant/stable phoneme should also be explored to further investigate this possible iconic connection with their semantic field: “A Process Leading to a Permanent/Stable Change”. I have limited this exploratory study to tricononsonantal C–C–C roots and additional studies may include biconsonantal C–C or larger C–C–C–C roots as well. It is my hope that this and future studies will support the hypothesis that although the signals and meanings of individual linguistic signs may be fundamentally arbitrary, there may be systematic and synergetic communication- and humanoriented forces connecting signals and meanings in a non-arbitrary way to make their classification more economical and motivated.

Notes * The title of this chapter is a variation of the name of the volume I co-edited with Ellen Contini-Morava: Between Grammar and Lexicon (Contini-Morava and Tobin 2000) focusing on the close interaction between grammar and lexicon. Much of the data was collected for a seminar in Language and Communication at the Ben-Gurion University of the Negev by my student Iris Einhorn. The idea of postulating a functional connection between phonologically related triconsonantal (CCC) roots and semantically related lexical systems underlying this chapter was inspired by the work of my late friend and colleague, Sàndor G. J. Hervey of the University of Saint Andrews, semiotician and linguist par excellence, whose “semiotic perspectives” include the notions of “economy and motivation in semiotic systems” as well as “mnemonic-classificatory economy” (Hervey 1982, 1985, 1988). I would like to dedicate this chapter to his memory. 1. I am presenting an (oversimplified) panchronic view of Hebrew phonetics and phonology whereby certain unavoidable diachronic, orthographic and dialectal inconsistencies are present. These include the biblical occlusive-spirant allophonic (today basically phonemic) alternations [p-f, b-v, k-x] (which have also been partially maintained in Israeli Hebrew), which I will transcribe as six separate phonemes, as well as the similar [g–ɣ, d–ð, t–θ ] alternations which have been lost and will be transcribed as stop phonemes only. Orthographically, each pair is represented by a single letter in the Hebrew consonantal syllabary writing system. I will also maintain the pharyngeal consonants het /ħ/, ayin /ʕ/ and the glottal stop aleph /ʔ/, which do not necessarily appear in Israeli Hebrew, and the (lip + posterodorsal) or (postero-dorsal + uvula) phonemes /w, q/, even though /w/ has been replaced by /v/ and /k/ and /q/ have merged into /k/. I transcribe emphatic ‘tet’ and non-emphatic ‘taf’

37

38

Yishai Tobin as /t’-t/, and ‘samekh’ and ‘sin’ are represented as /s, S/ respectively and ‘ʃin is transcribed as /ʃ /. It also should be noted that I have viewed the historical (and Oriental or Sephardic) Hebrew /r/ phoneme as an apical trill although in Israeli Hebrew it usually is pronounced as a voiced posterodorsal fricative /ɣ/. I also feel that there may be an iconic connection between the trilled/vibrant/mobile /r/ phoneme and the concept of “A Change in Structure” and it may very well be possible that stable phonemes such as /l/ may be related to different semantic fields in an iconic way as well. 2. The central role of the C–C–C root in the structure of, and in the process of word formation in the Hebrew language, is a well-known and recognized phenomenon (e.g. Berman 1987, Ephratt 1997, Ravid 1995, Rosén 1977). Linguists and psycholinguists differ regarding the “mental representation” they postulate for roots versus words and the role they play in the acquisition, perception, and productivity of the Hebrew lexicon (e.g. Aronoff 1976, Bat-El 1994, Bolozkoy 1978, 1999, Feldman et al 1995, Frost et al 1997, 2000, Schwarzwald 1984, Shimron (2003), Walden 1982). Although I am limiting this study to C–C–C roots it may be possible to explore this phenomenon in biconsonantal C–C or quadriconsonantal C–C–C–C roots as well. 3. The theory of Phonology as Human Behavior was developed by William Diver (1979) in an analysis of the non-random distribution of certain classes of initial consonant clusters in English which he later expanded to explain the non-random combinations of vowels and consonants in English and in language in general (Diver 1995). The theory has been extended further to explain the combinatory phonology of consonant and vowel phonemes in other languages such as Italian (1987), Hebrew (Tobin 1990b,c), Urdu (Aronoff 2002, Hameed (this volume), Jabeen 1993), Mewati (Fatihi 1987) and Spanish (Flores 1997). The theory has also been employed to explain the non-random distribution of initial consonant clusters in forty-two different languages representing diverse language families (Tobin 2002a), phonological reduplication in Hebrew roots, applied to the areas of developmental and clinical phonology and prosody in Finnish (Moore 1991a, b, 1993, Moore and Korpijaakko-Huuhka 1996, Moore and Rosenberg-Wolf 1998), Japanese (Tobin and Mikado 2001), and Hebrew, and compared and contrasted to other functional and quantitative oriented phonological theories (Tobin 1988b, 1995, 1997a, b, 1999, 2000, 2001, 2002b). 4. I will not deal with the issue of whether these additional and separate dictionary entries actually represent the same sign or “legitimate” homonyms here. I will only hypothesize if they are suitable for the semantic field of “A Change in Structure” postulated for the (C-r-C) root system. 5. From the point of view of the theory of Phonology as Human Behavior the number of roots in the sets of bilabial /C–C–C/ root patterns is directly proportionate with the effort needed to produce them: /p/ which requires the control of only one set of articulators has the most roots (31) followed by /b/ (17) which requires the control of two sets of articulators, followed by /m/ (14) and /w/ (1) which require the control of three sets of articulators. 6. From the point of view of the theory of Phonology as Human Behavior the number of roots in the sets of mobile apical stop /C–C–C/ root patterns is directly proportionate with the effort needed to produce them: /t/ which requires the control of only one set of articulators has the most roots (10) followed by /t’/ (7) and /d/ (6) which require the control of two sets of articulators. There are no nasal apical C-r-C/ which require the control of three sets of articulators. In Israeli Hebrew both /t/ and the historical ejective /t’/ have merged to nonejective /t/ even further supporting the theory.

Hebrew triconsonantal roots revolving around /r/ 7. From the point of view of the theory of Phonology as Human Behavior the number of roots in the sets of stable apical fricatives and the sole voiceless apical affricate in the /C–C– C/ root patterns is directly proportionate with the effort needed to produce them: /s/ (11) and /S/ (11) which require the control of only one set of articulators have the most roots (22) while the less favored voiced /z/ which requires the control of two sets of articulators and the disfavored affricate /ts/ each have 11 instances respectively. 8. From the point of view of the theory of Phonology as Human Behavior the number of roots in the sets of the anterodorsal stable voiceless fricative and the sole voiced anterodorsal glide in the /C–C–C/ root patterns is directly proportionate with the effort needed to produce them: /ʃ / (11) which requires the control of only one set of articulators is slightly favored (11) while the less favored voiced glide /j/ which requires the control of two sets of articulators but is easier to make has 10 instances. 9. From the point of view of the theory of Phonology as Human Behavior the number of roots in the sets of the posterodorsal mobile voiceless stops and the sole voiced posterodorsal mobile stop in the /C–C–C/ root patterns is directly proportionate with the effort needed to produce them: /k/ (11) and /q/ (21) which require the control of only one set of articulators are favored (32) over the voiced stop /g/ which requires the control of two sets of articulators and has 17 instances. In Israeli Hebrew both /k/ and the /q/ have merged to /k/ even further supporting the theory. 10. From the point of view of the theory of Phonology as Human Behavior the number of roots in the sets of mobile pharyngeal fricatives in the /C–C–C/ root patterns is directly proportionate with the effort needed to produce them: /ħ/ (25) which requires the control of only one set of articulators is slightly favored over the voiced stable pharyngeal fricative /ʕ/ which requires the control of two sets of articulators and has 23 instances. 11. From the point of view of the theory of Phonology as Human Behavior the number of roots in the sets of the laryngeal voiceless laryngeal stop and fricative in the /C–C–C/ root patterns is directly proportionate with the effort needed to produce them: the laryngeal mobile stop /ʔ/ which requires less control of articulators, is easier and acquired earlier, and is favored in initial position is favored (25) over the more difficult voiceless stable pharyngeal fricative /h/ and has only 4 instances.

References Abel, Carl. 1884. Über den Gegensinn der Urworte. Leipzig. Alcalay, Reuben. 1979. The Complete Hebrew/English Dictionary. Ramat-Gan/Jerusalem: Massada. Andrews, Edna and Tobin, Yishai (eds). 1996. Towards a Calculus of Meaning: Studies in Markedness, Distinctive Features and Deixis. Amsterdam/Philadelphia: John Benjamins. Aphek, Edna and Tobin, Yishai. 1988. Word Systems in Modern Hebrew: Implications and Applications. Leiden/New York: E. J. Brill. Aphek, Edna and Tobin, Yishai. 1989. “The Agnonian text: A study of polaric semiology.” In Tobin (ed.) 1989a, 215–233. Aphek, Edna and Tobin, Yishai.1991. “Semantic polarity and the origin of language.” In Studies in Language Origins Volume II, W. von Raffler-Engel, J. Wind and A. Jonker (eds), 263–284. Amsterdam/Philadelphia: John Benjamins.

39

320 Yishai Tobin Aronoff, Mark. 1976. Word Formation in Generative Grammar. Cambridge, MA: MIT Press. Aulanko, R. and Leiwo, M. (eds.) 1991. Studies in Logopedics and Phonetics 2. Publications of the Department of Phonetics, University of Helsinki, Series B: Phonetics, Logopedics and Speech Communication 3. Azim, Abdul.l. 2002. “Problems of aspiration in modern standard Urdu.” In Reid, Otheguy and Stern (eds), 273–307. Bat-El, Outi. 1994. “Stem modification and cluster transfer in modern Hebrew.” Natural Language and Linguistic Theory 12: 571–596. Ben-Ezra. 1921. “Antithetical words.” Ha-Doar 22: 357. (in Hebrew). Berman, Ruth. 1987. “Productivity in lexicon. New-word formation in modern Hebrew.” Folia Linguistica 21: 227–254. Beöthy, E. and Altman, G. 1984a. “The diversification of meaning of Hungarian verbal prefixes II.” Finnisch-Ugrische Mitteilungen 8: 29–37. Beöthy, E. and Altman, G. 1984b. “Semantic diversification of Hungarian verbal prefixes III.” Glottometrika 7: 45–56. Blau, Armin. 1928. Über den Gegensinn der Worte im Hebraischen. Berlin: Itzkowski & Co. Bolozkoy, Shmuel. 1987. “Word formation strategies in the Hebrew verb system.” Afroasiatic Linguistics 5 (3): 1–26. Bolozkoy, Shmuel. 1999. Measuring Productivity in Word Formation: The Case of Israeli Hebrew. Leiden/New York: E. J. Brill. Clark, Herbert. 1973.“Space, time, semantics and the child.” In Cognitive Development and the Acquisition of Language, T. F. Moore (ed), 18–33. New York: Academic Press. Contini-Morava, Ellen and Sussman Goldberg, Barbara (eds). 1995. Meaning as Explanation: Advances in Linguistic Sign Theory. Berlin/New York: Mouton de Gruyter. Contini-Morava, Ellen and Tobin Yishai (eds). 2000. Between Grammar and Lexicon. Amsterdam/Philadelphia: John Benjamins. Davis, Joseph. 1987. “A combinatory phonology of Italian.” Columbia University Working Papers in Linguistics 8: 1-99. Diver, William. 1974. “Substance and value in linguistic analysis.” Sémiotext(e) 1 (2): 11–30. Diver, William. 1975. “Introduction.” Columbia University Working Papers in Linguistics 2: 1–20. Diver, William. 1979. “Phonology as human behavior.” In Psycholinguistic Research: Implications and Applications, Dorothy Aaronson and Robert Reiber (eds), 161–186. Hillside NJ: Lawrence Erlbaum. Diver, William. 1995. “Theory.” In Contini-Morava and Sussman Goldberg (eds), 43–114. Ephratt, Michal. 1997. “The psycholinguistic status of the root in modern Hebrew.” Folia Linguistica 31: 77–103. Fatihi, A.R. 1987. Economy of Articulation in Mewati Phonology. Ph.D. Dissertation, Aligarh Muslim University, U.P., India. Feldman, L. B., Frost, R. and Pinni, T. 1997. “Decomposing words into their constituent morphemes: Evidence from English and Hebrew.” Journal of Experimental Psychology: Learning, Memory and Cognition 21 (4): 947–960. Flores, Nydia. 1997. “The distribution of post-vocalic phonological units in Spanish.” Paper presented at the Fifth International Columbia School Conference on Linguistics, Rutgers University, February 15, 1997. Freud, Sigmund. 1910. “Über den Gegensinn der Urworte.” Jahrbuch für psychopathologische Forschungen 2: 179–184. Freud, Sigmund. 1958. On Creativity and the Unconscious. New York: Harper/Row.

Hebrew triconsonantal roots revolving around /r/ Freud, Sigmund. 1976. The Interpretation of Dreams. Harmondsworth: Penguin. Frost, Ron, Deutsch, A. and Forster, K.I. 2000. “Decomposing morphologically complex words in nonlinear morphology.” Journal of Experimental Psychology: Learning, Memory and Cognition 26 (3): 751–765 García, Erica, van Putte, Florimon, and Tobin, Yishai. 1987. “Cross linguistic equivalence, translatability, and contrastive analysis.” Folia Linguistica 21: 373–405. Givón, Talmy. 1979. On Understanding Grammar. New York: Academic Press. Gordis, Robert. 1936. “Studies in Hebrew roots of contrasted meanings.” Jewish Quarterly Review 27: 33–58. Hameed, Shahana. (this volume). “Interaction of physiology and communication in the makeup and distribution of stops in Lucknow Urdu.” Paper presented at the Sixth International Columbia School Conference on Linguistics, Rutgers University, October 11, 1999. Hervey, Sándor J. G. 1982. Semiotic Perspectives. London: George Allen and Unwin. Hervey, Sándor J. G. 1985. “Mnemonic classificatory economy.” La Linguistique 21: 141–158. Hervey, Sándor J. G. 1988. “Economy and motivation in semiotic systems.” La Linguistique 24: 27–38. Jabeen, Shazi Shah. 1993. Economy of Articulation in the Phonology of Bihar Urdu (as Spoken in and around Gaya. Ph.D. Dissertation, Aligarh Muslim University, U.P., India. Jakobson, Roman. 1941/1968. Child Language, Aphasia, and Phonological Universals. The Hague/Paris: Mouton. (Kindersprache, Aphasie und allgemeine Lautgesetze, Uppsala Universiteits arsskift, 1941: 1–83). Jakobson, Roman. 1971. Selected Writings I, Phonological Studies. The Hague/Paris: Mouton. Jakobson, Roman and Waugh, Linda. 1979. The Sound Shape of Language. Berlin/New York: Mouton de Gruyter. Klein-Andreu, Flora. 1983. Discourse Perspectives on Syntax. New York: Academic. Kovaćs, Ferenc. 1961. “A propos d’une loi sémantique.” Acta Linguistica Hungarica II (3–4): 405–411. Kronasser, Heinz. 1952. Handbuch der Semasiologie. Heidelberg: Carl Winter Universitätsverlag. Krylov, J. K. 1982. “Eine Untersuchung statistischer Gesetzmässigkeiten auf der paradigmatischen Ebene der Lexik natürlicher Sprachen.” In Studies on Zipf ’s Law, H. Guiter and H. Arapov (eds), 234–262, Bochum: Brockmeyer. Landau, E. 1896. Die gegensinnen Wörter im Alt- und Neuhebräisches sprachvergleichend dargestellt. Berlin. Makkai. Adam. 1972. Idiom Structure in English. The Hague/Paris: Mouton. Mandelbrot, Bênoit. 1953. “An information theory of the statistical structure of language.” In Communication Theory, W. Jackson (ed), 486–502. London: Unwin Hyman. Menzerath, P. 1954. Die Architektonik des deutschen Wortschatzes. Bonn: Dumier. Moore, Kate. 1991a. “Speech rate, phonation rate, and pauses in cartoon and sports narrations.” In R. Aulanko and M. Leiwo (eds), 135–143. Moore, Kate. 1991b. “A taxonomy of pauses in Finnish.” In R. Aulanko and M. Leiwo (eds), 145–150. Moore, Kate. 1993. “Developmental disfluencies in preschool children.” In Nordic Prosody VI, Bjorn Granstrom and Lennart Nord (eds), 173–181. Stockhom: Almquist & Wiksell International. Moore, Kate and Anna-Maija Korpijaakko-Huuhka. 1996. “The clinical assessment of flu-

32

322

Yishai Tobin ency in Finnish.” In Advances in Clinical Phonetics, Martin J. Ball and Martin Duckworth (eds), 171–196. Amsterdam/Philadelphia: John Benjamins. Moore, Kate and Rosenberg-Wolf, Capita. 1998. “Perceptions of hesitations in speech.” In Nordic Prosody VII, Stephen Werner (ed), 195–269. Frankfurt/New York: Peter Lang. Morrison, Ben. 1954. Wonderful Words. Johannesburg: City Book Agency & Ellams. Noeldoeke, Th. 1910. “Wörter mit Gegensinn”. Neue Beiträge zur Semitischen Sprachenwissenschaft: 68–108. Pike, Kenneth. 1967. Language in Relation to a Unified Theory of the Structure of Human Behavior. The Hague/Paris: Mouton. Ravid, Dorit. 1995. Language Change in Child and Adult Hebrew. Oxford /New York: Oxford University Press. Reid, Wallis, Otheguy, Ricardo, and Stern, Nancy (eds). 2002. Signal, Meaning, and Message. Amsterdam/Philadelphia: John Benjamins. Rosén, Haiim. 1977. Contemporary Hebrew. The Hague/Paris: Mouton. Saussure, Ferdinand de 1916/1959. Cours de Linguistique Générale / A Course in General Linguistics. Paris: Payot, New York: Philosophical Library. Schwarzwald, Ora R. 1984. “Analogy and regularization in morphophonemic changes: The case of weak verbs in post-biblical and colloquial modern Hebrew.” Afroasiatic Linguistics 9 (2): 87–100. Shimron, Joseph (ed.). 2003. The Processing and Acquisition of Root-Based Morphology. Amsterdam/Philadelphia: John Benjamins. Tobin, Yishai (ed.). 1988a. The Prague School and its Legacy. Amsterdam/Philadelphia: John Benjamins. Tobin, Yishai. 1988b. “Two quantitative approaches to phonology: A contrastive analysis.” In Beiträge zur quantitativen Linguistik, Hermann Bluhme (ed), 71–112. Tübingen: Gunter Narr Verlag. Tobin, Yishai (ed.). 1989a. From Sign to Text: A Semiotic View of Communication. Amsterdam/ Philadelphia: John Benjamins. Tobin,Yishai. 1989b.“The ayin-dalet system in modern Hebrew: a sign-oriented approach.” In Sprechen mit Partikeln, H. Weydt (ed), 391–402. Berlin/New York: Mouton de Gruyter. Tobin, Yishai. 1990a. Semiotics and Linguistics. London/New York: Longman. Tobin, Yishai. 1990b. “A combinatory phonology of the Hebrew triconsonantal (CCC) root system.” La Linguistique 26: 99–114. Tobin, Yishai. 1990c. “Principles for a contrastive phonotactics: The Hebrew triconsonantal (CCC) root system a case in point.” Papers and Studies in Contrastive Linguistics 26: 137–153. Tobin, Yishai. 1991a. “Process and result in the Hebrew infinitive: A study in linguistic isomorphism.” In Studies in Hebrew and Aramaic Syntax, K. Jongeling, M. Van den Berg, and L. Van Rompay (eds), 194–210. Leiden/New York: E. J. Brill. Tobin, Yishai. 1991b. “Sign-oriented linguistic theory: the search for isomorphic unity.” Proceedings of LP ‘90: Linguistics and Phonetics: Prospects and Applications, Palek Bohumil and Janota Premsyl (eds), 112–130. Prague: Charles University Press. Tobin, Yishai. 1993. Aspect in the English Verb: Process and Result in Language. London/New York: Longman. Tobin, Yishai. 1994/1995. Invariance, markedness and distinctive feature analysis: A contrastive study of sign systems in English and Hebrew. Amsterdam/Philadelphia: Benjamins (1994/hardcover) and Be’er Sheva: Ben-Gurion University of the Negev Press (1995/paperback).

Hebrew triconsonantal roots revolving around /r/ Tobin, Yishai. (ed.). 1995. Phonology as Human Behavior: Theoretical implications and cognitive and clinical applications. dibur u-ʃmiya (Speech and Hearing Disorders) 18 (Special Issue on Phonology). Tel-Aviv: The Israel Speech and Hearing Association. (in Hebrew). Tobin, Yishai. 1996. “Invariance, markedness, and distinctive feature theory: The modern Hebrew verb.” In Andrews and Tobin (eds), 347–379. Tobin, Yishai. 1997a. Phonology as Human Behavior: Theoretical Implications and Clinical Applications. Durham, NC/London: Duke University Press. Tobin, Yishai. 1997b. “Developmental and clinical phonology: Roman Jakobson and beyond.” Acta Linguistica Hafniensia 29: 127–166. Tobin, Yishai. 1999. “Developmental and clinical phonology: The Prague school and beyond.” Travaux du Cercle Linguistique de Prague, Nouvelle Série/Prague Linguistic Circle Papers. Volume 3: 53–68. Tobin, Yishai. 2000. “Comparing and contrasting Optimality Theory with the Theory of Phonology as Human Behavior.” The Linguistic Review 17 (2–4): 291–301. Tobin, Yishai. 2001. “Trying to ‘make sense’ out of phonological reduplication in Hebrew.” In Proceedings of LP 2000: Item Order: Its Variety and Linguistic and Phonetic Consequences, Bohumil Palek and Osamu Fujimura (eds), 227–260. Prague: Charles University Press. Tobin, Yishai. 2002a. “Phonology as human behavior: Initial consonant clusters across languages.” In Reid, Otheguy, and Stern (eds), 191–255. Tobin, Yishai. 2002b. “Phonology as human behavior: Theoretical implications and cognitive and clinical applications.” In Clinical Linguistics: Theory and Applications in Speech Pathology and Therapy, Elisabetta Fava (ed), 3–22. Amsterdam/Philadelphia. John Benjamins. Tobin, Yishai and Mikado, Haruko . 2001. “An Analysis of Japanese Speech Errors Based on the Theory of Phonology as Human Behavior.” Paper given at the Second Malaysian International Conference on Languages, Literatures, and Cultures. Kuala Lumpur, April 18, 2001 Traugott, Elizabeth C. 1974. “Explorations in linguistic elaboration: Language change, language acquisition, and the genesis of spatio-temporal terms.” In Historical Linguistics, I . J. Andersen and C. Jones (eds), 263–314. Amsterdam: North Holland. Traugott, Elizabeth C. 1975. “Spatial expressions of tense and temporal sequencing: A contribution to the study of semantic fields.” Semiotica 15 (3): 205–230. Traugott, Elizabeth C. 1978. “On the expression of spatio-temporal relations in language.” In Universals of Human Language, Volume 3, Word Structure, Joseph Greenberg (ed), 369– 400. Stanford: Stanford University Press. Waugh, Linda and Rudy, Stephen (eds.). 1991. New Vistas in Grammar: Invariance vs. Variation. Amsterdam/Philadelphia: John Benjamins. Walden, Zviya. 1982. The Roots of Roots. Ph.d. Dissertation, Harvard University. Wierzbicka, Anna. 1972. Semantic Primitives. Frankfurt: Athenaum. Wierzbicka, Anna. 1980. Lingua Mentalis. New York: Academic Press. Zipf, George Kingsley 1949. Human Behavior and the Principle of Least Effort. Cambridge, MA: Addison-Wesley.

323

CHAPTER 13

Length of the extra-information phrase as a predictor of word order A cross-language comparison Ricardo Otheguy Graduate Center, City University of New York

Betsy Rodríguez-Bachiller Kean University

Eulalia Canals Graduate Center, City University of New York

1. Introduction The theoretical statements and analytical practices of the late William Diver and of the Columbia School in general can be interpreted as embodying the proposition that syntactic observations can be accounted for without postulating a syntactic component (Diver 1995; Huffman 2001). While it is recognized that one can observe certain words appearing in certain orders and displaying certain restrictions of co-occurrence, these observations do not require the postulation of syntactic constructs in the underlying system, which consists exclusively of signals and meanings. There are thus in the analyses of the Columbia School (henceforth CS) no rules of order, no base phrase markers, no rules providing for subsequent alterations of base orders, and in general, no mechanisms to specify what goes with what and in what order. And unlike in other functional schools, in CS practice so far, there are also no constructions. How then, does this a-syntactic system account for alternations in syntactic orders? How does it explain, for example, that adverbial phrases, such as in modest and vernacular forms can appear either before or after direct object NPs, as shown in (1a) and (1b)? (The initials after the examples remit to the Corpus list given before the Reference list at the end of the chapter.) (1) a.

The society of Grand Isle preserves in modest and vernacular forms relics of courtly society . . . RIN x.12

326

Ricardo Otheguy, Betsy Rodríguez-Bachiller, and Eulalia Canals

b. The society of Grand Isle preserves relics of courtly society in modest and vernacular forms . . . (modified from RIN x.12) This chapter is an attempt to provide a CS account for items like (1a) and (1b), to make cross-linguistic predictions based on the account, and to thus illustrate how an explanation of syntactic phenomena might be provided in the radically austere vision of linguistic competence put forth by Diver.

2. Syntactic orders explained under the Columbia School To our knowledge, syntactic orderings in the CS view have only been explained, and can only be produced, by two elements, as outlined here. Determinants of order in language: a. The ordering or position of formatives can itself constitute a signal. b. General cognitive-iconic principles can determine, or incline users toward, certain orderings and groupings. For an example of (a), order or positional signals, there is the well-known case of the English system of Degree of Control, where placing a word inferred to be a Participant before a word inferred to be an Event (i.e., an occurrence) signals that the participant is a High Controller, whereas placing it after the Event signals that it is a Low Controller. (We avoid the more familiar formulation, that pre-verbal NPs in English are generally subjects and post-verbal ones are generally objects, to remind the reader that, in keeping with CS tenets, the system of Degree of Control is intended as a notional, not a syntactic, construct.) The English system of Degree of Control is outlined here. Example of a positional signal: In English, the positioning of: Participant – (Event) – Participant signals: Higher Controller (C1) – (Event) – Lower Controller (C2) Example (1), repeated here as (2), provides an example of the English system of Degree of Control: The society of Grand Isle (C1) preserves (E) in modest and vernacular forms relics of courtly society (C2) . . . RIN x.12 b. The society of Grand Isle (C1) preserves ( E) relics of courtly society (C2) in modest and vernacular forms . . . (modified from RIN x.12)

(2) a.

In (2), the relative ordering of the two Participants, the society of Grand Isle and relics of courtly society, is not specified by a phrase marker; the positional signal of the system of Degree of Control explains why the author didn’t write them in the

Word order and length of extra information

opposite order. And the same is true for the ordering of these two Participants with respect to preserves. (In our discussion from now on, we will ignore the first Participant, in this case The society of Grand Isle, and concentrate on the relation between the Event (E) word and the second Participant (henceforth, P), in this case, between preserves and relics of courtly society.) For an example of the second CS type of explanation for syntactic order, general cognitive-iconic principles, let us take the principle outlined as follows: Example of a cognitive-iconic principle: Group together signals whose meanings form cognitive units: keep together what goes together. An example of this principle at work appears in the same Grand Isle example, repeated below as (3). (We use the letter E to refer to the Event word, the traditional verb; the letter X to refer to the extra information phrase, the traditional adverbial phrase; and the letter P to refer to the second Participant.) Order E X P E = Event X = extra information about the Event P = the second Participant (3) The society of Grand Isle preserves (E) in modest and vernacular forms (X) relics of courtly society (P) . . . RIN x.12 Notice the location of the phrase in modest and vernacular forms, which provides extra information (X) about the Event preserves. In the CS view, the relative ordering of these elements does not require specification through a phrase marker either. The author wrote preserves and in modest and vernacular forms in this order because it makes processing sense to place the extra information phrase where we find it. The most natural place for Extra information about an Event is as near as possible to the Event. Although in principle this functional consideration would be compatible with placement either directly before or directly after the Event, placement directly before the Event has the disadvantage of separating the Event from the first Participant. Therefore, all else being equal, placement directly after the Event is preferable.

3. Competition between signals and cognitive principles One useful way to support the reality of positional signals and cognitive principles of the sort just illustrated is to study these two types of constructs in situations of competition, that is, in situations where the signal calls for one ordering and the principle calls for another. We study this competition from a cross-linguistic perspective, comparing the situation just outlined for English with that of languages where, by hypothesis, there is no positional signal equivalent to that of the English

327

328

Ricardo Otheguy, Betsy Rodríguez-Bachiller, and Eulalia Canals

system of Degree of Control. In such languages, as for example Spanish, the ordering of Events, Participants and extra information (or, in more familiar terms, the ordering of verb, object NP, and adverbial phrase) is, by hypothesis, governed by general cognitive principles alone. Thus, in such languages, unlike in English, the two principles do not compete. These cross-linguistic differences yield testable predictions that can provide empirical support for the rule-less analysis of syntactic alternations, especially in our comparative data from translations. Competition in English between the system of Degree of Control and the cognitive principle of keeping together what goes together is unavoidable. Quite obviously, it is better for the system of Degree of Control to have the second Participant right after the Event, as outlined below, making the positional signal maximally transparent: Preferred ordering for enhancing the transparency of the positional signal of Degree of Control: E P X. But just as obviously, such placement competes with the interest of the grouping principle, which calls for the Extra information to also be placed right after the Event, as outlined here: Preferred ordering for ease of processing the eXtra information about the Event: E X P. In our example about Grand Isle in (2a), the eXtra-information phrase in modest and vernacular forms is in a natural place for processing it together with preserves, but in a disadvantageous place for the positional signal, which would have been more transparent as in (2b). The problem is outlined as follows: Natural and unnatural placement of X: In the order E X P: • X is in its natural place, E X, preserving grouping of X with E. • But X is interrupting the E P positional signal of Degree of Control. In the order E P X: • X is not in its natural place, breaking grouping. • But X is not interrupting the E P positional signal of Degree of Control. Given that in the CS view language systems have only one level of representation, reflecting the one temporal dimension that language users work with, and given that this temporal dimension has to do most of the jobs that the lexicon and the grammatical morphology do not do, competition for a certain slot of the type outlined above is to be expected, and its existence makes sense within a CS approach. The competing claims of two positional principles, and the way they arise in English but, as we shall see, do not arise in other languages, serves as the basis for the predictions in this study. In studying the positional signal of Degree of Control in English and the second

Word order and length of extra information

Participant’s need to be next to the verb, and its competition with the extra information phrase that is also best placed in the same slot, we are setting aside three other functional principles that have been brought forth in the literature as motivators of word order and that are highly compatible with the CS approach. We know from the research by Arnold, Wasow, Losongco and Ginstrom (2000) and others that speakers appear to be striving to put short or light elements before long or heavy ones, and elements containing old or known information before those containing new information. And we know from the work of Hawkins (1994) that the traditional ‘constituent’ appears to be a relevant unit (that is, speakers group signals into meaningful chunks) and that these constituents are ordered in such a way as to facilitate the quick apprehension of the messages intended by utterances. These three principles are almost certainly interacting with the two discussed here (the use of position as a signal, the use of position to keep together what goes together). All of them will eventually have to be taken into account if the CS approach is to succeed in explaining syntactic phenomena without resort to formal syntactic constructs, but such an accounting falls outside the scope of this study.

4. Predictions regarding competition between signals and principles Our first prediction, as outlined below, is that speakers of English will place the eXtra information about the Event in its natural place right after the Event (producing the order E X P) only when this eXtra information phrase is relatively short,¹ so as not to interfere too much with the E P positional signal. A longer eXtra information phrase placed after the Event should tend to interfere with the E P positional signal. Thus the natural position should be avoided for longer Xs, such Xs being placed instead in an unnatural position, after the Participant (producing the order E P X).² Prediction One: In cases of E X P: • X will tend to be shorter. In cases of E P X: X will tend to be longer. An example of shorter X in its natural position is the one in (2a). An example of longer X placed in its unnatural position, to enhance the transparency of the positional E P signal of Degree of Control, is presented in (4): (4) E P X Yet neither does Chopin punish (E) her protagonist (P) with a terrible death, as Flaubert did his Madame Bovary, in the novel to which The Awakening has always been compared (X). RIN vii.25 Given the length of the X here, the signal of the system of Degree of Control would

329

330

Ricardo Otheguy, Betsy Rodríguez-Bachiller, and Eulalia Canals

be placed at too great a risk if the X were kept next to punish, as in (5): (5) E X P Yet neither does Chopin punish (E) with a terrible death, as Flaubert did his Madame Bovary, in the novel to which The Awakening has always been compared (X) her protagonist (P). (modified from RIN vii.25) In the data we report on here, our prediction is supported. But we have also learned that things are more complicated than we thought, for the length of the X also interacts with the length of the second Participant. For the speaker to favor the system of Degree of Control, that is, for the X to be in its unnatural position behind the P, the P itself cannot be too long, or otherwise the association between the E and the X will itself be lost due to excessive distance. So the generalization is actually about the length of both the Extra information and the length of the Participant, as follows: Generalization One : In cases of E X P: • X will tend to be shorter. • P will tend to be longer. In cases of E P X: • X will tend to be longer. • P will tend to be shorter.

5. Statistical predictions for English Counts were made on two English texts, one essay and one fiction (RIN and CPF-E, see details under Corpus list before the References below). Our first count was of how many instances we found in our texts of an X that is longer than a P in each of the two types of orders. Table 1(E) confirms our prediction. In the columns on the left, we see that, when the X is in its unnatural position, not interrupting the positional signal of the system of Degree of Control (E P X), the X is longer than the P in Table 1(E). Occurrences of X longer or shorter than P EPX X not interrupting

EXP X interrupting

N

N

%

%

X>P X≤P

111 12

90 10

3 16

16 84

Total

123

100

19

100

Note: Texts: RIN, CPF-E

Word order and length of extra information Table 2(E). Average length of X and P in words

X P

EPX X not interrupting

EXP X interrupting

6.1 2.0

2.4 4.4 3.7 2.4

X diff P diff Note: Texts: RIN, CPF-E

90% of cases. When, in the column on the right, the X is in its natural position next to the Event word but interrupting the positional signal (E X P), it is longer than the P in only 16% of the cases studied. Our second count, perhaps the most compelling, is of the average length, measured in words, of the X and the P in the two orderings. Table 2(E) shows the situation clearly. In the column on the left, where the X is not interrupting, the X is longer and the P is shorter. On the right side, where the X is interrupting, the X is shorter and the P is longer. We repeated this count using three English fiction texts, one English original and two translations from Spanish and Catalan (SFZ, GTP, RPD-E). (See Table 2.1(E).) Some readers may question the legitimacy of using translations from Spanish and Catalan as data for English, especially since the contrast between English and Spanish is the major focus of our chapter, but we have observed that the pressures of the system of Degree of Control in English are quite robust, despite the possibility of English translations being influenced by Spanish originals and thus being perhaps not as natural as texts originally produced in English. Table 2.1(E). Average length of X and P in words

X P

EPX X not interrupting

EXP X interrupting

3.2 2.4

2.2 5.5

Note: Texts: SFZ, GTP, RPD-E

The pattern in these three texts remains the same. In all cases, the predictions are met, as the interrupting Xs are consistently shorter than the non-interrupting ones; and similarly the Ps with non-interrupting Xs are shorter than the Ps with interrupting ones. A third way of showing the same effect is to compare the ratio of X to P in the two kinds of orderings. Table 3(E) shows the results. When the X is not interrupting, it is on average 2.6 times longer than the P. When it is interrupting, the ratio, as expected, is less than one. This difference can also be expressed as the distance from one, which is the ratio of two equal quantities. The ratio of the length of Xs to Ps in

33

332

Ricardo Otheguy, Betsy Rodríguez-Bachiller, and Eulalia Canals Table 3(E). English. Ratio of length of X to length of P, in total words EPX X not interrupting

EXP X interrupting

Ratio X / P

Ratio X / P

N R

651 / 255 2.55

51 / 96 .53

R diff

2.02

Note : Texts: RIN, CPF-E

the EPX ordering is 1.55 (greater than one), whereas in EXP order the ratio is 0.47 (less than one). These results suggest that users of English, at least the authors of our texts, are mindful of the cognitive principle, but are restrained at the same time by the need to keep the transparency of the word-order signal of Degree of Control.

6. Statistical predictions for Spanish It follows from our analysis that in a language where lower Control is not signaled by position after the Event (where it is signaled, for example, by accusative morphology, or just not signaled at all), the cognitive principle would have free rein in that it would not be competing with a signal of Degree of Control. Under such circumstances, one would expect that phrases giving extra information would appear right after the Event in the preponderance of cases. There is in Spanish a system of Degree of Control, but it does not involve the ordering of lexical items as in English. Alongside instances where the Spanish order in two-participant events is similar to that of English (in traditional terms, NP V NP), it is not uncommon to find cases of V NP NP, where the Event word is followed by the participants. (6) La voluntad de trascender le impidió a Martí detenerse en la especulación, y como en él se produjo la extraordinaria síntesis de una gran inteligencia para el análisis abstracto y para concretar fórmulas de comportamiento, tienen (E) sus ideas (P1) un mérito singular (P2). RMP-S 22.5 . . . have (E) his ideas (P1) a merit singular (P2) ‘Martí’s overriding desire to affect reality kept him away from pure speculative thought: he constantly strove to reduce abstract thought to concrete formulae of conduct, and his ability to do so was singular.’ RMP-E 23.7 In this example we first get the Event (tienen, ‘have’), then the inferred high Controller (sus ideas, ‘his ideas’), then the inferred low Controller (un mérito singular, ‘a

Word order and length of extra information

singular merit’). In Spanish, no signal appears to be limiting the freedom of speakers to place Participants in different positions. Our prediction should be, then, that the effects we found in English in the counts reported above should not be found in Spanish. We try to make this point using different terms in the two languages for the phrases with extra information placed between the Event and the Participant. In Spanish, we say these Xs are intervening, but not interrupting, as they are in English. Our prediction should be that it will be possible to find Xs in this intervening position in Spanish even when they are quite long. And the interaction with the length of the P that we saw in English should also not be present in Spanish. We made the same count in Spanish texts as we did in English³ (we do not offer a comparison with English Table 2.1(E).) Table 1(S). Occurrences of X longer or shorter than P EPX X not intervening N

EXP X intervening %

N

%

X>P X≤P

13 5

72 28

3 11

21 79

Total

18

100

14

100

Note : Text: RMP-S

Table 2(S). Average length of X and P in words

X P

EPX X not intervening

EXP X intervening

6.3 2.6

3.6 5.8 2.7 3.2

X diff P diff Note : Texts: RIN, CPF-E

Table 3(S). Spanish. Ratio of length of X to length of P, in total words EPX X not intervening

EXP X intervening

Ratio X / P

Ratio X / P

N R

114 / 48 2.38

51/82 .62

R diff

1.76

Note : Text: RMP-S

333

334

Ricardo Otheguy, Betsy Rodríguez-Bachiller, and Eulalia Canals

The tables do not show a straightforward confirmation of our expectations. First off, we notice that we have less data for Spanish, and this may be part of the problem. But that aside, notice that we get similar effects in Spanish to those we saw in English, contrary to our prediction. As in English in Table 1(E), in Spanish in Table 1(S) we notice that Xs that are longer than Ps appear mostly in situations where the X is not between the Event and the Participant. For Tables 2(E) and 2(S), note that we get similar results, the intervening Xs of Spanish, like the interrupting Xs of English, are shorter than the non-intervening ones, and the length of the Participant also works as in English. Finally in Tables 3(E) and 3(S), notice that the ratios of X to P in the two orders in Spanish follow the same pattern as in English. But now notice something else in these three tables. The effects go in the same direction, but they are not of the same magnitude. Spanish is, so to speak, a more tolerant version of English. Look at the right hand column of each pair of tables. In Tables 1(E) and 1(S), where English permits only 16% of Xs longer than Ps, Spanish allows 21%. In Tables 2(E) and 2(S), where the average interrupting English X is only 2.4 words in length, the average Spanish X is 3.6 words long. Again in Tables 2(E) and 2(S), note the two lines at the bottom of the tables showing differences between the Xs and the Ps. The differences in English are larger. In Spanish, the differences are still there, but they are smaller. The same can be said of Tables 3(E) and 3(S). Notice the last two lines at the bottom. The ratios behave similarly in the two columns in both languages, but the differences are bigger in English. Clearly, we have to gather more Spanish data. But if these data hold up with a larger N, we are seeing something interesting. Users of Spanish are intolerant of long Extra information between Event and low Participant. But they are not quite as intolerant as users of English. The need to keep the Event and the Participant together seems to be operative also in Spanish. A long intervening X is avoided by users of Spanish, because they too want to have some transparency in the connection between Event and Participant, even if there is no positional signal. But the avoidance of long intervening Xs is not as great as the avoidance of long interrupting Xs by users of English.

7. Further statistical predictions Now that we have discussed the comparisons where the difference between English and Spanish seems to be only a matter of degree, let us go over comparative results where the contrast is greater. We made a prediction regarding the disposition by speakers of these two languages to use these kinds of Xs in the first place. Since Xs after Event words are only intervening, not interrupting, in Spanish, speakers of Spanish should be happy to have them. Since they are interrupting in English, speakers of English should be wary of them. And this is exactly what we find. Consider first Table 4. This table asks the question whether both languages are equally willing to add Extra information to their Events. The answer, as expected, is

Word order and length of extra information Table 4. Presence vs. absence of X English N X is absent X is present

Spanish %

N

%

148 190

44 56

25 32

44 56

338

100

57

100

Note: Texts: RIN, CPF-E, RMP-S

yes. Table 4 tabulates all cases of an Event with a signaled or inferred lower Controller, and shows both languages to be the same. In both languages, 56% of all Events with a lower Controller have Extra information. But look now at Table 5. Here we see something revealing. Only eight percent of all English Xs appear in between Event and Participant. But 44% of Spanish Xs are in that position. Table 5 shows very clearly that speakers of Spanish are much more willing to have intervening Xs than speakers of English are willing to have interrupting ones. Spanish, as we saw in Tables 1, 2 and 3, differs in subtle ways from English in how it handles the Xs placed in their natural position. But it differs from English in dramatic ways in the decision of whether to have those Xs in the first place. So we have perhaps our second generalization: Generalization Two : In a language where a cognitive principle is in competition with a grammatical signal, it will operate less strongly than in a language where the principle faces no competition. English phrases with eXtra information act, so to speak, like capitalists who say they love competition but in fact concentrate their investment in areas where they have a monopoly. These phrases simply avoid entering the market, yielding the spot after the Event, most of the time, to the lower Controller. In contrast, Spanish phrases, facing no competition, occupy this natural spot on a regular basis. Table 5. Placement of X English N Not interrupting E P X Interrupting/intervening E X P Note: Texts: RIN, CPF-E, RMP-S

Spanish %

N

%

174 16

92 8

18 14

56 44

190

100

32

100

335

336

Ricardo Otheguy, Betsy Rodríguez-Bachiller, and Eulalia Canals

8. Support for the analysis from the behavior of translators To test further the results in Table 5, we made a qualitative study of translators who render Spanish texts into English, to see what it is that reluctant English users of interrupting Xs do when faced with the texts of willing Spanish users of Xs which are simply intervening. In some cases, the English translator will keep the intervening Spanish X, turning it into an interrupting English one, as in (7). (7) Creía Martí que ¨cada ser humano lleva (E) en sí (X) un hombre ideal (P), lo mismo que cada trozo de mármol contiene en bruto una estatua tan bella como la que el griego Praxíteles hizo del dios Apolo¨. RMP-S 20.26 He believed that every human being has (E) within him (X) an ideal man (P), just as every piece of marble contains in a rough state a statue as beautiful as the one that Praxiteles the Greek made of the god Apollo. RMP-E 23.1 Here the Spanish Extra information, en sí, can be turned into the kind of short, twoword X that makes English speakers willing to interrupt. So the translator is faithful to the original. But this is not what usually happens. More often, the translator takes the intervening Spanish X and, in constructing the English version, moves it out of the way, as in (8). (8) a.

Durante sus años en Estados Unidos, mientras escribía para los periódicos de la Argentina, Venezuela y México, Martí analizó (E) con notable acierto (X) la vida norteamericana (P). RMP-S 18.24 During the years he spent in the United States, Martí analyzed (E) American society (P) with clarity and insight (X) as a correspondent for the most influential newspapers of Argentina, Venezuela and Mexico. RMP-E 19.26 b. . . . y la bala desbarató el armario del cuarto, atravesó la pared de la sala, pasó con un estruendo de guerra por el comedor de la casa vecina y convirtió (E) en polvo de yeso (X) a un santo de tamaño natural en el altar mayor de la iglesia, al otro extremo de la plaza (P). GCM 10.9 . . . and the bullet wrecked the cupboard in the room, went through the living room wall, passed through the dining room of the house next door with the thunder of war, and turned (E) a life-size saint on the main altar of the church on the opposite side of the square (P) to plaster dust (X). GCD 4.

Notice here that con notable acierto and to plaster dust are intervening, but that the translators were not willing to write the interrupting versions shown in (9).

Word order and length of extra information

(9) a.

During the years he spent in the United States, Martí analyzed (E) with clarity and insight (X) American society (P) as a correspondent for the most influential newspapers of Argentina, Venezuela and Mexico. b. . . . and the bullet wrecked the cupboard in the room, went through the living room wall, passed through the dining room of the house next door with the thunder of war, and turned (E) to plaster dust (X) a life-size saint on the main altar of the church on the opposite side of the square (P).

Sometimes the reluctance to interrupt applies even to a two-word X. And the solution is not always to move the X behind the Participant, but to put it in front of the Event, where it does not interrupt either, as in (10). (10)

Martí censuró (E) con severidad (X) el materialismo, los prejuicios, la soberbia expansionista, la corrupción política (P); y con entusiasmo aplaudió el amor a la libertad, la tolerancia, el espíritu igualitario y la práctica de la democracia. RMP-S 20.4 Marti roundly (X) censured (E) materialism, prejudice, expansionist arrogance, and political corruption, (P) and enthusiastically applauded love of liberty, tolerance, egalitarianism, and the practice of democracy. RMP-E 21.7 Cf. Marti censured with severity materialism. Marti censured roundly materialism.

Notice that the translator avoids the interrupting solution for the translation of con severidad. He makes a lexical adjustment and places the X before the Event. The distaste for an interrupting X is so great for this user of positional signals of Degree of Control, that in (11) he engages in the oldest trick in the translator’s bag: don’t translate. (11) Martí nació en La Habana en 1853, y a los diecisiete años fue deportado por su oposición al régimen español. Después de denunciar (E) ante el mundo (X) los horrores del presidio político (P) en Cuba, y de concluir sus estudios en la Universidad de Zaragoza, se estableció en México, dando inicio a su brillante carrera de escritor. RMP-S 18.8 Martí was born in Havana in 1853. At seventeen he was exiled to Spain for his opposition to colonial rule. There he published a pamphlet exposing (E) the horrors of political imprisonment (P) in Cuba, which he himself had experienced. Upon graduating from the University of Saragossa, he established himself in Mexico, where he began his literary career. RMP-E 19.8 Cf: After denouncing before the world the horrors of political imprisonment in Cuba . . .

337

338

Ricardo Otheguy, Betsy Rodríguez-Bachiller, and Eulalia Canals Table 6.1. Treatment by translator of all Xs, interrupting or intervening (E X P) Spanish to English translation N % Faithful translation Changed translation

3 11 14

21 79 100

Note : Texts: CPF-E, S; RMP-S, E

The original Spanish of (11) says that he denounced before the world the horrors of political imprisonment. The translator has gone through several contortions to get most of the other information in. But this particular X, which would now be three words long in English, was seemingly too much for his need to keep the positional signal transparent, maintaining the connection between the Event exposing and the Participant the horrors of political imprisonment. The Italians say, Traduttore, Traditore. What we are showing are some of the reasons why translators are turned by the grammar into such notorious traitors. This penchant in the translator to move the intervening Spanish X out to some place where it will not end up as an interrupting English X is quite general. Consider Table 6.1. Of the 14 intervening Xs that the translator found in our Spanish text, only three survived as interrupting Xs in English. The other eleven, 79% of the total, were either moved or dropped altogether. Since a comparison group is needed, we have started to make sure that translators going the other way are not doing the same thing, and they are not. The five examples we have checked of interrupting Xs in an English original stay the same in Spanish, as shown in Table 6.2.

9. Conclusion In conclusion, the distribution of Xs and Ps in English is exactly as one would expect it to be in a language where both the eXtra information and the lower Participant are preferably placed next to the Event. When we then compare English with a language where, by hypothesis, this problem does not exist, we get results that fit very well with our understanding of the situation. Users of Spanish fill their texts Table 6.2. Treatment by translator of all Xs, interrupting or intervening (E X P) English to Spanish translation N % Faithful translation Changed translation Note : Texts: CPF-E, S; RMP-S, E

2 3

40 60

5

100

Word order and length of extra information

with intervening eXtra information. Users of English use such Xs sparingly. When translators are faced with these intervening Spanish Xs, they usually move them or drop them altogether. Charles Darwin said that we must pay special attention to those of our predictions that produce results that contradict our thinking or that we do not fully understand. Write them down right away, he said, for otherwise you will forget them. In that spirit, we should not long dwell on what we think may be confirming results in Tables 1E, 2E, and 3E, where users of English handle the length of Xs and Ps as predicted, nor on Table 4, where we show the relative reluctance of English speakers to use interrupting Xs as predicted. Moreover, we should not pay too much attention to the confirmation of our suspicions about the treacherous ways of translators. Instead, we should remember, and continue to try to understand better, the parallel tables in 1 through 3, where we still do not completely understand what is going on. We offer our speculation about the possibility (which would have to be explored in a further study) that there may be a cognitive ranking of notional elements such that an inferred Participant outranks extra information from the point of view of grouping with the Event, all else (such as length considerations) being equal. Still, our study points the way to the possibilities for the kind of syntactic analysis that becomes necessary given the reluctance of CS scholars to accept the traditional syntactic component and the notion of syntactic rules. If one accepts the argument that these constructs are aprioristic and unjustified, one then becomes responsible for explaining how the syntactic observations that these constructs typically cover (order, selection, etc.) are to be explained. We suggest that positional signals and general cognitive principles of the sort discussed here may do the job, especially if we also bear in mind compatible efforts along similar lines by linguists and psycholinguists like Arnold et al., Hawkins, and the many sources they discuss. The present study suggests that positional signals and cognitive grouping principles may be quite real, as they produce solid predictions for a language that appears to operate with both as well as reliable comparative predictions when a language that has only one is brought into the analysis.

Corpus CPF-E = Christie, Agatha. Passenger to Frankfurt. 1970. New York: Pocket Books (Simon & Schuster). CPF-S = Christie, Agatha. Pasajero a Frankfurt. 1990. (Trans. Ramón Margalef Llambrich). Barcelona: Editorial Molino. GCD = García Márquez, Gabriel. Chronicle of a Death Foretold. 1983. (Trans. Gregory Rabassa). New York: Knopf. GCM = García Márquez, Gabriel. Crónica de una muerte anunciada. 1981. Barcelona: Plaza y Janes Editores. GTP-E = García Márquez, Gabriel. 1978. “The Sea of Lost Time.” In Innocent Eréndira, and Other Stories. (Trans. Gregory Rabassa). New York: Harper & Row.

339

340 Ricardo Otheguy, Betsy Rodríguez-Bachiller, and Eulalia Canals

GTP-S

= García Márquez, Gabriel. “El mar del tiempo perdido.” 1981. In La increíble y triste historia de la cándida Eréndida y su abuela desalmada. Barcelona: Bruguera. RIN = Robinson, Marilynne.1992. “Introduction” in Kate Chopin. The Awakening. New York: Bantam Books. RMP-E/RMP-S = Ripoll, Carlos. (Ed.) 1980. José Martí: Pensamientos sobre la libertad, la justicia, el arte y la moral: Antología bilingüe. José Martí: Thoughts on Liberty, Social Justice, Government, Art, and Morality: A Bilingual Anthology. New York: Eliseo Torres & Sons, Las Americas Publishing. RPD-C = Rodoreda, Merçé. La plaça del Diamant: Novel·la. 1962. Barcelona: Club Editor. RPD-E = Rodoreda, Merçé. The Time of the Doves: A Novel. 1980. (Trans. David Rosenthal). New York: Taplinger Publishing Co. SPZ-C = Pierre Salinger. Franny and Zooey. 1999. (Trans. David Rosenthal). Barcelona: Edicions 62. SFZ-E = Pierre Salinger. Franny and Zooey. 1963. SPZ-S = Pierre Salinger. Franny and Zooey. 1978. (Trans. Maribel de Juan).Barcelona: Bruguera.

Notes 1. By “length” we mean the number of words in the respective “constituent.” 2. Counting the number of words is an approximate way to count complexity of a constituent in terms of processing. If a constituent is relatively “flat” (in traditional terms), i.e., not including embedded clauses or the like, then it may be easier to process, hence not as intrusive as one that requires more attention, such as the X in example (4). By and large, more complex constituents will also be longer, thus counting the number of words is an indirect measure of complexity. 3. Instances of both Participants occurring together after the Event are rare. For our counts, we included only Extra information following the Event or the second Participant.

References Arnold, Jennifer, E., Wasow, Thomas, Losongco, Anthony and Ginstrom, Ryan. 2000. “Heaviness vs. newness: The effects of structural complexity and discourse status on constituent ordering.” Language 76 (1): 28–55. Diver, William. 1995.“Theory.” In Meaning as Explanation: Advances in Linguistic Sign Theory, Ellen Contini-Morava and Barbara Sussman Goldberg (eds.), 43–114. Berlin: Mouton de Gruyter. Hawkins, John A. 1994. A Performance Theory of Order and Constituency. Cambridge: Cambridge University Press. Huffman, Alan. 2001. “The Linguistics of William Diver and the Columbia School.” Word 52: 29–68.

CHAPTER 14

Word-order variation in spoken Spanish in constructions with a verb, a direct object, and an adverb The interaction of syntactic, cognitive, pragmatic, and prosodic features* Francisco Ocampo University of Minnesota

1. Introduction In this chapter I consider word order variation in constructions with a transitive verb, a lexical direct object¹ and an adverb in spoken Spanish. This variation may be observed in examples (1) to (5), taken from a corpus of informal conversations: (1) entonces agarró la libreta 17a9² then he.grabbed the notebook ‘Then he took the notebook.’

ADV V DO

(2) las nietas las tiene acá 6a18 the grand-daughters them you.have here ‘Your grand-daughters are living here.’

DO V ADV

(3) pero ahora Thompson no conseguí 1a22 but now Thompson no I.got ‘But this time I couldn’t get Thompson [apples].’

ADV DO V

(4) te daban la caja así 14b16 to-you they.gave the box like.this ‘They gave you the box like this.’

V DO ADV

(5) lo criticaban mucho a Reagan 17a16 to.him they.criticized a.lot to Reagan ‘They criticized Reagan a lot.’

V ADV DO

This variation is not random. There is a correlation between these word orders and a number of syntactic, cognitive, and pragmatic factors.

342

Francisco Ocampo

2. Speakers and data This work is based on a total of 275 examples taken from a corpus of 20 hours of informal conversations encompassing 31 middle class speakers born and raised in La Plata, Argentina. The number and type of constituents in the clause affect word order (Ocampo 1995a), therefore only constructions with a direct object, a verb and an adverb were covered; constructions containing additional constituents were not taken into account. In order to minimize hidden factors, only declarative independent clauses were included. Constructions with noun phrases, which function like adverbs, like los sábados ‘on Saturdays’, were excluded. Moreover, the negative adverbs tampoco ‘neither’, nunca ‘never’, and nada ‘nothing’, were left out as they present particular problems which need further research. I considered direct objects those noun phrases that could be substituted by the direct object clitics lo, la, los, las. In the data, direct objects may co-occur with a clitic as seen in example (2) where the direct object has the coreferential clitic las, and in example (5) where the direct object has the coreferential clitic lo. Silva-Corvalán (1984) regards these cases a variable phenomenon of object-verb agreement, which is in the process of diffusion. As a consequence, in my analysis I do not assume fundamental syntactic differences between examples with clitics and examples without them.³

3. Correlations The different factors that correlate with word order operate on two levels. The first level includes cognitive factors some of which have grammaticalized and are now part of the morphology and syntax of spanish. This level can be called cognitivegrammatical. The second level includes pragmatic-discourse factors and refers to the use of language in conversation. Both levels operate simultaneously. First I will refer to cognitive and grammatical factors and I will state their correlation with word order. Then I will convey how this word order is affected by pragmatic and discourse factors. Syntactically, the constructions analyzed here have two types of argument. The direct object is considered a core argument because the relation between the noun phrase and the verb is present in the semantics of the verb. On the other hand, the adverb is considered an oblique argument since the relationship is expressed outside the verb: it is present in the semantics of the adverb. I will now refer to the cognitive and syntactic factors that correlate with the word order of the direct object. The factor that correlates with the position of the direct object is its topicality (Ocampo 2003). Direct objects appear in postverbal position when they have low topicality and in preverbal position when their topicality is high. Chafe (1976: 50) expresses that “the topic sets a spatial, temporal, or individual

Word-order variation in spoken Spanish

framework within which the main predication holds”. Lambrecht (1994: 127) states “A referent is interpreted as the topic of the proposition if in a given discourse the proposition is construed as being about this referent, i.e. as expressing information which is relevant to and which increases the addressee’s knowledge of this referent.” Instead of topic I will utilize the label topicality to refer to this notion because this term allows establishing gradations, while topic refers to discrete categories. A cognitive factor associated with topicality is the assessment by the speaker of the status new or given of the referent in the mind of the hearer. Noun phrases with high topicality have given referents most of the time. The preverbal position of a NP with high topicality may be motivated by the cognitive principle of advantage of first mention (Gernsbacher and Hargreaves 1992). If we take the previous definition of the topic (i.e. the referent which the proposition expressed by the sentence is about), then the initial position of the topic is motivated by the fact that “first mentioned participants form the foundation of their sentence-level structures, and, therefore, the remainders of the sentence are represented vis à vis those initial participants” (Gernsbacher and Hargreaves 1992: 89.) Given this connection between high topicality, syntax, and cognition, I list the notion of topic under cognitive and syntactic factors.⁴ The second cognitive-grammatical factor that correlates with word order is adverb type and adverb function (Ocampo 1995b and 2001). As previously stated, adverbs are oblique arguments. Their syntactic relationship with the verb is established by factors outside the verb, in this case within the adverb. It is for this reason that adverb type correlates with word order. There is an adverb class whose members occupy a postverbal position when the pragmatic function of the utterance is to convey information only. These adverbs have left scope i.e. they modify constituents that precede them. As the majority of the adverbs in the data belong to this class, these postverbal adverbs may be considered the unmarked case. Another class whose members also appear postverbally when the pragmatic function is to convey information is immediately postverbal adverbs. The adverbs of this class present in the data are listed in (6): (6) Immediately postverbal adverbs: mal ‘bad’, bien ‘well’, más ‘more’, mucho ‘a lot’, poco ‘little’, tanto ‘so much’, bastante ‘enough’, menos ‘less’. These adverbs appear immediately after the verb. Their peculiarity is that first it is not possible to place any item between the verb and some of these adverbs (mucho,⁵ tanto, bastante), and secondly, any adverb placed between the other adverbs (mal, bien, más, poco, menos) and the verb, will not modify the verb but the immediately postverbal adverb, forming a cluster: (7) canta [espantosamente mal] he.sings [dreadfully bad ‘He sings dreadfully bad.’

343

344 Francisco Ocampo

With the exclusion of mal and bien, the rest of the adverbs are quantifiers. Kovacci (1999) also finds that these quantifiers occupy an immediate postverbal position. A. Ocampo (1999) hypothesizes that these adverbs communicate properties basic to the verbal situation, they have grammaticalized and have become almost like affixes. As they form a whole unit with the verb they do not allow the insertion of constituents that do not cluster with them. There is in the data another class of adverbs whose members have a postverbal position when the pragmatic function is to convey information. These adverbs have left scope exclusively and modify the whole construction. For this reason they always appear last in Rioplatense Spanish. The adverbs found in the data are listed in (8): (8) Exclusively left scope adverbs: nomás ‘only’, nada más⁶ ‘no more’ The fourth adverb type that correlates with word order is aspectual adverb. Aspectual adverbs present in the corpus are listed in (9): (9) Aspectual adverbs: ya ‘already’, todavía ‘still’, aun ‘still’, siempre ‘always’, casi ‘almost’ These adverbs have right scope, i.e. they modify constituents that follow them. For this reason they appear in preverbal position when the pragmatic function of the utterance is to convey information. The adverbs in (9) have in common that they convey aspect in their prototypical meaning. Comrie (1981: 3) states that “aspects are different ways of viewing the internal temporal constituency of a situation.” Perfective aspect presents the totality of the situation as a single unanalyzable whole, whereas imperfective aspect makes reference to the internal temporary constituency of the situation. Ya indicates that a situation has been accomplished, or has started, at the moment of speaking. As this adverb presents the beginning or the end of a situation as a whole, it conveys perfective aspect.⁷ Todavía and aún stipulate that a situation has not been completed, or has not started, at the moment of speaking. Casi expresses that, although very near materialization, a situation has not been accomplished at the moment of speaking. Siempre conveys that a situation does not finish: it continues or repeats itself. As these adverbs make reference to a situation in progress, they convey imperfective aspect (Ocampo 1995b: 77). A. Ocampo (1999) hypothesizes that the relationship between these adverbs and the verb is similar to the one that exists between an adjective and a noun in Spanish. In general, adjectives appear after the noun in Spanish. However, an adjective may appear before the noun when it is an epithet: when the meaning that it conveys is also part of the meaning of the noun. (As, for example, in the construction la blanca nieve ‘the white snow’, because snow is prototypically white.) As aspect is an inherent component of the verb these adverbs behave like epithets and they appear in preverbal position. A function that adverbs may have and that correlates with word order is sentential adverb (Barrenechea 1977; Ifantidou-Trouki 1993). These adverbs are not

Word-order variation in spoken Spanish

in construction with the verb but with the whole sentence and they appear at the beginning of the construction. In her categorization of adverbs according to their function, Kovacci (1999: 737) distinguishes “frame adverbs” that set a temporal or a spatial frame with respect to the whole predication. She notes the preverbal position of these adverbs when they are external to the predicate. In my data, sentential adverbs are not restricted to temporal or spatial ones.⁸ Any adverb type may also appear in preverbal position when it functions as a link, relating the proposition of the sentence to the previous context (Ocampo 1995b, 2001). Although this notion of link is not a class but a function and, given the appropriate context, any adverb can be a link, in the data there are adverbs which appear more frequently as links because of their meaning: además ‘besides’, después ‘later’, entonces ‘then’, luego ‘after’. Examples of all the previously mentioned adverb classes and functions appear in Section 4 below. These cognitive and syntactic factors: topicality, status of the referent, adverb type and adverb function; operate simultaneously with a discursive-pragmatic factor. Cognitive-syntactic factors correlate with word order when the pragmatic function of the utterance is to convey information only. This is to be understood in a very broad sense as communicating something. This pragmatic function and the corresponding word order can be considered unmarked. Conveying information is the pragmatic function present in the majority of the constructions in the data, as shown in Table 1. We can see here that 232/275, 84% of the constructions in the data, convey this pragmatic function. I label informational word order the word order that has the pragmatic function of conveying information only. As will be shown in the following sections, when a construction, besides conveying information, has an additional pragmatic function, it experiences a variation in its informational word order. There is also a prosodic factor, primary stress placement, which correlates with the pragmatic function of the utterance. Table 2 A and B present the word orders found in the data, the pragmatic functions, and the number of tokens for each case.⁹ In the tables I label prev all the adverbial types and functions that occupy a preverbal position when the utterance has the pragmatic function of conveying information: aspectual, sentential, and link. I label post all the adverbial types and functions that occupy a postverbal position when the utterance has the pragmatic function of conveying information: unmarked adverbs, immediately postverbal adverbs and exclusively left scope adverbs.¹⁰ In the Table 1. Number of constructions and pragmatic function Pragmatic function

Tokens

%

Convey information Other functions Unclear cases

232 36 7

84 13 3

Total

275

100

345

B

N

A DOtop V ADVpost DOtop V ADVpost DOtop ADVpost V DOtop V ADVpost

ADVprev DO V V DO ADVprev V DO ADVprev 157

2 3 5 5

ADVprev V DO 138 DO ADVprev V 1 ADVprev V DO 3

Word orders and numbers of tokens

Table 2A. Word orders for pragmatic functions

Pragmatic functions Convey info DO contrast ADV contrast DO highlighted ADV and DO highlighted Dev. from expectation Hypothesis/deduction Unclear cases TOTAL

A DO V ADVpost ADVpost V DO ADVpost V DO ADVpost V DO

N 9 1 2 1

C

8

N

V ADVpost DOatop 4

D

C

10

N

V ADVpost DO

D

10

N

N

ADVprev DOtop V

2

N

V ADVip DO

4

DOtop V ADVprev

B

4

10

N

V DO ADVls

13

49

2

3 7 1

10

DO V ADVls

10

1

2

2 64

7

DO ADVls V

Word orders and numbers of tokens

Table 2B. Word orders for pragmatic functions

Pragmatic functions Convey info DO contrast ADV contrast ADV highlighted DO highlighted Dev. from expectation Multiple factors Unclear cases TOTAL

Word-order variation in spoken Spanish

tables direct objects have low topicality, unless explicitly marked with the label top. In the remainder of the exposition, I will first mention the interaction of the previously mentioned factors and word order when the pragmatic function of the utterance is to convey information. Then, I will show how this word order changes when the utterance has an additional pragmatic function. Finally, I will state some conclusions.

4. Word orders 4.1. Convey information As shown in Table 2A, column A, a construction with a low-topicality direct object and an aspectual adverb, sentential adverb, or an adverb that functions as a link has an ADV V DO word order, when it has the pragmatic function of conveying information, as schematized in (10): (10) ADV aspectual V DO sentential link

This order follows the previously stated correlations: these adverb types and functions appear in preverbal position and the direct object with low topicality is in postverbal position. Example (11) shows a construction with an aspectual adverb (listed in (9)):¹¹ (11) y los hijos se () fueron con ella () ahora una de las chicas, la chica and the children cl went with her now one of the girls the girl ‘And her children left with her. Now one of the girls, the girl married,’ se casó, ya tie^ne un neni’to 9b18 cl married already she has a little.baby ‘she already has a little baby.’ We see that the utterance conveys information only. Primary stress falls on the focus, in this case the direct object. In (12) we see a construction with a sentential adverb: (12) yo tuve que explicarle, porque no los conocía. Só”lo conocía I had that to.explain.him because not them he.knew only he.knew voso’tros.¹² 22b19 you.plural ‘I had to explain him because he didn’t know them. He only knew “you plural”.’ Here the adverb sólo ‘only’ modifies the whole sentence. Example (13) shows a construction with an adverb functioning as a link:

347

348

Francisco Ocampo

(13) este y: bueno, estuvo: — de La Plata, se cambió a San Luis. hum and well, she.was from cl she.changed to ‘Hum and well, she was in — from La Plata she moved to San Luis.’ Ahí” le habí”an dado cá’tedras 11b9 there to.her they.had given professorships ‘There she was given a teaching position.’ The construction conveys information only. The deictic adverb ahí links the proposition to the previous context. Primary stress falls on the focus. In the data, some adverbs with right scope, may modify exclusively the constituent to their right forming a cluster with it, as in the following example: (14) querí”an [sólo a é’sas] 17b16 they.loved [only to them.feminine ‘They loved only these girls.’ The adverb sólo modifies the prepositional phrase a ésas. Notice that the adverb does not receive stress. In this case the construction can be considered formed by two constituents, a verb and a direct object. As these constructions are outside the scope of this chapter, which deals with three-constituent constructions, they were not included in the examples analyzed. Their word order follows the regular pattern that two-constituent constructions with a verb and a low-topicality direct object take when they have the pragmatic function of conveying information (Ocampo 1995a). As shown in Table 2, column B, a construction with a high-topicality direct object and a post adverb has a DOtopic V ADVpost word order, exemplified in (15): (15) y el ti’po lo metie^ron ade”ntro 6a14 and the guy him they.put inside ‘And the guy they made him come inside.’ This construction is part of an oral narrative where the speaker recounts his father’s cremation: (16) y después, este, tuve que ir, [. . .] con mi un — con un — con otro hermano mío, a ver, allá al al crematorio.[. . .] mi hermano estaba medio muerto, porque mi hermano de: de:: decía no, yo no quiero ver. Porque a uno lo obligan a ver. [. . .] Mirá nosotros vamos a decir que a nosotros el médico nos prohibió (0.8) porque e somos cardíacos. Tonces te quedás acá y no entrás. Me llaman a mí, me llaman en ese momento de la oficina para:: firmar un coso, y lo e lo meten a éste, lo hacen entrar. Y el tipo vio cuando: cuando ap — cuando aplastaban las cenizas con un rodi:llo, todo eso (que hacen). Una cosa maca:bra es [. . .] Y después cuando vienen las cenizas el tipo la aplasta y pasan un rodi:llo, todo lo lo hacen — muelen. Era desagradabilísimo.[. . .] Y el tipo lo metieron adentro y vio eso. Y le cuento que quedó mal, mi hermano quedó mal. Era mi hermano menor. (1) Quedó como la mona el tipo.

Word-order variation in spoken Spanish 349

‘And then hum I had to go [. . .] with my — with one — with one of my brothers to watch there to the crematory. [. . .] my brother was very upset, because my brother s s said hell no, I don’t want to watch. Because they make you watch. [. . .] look, we will tell them that the doctor forbade us (0.8) because we have a heart condition. So you stay here and you don’t come in. They call me, I was called in that very moment from the crematory’s office to: to sign something, and they make the guy they put him inside, they make him come in. And the guy saw when when fl — when they flattened the ashes with a roller, all those things (that they do). It‘s really macabre. [. . .] And then when the ashes came the guy flattens them and they pass a roller, everything they do — grind. It was extremely unpleasant [. . .] And the guy they made him come inside and he watched that. And let me tell you that he was really upset, my brother was really upset. He was my youngest brother. (1) The guy freaked out completely.’ We see that the DO is the topic of the construction and that the utterance has the pragmatic function of conveying information. The word order of the construction follows the stated correlation: direct objects with high topicality appear in preverbal position and post adverbs show in postverbal position. Table 2A, column C, indicates that a construction with an aspectual adverb or a sentential adverb or an adverb that has a linking function, and a direct object with high topicality has an ADV DO V word order. This is exemplified in (17): (17) ADV aspectual DOtopic V sentential link

E”nto”nces, a e’sa prue” la la le’en:, 12a11 then to this test it it they.read ‘Then, this test is read.’ The context for this construction is the following (18): (18) Vos, para presentarte a un concurso tenés, un: código () e formado por una letra y un número. No firmás no figurás, no sos nadie, sos un número. () Es decir que: — tu prueba, no se puede relacionar con vos, () no es cierto? [. . .] entonces ya es bien objetiva, se supone, la este:m () e la calificación. () Entonces, a esa prueba la la leen:, la analizan ‘To compete for the position you have a code made by a letter and a number. You don’t sign, your name doesn’t show up, you are nothing, you are a number. () In other words your test can’t be related to you, () you see. [. . .] so the result is supposed to be very objective. () Then, this test is read, analyzed’ Here the adverb entonces links the content of the sentence to the previous context. The referent of the direct object is the topic of the construction and it receives a primary stress. In the data, preverbal direct objects with high topicality receive first or

350

Francisco Ocampo

secondary stress. In this example, the verb receives an additional primary stress because it is the focus. In this example, the functions of link and topicality correlate with the preverbal order of both constituents. What must be explained now is why the adverb takes the first position and the DO the second one. In my opinion this is a matter of iconicity. The linking adverb appears first because it is in contact with the previous discourse context. Notice that this function of link has features in common with prepositions, which also appear in first position in Spanish. Shown in Table 2A, column D, there are in the data four cases of a construction with a postverbal adverb and a direct object whose referent has low topicality in the sentence but is an ongoing conversation topic. In all the cases the referents are given. The word order is V ADVpost DOantitopic, exemplified in (19): (19) lo agarró” así al pe’rro 9b10 it she.grabbed like.this to.the dog ‘She grabbed the dog like this.’ This construction is part of an oral narrative about the death of a favorite pet: (20) vivíamos frente a Plaza Matheu. Elena cruzó que creo que iba a la farmacia (de Tasarino) que estaba allá del otro lado la () de la Plaza iba. () Entonces Milord la cor — la: siguió entonces le dije — te dije que te vuelvas a casa, que no podés venir conmigo, que es muy tarde — dijo Elena. Volvete a casa () y él se volvió, le hizo caso a Elena. Se volvió. Y al cruzar de la Plaza Matheu a la vereda de la casa nuestra, un auto que pasaba () ella cuando sintió la frenada del coche, se vino corriendo que no fue a la farmacia ni nada. Y () lo agarró así al perro ‘Our home was facing Matheu Square. Elena crossed the street, I think she intended to go to (Tasarino’s) pharmacy that was over there, on the other side of the Square. So Milord ran — followed her and she said — I told you to go back home, you can’t come with me, it’s too late, Elena said. Go back home () and he went back, he obeyed her. He went back. And while he was crossing from Matheu Square to the sidewalk of our home, a car that was passing by () when she heard the noise of the breaks she came back running, she didn’t go to the pharmacy, nothing. And () she grabbed the dog like this.’ As we see in (19) the general topic of the conversation is the dog, but the sentence in question is about Elena. If we compare (15) and (16) with (19) and (20) we can see that in both cases the topic referent has no agency. In (15) and (16) the agent is vague, and the sentence is about the DO referent, which is the only salient element in the construction. In (19) and (20), on the other hand, two referents compete for topicality: the agent Elena, the topic of the sentence, and the DO el perro ‘the dog’, the topic of the conversation. This construction-type has been mentioned by some authors, Chafe (1976) and Lambrecht (1994) among them. These two authors call antitopic the function of the

Word-order variation in spoken Spanish Table 3. Relationship between topicality and status new/given of the DO referent in constructions with the pragmatic function of conveying information High topicality (N = 21)

Low topicality (N = 211)

Tokens 21

Tokens 211

% 9

Referent new

Referent given

Tokens 2

Tokens 19

% 10

% 90

% 91

Referent new

Referent given

Non referential NP

Tokens 124

Tokens 75

Tokens 12

% 59

% 36

% 6

DO in this construction. Differently to the antitopics described by Lambrecht for French, in my data the direct objects receive stress. As previously stated, a cognitive factor associated with topicality is the assessment by the speaker of the status new or given of the referent in the mind of the hearer. I utilize for the analysis Prince 1981’s typology. I considered new referents her categories brand-new, brand-new anchored, inferrable and unused, and given referents her categories textually evoked and situationally evoked. The relationship between topicality and cognitive status of the direct object referent when the construction has the pragmatic function of conveying information is shown in Table 3. Here we can see that the vast majority of direct objects have low topicality (211/232; 91%), and that only 21/232; 9% of the cases show high topicality. This coincides with the known observation that unmarked direct objects have low topicality and appear in postverbal position. Among the 21 cases with high topicality, the majority (19/21; 90%) exhibits a given referent. We can also see that among direct objects with low topicality, the majority (124/211, 59%) has new referents. Lambrecht (1988: 146) states that human language is subject to a general cognitive constraint imposed on the simultaneous performance of two independent tasks: (a) to determine the intended referent, and (b) to process propositional information about this referent. The newer the referent, the heavier the cognitive effort in the first task. As direct object referents with high topicality present the hearer with a heavy second task, to ease the cognitive effort these constructions will carry a lighter first task in the form of a given direct object referent. Finally, from Table 3 it is also possible to calculate that among constructions with a verb, an adverb and a direct object, that have the pragmatic function of conveying information, the most common type is the one that contains a direct object with low topicality and a new referent 124/232, 53%. As Table 2B, column A shows, when a construction contains a low-topicality direct object and a post adverb, both constituents, predictably, appear in postverbal position. The unmarked word order is V DO ADVpost, with the DO closest to the verb, as seen in (21):

35

352

Francisco Ocampo

(21) entonces, los domingos, cocina el padre, hace el asado, tie”ne so the Sundays cooks the father he.prepares the roast he.has su parrilla afue’ra 19b11 his barbecue outside ‘So, on Sundays the father cooks, he prepares the roast, he has his barbecue outside.’ Here we see that the utterance conveys information only, the direct object has low topicality and the adverb is postverbal. Primary stress falls on the focus. Table 2B, column B illustrates that a construction with a low-topicality direct object and an exclusively left scope adverb has a V DO ADVleft scope word order: (22) ella no trabaja con harina blanca, apenas muy po:co [. . .] y usa she no works with flour white hardly very little and she.uses ace’i:te nomá^s oil only ‘She doesn’t work with white flour, just a little, and she uses only oil.’ 26b31 The adverb nomás modifies the whole construction and has left scope only. For this reason it appears in final position. In Table 2B, column C we can see that constructions with a low-topicality direct object and an immediately postverbal adverb have a V ADVimmediately postverbal DO word order, as in (23): (23) hizo cursos de traducción. Terminó. Pero no — no ha’bla bie^n he.did courses of translation he.finished but no no he.speaks well el idio”ma 6b4 the language ‘He took translation courses. He completed them. But he doesn’t speak well the language.’ Here the direct object has low topicality and the utterance conveys information only. Primary stress falls on the focus, the verb, which is also highlighted. Table 2B, column D indicates that there are in the data 10 cases of constructions with a low-topicality DO and a postverbal adverb that do not follow the ordering shown in (21) V DO ADVpost. Instead they exhibit a V ADVpost DO word order, as seen in (24) (24) Pero yo no he chupa:do, no he trasnocha:do, no he hecho — but I no have boozed no I.have stayed.up.all.night no I.have done he hecho así una vi’da má”s o menos regula’r. (. . .) No he I.have done like.this a life more or less regular no I.have comido cosas ra:ras. 6b21 eaten things strange

Word-order variation in spoken Spanish

‘But I haven’t boozed, I haven’t stayed up all night, I haven’t done — I have lived like a fairly normal life. (. . .) I haven’t eaten unhealthy food.’ Notice that this word order follows the correlations stated in Section 3: both constituents appear in postverbal position. The word order variation exemplified in (23) is the result of a different syntactic alignment, and conveys subtle differences in meaning. In (23) the adverb clusters with the NP, and this is reflected in the prosody: the adverb does not receive stress. If the adverb were placed after the DO it would have retained its original meaning ‘like this’. Instead, here its word order results in a change of meaning. In other examples of this construction-type in the data the adverb clusters with the verb resulting also in fine meaning changes. Differently with V DO ADVpost constructions, in these V ADVpost DO constructions the adverb does not receive primary stress and is not focus.

4.2. Additional pragmatic functions In the previous section I have established correlations between cognitive, grammatical, and prosodic factors and word order when the pragmatic function of the utterance is to convey information only. As previously stated, these informational word orders can be considered unmarked. When, besides conveying information, these same construction-types contain an additional pragmatic function, this function correlates with a variation of the corresponding informational word order of the construction. As shown in Table 2, there are in the data four pragmatic functions, besides the one of conveying information: contrast, highlight, deviation from expectation, and hypothesis/deduction. I will now address these correlations.

4.2.1. Contrast

Contrast is the pragmatic relationship between the rest of the construction and a constituent that stands in opposition to a closed numbers of alternatives, members of the same semantic set. What is asserted is which candidate is the correct one (Chafe 1976; Silva-Corvalán 1983). In spoken Spanish the contrastive constituent appears first and receives primary stress¹³ (Ocampo 1995a, b, 2001). Table 4 comTable 4. Comparison between informational word order and Contrast Convey information

Contrast

ADVprev V DO V DO ADVpost V DO ADVpost ADVprev V DO DOtop V ADVpost DOtop V ADVpost

DOc ADVprev V DOc V ADVpost ADVpostc V DO ADVprevc V DO DOtopc V ADVpost DOtop ADVpostc V

353

354

Francisco Ocampo

pares contrast with conveying information. As shown in Table 4, in the constructions analyzed here the contrastive constituent appears preverbally, the other constituents do not undergo word order change. If the contrastive constituent already occupies the first position when the construction conveys information, then there is no word order change, as is the case of constructions with a preverbal adverb and a low-topicality direct object, and constructions with a high-topicality direct object and a postverbal adverb. In constructions with a high-topicality direct object and postverbal adverb, both constituents appear preverbally. The DO topic appears in first place, the adverb in second place. It seems that topicality wins in the competition for first place, at least for this type of construction. Contrast is exemplified in (25): (25) ADVpost contrast V DO lo que faltan un poco son las salsas, no? p ha – tiene poquitas salsas. (0.6) what miss a bit are the sauces no it.has few sauces ‘What is missing are the sauces, isn’t it? It has few sauces.’ Allá’ tienen más sa:’lsas. 23a9 there they.have more sauces ‘Over there they have more sauces.’ The speaker comes from a lunch in a new restaurant in La Plata, which has a salad bar. As she had visited the United States a short time before the conversation, she contrasts La Plata with the United States, stating that in the U. S. salad bars have more sauces to choose from. Notice that the adverb receives primary stress.

4.2.2. Highlight

When a single constituent is highlighted in any way, it appears in first position and receives primary stress, as seen in (26): (26) somos un partido de: de élite, no no podemos mezclarnos así con we are a party of of elite no no we.can mix.ourselves like.this with la gente () yo te explico () sí sí he [se ríe] () he y si somos the people I to.you explain yes yes ha [laughs] ha and if we.are pocos. No no t m. Ni vamos a figurar, do”s o tres vo’tos saca”mos few no no no will.go to appear two or three votes we.will.take nada má”s. 22a18 nothing more ‘We are an elite party, no no we can’t mix with the populace. Let me explain you () yes yes ha [laughts] () ha we are so few. We don’t. We are not even going to figure [in the vote computation], we will get two or three votes only.’ Here the speaker is talking about the upcoming presidential election. He makes fun of the fact that his party is not a popular one. He highlights the fact that the number of votes that they will receive will be extremely low. The direct object ‘two

Word-order variation in spoken Spanish Table 5. Comparison between informational word order and highlight Convey information

Highlight

V DO ADVpost V DO ADVls DOtop V ADVpost ADVprev V DO

ADVposthighl V DO DOhighl V ADVls DOtophighl V ADVpost ADVprevhighl DOhighl V

or three votes’ is highlighted and appears first. Notice that votos, the nucleus of the noun phrase, besides receiving primary stress is perceived as prosodically salient. Table 5 shows a comparison between the respective word orders of conveying information and highlight. We can see that the highlighted constituent occupies the first position and the other constituents do not change their word order. In the data highlighted constituents receive primary stress. Similarly with contrast, if the highlighted constituent already occupies the first position when the utterance conveys information, the word order remains the same. In constructions with a prev adverb and a low-topicality direct object, if both constituents are highlighted, both receive primary stress and appear preverbally. The prev adverb appears in first position and the DO in second position.

4.2.3. Deviation from expectation

An utterance expresses the pragmatic function deviation from expectation when it conveys a message not expected, given the previous discourse, knowledge of the world or the culture, knowledge of other discourse participants, etc. This pragmatic function includes two related notions (i) unexpected or surprising news, and (ii) denial of a previous assumption. In the first case, what the utterance communicates is not expected. In the second case, something in the previous discourse has led the speaker to believe that the hearer has a false assumption. Therefore, the speaker tries to clarify the situation (Ocampo 1995a). Primary stress falls on the focus of the surprising news or the clarification. For example, we can see in (10) that a construction with an aspectual adverb and a low-topicality direct object has a word order ADVaspectual V DO when the utterance has the pragmatic function of conveying information. When this same type of construction has the additional pragmatic function of deviation from expectation its word order is V DO ADVaspectual, as in (27): (27) S: ¿una batata? a sweet.potato ‘A sweet potato?’ A: te”ngo u’na, todaví^a 2a16 I.have one still ‘I still have one.’ This oral exchange happens during dinner. Speaker S, believing that speaker A has finished her roasted sweet potatoes, offers her another one. Speaker A denies the

355

356

Francisco Ocampo Table 6. Comparison between informational word order and deviation from expectation Convey information

Deviation from expectation

ADVprev V DO ADVprev DOtop V V DO ADVpost

V DO ADVprev DOtop V ADVprev ADVpost V DO

previous assumption of S, stating that she still has one sweet potato. Notice that primary stress falls on the focus of the clarification, the pronoun una ‘one’. Table 6 shows a comparison between word orders in constructions with the pragmatic functions of conveying information and deviation from expectation. In this table we see that in constructions conveying deviation from expectation the adverb occupies an unexpected position: the prev adverb appears postverbally and the post adverb appears preverbally. The word order of the other constituents does not change. The resulting word order could be an instance of iconicity: the unexpected position of the adverb parallels the unexpected information conveyed.

4.2.4. Hypothesis/deduction

The expression of a hypothesis or a deduction by a construction correlates with a variation in its informational word order, as exemplified in (28): (28) V: Tonces eran, como las once y pico de la noche, so they.were as the eleven and something of the night ‘So it was like past eleven pm.’ A: no tenían a’uto, todaví”a 19a4 no you.had car still ‘you still didn’t have a car.’ V: no t — sin auto, llamamos un taxi no we.h without car we.called a taxi ‘We didn’t h — without a car, we called a taxi’ This is part of an oral narrative where speaker V, who is pregnant, is telling about a night she and her husband rushed to the hospital mistakenly believing that their child was about to be born. Speaker A knows that V’s mother recently lent her a car, but she is not sure if V and her husband had the car that night. This is why her statement is hypothetical. Primary stress falls on the focus. Her assumption is confirmed by speaker V. Table 2A, column A, shows five cases of this pragmatic function, all for the same construction-type. The word order is V DO ADVprev , and the corresponding informational word order is ADVprev V DO. We see that the prev adverb occupies a postverbal position for this construction-type when it conveys hypothesis or deduction. Primary stress falls on the focus, which in the examples of the data is the direct object or the verb.

Word-order variation in spoken Spanish

4.2.5. Multiple factors

Table 2B, column B shows one token of a case where more than one factor is at play. The construction is the following: (29) S: es helado: este, prefabricado, eh? no: — no es helado is ice cream hum concentrated ha? no no is ice cream ‘It is artificial ice cream ha? It is not ice cream.’ A: ¿es el — que — se hace? (. . .) ¿Royal, pero de — vos ¿qué le is the.one that is made but from you what to.it agregás? you.add ‘is it the one — that — you prepare (. . .) Royal but from — what do you add to it?’ S: (. . .) le’che nomá^s le po”ngo 2b17 milk only to.it I.put ‘I add only milk.’ In Table 2 this construction-type has a V DO ADVls informational word order. Here the word order is DO ADVls V. As there are various factors that correlate with this word order and there is only one token, the explanation is ad hoc. The direct object receives primary stress and is perceived as salient. This plus the context triggers an interpretation of deviation from expectation: the ice cream is not prepared by adding many ingredients but only milk. In the absence of data, I hypothesize intuitively that the same message of deviation from expectation could have been conveyed if the adverb were in postverbal position and the DO highlighted: (30) leche le pongo nomás milk to.it I.put only ‘I add only milk.’ The difference with (30) is that in (29) the adverb clusters with the direct object reinforcing the information that milk is the only ingredient. Notice that the adverb receives tertiary stress, which facilitates clustering. The explanation for this construction is only a supposition, since it is not possible to find out at this time if these factors are really the ones that influence word order. As shown by Table 1, the correlations listed are supported by the majority of the data, with only 3% (7/275) unclear examples. These seven tokens belong to three categories. In one case there is not enough context in the data to explain the word order. In three cases, the construction conveys a pragmatic function but I could not ascertain what this function is. Finally, there are three cases that contradict the correlations stated in Section 3.

357

358

Francisco Ocampo

5. Conclusions I have tried to demonstrate that the word order of constructions with a verb, a direct object and an adverb correlates with cognitive and syntactic factors when the pragmatic function of the utterance is to convey information only. These word orders can be considered unmarked. When the utterance, besides conveying information, carries an additional pragmatic function, there is a variation of the corresponding informational word order. This issue of word order illustrates the conjoined performance of two factors, one language internal and the other language external. DuBois (1985), with the labels internal linguistics and external linguistics, has mentioned this same conjoined performance. As stated above, I have shown the existence of a correlation between word order and other factors. However, this correlation — although it advances our knowledge of the subject — by itself does not constitute an explanation in terms of a sign-oriented approach to language analysis: ‘The explanation for the distribution of forms consists in identifying the signs that these forms represent, establishing their meanings, and demonstrating the link between the meanings and the messages being communicated’ (Contini-Morava 1995: 4). Therefore, in order to establish an explanation for Spanish we have to find out if it is possible to identify forms and meanings in the elements analyzed in the previous sections. As there is no requirement in linguistic sign theory that the sign should be a piece of morphology, word order can be a candidate for a signifier (Contini-Morava 1995; Diver 1995; Gvozdanović 1995). Now the task is to identify a particular meaning for each word order so that it reaches the category of sign. Possible candidates are the pragmatic functions postulated above. As I have shown, there is a systematic correlation between word order and pragmatic function. Consequently, we could propose a sign formed by word order as signifier and pragmatic function as meaning. Unfortunately, the issue is not that simple. If we look at Tables 2A and 2B we can see that the same word order may correlate with more than one pragmatic function. For example, V DO ADVprev correlates with the pragmatic functions of deviation from expectation and hypothesis/deduction. As this is not an isolated event, it undermines the tendency towards one form one meaning. What happens is that pragmatic functions are not meanings but messages: they are inferences obtained from word order, stress, perhaps tonal contour, and contextual factors. As I do not have other candidates for meanings, it is not possible to solve the question at this time and further research is needed.

Notes * I would like to thank the participants of the 6th International Columbia School Conference for their comments. Special thanks to the anonymous reviewer of the manuscript. His/ her useful comments made me rethink crucial parts of my work. In any case, all errors that remain are mine.

Word-order variation in spoken Spanish 1. This category excludes stressed personal pronouns (yo, tú, él, etc.), and unstressed personal pronouns (clitics lo, la le, etc.). Nouns and the rest of the pronouns are included. 2. The numbers and letters after each example indicate its location in the corpus. 3. Of course, the possibility remains that further research might reveal differences in use between these examples, but I have to work with the knowledge available at the present time. 4. I do not intend to deny with this choice of label the pragmatic connection between topicality and context. 5. One exception to this is an utterance of the type Juan trabaja muy mucho literally ‘John works very a lot’ that I have occasionally heard from uneducated rioplatense speakers. 6. I have included in the category this adverbial phrase because it has lexicalized and behaves like a simple adverb. Notice that this is also the case with the first adverb listed nomás. 7. The central meaning of ya is that of a discrete deictic marker. This meaning in combination with the context conveys a message of aspect. (A.Ocampo and F. Ocampo 2000). 8. Kovacci (1999) also describes other types of adverbs that have this sentential function, but she lists them under various categories (i.e. external to the dictum, modus adverbs). 9. The categories listed under the heading ‘pragmatic function’ will be explained and illustrated below in Section 4. 10. Keep in mind that prev adverbs may appear in postverbal position and post adverbs may appear in preverbal position in marked constructions, as it will be shown in Section 4.2. 11. The conventions used in the transcription are as follows. Utterances appear between parentheses when I am not sure of the accuracy of the transcription. Three periods between parentheses ‘(. . .)’ mean that it was impossible to ascertain what the speaker said. Three periods between square brackets ‘[. . .] ’ indicate that data is omitted. A hiatus of any kind in the speech flow in marked by ‘-’. A pause is indicated by parentheses ‘()’. A number between parentheses ‘(1.4)’ shows the length of the pause (in seconds and tenths of seconds). Words or utterances underlined mark that this portion is perceived as salient (strong primary stress, or high pitch, or uttered with a louder voice, etc.). Lengthening is indicated by‘:’ after a vowel or a consonant. The symbol ’ stands for primary stress,” stands for secondary stress, ^ for tertiary stress. Any additional information appears between square brackets ‘[risas]’h. The construction relevant for the analysis is shown in bold characters. 12. Although this is a personal pronoun, the speaker uses it as an illustration and it functions as a lexical NP. 13. Other constituents in the same construction may also receive primary stress, but the contrastive constituent will always receive primary stress.

References Barrenechea, Ana María. 1977. “Operadores pragmáticos de actitud oracional: los adverbios en -mente.” In Estudios sobre el español hablado en las principales ciudades de América, Juan M. Lope Blanch (ed.), 313–332. México: UNAM. Chafe, Wallace. 1976. “Givenness, contrastiveness, definiteness, subjects, topics and point of view.” In Subject and Topic, Charles Li (ed.), 26–55. New York: Academic Press. Comrie, Bernard. 1981. Language Universals and Linguistic Typology. Chicago: University of Chicago Press.

359

360 Francisco Ocampo Contini-Morava, Ellen. 1995. “Introduction: On linguistic sign theory.” In Contini-Morava and Goldberg (eds.), 1–39. Contini-Morava, Ellen, and Barbara Sussman Goldberg (eds). 1995. Meaning as Explanation. Advances in Linguistic Sign Theory. Berlin: Mouton de Gruyter. Diver, William. 1995. “Theory.” In Contini-Morava and Goldberg (eds.), 43–114. DuBois, John. 1985. “Competing motivations.” In Iconicity in Syntax, John Hyman (ed.), 343– 365. Amsterdam/Philadelphia: John Benjamins. Gernsbacher, Morton Ann and David Hargreaves. 1992. “The privilege of primacy. Experimental data and cognitive explanations.” In Pragmatics of Word Order Flexibility, Doris Payne (ed.), 83–116. Amsterdam/Philadelphia: John Benjamins. Gvozdanović, Jadranka. 1995. “Remarks on sign-oriented approaches to language analysis.” In Contini-Morava and Goldberg (eds.), 169–184. Ifantidou-Trouki, Elly. 1993. “Sentential adverbs and relevance.” Lingua 90: 69–90. North Holland. Kovacci, Ofelia. 1999. “El Adverbio.” In Gramática descriptiva de la lengua Española, Ignacio Bosque and Violeta Demonte (eds.), 705–786. Madrid: Espasa Calpe. Lambrecht, Knud. 1988. “Presentational cleft constructions in spoken French.” In Clause Combining in Grammar and Discourse, John Hyman and Sandra Thompson (eds.), 135– 179. Amsterdam/Philadelphia: John Benjamins. Lambrecht, Knud. 1994. Information Structure and Sentence Form: Topic, Focus, and the Mental Representation of Discourse Referents. Cambridge: CUP. Ocampo, Alicia. 1999. “Variación en la posición preverbal/postverbal en la categoría adverbio: Una interpretación semántica.” Paper presented at the XII Congreso Internacional de la Asociación de Lingüística y Filología de la América Latina, Santiago de Chile, Chile. Ocampo, Francisco. 1995a. “The word order of two-constituent constructions in spoken Spanish.” In Word Order in Discourse, Pamela Downing and Michael Noonan (eds.), 425–447. Amsterdam/Philadelphia: John Benjamins. Ocampo, Francisco. 1995b. “Pragmatic factors in word order: Constructions with a verb and an adverb in spoken Spanish.” Probus 7: 69–88. Ocampo, Francisco. 2001. “Word order variation in constructions with two adverbs in spoken Spanish.” In Adverbial Modification, Reineke Bok-Bennema, Bob de Jonge, Brigitte Kampers-Manhe, and Arie Molendijk (eds.), 13–29. Amsterdam: Rodopi. Ocampo, Francisco. 2003. “The expression of topic in spoken Spanish: An empirical study.” In A Romance Perspective on Language Knowledge and Use, Rafael Núñez-Cedeño, Luis López, and Richard Cameron (eds.), 195–208. Amsterdam/Philadelphia: John Benjamins. Ocampo, Alicia and Francisco Ocampo. 2000. “Un hito en el discurso: significado y mensajes de ya. Evidencia del Español Rioplatense.” Foro Hispánico 17. Estudio Analítico del Discurso, Robert de Jonge (ed.), 83–94. Amsterdam/Atlanta: Rodopi. Prince, Ellen. 1981. “Toward a taxonomy of given-new information.” In Radical Pragmatics, Peter Cole (ed.), 223–255. New York: Academic Press. Silva-Corvalán, Carmen. 1983. “On the interaction of word order and intonation: Some OV constructions in Spanish.” In Discourse Perspectives on Syntax, Flora Klein-Andreu (ed.), 117–140. New York: Academic Press. Silva-Corvalán, Carmen. 1984. “Semantic and pragmatic factors in syntactic change.” In Historical Syntax, J. Fisiak (ed.), 555–573. Berlin: Mouton.

CHAPTER 15

Estrategias discursivas como parámetros para el análisis lingüístico Angelita Martínez Universidad de Buenos Aires

In the distribution of third-person clitic pronouns, Spanish spoken in northwest Argentina exhibits variability not found in standard Argentinian, even though the pronouns’ basic meanings appear to remain constant. (The meanings have been hypothesized to be: le, MORE active participant in event named by verb stem and lo, LESS active participant in event named by verb stem [García 1975; Martínez 1999, 2000].) Speakers of standard Argentinian Spanish use le in three-participant messages and in a limited set of two-participant messages with a few verbs such as ayudar and invitar, whereas substrate speakers of both Guaraní and Quechua use le more freely in two-participant messages. In this chapter, we investigate the motivation underlying the increased frequency of le in Quechua-substrate speakers, which we suspect may arise from the influence of the Quechua nominal suffix -ta, whose use most closely matches that of lo, but on occasion corresponds to le. (For a discussion of Guaraní-influenced Spanish, where motivation of a different sort obtains, see Martínez 1996, 1999.) This chapter relies on two corpora of oral narratives to quantitatively analyze the distribution of these forms in order to support our qualitative predictions about strategies for their use. In 41 stories of comic deception, we have found evidence of discourse strategies that explain how narrators make use of these forms’ meanings to heighten suspense by foreshadowing the unexpected deceit by a central character. Le (MORE active), used to refer to the second participant in a two-participant message rather than the usual lo, alerts the listener that the second participant is about to play a more powerful role than anticipated in the denouement.

1. Introducción Es ya un hecho aceptado que en la “variación sintáctica” lo que se halla en juego es la alternancia de formas no equivalentes (Bolinger 1977:4) que permiten al hablante decir cosas distintas de un mismo referente (García 1985). Esta postura contrasta con la teoría sociolingüística tradicional que centraba la oposición de las variantes en sus valores sociales o estilísticos presuponiendo identidad de valor referencial.

362

Angelita Martínez

La relación entre la “variación sintáctica” así entendida como fenómeno comunicativamente motivado y el género del discurso ha sido examinada por algunos analistas de la lengua en uso (Diver 1987; Tobin 1990; García 1996). Con este trabajo pretendemos dar evidencia de que, dentro de ese campo, los textos en los que se advierte la presencia de “sustrato” lingüístico constituyen una prometedora fuente de análisis. En efecto, si estamos dispuestos a no buscar obstinadamente explicaciones a partir de nuestras propias categorizaciones del mundo y restricciones al universo del discurso (Kerbrat-Orecchioni 1986) se nos abre un iluminador camino hacia cómo otras comunidades categorizan la experiencia, que puede ayudarnos a entender el comportamiento lingüístico de sus hablantes. Específicamente queremos demostrar que el uso variable idiosincrásico de los clíticos lo y le para remitir acusativo en narrativas de la región argentina de “sustrato” quechua, si bien responde a los mismos “significados básicos” comprobados para la variedad rioplatense y otras variedades americanas (García 1975; García y Otheguy 1983), evidencia la expansión de la forma le debido a estrategias particulares de uso con que los hablantes explotan dichos significados. En las narrativas estudiadas, por ejemplo, la forma le constituye una manera de evaluar como “más activo” al personaje en el momento en que, mediante un engaño, provoca el cambio de rumbo en el relato. Con esta estrategia el narrador obtiene el rédito comunicativo de crear suspenso. Es probable que las características de la “lengua de sustrato” tengan influencia en este uso expandido de la forma le. En efecto, la lengua quechua posee un sufijo nominal -ta que se emplea para remitir a lo que en castellano correspondería al acusativo pero, también, en algunos casos, al dativo. Este hecho podría ser el disparador de la expansión aludida en tanto implica una mayor flexibilidad de lo que en la variedad rioplatense se categoriza como acusativo o dativo.

2. Delimitación del problema El uso variable de los pronombres clíticos lo, la y le en variedades de español habladas en la Argentina en zonas de contacto con lenguas aborígenes, ya sea por bilingüismo o por “sustrato”, manifiesta diferencias respecto del rioplatense estándar. Los análisis que hemos llevado a cabo anteriormente (Martínez 1996; 1999) nos han permitido postular que estas variedades poseen el mismo sistema gramatical de base Caso, propio del rioplatense. Comprobamos, además, que no se encuentra el mismo tipo de “desvío” en cada una de las regiones en cuestión; muy por el contrario, los datos muestran que cada zona presenta características de uso peculiares. En algunos casos, sin embargo, la diferencia radica, a primera vista, en la frecuencia de uso de las formas. Por ejemplo, el empleo variable de los pronombres lo(la) vs. le para señalar acusativo en situaciones de dos participantes, tan generalizado en los relatos populares recogidos en la región guaranítica¹, se halla presente también en el noroeste argentino (NOA)² pero en una proporción mucho menor.

Estrategias discursivas

Dicho hecho no es del todo ajeno a la variedad rioplatense estándar, donde son comunes emisiones como: La música le/lo molesta o Juan le/lo ayuda. Pero, mientras en esta variedad los casos de “variación” se hallan restringidos a unos pocos verbos, en el NOA la alternancia se manifiesta con un número mucho mayor de bases verbales y no parece responder a los mismos parámetros que muestran tener influencia (García 1975) en la variedad rioplatense. Según los estudios que se conocen sobre el tema, en algunas zonas, tal como hemos adelantado, se observa un fuerte “sustrato” quechua, especialmente en el norte de la provincia de Salta y en Jujuy donde vive el pueblo colla, y si bien algunos investigadores opinan que allí los argentinos no hablan ya quechua³, otros piensan que dicha lengua se encuentra aún en vías de extinción (Cerrón-Palomino 1987: 71). En trabajos anteriores (Martínez 1999, 2000), hemos comprobado que el uso de los clíticos lo, la y le, en leyendas populares recogidas en Salta y Jujuy, responde a un sistema de Caso en el que, tal como se ha demostrado para el español rioplatense, lo (la) significa lo menos activo del evento mientras que le implica el grado intermedio de actividad en relación con el sujeto agente. En efecto, hemos podido demostrar que, en las leyendas de la región, la selección del clítico responde al significado etimológico de Caso, a partir del análisis de: a. las emisiones que configuran situaciones de tres participantes (Martínez 1999) b. las emisiones que configuran situaciones ambiguas respecto del número de participantes (Martínez 2000) c. las emisiones que configuran situaciones claras de dos participantes (Martínez 2000). Respecto de estas últimas, mostramos que la “variación” de las formas para señalar acusativo puede explicarse si se toman en consideración factores de orden discursivo. Específicamente en las leyendas analizadas, pudimos probar que el uso de le para señalar un acusativo activo se relaciona con el poder de transformación del referente y el episodio que narra la transformación. Por ejemplo: (1) Y entonce dice que un día había llegado un viejito. Dice que lo que voltiaba la gente él lo levantaba. Lo que él sacaba es que lo retaban y se reían de él. Dicen que anduvo todo el día el viejito viendo la riqueza y el orgullo de la gente y ha comenzao a buscar adónde alojarse, y naides quería darle posada. Entonce dice que había ido a la orilla de la ciudá y encontró una casita muy pobre, de un matrimonio pobre y que ahí había encontrado posada. Dice que este matrimonio era tan pobre, que no tenían qué darle de comer esa noche. Lo único que tenian, dice, que era una gallinita. Ésa la mataron y la prepararon y le invitaron. Entonces (el viejito era Dios) les dice: Se va a perder la ciudá. Van a venir temblores para castigar esta gente tan mala y orgullosa. Vamos. Lejos los voy a llevar. (VB, VII, 1382, 217)

363

364 Angelita Martínez

También pudimos comprobar que los referentes señalados mediante le eran entidades que, al transformarse, aumentaban su poder dentro del relato. Por ejemplo, en (1), donde se recrea el antiguo y generalizado mito de Dios disfrazado que pone a prueba a los hombres, Dios se presenta en el pueblo como un mendigo harapiento, solicitando ayuda en tanto que, posteriormente, se revela su identidad superior. Es en ese momento cuando el narrador lo señala mediante la forma más activa le. Con el fin de (des)confirmar nuestros resultados, procedimos, en esta oportunidad, al análisis de otro tipo de relatos, los cuentos, en los que también se observa la “variación” de los clíticos lo(la)/le, como, por ejemplo, en el siguiente episodio:⁴ (2)

Y (el burro) si ha ido y si acostau medio cerca nomah de la casa y áhi s’hecho el muerto. Y salen los Juancitoh chicoh y lo ven al burrito echau áhi cerca’la casa y le avisa despueh el zorrito chico al viejo. —Papá, mamita —dice—, fijesé áhi tenimo un lindo asau. Ahí’ta un burrito gordo, ‘ta gordo! —Eh! —dice— .¿Cierto hijito? —Sí! Salga papá, mire, mire, mire, aquí ‘stá! Sale el Juan y la Juana —Eh! Qui hermoso! Ahura si tenimoh asau —dice—. Este lo vamo a tené que tráir ¿Saben hijoh? Vamu a sacar todo loh lazoh pa traerlo dando vuelta, dando vuelta. Ustedeh van a tirar de una mano; voh de la patita, voh de otra patita; yo vi’a tirar de medio cuerpo y tu mamá que tire del cogote. Lo ramiemo, lo trímo aquí cerquita’la casa porque aquí ya vamo a estaquiarlo y vamos comer asau. Bueno, dice: —Traigan todoh loh lazoh, hijitoh, traigan todoh loh lazoh. Y han sacu loh lazoh hermosoh lazoh!, cimbau, piolah. Le han llenau en las manoh, en el cogote, en lah patah, medio cuerpo. Lo han llenau de lazoh. Meta tirá! Y le han dau vuelta para un lau, pa’l otro lau. Y’staba muerto, el burro, el burro, sin resollar, nada. Bien, bien. La vista apagada, lah orejah, todo, cáidah. En fin, el hecho que ‘staba muerto. Y l’han llevau y l’han llevau, así dando vuelta l’han llevau, hasta que han llegau cerquita la casa. Ya han ‘tau cerquita. —Bueno, ahí noah lo vamoh a dejar hijoh. —Vamoh a preparar loh cuchilloh —dici— pa pelalo y pa’asalo, pa que comamo asau. —Bueno, papá, bueno. Y apurau el hijo, todo y el viejo tamien apurau ya preparando elloh todo. Y lo han pelau y lo han estaquiau ya pa’sacá lo mejor pa’l asau. Ya preparando elluh todo ya. Y en eso que salen con cuchilloh, ya’taban meta afilá los cuchilloh,

Estrategias discursivas

pa’degollale así sacá lah parteh buenah. . .Que ha sentío el burrito. —(Ya, bueno, tan meta afilá cuchilloh. . .) Se levanta ‘l burrito, empieza a pegá una rebuznada: —Haah-juh, haah-juh-hahy. Y sale el burrito, por la playa se va, se escapa al galope. . . (FC, 28, 61)⁵ Tal como podemos ver, en este cuento la alternancia de los clíticos se halla presente. Ciertamente, en el episodio considerado, en emisiones de dos participantes, en las que el burro constituye el participante menos activo del evento, éste es señalado alternativamente mediante las formas lo y le. Ahora bien, puesto que en los cuentos las transformaciones casi no se manifiestan, creemos, fieles a los principios teóricos en los que hemos inscripto nuestro trabajo,⁶ que hallaremos algún otro factor, que, en la misma dirección cognitiva, favorezca el empleo de le para referir acusativos. Nuestra hipótesis, en este caso, es la siguiente: El uso variable de lo (la) vs. le para remitir acusativo, observado en los cuentos populares del NOA, no responde a las características del “leísmo” peninsular, sino a los valores de Caso de las formas. A su vez, los factores que influyen en dicho uso no son los mismos que los que se manifiestan en la variedad rioplatense. Son factores que, tal como hemos demostrado para el género leyenda, se relacionan con el devenir del relato y responden a características culturales propias de la comunidad. En este trabajo trataremos de demostrar que, en los cuentos del NOA, el uso acusativo del pronombre le en situaciones de dos participantes, está presente en las partes del relato en las que hay suspenso. Consideramos situación de suspenso aquella en que es posible que una entidad se comporte de manera no esperada y cambie el curso de los acontecimientos en el relato. Predecimos que la selección de le, debido a su significado de +actividad, se verá favorecida cuando el hablante señale tal tipo de entidades, que, dado su comportamiento, pueden ser conceptualizadas como referentes que cobran poder y por lo tanto, más activas.

3. El corpus Para el análisis del empleo de las formas en “variación” hemos utilizado las dos colecciones de cuentos populares que mencionamos a continuación: a. Fleming de Cornejo, Margarita, Relatos populares salteños (FC) b. Vidal de Battini, Berta, Cuentos y leyendas populares de la Argentina (VB)⁷ En ambos corpora, siguiendo nuestras pautas metodológicas, hemos trabajado con todos los cuentos de narradores que presentan, al menos una vez, el uso variable de los clíticos.⁸ Consideramos para nuestro análisis un total de 41 cuentos. Para intentar una explicación de este uso variable, redujimos nuestros corpora a las emisiones que configuran situaciones claras de dos participantes⁹ y que poseen verbos con los que se observa la “variación”, al menos una vez.

365

366 Angelita Martínez

4. El análisis de los datos Para proceder al análisis de estas emisiones hemos tenido en cuenta, en primer lugar, la frecuencia relativa del empleo variable de los clíticos en relación con los parámetros que tradicionalmente¹⁰ se han manifestado influyentes en distintas variedades del español; es decir: • • • •

El género del referente La índole —animado-no animado— del referente La índole —animado-no animado— del sujeto La índole —más o menos activa— de la base verbal

4.1. El género del referente Para poder probar nuestra hipótesis lo primero que debemos hacer es descartar el hecho de que la “variación” observada en el NOA corresponda a uno de los mentados “focos de “leísmo” en América” (Guitarte 1958:388); es decir, probar si en nuestro corpus el género —masculino o femenino— del referente es, como ha sido demostrado para Castilla (Klein-Andreu 1981; García 1986), el factor relevante en la selección del clítico y si los referentes animados masculinos favorecen el empleo de la forma le. Para ello medimos la frecuencia relativa de empleo de las formas en relación con el género masculino o femenino de los referentes animados. Si se tratase de un “foco de “leísmo” en América” a la manera castellana, deberíamos predecir que un referente masculino será conceptualizado más activo y por lo tanto más merecedor de ser señalado mediante la forma le.¹¹ Los resultados se presentan en la Tabla 1. Los porcentajes que se leen en la Tabla 1¹² nos indican que el género del referente no es el factor decisivo en la elección del clítico. Entidades masculinas y femeninas son señaladas mediante las formas le y lo con similar frecuencia. Si a esto añadimos la ausencia de laísmo en las emisiones que implican tres participantes, con referente femenino, como por ejemplo: (3) Y (el rey) la ve a la muchacha y se va y le da un abrazo (VB, IV, 872, 243) podemos concluir que el uso de le para referir acusativos es, al menos en la narrativa del NOA, diferente del que se ha impuesto en Castilla. Tabla 1. Frecuencia relativa de lo(la) vs. le para referentes acusativos en situaciones de dos participantes en relación con el género del referente Le Masculino Femenino OR = 1.53; χ2 = ins.

30 7

Lo(la) 16% 11%

155 55

84% 89%

Estrategias discursivas

4.2. La índole — animada–no animada — del referente En efecto, si, como estamos argumentando, lo que está en juego en la opción del hablante es el grado de actividad del referente, debemos probar si, al igual que en el Río de la Plata, un referente animado — dada su actividad intrínseca — favorece el empleo de le respecto de un referente no animado. Por ejemplo: (4) (El cóndor) se comprometió a llevarle (a la mujer) con la condición que cada vez que se pare, le dé una oveja. (VB, V, 996,126) En consecuencia, en la Tabla 2 medimos la frecuencia de uso de los clíticos teniendo en cuenta si el referente es o no animado y, de acuerdo con los significados de las formas, esperamos mayor frecuencia del clítico le para referir entidades animadas, por ser éstas intrínsecamente más activas que los objetos. Si bien los datos no son estadísticamente significativos, los resultados de la Tabla 2 muestran que, contrariamente con lo previsto, se observa un porcentaje relativo algo mayor de uso de le cuando los referentes son inanimados. El número negativo obtenido mediante el odds ratio revela que se contradice la hipótesis. Tabla 2. Frecuencia de uso de le vs. lo(la) de acuerdo con la índole animada–no animada del referente Le Referente animado Referente no animado

37 6

Lo(la) 15% 18%

210 27

85% 82%

OR = 0.79; χ2 = ins.

4.3. La índole — animada–no animada — del sujeto Si el sujeto controla al objeto, esperamos el uso de lo, de lo contrario, la forma favorecida será le. Sería de esperar, entonces, que los sujetos intrínsecamente menos activos, como por ejemplo, entidades no animadas, favorecieran la presencia de la forma le. En los corpora que estamos analizando, sin embargo, tal predicción no puede comprobarse dado que todos los sujetos son humanos o animales personificados.

4.4. La índole — más o menos activa — de la base verbal Dado que la “variación” observada se hace presente sólo con algunos verbos, la consideración del tipo de base lexical del verbo se hace necesaria. En efecto, el tipo de verbo que el hablante emplea en la emisión puede presuponer un objeto más o menos dependiente en el evento, es decir, que en las construcciones transitivas, según sea el carácter semántico del verbo, los objetos pueden intervenir de manera más o menos autónoma en la acción.¹³

367

368 Angelita Martínez

En efecto, el análisis de nuestros datos pone en evidencia que los verbos que implican un objeto independiente, tales como ayudar y (per)seguir —se infiere cierta acción de quien es ayudado o perseguido— presentan porcentajes relativamente altos de uso de le. Los porcentajes más bajos corresponden, congruentemente, a los verbos de los que se infieren objetos más afectados tales como matar y llevar. Predecimos entonces que las bases verbales que presuponen la actuación del objeto favorecerán el señalamiento del referente mediante la forma más activa le. En la Tabla 3 probamos dicha predicción midiendo el uso de los clíticos le vs. lo(la) para referir acusativos en situaciones claras de dos participantes según la índole de la base lexical del verbo, contrastando las bases lexicales ayudar y (per)seguir como activas frente a las que hemos considerado como verbos de objeto afectado: tocar, medir, echar, colgar, bajar, tantear, degollar, pechar, traer, sacar, matar, llevar, dejar, punzar, morder, atar, salvar y encontrar. Los verbos envidiar, sentir, ver, y buscar han sido considerados neutrales respecto del grado de afectación del objeto y, por lo tanto, han quedado fuera del análisis cuantitativo. Los resultados muestran que este factor es congruente con nuestra predicción basada en el significado de las formas. No explica, sin embargo, la elección del clítico. En efecto, con verbos activos la selección de le o lo se manifiesta equilibradamente mientras que, en un porcentaje considerable, en las emisiones con bases verbales que hemos considerado pasivas, se ha empleado la forma le. Inferimos entonces, a la luz de los resultados obtenidos, que, en el NOA, al menos dentro del mundo del cuento — así como en el de las leyendas — los parámetros puestos a prueba no influyen decisivamente en la selección de la forma en el contexto. Por el contrario, tal como puede observarse en la Tabla 2, el factor referencial contradice nuestra predicción. ¿Qué es entonces lo que motiva la “variación”? ¿Qué hace que estos hablantes señalen el mismo referente acusativo algunas veces mediante lo y otras mediante le? Puesto que los parámetros que tradicionalmente han mostrado influencia en la selección de los clíticos átonos, en distintas variedades del español, no son decisivos en la variedad del NOA, es de esperar que haya otras motivaciones que induzcan al hablante a optar por el uso variable observado. Intentaremos probar que tales motivaciones son congruentes con los significados postulados y están relacionadas con las características estructurales del tipo de relato que se manifiesta entre los hablantes de la comunidad del NOA. Tabla 3. Uso de le vs. lo(la) para referir acusativos en situaciones claras de dos participantes según la índole de la base lexical del verbo Le +Base activa −Base activa OR = 6.36; χ2 = 13.52; p < .001

6 30

Lo(la) 55% 16%

5 159

45% 84%

Estrategias discursivas 369

4.5. Parámetros discursivos: el devenir del relato En efecto, detengámonos a pensar en la naturaleza de los cuentos que se narran. Se trata de relatos en los que por regla general la resolución de la trama surge de la trampa, del engaño y la astucia de algún personaje. Hemos observado que, cuando se narra ese engaño, hay alguna entidad que modifica el transcurso de los acontecimientos y cambia el curso del relato. Este hecho se pone de manifiesto, por ejemplo, en la escena que sigue (que forma parte del cuento de ejemplo (2)), dentro del episodio en el que el burro engaña a los zorros para recuperar los lazos que el zorro había robado. (2) . . . —Bueno, ahí nomah lo vamoh a dejar hijoh. ... Y en eso que salen con cuchilloh, ya’taban meta afilá los cuchilloh, pa’degollale así sacá lah parteh buenah. . .Que ha sentío el burrito. (FC, 28, 61)¹⁴ Si observamos el desarrollo del relato podemos ver que el uso de la forma le remite al burro en el momento en que éste va a desengañar a sus enemigos probando que no está muerto y recuperando los lazos robados. O sea, el personaje va a provocar, mediante un suceso inesperado, el cambio de rumbo en los acontecimientos. Por otra parte, es la voz del narrador la que emplea la forma más activa para remitir al burro mientras que éste es señalado con lo cuando es referido por un personaje del cuento. En efecto, según se infiere del relato, la emisión“Bueno, ahí nomah lo vamoh a dejar hijoh” está en boca del zorro, uno de los personajes. Creemos que, en estos episodios, el “juego” de las variantes constituye una manera de adelantar que los sucesos que van a acontecer pueden cambiar el curso del relato. Coherentemente con el significado +activo de le, esa forma va a entrar en el campo de los acusativos para señalar a la entidad que cobra poder puesto que provoca la resolución del cuento por medio del engaño. Si nuestra argumentación es válida, se hace necesario considerar en qué “partes” del relato el narrador hace uso de la forma le para señalar lo que en la variedad rioplatense es un acusativo. Los episodios que narran el engaño del personaje y el consecuente cambio de rumbo de la narración deberían favorecer el empleo de la Tabla 4. Frecuencia relativa de uso de le vs. lo(la) en relación con el tipo de episodio narrado Le Episodio del engaño Otros episodios OR = 5.7; χ2 = 25.71; p < .001

32 11

Lo(la) 29% 7%

80 157

71% 93%

370

Angelita Martínez

forma le. Por ello medimos, a manera de control, en primer lugar, la frecuencia relativa de uso de los clíticos en nuestros corpora, según se encuentren o no en el episodio en el que se relata la burla. Esperamos que el uso de le se vea favorecido cuando se produce dentro del episodio que narra el engaño puesto que es allí donde es comunicativamente efectivo categorizar a una entidad como burladora. Tal como esperábamos, son los episodios que narran la burla aquellos en que se favorece la selección del clítico le.¹⁵ Debemos preguntarnos a continuación qué entidades son las referidas —en los episodios del engaño— mediante el clítico le. Porque de acuerdo con nuestra argumentación, la forma le debería señalar un referente que, por algún motivo, puede ser considerado relativamente más activo. En estos episodios que estamos analizando, el referente más apropiado para que se le asigne un “plus” de actividad, será la entidad que provoca la burla. Predecimos entonces, si nuestra argumentación es correcta, que los burladores favorecerán relativamente la selección de la forma le frente a otras entidades. Por ejemplo, la escena siguiente: (5) [. . .] —¡Socorro! ¡Socorro! ¡Que me muero di hambre! ¡Por favor, socorro! ¡Socorro! Oyó un hombre que pasaba por allí cerca ese clamor. Llegó hasta el lugar y encontró al tigre. Entonces al verlo, el tigre se alegró y le dijo: —¡Ay, por favor! ¡Señor hombre, haceme un favor! ¡Nunca más voy a volver hacer lo que hacía! ¡Perdoname! ¡Perdoname! ¡Haceme un favor! ¡Un favor! Es la última vez que me voy a portar mal. ¡Sacame esta piedra de encima! ¡Sacame esta piedra de encima! —Güeno —dice el hombre. Le sacó la piedra y le dejó libre. Cuando estuvo libre, se abalanzó sobre él y le dijo: —¡Te como! ¡Te como porque tengo hambre! (VB, III, 581,69) Para comprobar si esto es así, medimos la frecuencia de uso de los clíticos de acuerdo con el rol del personaje referido. Vemos que, de acuerdo con la Tabla 5, la selección de le es ampliamente favorecida cuando se trata de señalar a la entidad que, mediante astucia o condiciones especiales, provoca la burla. Si recordamos los valores de la Tabla 2, podemos explicar la frecuencia de la forma le para señalar referentes inanimados. En efecto, tales refeTabla 5. Frecuencia de uso de le vs. lo(la) en relación con el rol del personaje en el episodio del engaño Le Entidad que burla Otras entidades OR = 7.90; χ2 = 12.78; p < .001

29 3

Lo(la) 40% 7%

44 36

60% 93%

Estrategias discursivas

rentes son, en todos los casos, entidades que provocan un desenlace inesperado. Por ejemplo en el relato en que se utiliza un muñeco hecho con cera con el fin de atrapar a un conejo que hace “picardías”: (6) Entonces él (el conejito) ha querido morderle (al muñeco de cera). Le ha mordido y se quedó peor pegado. (VB, III, 670, 298) De todos modos, la frecuencia con que los hablantes seleccionan el clítico lo para referir a todas las entidades, incluso al burlador —en efecto, los datos muestran que los totales para lo son superiores al 50% en ambas celdas— nos lleva a pensar que la necesidad comunicativa que subyace a esta estrategia sólo se produce bajo ciertas condiciones, tal como intentaremos demostrar más adelante. Previamente, y a manera de control, observaremos la frecuencia relativa de uso de le para el burlador fuera de los episodios de burla. Podremos así saber si, como venimos argumentando, el mismo personaje adquiere un estatus especial únicamente cuando se lo categoriza en su papel de burlador, debido al poder que le confiere su manera de actuar. Para ello, en la Tabla 6 se mide la frecuencia de uso relativo de las formas en cuestión para señalar al burlador fuera y dentro del episodio del engaño. El resultado de la tabla permite asegurarnos de que la estrategia de empleo de le, para destacar a un personaje, se lleva a cabo en el episodio del engaño porque es allí donde cumple una función comunicativa relevante: alertar al oyente sobre los sucesos extraordinarios que vendrán. Conviene advertir que la única emisión en la que se emplea le para el burlador fuera del episodio del engaño posee el verbo ayudar, una de las pocas bases léxicas que en español rioplatense presenta uso variable. La actividad que se desprende del objeto (quien recibe ayuda) es innegable y por lo tanto la selección del hablante es coherente con el significado más activo de la forma le: (7) El zorro olvidó todo y le quiso ayudar (al conejo). (VB, III, 670, 300) Corresponde ahora averiguar, dentro de los episodios del engaño, quién refiere a la entidad burladora, es decir quién es el enunciador. Puesto que suponemos que la estrategia que estamos estudiando constituye una evaluación del narrador, predecimos que el uso de le se verá favorecido cuando el enunciador es el narrador. Para probar si es así, medimos la frecuencia de uso de los clíticos que refieren a la entidad burladora de acuerdo con el rol del enunciador. Tabla 6. Frecuencia de uso de le vs. lo(la) cuando el referente es el personaje burlador en relación con su presencia dentro y fuera del episodio del engaño Le Episodio del engaño Otros episodios OR = 7.52; χ2 = 4.44; p < .05

29 1

Lo(la) 40% 8%

44 11

60% 92%

37

372

Angelita Martínez Tabla 7. Frecuencia de uso de le vs. lo(la) cuando refieren al burlador en relación con el rol del enunciador Le Narrador Personaje

26 3

Lo(la) 44% 21%

33 11

56% 79%

OR = 2.88; χ2 = ins.

Los resultados de la Tabla 7 apoyan nuestra predicción. El valor de odds ratio indica que la polarización es evidente si bien el nivel de significatividad no es inferior al 5%. De todos modos, si observamos las tres emisiones en las cuales el enunciador no es el narrador y, a pesar de ello, se selecciona la forma le, comprobamos que el personaje enunciador está tratando de engañar a otro y, justamente, el referente del clítico es el instrumento que provocará la tragedia. Veamos un ejemplo: (8) Pinta guagüita, pinta guagüita, pinta guagüita. Después usté ya va a tantearle al horno. (FC, 64, 30) En el cuento al que pertenece (8), la perdiz induce al zorro a cocinar en el horno a los hijos de éste, haciéndole creer que el calor, en lugar de matar a los zorritos, les otorgará una piel manchada (con “pintas”), a la manera del plumaje de las perdices, cuyos colores el zorro envidiaba. Creemos que, en estos casos, el narrador hace, a través de las palabras del personaje, en el momento de mayor fuerza dramática del cuento, un guiño de complicidad, dirigido al oyente. Ahora bien, si queremos sostener nuestra hipótesis de que la “[variación]” que estamos estudiando constituye, en los relatos del NOA, una estrategia mediante la cual se anuncia al oyente que una entidad va a comportarse en forma no esperada, resulta congruente predecir que esas entidades serán referidas mediante le en la narración de los antecedentes del episodio más que en las consecuencias del mismo, puesto que no tendría sentido anunciar algo que ya se ha consumado. Por ello vamos a medir la frecuencia de uso de los clíticos para referir a la entidad que provoca la burla, en relación con la parte del relato del engaño en la que se encuentra el clítico. Consideraremos antecedentes a la narración de los hechos a partir del momento en que se prepara el engaño y consecuencia a la narración de los hechos inmediatos que se derivan del engaño. Tabla 8. Frecuencia relativa de uso de le vs. lo(la) para señalar al burlador–narrador, de acuerdo con la parte del relato en que aparecen las formas Le Antecedentes Consecuencias OR = 14.28; χ2 = 9.94; p < .01

25 1

Lo(la) 54% 8%

21 12

46% 92%

Estrategias discursivas

En la Tabla 8 consideramos la frecuencia relativa del uso de las formas para señalar al burlador, según la parte del episodio del engaño. La tabla nos muestra que, coherentemente con nuestra predicción, el empleo de le se produce, preferentemente, cuando aun no se ha consumado el engaño. La “estrategia comunicativa” cobra relevancia en tanto “advertencia” al oyente y ésta solo puede ser exitosa si es previa a los sucesos advertidos.

4.6. ¿Desde dónde se puede crear suspenso? Tal como hemos planteado en nuestra hipótesis original, el uso variable de los clíticos le vs. lo(la) para remitir a un acusativo se manifiesta en el NOA como una estrategia discursiva relacionada con el devenir del relato. Debemos preguntarnos ahora a qué se debe que en aproximadamente el 50% de las opciones que deberían favorecer la forma le los hablantes han seleccionado, sin embargo, la forma lo. Creemos que la relevancia de señalar al burlador no es la misma en todos los relatos. En algunos casos, el burlador se presenta como un personaje conocido por sus burlas. Los cuentos recogen tradiciones orales que conviven en la comunidad y algunos personajes constituyen “prototipos”. En los cuentos, a veces, el narrador recuerda esas tradiciones. Por ejemplo: (9) Diz que era sobrino el zorro del tigre. El zorro li ayudaba a carniar al tigre. El tío era muy mezquino. El zorro le tenía que hacer picardías al tigre pa poder conseguir algunas carnecitas. (VB, I, 102, 247) (11) El zorro era sobrino del tigre. El Juan se portó siempre mal con el tío [. . .] El tío le daba castigos, pero el sobrino se burlaba siempre del tío. (VB, I, 106, 259) Pensamos que, en estos casos, el oyente ya está previamente prevenido sobre quién es el burlador y el narrador no necesita recordarlo. Predecimos, entonces, que la selección de le se verá favorecida cuando el burlador no es un personaje conocido como tal. En la Tabla 9 medimos la frecuencia relativa de uso de los clíticos teniendo en cuenta si la entidad que produce el engaño es un personaje reconocido explíciTabla 9. Frecuencia relativa de uso de los clíticos para señalar al burlador, en el episodio del engaño (antecedentes), en relación con la presencia o ausencia de anuncio de la burla en la narración Le Burlador no explicitado Burlador explicitado OR = 60; χ2 = 22.87; p < .001

24 1

Lo(la) 80% 6%

6 15

20% 94%

373

374

Angelita Martínez

tamente — por el narrador — como burlador o no lo es. Tal como esperábamos, cuando la burla no está previamente anunciada, el narrador recurre más frecuentemente a la forma le para señalar al burlador en el episodio del engaño.

4.7. El contexto como desambiguador Como es de esperar, hallamos algunos aparentes contraejemplos. Son aquellos casos en que el narrador selecciona la forma lo a pesar de ser el referente un candidato muy favorecido para la opción de le —se encuentra en los antecedentes o el clímax del engaño y es el burlador. Dichos casos pueden explicarse si atendemos a la situación contextual. Volvamos al ejemplo (2), fragmentos del cual se repiten aquí: (2) Y salen los Juancitoh chicoh y lo ven al burrito echau áhi cerca’la casa y le avisa despueh el zorrito chico al viejo. ... Y han sacu loh lazoh hermosoh lazoh!, cimbau, piolah. Le han llenau en las manoh, en el cogote, en lah patah, medio cuerpo. Lo han llenau de lazoh. ... Y en eso que salen con cuchilloh, ya’taban meta afilá los cuchilloh, pa’degollale así sacá lah parteh buenah. . .Que ha sentío el burrito. (FC, 28, 61) El referente de las formas verlo y degollarle es el mismo: el burro que va a burlarse de los zorros, y ambas formas son emitidas por el narrador dentro del episodio del engaño y antes de las consecuencias del mismo. Sin embargo, la elección de la estrategia de evaluación del referente como capaz de cambiar los acontecimientos es más eficiente, porque provoca mayor fuerza dramática, en el momento en que se aproxima el degüello y la burla del burro. Por otra parte, los zorros ven al burro y creen que está muerto, y el narrador apoya —mediante la selección de la forma lo— la perspectiva de los zorros. En cambio, cuando selecciona le se manifiesta omnisciente y anuncia lo inesperado, reforzando así el suspenso del episodio. Es interesante señalar además que, en el mismo evento comunicativo, el hablante privilegia le en “le han llenado en las manos, en el cogote, en las patas, medio cuerpo” vs lo en “lo han llenado de lazos”. En este caso, creemos que el uso de la forma le se debe a la ambigüedad en lo que concierne al número de participantes del evento. La emisión correspondiente puede ser categorizada como de tres participantes; en efecto, además del sujeto ellos y el objeto el burro, el narrador menciona las manos, el cogote, las patas y medio cuerpo, que constituyen, realmente, lo menos activo del evento. Es natural, entonces, que el hablante seleccione una forma más activa para señalar al burro como participante intermedio en cuanto al grado de actividad.

Estrategias discursivas

4.8. Experimento de control Para cerciorarnos de que la estrategia analizada se lleva a cabo para reforzar el suspenso de un episodio, hemos realizado el siguiente experimento de control: Hemos seleccionado seis cuentos, los hemos transcripto fielmente pero hemos obviado el uso de le. Es decir, para remitir a los acusativos siempre escogimos las formas lo y la. Estos cuentos transcriptos fueron entregados a 14 alumnos que se hallaban cursando el último año del profesorado en Castellano y Literatura en el Instituto del Profesorado Secundario Joaquín V. González, de la ciudad de Buenos Aires. Cada estudiante recibió los seis cuentos junto con la consigna de que los leyera atentamente y señalara los episodios en los que percibía que había suspenso. En 72 de los 84 relatos (85.7%) devueltos por los estudiantes hay coincidencia entre el episodio marcado por los estudiantes y el episodio que en los cuentos originales presenta el uso de la forma le.Si bien el número de cuentos utilizado para el experimento es pequeño, creemos que los resultados constituyen una prueba que apoya nuestra hipótesis.¹⁶

5. Conclusiones A partir de nuestro análisis creemos haber demostrado que también en los cuentos —como en las leyendas— del NOA el empleo variable de los clíticos de tercera persona lo, la, le responde a parámetros que se relacionan con el Caso y reflejan valores culturales de la comunidad. En efecto, el “significado básico” de las formas —le relativamente más activo que lo— es explotado, entre estos hablantes, como recurso narrativo. El narrador asigna a un referente —señalándolo mediante le— un “plus” de actividad cuando decide anticipar la intervención inesperada que tendrá impacto en el relato. Logra así satisfacer la necesidad de comunicar, mediante la evaluación del referente, que algo inesperado va a suceder — porque el participante referido es más poderoso que lo que se supone en el devenir del cuento— y crear suspenso. Esta estrategia es cognitivamente equivalente a la observada en el análisis de las leyendas recogidas en la misma región, al que nos hemos referido con anterioridad, en relación con entidades que cobran poder cuando se transforman y son relevantes para la resolución del conflicto. Anunciar lo inesperado, es, tradicionalmente, en la literatura, propio de distintos géneros literarios y no es ajeno al género narrativo pero, de hecho, el ingenio humano crea distintas estrategias para resolver el problema. El uso variable de los clíticos, que hemos tratado de explicar en este trabajo, indicaría que las restricciones al universo del discurso son relativas a los valores de la comunidad que las crea. Por otra parte, la creatividad que implica este empleo variable justificaría la escasa presencia de narradores que la utilizan. Puesto que serán aquellos que poseen mayor sensibilidad los más aptos para integrar el significado aprendido de las formas coherentemente en un contexto y derivar un mensaje que es sutil y complejo.

375

376

Angelita Martínez

Para concluir, creemos necesario referirnos a las implicaciones teóricas que trae aparejado este tipo de análisis: los aparentes desvíos de la formas lingüísticas respecto del uso estándar pueden responder a una categorización sistemática desde una perspectiva diferente que deriva de explotaciones inéditas que satisfacen nuevas necesidades comunicativas. Esto significa que, a pesar de los estudios formalizantes que han ocupado el campo de la lingüística de las últimas décadas, la sintaxis, debido a su motivación semántica, parece ser el dominio más creativo de la producción lingüística.

Fuentes de los ejemplos lingüísticos Fleming de Cornejo, M. 1988. Relatos folklóricos salteños. Recopilación de versiones del patrimonio tradicional de Salta. Salta. Vidal de Battini, Berta, 1984. Cuentos y Leyendas populares de la Argentina. Buenos Aires, Ediciones Culturales Argentinas, Secretaría de Cultura, Ministerio de Educación y Justicia.

Notes 1. La región guaranítica argentina, geográficamente fronteriza al Paraguay, comprende las provincias de Corrientes, Misiones, este de Chaco y Formosa, la zona extrema del nordeste de Santa Fe y una proyección en Entre Ríos. 2. Rojas (1984: 184–187) estudia el uso de los pronombres en el NOA a partir de grabaciones a hablantes nativos o de larga residencia en la región. Respecto de los clíticos lo, la y le señala discordancias de género y de número entre el pronombre y su referente y menciona que el empleo de le para la función acusativo se registra de manera reducida. Sobre un total de 780 casos de objeto directo con pronombres átonos, sólo en un 2% de los mismos la autora observa la forma le. Nuestros datos concuerdan, tal como se manifiesta en este artículo, respecto de la baja frecuencia de empleo de le para remitir acusativos. 3. Según Martorell de Laconi (1998:121), por ejemplo, en 1993 la Asociación Indígena de la República Argentina (AIRA) estimó en 72.000 los descendientes de quechuas y aymarás (collas). Estos no hablan el quechua, sino un español con muchas transferencias. Agrega la autora que los “quichuhablantes en el NOA no son nativos sino procedentes de Bolivia y Perú, en recientes inmigraciones, asentados en zonas marginales urbanas de las principales ciudades de la región”. 4. Los relatos se hallan organizados en episodios que constituyen trechos de discurso conectados en los que hay un tema y un referente que surge como central. Hemos considerado episodio del engaño a la parte del relato en que se cuenta la historia del engaño. Abarca los antecedentes al momento del engaño, el engaño propiamente dicho y las consecuencias del mismo. En general se halla encabezado por una emisión introductoria, tal como podemos ver en el ejemplo (2). 5. En los ejemplos correspondientes a FC se ha establecido un número (correlativo) para cada cuento y se indica el número de página. 6. El enfoque de este trabajo se basa en el análisis “microsintáctico”, firmemente enraizado en los presupuestos pragmáticos de toda habla (García 1995:70); toma en consideración el

Estrategias discursivas estudio de la lengua en uso como medio de comunicación entre los seres humanos y la interpretación del significado en el contexto, interesándose por cómo afecta la función comunicativa a la gramática de las lenguas (Reid 1995:115–6). Se nutre de los hallazgos de la Pragmática en lo que se refiere a los procesos comunicativos (Sperber y Wilson 1986), generando la búsqueda de parámetros lingüísticos y extralingüísticos, de orden pragmático y discursivo, para explicar la selección lingüística. 7. De los seis volúmenes de VB consultados, nos ocupamos de los cuentos que corresponden a las provincias de Salta y Jujuy porque, como ya hemos adelantado, son las zonas en donde se observa, a pesar de la baja frecuencia, mayor cantidad de usos variables. 8. Fieles a los principios metodológicos que guían nuestro trabajo, sólo hemos tenido en cuenta para nuestro análisis los relatos de los narradores que nos garantizan el empleo variable de los clíticos. Para ello hemos seleccionado, en primer término, los narradores que presentan “variación” al menos en un relato, para luego considerar todos los relatos correspondientes a esos hablantes. 9. Las situaciones ambiguas respecto del número de participantes —emisiones de acusativo con infinitivo (ACI) del tipo Juan le/lo hizo comer (carne) y emisiones con verbo llamar con y sin atributo del tipo A la niña le/la llamaban (Belleza del mundo)— no han sido consideradas dentro de los datos que estamos analizando. Si bien estas emisiones son infrecuentes en el corpus, en ellas, el uso variable de los clíticos responde a la posibilidad de conceptualizar la escena como de dos o de tres participantes, tal como corresponde a las variedades en que los factores que determinan la selección del clítico se relacionan con la sustancia del caso. En efecto, el uso de la forma le se ve favorecido cuando en las construcciones ACI se halla presente el objeto del infinitivo y el atributo en las emisiones con el verbo llamar. 10. Me refiero, entre otros, a los análisis de Hurst 1951; García 1975; Klein-Andreu 1981; García y Otheguy 1983; García 1990; Brânza 1999; Fernández Ordóñez 1999; Mauder 2000. 11. Hemos considerado los casos en que el referente es una entidad animada puesto que para el caso de los referentes inanimados el género es sólo gramatical y no se corresponde, al menos directamente, con un contenido semántico. 12. Para lo que se refiere a pruebas estadísticas utilizadas en este análisis, hemos seguido a Butler 1985. 13. Hay acuerdo entre casi todos los lingüistas en que, semánticamente, el grado de transitividad es un continuo, pero no se han establecido criterios claros que fijen límites entre verbos más o menos transitivos (cf., por ejemplo, la extensa clasificación semántica de Cano Aguilar 1987). Esto se debe a que la transitividad es un fenómeno complejo en el que se interrelacionan diferentes factores. Hurst (1951:75) alude a “the dynamic quality of the verb” como una característica poco “objetivable” pero necesaria en el análisis de ciertos usos pronominales. 14. Los clíticos que se señalan en itálica son aquellos que corresponden, tal como hemos adelantado, a verbos en que se manifiesta, al menos una vez, el uso variable. 15. Es interesante señalar que entre los usos de le que hemos categorizado como ‘otros episodios’ se encuentran la mayoría de las emisiones con bases verbales activas y en VB la referencia a una entidad que no engaña pero que se transforma en la resolución del relato: Y la otra se paró y le cayeron los huesos. Y salió una paloma volando, y (la mujer trabajadora) le dijo (a la haragana convertida en paloma) que ella le salvó, que ella se condenó por no haber terminado nunca las obras.(IX, 2413, 336)

377

378

Angelita Martínez 16. Agradezco a la profesora A.Vitale, quien generosamente me permitió el acceso a los estudiantes y suministró las pruebas.

Referencias Bolinger, D. 1977. Meaning and Form. London: Longman. Brânza, Mircea-Doru. 1999. Uso variable de los clíticos de 3º persona en el centro norte de España. Tesis doctoral. Leuven: Katholieke Universiteit. Butler, C. 1985. Statistics in Linguistics. Oxford: Basil Blackwell. Cano Aguilar, R. 1987. Estructuras sintácticas transitivas en el español actual. Gredos. Madrid. Cerrón Palomino, R. 1987. Lingüística Quechua. Cuzco: Centro de Estudios Rurales Andinos “Bartolomé de las Casas.” Company Company, C. 1995.“Cantidad vs. cualidad en el contacto de lenguas.” Nueva revista de Filología Hispánica 43 (2): 305–339. Contini-Morava, E. 1995. “Introduction: On linguistic sign theory.” En Meaning as Explanation: Advances in Linguistic Sign Theory, E. Contini-Morava y B. S. Goldberg (eds), 1–39. Berlin: Mouton de Gruyter. Diver, W. 1987. “The dual.” Columbia University Working papers in Linguistics 8: 100–114. Diver, W. 1995. “Theory.” En Meaning as Explanation: Advances in Linguistic Sign Theory, E. Contini-Morava y B. S. Goldberg (eds), 43–114. Berlin: Mouton de Gruyter. Fernández Ordóñez, I. 1999.“Leísmo, laísmo y loísmo.” En Gramática descriptiva de la lengua española, I. Bosque y V. Demonte (eds), 1317–1397. Madrid: Espasa Calpe. García, E.C. 1975. The role of theory in linguistic analysis: The Spanish pronoun system. Amsterdam: North-Holland. García, E.C. 1985. “Shifting variation.” Lingua 67: 189–224. García, E.C. 1986. “The case of Spanish gender.” Neuphilologische Mitteilungen 87: 165–184. García, E.C. 1990. “Bilingüismo e interferencia sintáctica.” Lexis 14 (2): 159–195. García, E.C. 1995. “Frecuencia (relativa) de uso como síntoma de estrategias etnopragmáticas.” En Lenguas en contacto en Hispanoamérica, Klaus Zimmermann (ed), 51–72. Madrid: Vervuert. Iberoamericana. García, E.C. 1996. “¿Cómo qué ‘qué’?” Hispanic Linguistics 8 (Spring): 59–93. García, E.C. 1997. “La portée de la variabilité.” En La variacion en syntaxe [Langue Française 115], F. Gadet (ed.), 30–47. Paris: Larousse. García, E. C. y R. Otheguy. 1983. “Being polite in Ecuador.” Lingua 61: 103–132. Hurst, D. A. 1951. “Spanish case: Influence of subject and connotation of force.” Hispania 34: 74–79. Kerbrat-Orechioni, C. 1986. La enunciación de la subjetividad en el lenguaje. Buenos Aires: Hachette. Klein-Andreu, F. 1981. “Distintos sistemas de empleo de ‘le’, ‘la’, ‘lo’: Perspectiva sincrónica, diacrónica y sociolingüística.” Thesaurus 36: 1–21. Malinowski, B. 1984 [1944]. Una teoría científica de la cultura. Madrid: Sarpe. Martínez, A. 1995. “Variación lingüística y etnopragmática: Dos caminos paralelos”. En II Jornadas de Lingüística Aborigen, Instituto de Linguística, Universidad de Buenos Aires, 427–437. Martínez, A. 1996. “Lenguaje, pensamiento y cultura: Uso de ‘le’ en la narrativa oral no estándar de Chaco y Formosa.” Hispanic Linguistics 8(1): 94–122.

Estrategias discursivas Martínez, A. 1999. “El hilo se corta por lo más delgado: Variedades dialectales en el uso de los clíticos.” Anuario de Lingüística Hispánica (Universidad de Valladolid) 13: 643–652. Martínez, A. 2000. “Las estrategias discursivas y la estructura de la lengua”. En Foro Hispánico. Estudio analítico del signo lingüístico. Teoría y descripción, 61–82. Amsterdam: Rodopi. Mauder, E. 2000, May. “Frecuencias inesperadas y decisiones desesperadas.” Comunicación presentada en el Congreso de ALFAL, Región Norte de Europa, Groningen. Palacios Alcaine, A. 1999. “Un caso de bilingüismo histórico: Aspectos lingüísticos de la obra de Santacruz Pachacuti.” Anuario de Lingüística Hispánica (Universidad de Valladolid) 12: 397–412. Reid, W. 1995. “Quantitative analysis in Columbia School theory.” En Meaning as Explanation: Advances in Linguistic Sign Theory, E. Contini-Morava y B. S. Goldberg (eds), 115– 152. Berlin: Mouton de Gruyter. Rojas, E. M. 1984.“Sobre las formas pronominales átonas en el habla del NOA.” Cuadernos de Literatura, Universidad Nacional del Nordeste, Chaco: 181–206. Sperber, D. y D. Wilson. 1986. La relevancia. Madrid: Visor. Tobin, Y. 1990. Semiotics and Linguistics. London: Longman.

379

Index of names Abel 290 Alcalay 296 Altman 293 Anderson 149 Andrews 291 Aphek 12, 291, 296 Arnold 329, 339 Aronoff 147, 149, 318 Azim 287 Barlow 233 Barrenechea 344 Bat-El 318 Bateson 202 Beard 149 Behagel 15 Belić 153 Ben-Ezra 290 Bender 262, 272 Beöthy 292, 293 Berman 239, 240, 241, 248, 318 Blau 291 Bok-Bennema 216 Bolinger 12, 193, 216, 361 Bolozkoy 318 Borkovskij 153 Bouma 202 Brânza 377 Bresnan 63 Brewer 152 Brown 200 Brugman 14 Buscha 182, 198 Butcher 266 Butler 377 Bybee 151, 152, 270 Cameron 230 Canals 12 Cano Aguilar 377 Casad 57 Cerrón-Palomino 363 Chafe 182, 268, 342, 350, 353 Chalmers 15 Chomsky 63, 159, 233 Choueka 242, 251, 252 Clark 123, 296 Claudi 270 Company Company 216 Comrie 344

Contini-Morava 1, 2, 8, 13, 15, 45, 124, 150, 152, 201, 216, 287, 291, 317, 358 Cook 63 Cresswell 233 Croft 186 Culicover 63 Davidson 126 Davis 2, 8, 9, 159, 165, 171, 172, 173, 287 Dewell 14 De Jonge 9, 10, 15, 207, 208, 216 Diver 8, 9, 11, 15, 51, 86, 105, 122, 127, 132, 149, 155, 156, 157, 168, 170, 171, 178, 216, 230, 287, 289, 291, 318, 325, 326, 358, 362 DuBois 358 Duden 180, 181, 182, 194, 198, 199, 200, 201 Einhorn 317 Elson 8, 151 Ende 189 Ephratt 318 Ernout 149, 153 Evans 8, 14, 122, 127 Fatihi 318 Fauconnier 4, 57, 63, 64, 65, 66, 71, 86 Feldman 318 Fernández Ordóñez 377 Fillmore 64 Fleming de Cornejo 365 Flores 318 Freud 290 Frost 318 García 8, 13, 86, 159, 160, 161, 164, 165, 169, 171, 172, 216, 296, 361, 362, 363, 366, 376, 377 García Márquez 208, 209, 210, 211, 212, 215, 216, 217 Genzmer 183, 184, 198 Gernsbacher 63, 343 Ginstrom 329 Givón 63, 85, 222, 296 Goffman 64, 202 Goldberg 8, 13, 291 Gordis 291 Gorup 161, 163 Greenberg 62, 63, 72, 85 Grice 95 Grimshaw 171 Guitarte 366

382

Index of names Gvozdanović 358 Haegeman 63 Haiman 4, 78, 85, 265 Halliday 194 Hameed 11, 318 Hargreaves 343 Haverkate 206 Hawkins 329, 339 Heidegger 189 Heine 270 Hentschel 182, 197, 198 Hervey 289, 317 Hooper 205 Hopper 9, 184, 270 Huffman 2, 5, 13, 40, 43, 45, 86, 94, 155, 159, 167, 168, 169, 233, 325 Hünemeyer 270 Hurst 377 Husain 287 Ichihashi-Nakayama 9, 11 Ifantidou-Trouki 344 Jabeen 287, 318 Jackendoff 43, 127 Jacobsen 261, 269, 270 Jakobson 291 Janda 15 Janssen 8, 45 Jing 9, 10 Johnson 8, 76, 104, 105, 106, 124 Joos 194 Junger 238 Kampers-Manhe 216, 217 Kemmer 222, 224, 233 Kendall 261, 267, 268 Kerbrat-Orecchioni 362 Kirsner 2, 7, 8, 9, 13, 14, 86, 116, 117, 118, 119, 120, 121, 122, 166, 170, 172, 178, 191, 201, 202 Klein 205, 206 Klein-Andreu 15, 291, 366, 377 Korpijaakko-Huuhka 318 Kovaćs 296 Kovacci 344, 345, 359 Kronasser 296 Krylov 292, 293 Kumashiro 57 Kuno 233 Kuznecov 153 Lakoff 1, 8, 63, 65, 73, 76, 104, 105, 106, 124 Lambrecht 182, 343, 350, 351 Landau 291 Langacker 1, 2, 3, 4, 5, 7, 9, 15, 28, 30, 33, 38, 43, 52, 57, 58, 63, 65, 72, 76, 86, 105, 106, 116,

123, 127, 183, 190, 201 Langdon 261, 266, 269 Lapesa 212 LaPolla 63 Leech 97, 123 Lehmann 270 Levinson 43, 123, 125, 126, 200, 222 Löbner 202 Losongco 329 Lyons 194 MacWhinney 171 Makkai 296 Mandelblit 242 Mandelbrot 292, 293 Martinet 287 Martínez 13, 361, 362, 363 Martorell de Laconi 376 Mauder 377 Menéndez Pidal 212 Menn 171 Menzerath 292, 293 Mikado 318 Mithun 266, 268 Molendijk 216 Moore 318 Morrison 291 Munro 261, 265, 266, 269 Newson 63 Nieuwenhuijsen 216 Nikiforidou 3, 52 Noeldoeke 291 Nunberg 64 O’Conner 220 Ocampo 2, 4, 12, 13, 342, 343, 344, 345, 348, 353, 355, 359 Oron 9, 10, 15, 239 Otheguy 4, 12, 13, 52, 55, 58, 122, 148, 150, 164, 165, 167, 169, 291, 362, 377 Pagliuca 270 Penny 152 Perkins 270 Perlmutter 63 Pike 296 Popper 9 Prince 351 Pütz 63, 69, 87 Quer 217 Radford 63 Ramsey 214 Ravid 318 Redden 269

Index of names Reddy 123 Reid 5, 6, 7, 8, 9, 13, 14, 15, 43, 45, 49, 122, 127, 148, 150, 156, 157, 158, 159, 160, 161, 164, 165, 166, 167, 169, 171, 172, 173, 184, 216, 233, 291, 377 Reinhart 233 Reuland 233 Robins 149, 153 Rodriguez-Bachiller 12, 123, 287 Rojas 376 Rosch 57 Rosen 63 Rosén 318 Rosenberg-Wolf 318 Rudy 291 Ruhl 5, 14, 93, 113 Sachs 259 Sag 63 Sagi 239 Sampson 14 Sasse 202 Saussure 52, 125, 132, 291 Schmid 70, 71, 72, 87 Schneider 183 Schwarzwald 318 Searle 190 Shermer 13 Shimron 318 Silva-Corvalán 342, 353 Slobin 15, 126 Smith 4, 74, 86 Smyth 13 Spaulding 218 Sperber 377 Stern 9, 10, 13, 148, 150, 172, 233, 243, 251, 252, 259, 291 Sussman Goldberg 8, 13 Sweetser 64, 75

Talmy 25 Tannen 190 Taylor 5, 8, 14, 30, 73, 93 Terell 179 Terrell 205 Tobin 9, 10, 12, 15, 122, 125, 235, 237, 238, 239, 242, 248, 258, 259, 287, 291, 292, 294, 295, 296, 317, 318, 362 Traugott 9, 270, 296 Turner 57 Tyler 8, 14, 122, 127 van Hoek 233 van Putte 296 van Schooneveld 9, 235 Van Valin, Jr. 63 Verhagen 14 Vet 216 Vidal de Battini 365 Walden 318 Wasow 63, 329 Watahomigie 262, 272 Watkins 266, 268 Waugh 291 Webelhuth 63 Weydt 182, 197, 198 Whaley 178 Whorf 106, 124 Wierzbicka 41, 296 Williams 13 Wilson 377 Winter 261, 266 Wolpert 14 Yamamoto 262, 263, 264, 272 Zipf 284, 292, 293

383

Index of subjects accusative 13 abstract settings 76, 86 acoustics 277 activity of participant 164, 169 adjective 26 advantage of first mention 343 adverbial phrase 325, 327, 328 adverbs aspectual 344 immediately post-verbal 343 left scope 343, 344, 352 linking function 345, 347, 348, 349, 350 right scope 344, 348 sentential 344 type 343 agent 251, 253 agglutinative 131, 148, 149 analytical responsibility 3, 8 anaphora 62, 66, 67, 72, 74, 75 antitopic 350 apertures glottal 278, 280, 287 supraglottal 280 appositive 224, 229 arbitrariness 114, 317 articulators 11, 278, 280 adroitness of 11, 281, 282, 283 glottis 278, 280 labium 282 supraglottal 278, 281 tongue, apex 281, 282 tongue, dorsum 282 tongue, medium 282 tongue, medium-dorsum mass 282 tongue, post dorsum 282 artificial data 5 aspect 201, 202, 241, 344 assertion 205, 207 attention 24, 160, 162 background information 184, 185 Binding Theory 233 binyanim 10, 235–260 Catalan 12, 331 cataphora 70, 81, 84 cataphoric es 68, 69, 77, 78, 74, 75, 84 it 61, 68, 81, 82, 83, 84 pronouns 4, 61, 62, 68, 69, 71, 72, 73, 78, 79,

80, 83, 85 categorization 22 causation 252 causative 236 child language 239 clitic pronouns 13, 159, 160, 161, 162, 163, 164 cognitive 327, 328, 332, 335, 339 cognitive-grammatical 342, 343 cognitive-iconic principle 12, 326, 327 effort 351 Cognitive Grammar 1, 2, 3, 4, 6, 8, 9, 15, 105, 116 cognitive linguistics 1, 8, 22, 42, 43, 63, 64 coherence 157, 159, 168 Columbia School linguistics 1, 2, 3, 4, 5, 7, 8, 9, 10, 12, 13, 14, 15, 93, 94, 96, 101, 105, 115, 116, 122, 148, 149, 150, 177, 178, 201 Columbia School phonological theory 287 see also Phonology as Human Behavior communication 101, 277, 278 communicative function 11, 12, 14, 101 instrument 5, 6, 9 load 284, 286 orientation 159, 168 system 101 communicative strategy, see strategy for the communicative exploitation of a meaning complexity 340 complex signs 55 compositional view of meaning 94 conception 242 conceptualization 6, 97, 98, 100, 102, 105, 108, 114, 237 conceptual continuum 113 distance 4, 10, 61, 77, 78, 79, 81, 82 packaging 71 conditional 177, 178, 179, 180, 181, 182, 183, 193, 196, 197, 199, 201 conjunction subordinating 9, 177, 182 temporal 179 connotation 103, 104 constituent 28, 329, 340 construction, see grammatical construction constructional schema 29, 52, 55 construction grammar 58 content requirement 22 context 5, 207

Index of subjects contextual relevance 207 contingency 185 controller 326, 332, 335 Control System 159, 168, 169, 223, 224, 233 degree of Control 12, 326, 328, 329, 330, 331, 332, 337 conventionalization 3, 49, 53, 101 Cora 57 core argument 342 correlation 159, 164, 165, 166, 168, see also quantitative data counterfactual 188, 193 creative usage 101 dative 13 dative versus accusative contrast 13 deixis 15, 160, 161 denotation — connotation 237 derivational morphology 10, 239 desinence 8 dialogue, constructed 190 directional 262, 263, 267, 268, 270, 271 direct object 342 discourse focus 184 frame 188, 189, 190, 201, 202 referent 158, 162 strategies 13 distribution 41 double subject constructions 3, 33 Dutch 86, 166, 170 economy classificatory 294 mnemonic 289 semiological 289 effort, economy of 318, 319 emphasis 80, 81, 219 empirical scope of linguistic analysis 9 encyclopedic knowledge 6 English 4, 5, 10, 12, 14, 51, 52, 62, 68, 70, 77, 79, 80, 81, 82, 86, 139, 147, 148, 149, 150, 155, 167, 170, 172, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 338, 339 English it 74, 75, 84 entrenchment 3, 49, 53 estrategia comunicativa 373 events discontinuity 15 intellectual 83 mental attitude 68 physical 83 process 248, 249, 251, 252, 253, 254, 256 propositional attitude verbs 64, 65, 68 result 248, 249, 252, 253, 254, 256 telicized 240, 258

evidentiality 261, 263, 267, 271 exocentric 36 expectation, deviation from 353, 355, 356, 357, 358 explanatory construct 115, 121 external linguistics 358 extraposition 84 extra information see information, extra falsifiability 9, 15 figure/ground 27 Focus 42, 54, 159, 160, 163, 165, 166, 167, 168 focus of contrast 353, 354 force-dynamic 25 foreground 4, 184 frame, see discourse frame French 161, 162, 163, 168, 351 frequency counts, see quantitative data functional similarities vs. structural differences 3 fusional 131, 145, 147, 148, 149 Generative Grammar 2, 13, 15, 233 German 4, 9, 62, 68, 69, 70, 77, 78, 79, 80, 81, 86, 87 given information, see information, given glottis, see articulators grammar, model of 22, 50 grammaticalization 9, 11, 262, 269, 270, 271, 272, 342, 344 unidirectionality in 270, 271 grammatical class 24 construction 28, 52, 54 markers 24 grammaticization, see grammaticalization Greek, Homeric 13, 127 Gricean Maxim of Quantity 95 Guaraní 35 head 25, 28 heavy vs. light constituents 329 Hebrew 10, 12, 235 260 highlighted constituent 354, 355 homonymy 5 Hualapai 11, 261 human factor in linguistic analysis 230, 231, 277, 292 hypothetical 180, 181, 182, 185, 186, 187, 188, 189, 190, 192, 193, 194, 196, 197, 198, 199, 200 iconicity 4, 15, 114, 235, 239, 242, 317, 318, 350 imagery 72 impersonal 160, 161, 162 indicative 10, 15, 199, 200, 205, 207, 210 Indo-European languages 9

385

386 Index of subjects inferential complexity 163, 172 inferential routines 3 inflection 8, 131, 148, 239 information 343, 344, 345, 347, 348, 349, 351, 352, 353, 354, 355, 356, 357, 358 extra 327, 328, 329, 332, 333, 339 given or old 186, 329, 343, 351 new 13, 329, 343, 351 structure 182 informational word order 345 instructional meaning, see meaning, instructional internal linguistics 358 interpretive abilities 45 intransitive 240, 241, 243, 258 Italian 161, 162, 163, 165, 171, 173 iteration 241, 243, 252, 253, 257, 258, 259 Jakobsonian style of semantic analysis 9 Japanese 33, 57 Kiowa 266 Korean 33 Kronasser’s Law 296 Krylov’s Law of Polysemy or Polylexy 293 landmark 27, 31 language, model of 47, 114 langue 291 Latin 15, 143, 144, 147, 153, 170, 172 leísmo 365, 366 lengua de sustrato 362, 363 Lexical Conceptual Structure 241 lexical item 122, 241, 257 lexical meaning 249 lexicon 239, 317 linguistic system 47, 52, 115, 125 locative 262, 267, 268, 271 Lucknow Urdu 11, 277 288 Luiseño 34, 57 Macedonian 133, 136, 137, 138, 140, 146, 147, 148, 151 Mandarin 33 Mandelbrot’s Law 293 markedness 186, 238, 259 meaning, see also semantic 155, 156, 157, 158, 159, 160, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 325, 327 grammatical, of morphemes 131, 136, 138, 139, 141, 142, 144, 145, 146, 147, 148, 149 instructional 10, 177, 184, 185, 201 instrumental concept of 5, 94 invariant 10, 12, 177, 178, 235, 237, 239, 243, 258, 295 least inappropriate relative to a given message 14

representational view of 94,196 schematic 41, 43, 46 meanings 3, 155, 156, 157, 158, 159, 163, 164, 165, 166, 167, 168, 169 envisaged as rough hints 5, 96 mental space 4, 61, 64, 65, 66, 67, 68, 69, 70, 71, 76, 81, 82, 83, 85 boundaries 82 builder 65, 67, 68, 69, 70, 71, 77, 79, 83 building verbs 68 designator 61, 67, 69, 72, 73, 77, 79, 85 Menzerath’s Law 293 message 155, 156, 157, 158, 161, 162, 163, 165, 168, 169 metaphor 101, 104, 105, 106, 113, 114, 115, 121, 123, 125, 126, 127 active 105 frozen 105 spatialization 78, 79 metonymy 38, 76 modality, see also subjunctive 182, 184, 187, 188, 191, 192, 193, 194, 196, 199, 200, 201, 205 non-assertion 207 modal verb 188, 193, 194, 195, 196, 198, 201 monosemy 5, 6, 45 motivation vs. prediction 11,15, 294, 317 negative data 9 negation in the main (or superordinate) clause 213 negation in main clause 214 in subordinate clause 213 network model 5, 9, 41, 46, 48 Newari 33 new information, see information, new nominalization 32 number 158, 159 object 27, 31 oblique arguments 342, 343 Ockham’s Razor 6, 8 oppositional structures 3, 43, 57 order, see word order paradigm 8, 132, 133, 135, 146 paraphrase relationship 181, 182 parole 292 participle 26 passive 243, 251, 252 patient 251, 253 phonesthemes 12 phonological grid 278, 280, 287 phonological units 277, 287 Phonology as Human Behavior 11, 277, 294, 318, 319 phrase marker 325, 326, 327

Index of subjects physiological factors 11, 277, 278, 283 interaction with communication 286 interaction with vision 285, 286 physiologico-acoustic substance 278 point of view of the speaker 261 Polish 141, 142, 152 politeness 200 polysemy 3, 5, 6, 8, 14, 41, 44, 73, 180 Pomo 266 portmanteau 131, 147, 150 positional signals, see word-order signals possessive 30, 32, 51 pragmatics 43, 342 pragmatic strengthening 270 prediction 15, 157, 158, 166, 168, 170, 171, 172, 326, 328, 329, 330, 331, 332, 333, 334, 339 in linguistics 11,15 prepositions 5 prescriptive rules 219, 220, 230, 231 processing 161, 163, 171, 172, 340 profiling 23 pronouns, see also anaphora, cataphora 10 quantitative data 8, 10, 12, 43, 156, 157, 158, 159, 160, 165, 168, 170, 188, 189, 194, 196, 201 quantitative approach to phonology 11 Quechua 13 raising 38 reciprocity 241, 243 reference point 29, 34, 76 reflexive 160, 161, 162, 163, 219, 221, 222, 229, 236, 240, 243, 249, 253, 258 relevance 208, 210, 215, 211 reported speech 212 representation vs. communication 101 representational view of meaning, see meaning, representational view of representational device 5 function 101 system 9, 14, 94, 101 result, see events, result Romance 8 Romanian 142, 144 Russian 4, 62, 65, 68, 69, 70, 86, 87, 147, 153 scene 169, 170 schematic meaning, see meaning, schematic schematizations 22 semantic, see also meaning bleaching 270 building blocks 5 composition 50, 53 domain 238, 242 field 290, 295, 316, 317, 318 opposition 14

substance 178, 163 Semantic Integrality 238 semiotic level 7 semiotic model 5 Semitic 9 sentence-oriented analysis 241, 243, 239 Serbian 144, 145, 147, 153 Serbo-Croatian 161, 163 shell noun 71, 72 sign, Saussurean 52 sign-based linguistics 1, 2, 5, 13, 239, 242, 291, 295 signal 155, 157, 158, 164, 166, 167, 170, 325, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 338, 339 signal-meaning units 3 significado básico 362, 375 Slavic 8 slippery slope argument 3, 13 space, see mental space Spanish 10, 12, 13, 131, 136, 138, 139, 141, 143, 144, 146, 147, 150, 152, 159, 161, 163, 164, 169, 170, 171, 328, 331, 332, 333, 334, 335, 336, 338, 339, 341 360 spatio-temporal-existential cline 295, 296 states 257 statistical data, see quantitative data strategy for communicative exploitation of a meaning 3, 6, 8, 9, 45,155, 156, 157, 159, 160, 161, 164, 165, 166, 167, 168, 169, 170, 373 subject 27, 31, 32, 38, 42 subjunctive, see also modality 10, 15, 181, 187, 188, 193, 194, 195, 196, 199, 200, 201, 205, 206, 207, 210 subordinate clause 4, 10, 25 sustrato, see lengua de sustrato Swahili 124 switch-reference 262, 264, 265, 266, 267, 268, 269, 270, 271 symbolic structure 21, 52 symbolization 21 synchronic or diachronic 11 synergesis in language 292, 295 isomorphic 295 syntactic paraphrase 181, 182, 199 syntagmatic distribution of phonological units 277, 282 system vs. use of language 114, 116, 125 systemic nature of language 121 temporality 179, 180, 182, 183, 193, 197, 200, 201, 202 tense 177, 180, 184, 186, 187, 189, 191, 192, 193, 197, 198, 200, 201, 202 text counts, see also quantitative data 8, 10

387

388

Index of subjects theoretical paradigms, comparison of 2 time 155, 157 tongue, see articulators topic 31, 37, 342, 343 topicality 342, 343, 345, 347, 348, 349, 350, 351, 352, 354 trajector 27, 31, 38, 54 transitive 242 triconsonantal (CCC) root system 12 “two-level” model of language 5 unidirectionality, see grammaticalization, unidirectionality in unmarked, see markedness Urdu, see Lucknow Urdu usage-based conception of distribution 41 use, see system vs. use of language validation techniques 8 value relations 112

variación sintáctica 361, 362, 363, 365, 366, 367, 368, 377 verb, see events, states viewing arrangement of speaker 183 vision as factor in phonological distributions 277, 286 voice 236 word 143, 145 word order 4, 12, 13, 58, 182, 326, 327, 328, 329, 330, 331, 332 word-order signals 4, 12, 58, 326, 327, 328, 329, 330, 334, 337, 338, 339 word-order variation 12 Yaqui 36 Yavapai 261, 267 Yuman 9, 261, 262, 266, 269, 270, 271 Zipf ’s Law 292, 293

In the series STUDIES IN FUNCTIONAL AND STRUCTURAL LINGUISTICS the following volumes have been published thus far or are scheduled for publication: 43 ANDREWS, Edna and Yishai TOBIN (eds.): Toward a Calculus of Meaning. Studies in markedness, distinctive features and deixis. 1996. xxviii, 432 pp. 44 JESSEN, Michael: Phonetics and Phonology of Tense and Lax Obstruents in German. 1999. xx, 394 pp. 45 MARTÍN-VIDE, Carlos (ed.): Mathematical and Computational Analysis of Natural Language. Selected papers from the 2nd International Conference on Mathematical Linguistics (ICML ’96), Tarragona, 1996. 1998. xviii, 391 pp. 46 ŠTEKAUER, Pavol: An Onomasiological Theory of English Word-Formation. 1998. x, 192 pp. 47 MARTÍN-VIDE, Carlos (ed.): Issues in Mathematical Linguistics. Workshop on Mathematical Linguistics, State College, PA, April 1998. 1999. xii, 214 pp. 48 REID, Wallis, Ricardo OTHEGUY and Nancy STERN (eds.): Signal, Meaning, and Message. Perspectives on sign-based linguistics. 2002. xxii, 413 pp. 49 HLADKÝ, Josef (ed.): Language and Function. To the memory of Jan Firbas. 2003. x, 339 pp. 50 DUŠKOVÁ, Libuše (ed.): Dictionary of the Prague School of Linguistics. Translated from the French, German and Czech sources. Author: Josef Vachek †. In collaboration with Josef Dubský. Translated by Aleš Klégr, Pavlína Šaldová, Markéta Malá, Jan Čermák, Libuše Dušková. 2003. x, 216 pp. 51 CONTINI-MORAVA, Ellen, Robert S. KIRSNER and Betsy RODRÍGUEZ-BACHILLER (eds.): Cognitive and Communicative Approaches to Linguistic Analysis. 2004. viii, 388 pp. 52 GORLACH, Marina: Phrasal Constructions and Resultativeness in English. A sign-oriented analysis. 2004. x, 150 pp. 53 EDDINGTON, David: Spanish Phonology and Morphology. Experimental and quantitative perspectives. 2004. xvi, 197 pp.