Conversation Analysis: Comparative Perspectives (Studies in Interactional Sociolinguistics)

Citation preview

This page intentionally left blank

Conversation Analysis “Conversation analysis” is an approach to the study of social ­interaction that focuses on practices of speaking that recur across a range of contexts and settings. The early studies in this tradition were based on the analysis of English conversation. More recently, however, conversation analysts have begun to study talk in a broader range of communities around the world. Through detailed analyses of recorded conversations, this book examines differences and similarities across a wide range of languages including Finnish, Japanese, Tzeltal Mayan, Russian and Mandarin. Bringing together interrelated methodological and analytic contributions, it explores topics such as the role of gaze in question-and-answer sequences, the organization of repair, and the design of responses to assessments. The emerging comparative perspective demonstrates how the structure of talk is inflected by the local circumstances within which it operates. jac k si dn e l l is an Associate Professor in the Department of Anthropology at the University of Toronto.

Studies in Interactional Sociolinguistics editors Paul Drew, Majorie Harness Goodwin, John J. Gumperz, Deborah Schifrin   1. Discourse Strategies John J. Gumperz   2. Language and Social Identity edited by John J. Gumperz   3. The Social Construction of Literacy Jenny Cook-Gumperz   4. Politeness: Some Universals in Language Usage Penelope Brown and Stephen C. Levinson   5. Discourse Markers Deborah Schiffrin   6. Talking Voices: Repetition, Dialogue, and Imagery in Conversational Discourse Deborah Tannen   7. Conducting Interraction: Patterns of Behaviour in Focused Encounters Adam Kendon   8. Talk at Work: Interaction in Institutional Settings edited by Paul Drew and John Heritage   9. Grammar in Interaction: Adverbial Clauses in American English Conversations Cecilia E. Ford 10. Crosstalk and Culture in Sino-American Communication Linda W.L. Young (with a foreword by John J. Gumperz) 11. AIDS Counselling: Institutional Interaction and Clinical Practice Anssi Perakyla 12. Prosody in Conversation; Interactional Studies edited by Elizabeth Couper-Kuhlen and Margret Selting 13. Interaction and Grammar edited by Elinor Ochs, Emanual A. Schegloff and Sandra A. Thompson 14. Credibility in Court: Communicative Practices in the Camorra Triales Marco Jacquement 15. Interaction and the Development of Mind Anthony J. Wootton 16. The News Interview: Journalists and Public Figures on the Air Steven Clayman and John Heritage 17. Gender and Politeness Sara Mills 18. Laughter in Interaction Philip Glenn 19. Matters of Opinion: Talking about Public Issues Greg Myers 20. Communication in Medical Care: Interaction between Primary Care Physicians and Patients edited by John Heritage and Douglas Maynard 21. In Other Words: Variation in reference and narrative Deborah Schiffrin 22. Language in Interaction in an urban school Ben Rampton 23. Discourse and Identity edited by Anna De Fina, Deborah Schiffrin and Michale Bamberg 24. Reporting Talk: Reported Speech in Interaction edited by Elizabeth Holt and Rebecca Clift 25. The Social Construction of Literacy, 2nd edition by Jenny Cook-Gumperz 26. Talking Voices, 2nd edition by Deborah Tannen 27. Conversation Analysis: Comparative Perspectives edited by Jack Sidnell

Conversation Analysis Comparative Perspectives Edited by

Jack S i dn ell University of Toronto

CAMBRIDGE UNIVERSITY PRESS

Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo, Delhi, Dubai, Tokyo Cambridge University Press The Edinburgh Building, Cambridge CB2 8RU, UK Published in the United States of America by Cambridge University Press, New York www.cambridge.org Information on this title: www.cambridge.org/9780521883719 © Cambridge University Press 2009

This publication is in copyright. Subject to statutory exception and to the provision of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published in print format 2009 ISBN-13

978-0-511-63432-1

eBook (EBL)

ISBN-13

978-0-521-88371-9

Hardback

Cambridge University Press has no responsibility for the persistence or accuracy of urls for external or third-party internet websites referred to in this publication, and does not guarantee that any content on such websites is, or will remain, accurate or appropriate.

Contents

List of figurespage vii List of tablesix List of contributorsxiii Transcription conventionsxv Part I  Introduction 1

Comparative perspectives in conversation analysis Jack Sidnell

Part II  Repair and beyond 2

Repetition in the initiation of repair Ruey-Jiuan Regina Wu

3

A cross-linguistic investigation of the site of initiation in same-turn self-repair Barbara Fox, Fay Wouk, Makoto Hayashi, Steven Fincke, Liang Tao, Marja-Leena Sorjonen, Minna Laakso, and Wilfrido Flores Hernandez

4

Repairing reference Maria Egbert, Andrea Golato, and Jeffrey D. Robinson

Part III  Aspects of response

1 3

29 31

60

104

133

5

Projecting nonalignment in conversation Anna Lindström

135

6

Two answers to inapposite inquiries Trine Heinemann

159

vi

Contents 7

Gaze, questioning, and culture Federico Rossano, Penelope Brown, and Stephen C. Levinson

8 Negotiating boundaries in talk Makoto Hayashi and Kyung-eun Yoon Part IV  Action formation and sequencing 9

Alternative responses to assessments Marja-Leena Sorjonen and Auli Hakulinen

10 Language-specific resources in repair and assessments Jack Sidnell 11 Implementing delayed actions Galina B. Bolden Part V  Conclusion 12 One perspective on Conversation Analysis: Comparative Perspectives Emanuel A. Schegloff

187

250

279 281

304 326

355 357

Bibliography 407 Index436

Figures

4.1 4.2 4.3

Open-class repair German “Wa?” page 108 Open-class English “What?” 109 Underspecified referent repair initiation (URRI) German “Was.” 110 4.4 URRI German “Was denn.” 110 4.5 URRI English “What.” 111 7.1 Location of three cultures 204 7.2 Gaze toward other participant, by language 208 7.3 Gaze behavior across the three cultures clustered by participation role 209 7.4 Distribution of actions by languages 214 7.5 Distribution of percentages of each action in which the Q-speaker looks at the addressee, 215 by language 7.6 Distribution of percentages of each action in which the Q-recipient looks at the Q-speaker, 217 by language 7.7 Italian dyads 227 7.8 Tzeltal dyads 228 7.9 Yélî Dnye dyads 229 9.1 Responses to assessments with a verb repeat282

Tables

3.1 Post-beginning and pre-completion repair initiation 3.2 Recognizable completion and repair initiation in English 3.3 Recognizable completion and repair initiation in Indonesian 3.4 Recognizable completion and repair initiation in Finnish 3.5 Recognizable completion and repair initiation in Bikol 3.6 Recognizable completion and repair initiation in Japanese 3.7 Recognizable completion and repair initiation in Sochiapam Chinantec 3.8 Recognizable completion and repair initiation in Mandarin 3.9 Site of initiation of simple recycling repair initiated in monosyllabic words 3.10 Site of initiation of simple replacement repair initiated in monosyllablic words 3.11 Site of initiation in recycling repairs in words of three or more syllables 3.12 Site of initiation in replacement repairs in words of three or more syllables 3.13 Frequency of monosyllabic words for each repair type 3.14 Frequency of multisyllabic words for each repair type

page 68 75 77 77 78 78 79 79 81 82 84 84 86 86

x

Tables

3.15 Site of initiation of recycling repairs in bisyllabic words 89 3.16 Percentage of recycling repairs initiated in bisyllabic words 90 3.17 Recycling repairs: Bisyllabic words by syntactic class 91 3.18 Recycling repairs: Multisyllabic words by syntactic class 92 3.19 English replacement repairs by word type for monosyllabic words 93 3.20 Replacement repairs by word type for monosyllabic words 94 3.21 Word length in recycling repairs 95 7.1 Instances of Q-speaker and Q-recipient gaze toward other participant, by language 207 7.2 Logistic regression analysis predicting gaze toward other participant in relation to role (Q-recipient as reference group), by language 209 7.3 Logistic regression analysis predicting Q-speaker gaze toward Q-recipient in relation to language (Italian and Tzeltal as reference group) 210 7.4 Logistic regression analysis predicting Q-recipient gaze toward Q-speaker in relation to language (Italian and Tzeltal as reference group) 211 7.5 Instances of mutual gaze by language 213 7.6 Logistic regression analysis predicting mutual gaze in relation to language (Italian and Tzeltal as reference group) 213 7.7 Questions in which speakers look at recipients and recipients do not look back, by language 214 7.8 Distribution of actions performed by questions, by language 214 7.9 Q-speaker looking at Q-recipient, by action and by language 215 7.10 Q-recipient looking at Q-speaker, by question type and by language 217 7.11 Distribution of response types by language 218 7.12 Distribution of responses to questions, by language 218

Tables

xi

7.13 Distribution of questions that do not get responded to in which the recipient does not look at the speaker 220 7.14 Logistic regression analysis predicting no response 221 7.15 Logistic regression analysis predicting no response if Q-recipient is not gazing, by language 222 7.16 Q-speaker gaze, by language 222 7.17 Q-recipient gaze, by language 223 7.18 Questions in which lack of Q-speaker gaze toward Q-recipient motivated by question content or competing activity 224 7.19 Questions in which lack of Q-recipient gaze toward Q-speaker motivated by question content or competing activity 224 11.1 Transcription symbols 352 12.1 Distribution of “so”- and “oh”-prefaced sequence initiators388

Contributors

Galina B. Bolden, Department of Communication, Rutgers University, New Brunswick, N.J., USA Penelope Brown, Language Acquisition Group, Max Planck Institute for Psycholinguistics, Nijmegen, The Netherlands Maria Egbert, Department of Business Communication and Information Science, University of Southern Denmark, Denmark Steven Fincke, Nuance Communications, Cambridge, Mass., USA Barbara Fox, Department of Linguistics, University of Colorado, Boulder, Col., USA Andrea Golato, Department of Germanic Languages and Literatures, University of Illinois at Urbana-Champaign, Ill., USA Auli Hakulinen, Department of Finnish Language and Literature, University of Helsinki, Finland Makoto Hayashi, Department of East Asian Languages and Cultures, University of Illinois at Urbana-Champagne, Ill., USA Trine Heinemann, Department of Business Communication and Information Science, University of Southern Denmark, Denmark Wilfrido Flores Hernandez, Department of Applied Language Studies and Linguistics, University of Auckland, New Zealand Minna Laakso, Department of Speech Sciences, University of Helsinki, Finland Stephen C. Levinson, Language and Cognition Group, Max Planck Institute for Psycholinguistics, Nijmegen, The Netherlands

xiv

Contributors

Anna Lindström, School of Humanities, Education and Social Sciences, Örebro University, Sweden Jeffrey D. Robinson, Department of Communication, Portland State University, Oregon, USA Federico Rossano, Department of Development and Comparative Psychology, Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany Emanuel A. Schegloff, Department of Sociology, University of California, Los Angeles, Calif., USA Jack Sidnell, Department of Anthropology, University of Toronto, Canada Marja-Leena Sorjonen, Research Institute for the Languages of Finland, Finland Liang Tao, Department of Linguistics, Ohio University, Athens, Ohio, USA Fay Wouk, Department of Applied Language Studies and Linguistics, University of Auckland, New Zealand Ruey-Jiuan Regina Wu, Department of Linguistics and Asian/ Middle Eastern Languages, San Diego State University, Calif., USA Kyung-eun Yoon, Department of African and Asian Languages and Literatures, University of Florida, Gainesville, Fla., USA

Transcription conventions (from Schegloff 2000)

I.  Temporal and sequential relationships Overlapping or simultaneous talk is indicated in a variety of ways. [

Separate left square brackets, one above the other on two successive lines with utterances by different speakers, indicates a point of overlap onset, whether at the start of an utterance or later.

] ]

Separate right square brackets, one above the other on two successive lines with utterances by different speakers indicates a point at which two overlapping utterances both end, where one ends while the other continues, or simultaneous moments in overlaps which continue.

=

Equal signs ordinarily come in pairs – one at the end of a line, and another at the start of the next line or one shortly thereafter. They are used to indicate two things:



(1) If the two lines connected by the equal signs are by the same speaker, then there was a single, continuous utterance with no break or pause, which was broken up in order to accommodate the placement of overlapping talk.



(2) If the lines connected by two equal signs are by different speakers, then the second followed the first with no discernable silence between them, or was “latched” to it.



A single equal sign indicates no break in an ongoing piece of talk, where one might otherwise expect it, e.g., after a completed sentence.

(0.5) Numbers in parentheses indicate silence, represented in tenths of a second; what is given here in the left margin indicates 0.5 seconds of silence. Silences may be marked either within an utterance or between utterances.

xvi (.)

Transcription conventions A dot in parentheses indicates a “micropause,” hearable but not readily measurable without instrumentation; ordinarily less than 0.2 of a second.

II.  Aspects of speech delivery, including aspects of intonation The punctuation marks are not used grammatically, but to indicate intonation. The period indicates a falling, or final, intonation . contour, not necessarily the end of a sentence. Similarly, a ?  question mark indicates rising intonation, not necessarily a ,  question, and a comma indicates “continuing” intonation, not ¿  necessarily a clause boundary. The inverted question mark is used to indicate a rise stronger than a comma but weaker than a question mark. ::

Colons are used to indicate the prolongation or stretching of the sound just preceding them. The more colons, the longer the stretching. On the other hand, graphically stretching a word on the page by inserting blank spaces between the letters does not necessarily indicate how it was pronounced; it is used to allow alignment with overlapping talk.

-

A hyphen after a word or part of a word indicates a cut-off or self-interruption, often done with a glottal or dental stop.

word

Underlining is used to indicate some form of stress or emphasis, either by increased loudness or higher pitch. The more underlining, the greater the emphasis.

word

Therefore, underlining sometimes is placed under the first letter or two of a word, rather than under the letters which are actually raised in pitch or volume.

WOrd

Especially loud talk may be indicated by upper case; again, the louder, the more letters in upper case. And in extreme cases, upper case may be underlined.

o

The degree sign indicates that the talk following it was markedly quiet or soft.

o

wordo

When there are two degree signs, the talk between them is markedly softer than the talk around it.

Combinations of underlining and colons are used to indicate intonation contours: _: If the letter(s) preceding a colon is/are underlined, then there is an “inflected” falling intonation contour on the vowel (you can hear the pitch turn downward). : If a colon is itself underlined, then there is an inflected rising intonation contour on the vowel (i.e., you can hear the pitch turn upward)

Transcription conventions

xvii

↑ ↓

The up and down arrows mark sharper rises or falls in pitch than would be indicated by combinations of colons and underlining, or they may mark a whole shift, or resetting, of the pitch register at which the talk is being produced.

> < < > <

The combination of “more than” and “less than” symbols indicates that the talk between them is compressed or rushed. Used in the reverse order, they can indicate that a stretch of talk is markedly slowed or drawn out. The “less than” symbol by itself indicates that the immediately following talk is “jump-started,”, i.e. sounds like it starts with a rush.

hhh (hh)

Hearable aspiration is shown where it occurs in the talk by the letter h – the more h’s, the more aspiration. The aspiration may represent breathing, laughter, etc. if it occurs inside the boundaries of a word, it may be enclosed in parentheses in order to set it apart from the sounds of the word. If the aspiration is an inhalation, it is shown with a dot before it.

.hh

III.  Other markings (( ))

Double parentheses are used to mark transcriber’s descriptions of events, rather than representations of them: ((cough)), ((sniff)), ((telephone rings)), ((footsteps)), ((whispered)), ((pause)), and the like.

(word)

When all or part of an utterance is in parentheses, or the speaker identification is, this indicates uncertainty on the transcriber’s part, but represents a likely possibility.

(lit/bit)

Where alternate hearings are possible these are enclosed in parentheses and separated by a back slash.

( )

Empty parentheses indicate that something is being said, but no hearing (or, in some cases, speaker identification) can be achieved.

IV.  Multi-linear transcription conventions Many of the transcriptions in this book are of talk in languages other than English and include one or two lines of glossing. In three-line transcripts the first line is a broad phonetic representation of the talk in the original language. The second line is a morpheme-by-morpheme gloss using a combination of word-for-word translation and abbreviations such as ASP to indicate particles and other functional items that do not admit of a direct translation into English. Authors who use these abbreviations in the transcripts they present include a key explaining them at the conclusion of the chapter. The third line presents an idiomatic English gloss – an attempt to get as close as possible to the contextual sense or meaning of the utterance. The following example from the chapter by Wu illustrates:

xviii

Transcription conventions 01 L:  wo dou hai mei jian guo ta.

I

all still N see ASP he

‘I haven’t met him yet.”

In two line transcriptions the morpheme-by-morpheme gloss is not included. The following example is from the chapter by Sidnell:

01 C:  him mada biilongz tuu Hamilton.



his mother is from Hamilton.

Part I

Introduction

1 Comparative perspectives in conversation analysis Jack Sidnell

 Introduction Comparison of the various ways in which people talk in different sociocultural and linguistic communities can easily lead to two, apparently contradictory, conclusions. On the one hand, the diversity of conduct is striking. People speak different languages, they are oriented to markedly different sociocultural norms of posture and tone, they inhabit very different social (as well as economic, political, etc.) worlds and thus find occasion to talk in vastly different “contexts.” One need only compare, for instance, the Mayan  villagers of Tenejapa with Yélî Dnye   speaking Rossel  Islanders to see this diversity (as Rossano, Brown   and Levinson  do in this volume). On the other hand, the commonalities are what are remarkable. Everywhere turns-at-talk are constructed and opportunities to speak distributed, courses of action are launched and ­co-ordinatively managed, troubles of speaking, hearing and understanding are located and their repair attempted. These commonalities suggest that, for all the diversity we see, people everywhere encounter the same sorts of organizational problems and make use of the same basic abilities in their solutions to them – a capacity for reading other’s intentions, anticipating and projecting actions, calculating inferences and processing information available to them (see Levinson  2006, Schegloff  2006). If these abilities and problems appear universal  and generic, the particular ways in which they are implemented or solved is anything but. After all, whatever happens in interaction happens through the medium of some specific set of locally available semiotic resources. Over the past forty years, conversation analysts

4

Jack Sidnell

have shown that actions in talk-in-interaction are formed through the use of distinctive prosodic patterns, lexical collocations, wordorder  patterns as well as language-specific  objects such as English “oh” (Heritage  1984), Mandarin “ a” (Wu  2004), Finnish “ nii” (Sorjonen  2001) and so on (see Stivers  and Sidnell  2005 for an overview). Of course, these semiotic resources vary significantly and systematically across different languages and communities. Some languages such as Vietnamese and Mandarin use a set of distinct tonal contours to signal differences of lexical meaning; others do not. Some languages are, like English, left-headed, whereas others, such as Korean  and Japanese , are right-headed. In some languages, such as Vietnamese again, words are typically made up of one morpheme having a single meaning whereas in others words often consist of several morphemes some of which may embody multiple meanings (e.g., -ó in Spanish “habló” expresses indicative mode, third person, singular, past tense, and perfective aspect). Because every turn-at-talk is fashioned out of the linguistic resources of some particular language, the rich and enduring semiotic structures of language must be consequential in a basic way for social interaction. So, although the problems are generic and the abilities apparently universal , the actual forms that interaction takes are shaped by and adapted to the particular resources that are locally available for their expression. The chapters in this book contain detailed analyses of talk in English, Finnish , German , Italian , Japanese , Mandarin , Tzeltal  (Mayan ), Russian , Swedish and Yélî Dnye   and provide the beginnings of an answer to the question of how the local resources  of these particular languages and other semiotic systems shape, constrain, torque or inflect the otherwise generic and universal  underlying organization of talkin-interaction. In recent work, both Schegloff  (1996c) and Fox  (2007) have argued that over time the organizational contingencies of interaction significantly shape (though do not determine) grammar. This book offers a complementary view in which, at some given point in time, the semiotic resources of any particular language – especially grammar – essentially define the possibilities for social action accomplished through talk. Notice, that this way of putting things largely obviates the theoretical syntactician’s appeal to a “deep” level of linguistic structure or “logical form” at which such “surface” differences between

Introduction

5

languages more or less completely evaporate. For social interaction does not work with such “underlying” mental representations; it works with the actual, observable surfaces of grammar and prosody  as well as the particular words and phrases that constitute possibly complete turn units and recognizable actions. As we will see in this chapter, it is this surface patterning of language that has consequences for turn-taking , repair and other aspects of the organization of talk-in-interaction. The linguistics that forms a natural companion to conversation analytic studies is thus the descriptive and empirical tradition that investigates not mental models but rather actual living languages in all their peculiarity and nuance often from a comparative perspective.1 In fact, recent work in interactional linguistics has emphasized and shown the value of just this connection (see Ford et al.   2002 for an overview). Some of the complexity of the relationship between (1) the specific semiotic resources of a particular language, (2) language and culture-independent principles, and (3) the contexted courses of action within which language is actually embedded can be seen through a consideration of repetition . Turns designed to show that they are repeats of something that someone has just said are apparently found in all communities and in all languages. But now consider that, strictly speaking, “unmarked” repetition of what someone has just said is interactionally impossible. That is to say, one cannot simply and solely repeat what someone has just said without adding something to it. While it’s possible for a second speaker to do a repeat in such as way as to make it hearable as saying “exactly” what another just said (for example, by preserving not only the words another speaker has used but also the original deictic  forms as well the intonation and other aspects of prosody ), to do so is to do something in addition to repetition. Repetition  of this kind is hearable as a verbatim repeat and thus possibly as “mimicry.” This points to the fact then that even in repeating what another has said, the speaker necessarily draws on the semiotic resources of some particular language to mark, and thus to frame, the repeat in some way. 2 Typically then, a repeat involves the use of some intonational or periphrastic marking which shows to what end this bit of talk is being repeated. One very common thing is for the repeat speaker to produce his or her talk with what is typically known as “question

6

Jack Sidnell

intonation” as in the following case from an English telephone call in which Ben has called for someone he refers to first as “Mary” and subsequently, after Ann indicates that she does not recognize anyone by that name, as “the tax lady” (Jefferson  1972, Schegloff et al.   1977). (1)  XTR1 01 ((click)) 02 Ann: Hello:¿ 03 Ben: hHello, Ma:ry? 04 (0.2) 05 Ann: No: 06 (0.3) 07 Ben: No, not Ma:ry?= 08 Ann: =No, it’s not Ma:ry. There’s no Mary he:re 09 (.) 10 I don’ think:hh 11 (3.0) 12 Ben: The tax lady: 13 (0.2) 14 Ann: → The tax lad(h)y::? 15 Ben: Ya hhh= 16 Ann: =Nah. (.) Wha number were you callin’.

Another common intonation-marking for repeats in English involves stressing the first syllable of a multi-syllabic word. In (2) below Ann and Bev are discussing the diet and regime of a woman who is pregnant. (2)  YYZ_T1A_A&D 2.03 01 Bev: anyway [.hhhh [I know. 02 Ann: 03 (0.2) 04 Ann: an’ i- (.) gra:vol an all these (s[ ) 05 Bev: → [gr↑a:vo:l 06 Ann: She takes gravol al- everyda:y.

Notice that whereas in Extract 1 the repeat elicits from Ben a confirmation (“Ya”   at line 15) in Extract 2 it does not: instead, Bev continues with the telling. This suggests that in Extract 2 Bev uses the repeat to do something other than check her hearing of “gravol.” While these forms of marking appear to be quite common across a wide range of languages, more obviously language-specific 

Introduction

7

resources are used in repeats too. Stivers  (2005) for instance has described one kind of repetition  in English that involves, among other things, expanding the pronoun + verb contraction of the original utterance (It’s → It is). Consider the following case from a conversation between co-workers. (3)  YYZ 27 Clare: Just we’ve got- Michael and I did the resourcing for next 28 → week and it’s just- it’s=just ughu(h)h(hh) .hh and it’s a 29 short wee:k so 30 (.) 31 Alice: → Yeah:- Oh:yeah it is a short week= 32 Clare: Ye[ah 33 Alice: [.hh[hh 34 Clare: [so (.) I’m rilly rilly sorry,

Indeed, for English, next-turn repeats take a great variety of forms and are deployed in a wide range of actions such as confirmation  (Schegloff  1996b), other-initiation of repair (Schegloff et al.   1977), correction (Schegloff et al.   1977; Goodwin  1984), displays of appreciation (Jefferson  1972) and agreement  (Pomerantz  1984). Coupled with the particular semiotic resources of English grammar, prosody  and lexis, the apparently universal  practice of ­repetition  is employed to effect a range of interactional outcomes. A number of the chapters in this book consider practices that involve repetition  including Hayashi  and Yoon on Japanese  and Korean , Sorjonen  and Hakulinen  on Finnish , Wu  on Mandarin , Heinemann  on Danish , Bolden  on Russian . These studies provide further evidence of the ways in which an apparently universal  aspect of interactional organization is shaped and, in some ways, constrained by the semiotic resources of a given language thus providing unique possibilities for social action. 3 A comparative perspective in conversation analysis The chapters collected in this book examine the organization of social interaction and, more specifically, talk-in-interaction, from a comparative, cross-linguistic perspective. Although some of them draw at points on a number of other traditions (particularly functional/interactional linguistics ), the analyses they present are largely

8

Jack Sidnell

developed within the framework of conversation analysis (hereafter CA) – an approach to talk and social interaction which emerged in the writings and lectures of Harvey Sacks , Emanuel Schegloff  and Gail Jefferson .4 CA takes not language per se as its focus but rather the practical activities in which language (along with gesture , gaze and other aspects of bodily comportment) is deployed, that is, talk-in-interaction. However, comparison of such practical activities across different languages brings the relevance of grammatical  (and other) structures into view as the chapters in this book demonstrate. In this chapter, I sketch some of the conceptual background of comparative studies in CA. I begin with a consideration of previous studies distinguishing the approach taken within CA from that known as cross-cultural  pragmatics. I then move to consider the CA methodology. Here I note a similarity between structuralist or distributionalist methods in linguistics and the CA approach of using collections. I argue that the former, but not the latter, essentially undermine the very possibility of comparison. Structuralism, taken to its logical conclusion, reveals the particularity of any system (e.g., of pronouns, or verbal tenses) and the elements of which it is composed. Since each element is defined by its relation to all the others, each element is a unique outcome of the particular system in which it is embedded. The result, as is well known, are accounts which are wholly hermetic and incapable of being compared to one another. In contrast, CA focuses on generic interactional problems which find solutions in the local resources  of particular languages and social systems. For example, turn-taking  and repair address interactional problems which must be solved in any community if its members are to coordinate their actions through talk-in-­interaction. This approach actually encourages comparison: by looking across different languages/communities we can see the way in which the same interactional problem is solved through the mobilization of different resources (Schegloff  2006). Conversation analysts are keenly aware of the method ological and theoretical issues raised by a comparative perspective. In an early co-authored paper, Schegloff  and Sacks  (1974: 234–235) wrote: The materials with which we have worked are audiotapes and transcripts of naturally occurring interactions […] with differing numbers of participants and different combinations of participant attributes. There is a

Introduction

9

danger attending this way of characterizing our materials, namely, that we be heard as proposing the assured relevance of numbers, attributes of participants, etc., to the way the data are produced, interpreted, or analyzed by investigators or by the participants themselves. Such a view carries considerable plausibility, but for precisely that reason it should be treated with extreme caution, and be introduced only where warrant can be offered for the relevance of such characterizations of the data from the data themselves. […] The considerations just adduced […] restrain us from further characterizing it here.

And the authors go on: For example, they restrain us from characterizing our findings as relating to “some general features of conversation rules in American English” – a suggestion offered by Dell Hymes  (personal communication), for that suggests implicitly an ethnic or national or language identification as a relevant putative boundary for both our materials and findings. We cannot offer a warrant for asserting such a boundary and we suspect others cannot either.

The passage sums up quite succinctly the distinctive position which conversation analysts have adopted toward what are referred to in this passage as “participant attributes.” Whereas other approaches such as linguistic anthropology  and sociolinguistics  typically treat participant attributes as self-evident features of “context” or, alternatively, as “external variables” to be correlated with aspects of speech behavior, CA shifts the burden of evidence by requiring researchers to show that putative “characteristics” of the participants (such as race, ethnicity, gender) have some “procedural consequentiality” or demonstrable relevance for the participants themselves in terms of the specific ways in which the interaction is organized.5 In the passage quoted above, Schegloff  and Sacks  characterized this as a reluctance to invoke a boundary where none seemed to be warranted by the data being examined. Importantly, Schegloff  and Sacks  were not denying the existence of differences either between groups within a society or between societies. Instead, what they were pointing to were the method ological and theoretical issues involved in making claims about the relevance of such differences for the organization of interaction. Within CA then, researchers have emphasized the various ways in which “participant attributes” are picked out and made relevant to the talk of the moment (see for instance Goodwin  1987; Schegloff  1991, 1992b; Sacks  1995).

10

Jack Sidnell

There were deep underlying issues at stake here – ones that Sacks  had initially addressed in his work on membership categorization  devices and that Moerman (1974, for instance) had applied to problems of ethnic categorization in anthropology . As Goodwin puts it, “showing that a category can be accurately applied to a participant does not demonstrate either that the participants themselves are dealing with each other in terms of such a category, or that it is in fact a relevant feature of the activity being studied” (1987: 120). A speaker (or recipient) may be Catholic, six feet tall, male, a dentist, a vegetarian or whatever else without this being in any way relevant to the way he talks on some particular occasion. Sacks noted in his lectures that all societies appear to divide people into groups based on at least two axes of difference: gender and age (Schegloff  2002c). It was not an interest in universals of social organization that prompted Sacks ’ noticing of this. Rather, the existence of at least two axes of difference meant that any given individual could be described or categorized  in more than one way. As such, any description of a participant (as a male, a child, etc.) required some “warrant” or motivation, some evidence that this and not some alternative was the relevant descriptive category. It is not difficult to see how these considerations bear on the problems of comparison to which the chapters in this book are addressed. How can we, for instance, warrant a description of the data as “requests  in Polish,” “French compliments”  or whatever where, typically, the participants do not display any overt orientation to the relevance of the language being used or their ethnicity?6 This is a particularly difficult problem and this book does not offer a single, conclusive solution to it. Rather, different authors take somewhat different approaches. One possible solution is discussed in this introduction: this involves linking the practices of talk-in-interaction to particular, distinctive aspects of the language being spoken thus warranting a characterization of those practices in terms of the language employed. The idea then is that particular languages provide specific resources and also establish unique constraints for the organization of interaction. Grammar, culture, and the organization of repair A few years after the Schegloff  and Sacks  paper cited above was published, Moerman  (1977) addressed the issue of comparison

Introduction

11

in a study of repair in Thai -Lue materials. If differences could be located between the Thai-Lue and English materials, these might be related to differences in language structure, social arrangement or, perhaps, culture. However, in the main, the study did not reveal significant differences in the organization of repair and Moerman (1977: 875)  was led to conclude: “the detailed, systemic, and massive parallels between” Thai and American English corpora “support the claim that the domain described by Sacks , Schegloff  and Jefferson is conversation – without respect to the language, nation, class, or culture in which it occurs.” In contrast to work in both cross-cultural pragmatics  and linguistic anthropology, early comparative research in CA provided evidence of massive, underlying structural similarities in interaction from very different communities. Whereas both crosscultural  pragmatics and linguistic anthropology tended to work from the assumption of significant differences between communities/cultures, conversation analysts began with a focus on generic interactional problems which had to be, in some way, solved in any community. For instance, the phenomena to which a conversational repair mechanism is addressed – misunderstandings, misspeakings and the like – are universal  and ubiquitous. If participants are to engage in coordinated action they will necessarily require some mechanism for dealing with these issues. At a finer level of detail, misunderstandings and misspeakings seem to take similar forms in every community. Misunderstandings arise because the recipient does not know a word the speaker has used or did not hear it adequately. Misspeakings arise when a speaker uses the wrong word or cannot find the one they need and so on. Given that everywhere the issues are more or less the same, it is hardly surprising to find that the solutions developed for dealing with them are also strikingly similar. This is not, however, to say that there are no differences. Here, I will briefly note a few of these staying with the problem of repair; below (see “Mobilization of Local Resources to Solve Generic Problems of Interaction” ) I discuss some claims for profound differences in turn-organization and turn-taking  between Japanese  and English. In a comparison of repair in English and German, Egbert (1996: 587) argued that “repair is sensitive to the linguistic inventory of a given language” in several different ways. First, Egbert showed that grammatical  markings present in German  but not English (such

12

Jack Sidnell

as articles marked for gender,  i.e. “das” and “der”) were subject to repair. Second, she observed that case-marking in German can serve as a resource for framing, and thus locating, the repairable  item. Third, she showed that, in other-initiated  repair, grammatical congruities across turns, the marking of an item as masculine singular accusative for instance, may serve as a resource for linking repairable , repair initiator and the repair itself. In a study which I discuss in greater detail below, Fox et al.   (1996) noted three differences between self-repair  in English and Japanese  and link these to the very different “syntactic practices” of the two languages. First, similar to Egbert ’s finding for German, Fox  et al. find that certain kinds of morphological repair are possible in Japanese but not English. They illustrate with the following example: (4)  K: ja nanji goro ni kurida[shi-*]soo? Then what.time about OBL go.out “Then about what time (shall we) go out?”

The authors (Fox et al. 1996: 202) observe that in this example the “speaker K replaces only the inflection al ending of the verb with another. […] The first form that K produces in this example ­(kurida-shi) has the adverbial ending -shi. […] However K cuts off the verb and replaces -shi with the ‘cohortative’ ending -soo. In other words, K has replaced one bound morpheme with another.” In English, repair does not operate on inflection al endings. This difference can likely be accounted for in terms of verbal morphology in the two languages. The Japanese  verb endings -shi and -soo are full syllables consisting of a consonant and a vowel. In contrast, English verb endings (with the exception of -ing and -s in certain contexts) are, typically, pronounced as a single consonant sound. They are thus unpronounceable independently of the word to which they are attached. This prevents their production as independent units in self-repair  as in the following invented example (from Fox  et al. 1996: 202). (5)  She look[ed]-*s at the table.

Moreover, as the authors point out, Japanese  is an agglutinating language in which each morpheme has roughly a single grammatical  meaning. In contrast, English is characterized by a relatively

Introduction

13

greater degree of fusion in inflection al morphology (Comrie 1981). So, for instance, verbal -s carries the meaning of present, third person and singular. Bound verbal morphemes are thus semantically complex in English but not in Japanese . English verbal morphemes (or at least verbal -s) may also indicate agreement with the subject (third person) of the clause and thus link back to something earlier in the utterance where as Japanese  verbal markers do not. The authors (Fox et al. 1996: 203) conclude that “these […] differences between English and Japanese  verb endings suggest to us that at a variety of levels verb endings in English are more tightly ‘bonded’ to the verb than are verb endings in Japanese  and hence less available for individual replacement  than are verb endings in Japanese.”  Another difference noted by these authors involves practices for delaying the production “of next noun” due in self-repair . The following illustrates one common practice in English: (6) on the back of his pickup truck [with a,*] (0.4) with a jack.

Here the speaker begins a prepositional phrase, initiates repair and then recycles the preposition and article before then producing the rest of the phrase. Recycling  the preposition (and article) thus “constitutes a procedure for delaying the production of the next item due. This procedure could, for example, be part of a word search , a request for recipient gaze, management of overlapping talk, and/or production of a dispreferred”  (Fox  et al. 1996: 204). Since Japanese  has postpositions rather than prepositions  (and no articles), this procedure for delay is not available. Fox et al.   show that Japanese  speakers often use a quite different procedure to effect the same result: employing a demonstrative  pronoun (translated as “that”) as a “place holder while the speaker looks for a lexically specific noun” (Fox et al. 1996: 206). The studies of Egbert  and Fox et al.   thus link language-specific  practices of repair  to particular grammatical  features of the language in question. Some of the variations in repair observed in the literature are accounted for not by reference to the grammar of the language but rather to the beliefs of the speakers. Ochs  (1984: 328–329) writes of the type of other-initiated  repair she terms “clarification ’: While clarification is a universal  activity, the manner in which clarification is accomplished varies cross-cultural ly. Prefer ences for

14

Jack Sidnell

accomplishing clarification are embedded in local principles of social order and local epistemologies. More specifically, I would like to suggest that both the conditions under which clarification takes place (what gets clarified, who participates in the activity of clarification, in which roles), and the discourse procedures speakers prefer to use, index members’ views of knowledge, particularly members’ views on the limits of knowledge (what can be known) and the paths to knowledge (how knowledge is acquired).

Ochs  goes on to contrast two basic practices in clarification  – a “minimal grasp strategy” in which the one initiating  clarification employs quizzical expressions, statements of non-understanding, WH-questions and other directives “to elicit from the child a reformulation of all or part of the unclear utterance or gesture”,  and an “expressed guess strategy” in which it is the one initiating ­clarification “who attempts a reformulation of the unclear act” (1984: 331). Ochs  argues that the more or less total absence of explicit guessing in the clarification practices of Western Samoans in favor of a minimal grasp strategy is linked to, among other things, local conceptions of knowledge: “ Samoans generally display a strong disprefer ence for guessing at what is going on in another person’s mind. This dispreference has reflexes in a range of verbal activities,” including those of clarification (Ochs  1984: 337, see also Ochs  1991). CA and Cross-Cultural Pragmatics  I want to briefly discuss here the way in which the approach taken in these chapters differs from that commonly known as cross­cultural pragmatics. One major set of differences is method ological. Work done under the rubric of cross-cultural pragmatics typically employs some kind of questionnaire method  in which respondents are asked to select the most appropriate or least appropriate response to some described situation. Alternatively respondents may be required to engage in a “discourse completion task.” In contrast, CA employs a method ology of observation and collection of naturalistic interactional data (see below). At an even more fundamental level, these two approaches exhibit quite different conceptualization of the object of study. Cross-cultural pragmatics  tends to focus on the different ways in which actions such as

Introduction

15

“promises,” “requests,” “invitations,” “refusals”   are managed and understood by recipients, often in terms of their implications for “face” and notions of politeness  (see for instance Blum-Kulka et al. 1989). In other words, actions such as “requests,” “invitations”  and the like are treated as though they were objective categories of social action providing a template through which one can view interactional data. CA adopts a different perspective. As Schegloff observed in an early paper  (1984: 30): “it is misleading to start to account for such categories  of action as questions, promises, and so on as the analytic objects of interest. They are commonsense, not technical, categories and should be treated accordingly.” Rather than begin with such “common sense categories of action” as promises and such, CA begins with relatively stable organizations of practice in talk-in-interaction, e.g., answering the phone with “hello?,” referring to persons by name or some alternative recognitional form (see Sacks and Schegloff 1979; Schegloff  1996a; Stivers  2007), multiple sayings such as “no no no” (Stivers  2004), etc.7 The practices of repair discussed in the previous section provide another example. Notice that with a format such as (partial repeat + WH word), a speaker may accomplish a wide range of actions such as expressing surprise in response to a news report (“They’re going where?”), challenging a previous speakers version of events (“You did what?”), requesting clarification (“ Move what”), etc. (see Schegloff  1997b). The contrast here between CA and cross-cultural  pragmatics is a decisive one. A focus on practices of speaking, rather than “action-types,” suggests that whatever action a given utterance (or part of an utterance) is understood by the participants to be doing is a contingent outcome. Participants understand some practice of speaking as doing a particular action by considering it in relation to the local sequential context in which it is produced (i.e. after a recognizable news report, an excuse, a suggestion, etc.). In that sense, categories  of action are themselves the product of local interpretive work. This perhaps accounts for a basic problem within cross-cultural pragmatics  of identifying comparable “actions” in the first place. That is, identifying invitations  in Russian , Japanese,  Samoan, and Greek so that they can then be compared necessarily involves a reification – a treatment of

16

Jack Sidnell

utterances as if each ­transparently ­conveyed a unique “illocutionary force.” Practices, structures, distributions Bloomfield (e.g., 1946, 1970) is often chided today for his distributional method  in which one accounts for the function of a given item by looking, one might say, dispassionately, – that is, without regard to its apparent “meaning” – at the positions in which it can be found to occur and the various alternatives for which it may substitute. This basic conceptual framework behind structuralism , in which relations in presentia and in absentia are more important than links between the sign and the object (designatum), has broad theoretical and method ological consequences. Perhaps most importantly, it offers an arguably more objective view than an impressionistic sense of what any given instance of a phenomenon “means” or is “doing.” I do not think it does CA any injustice to say that it employs a, broadly speaking, distributional method  – though it is one tempered by detailed attention to each particular case. Analysis is based on a collection of instances. Such a collection not only allows for the identification of an unmarked or default pattern (for discussion of this usage see Stivers , Enfield  and Levinson  2007; Stivers  2007; Enfield 2007), it also reveals the operation of underlying prefer ences or principles. Collections allow also for the identification of “deviant” instances in which the practice does not (for whatever reason) follow its normal course – a “deviance” to which participants themselves orient. Such deviant cases frequently reveal participants’ own orientations to the various contingencies involved in the execution of that practice – orientations which are typically unobservable in cases which conform to normative expectations and which are therefore literally unremarkable. In the context of an examination of practices for referring to persons, for instance, a collection of instances allows the analyst to distinguish cases in which just reference is being done from those in which something in addition to this is being accomplished. Consider, for instance, a case in which a spouse returns home from work and is met immediately with, “What did she say?”. Use of a locally subsequent form (the anaphor “she”) in a

Introduction

17

locally initial position (i.e. without any previous use of a name) can bring off that this has “been on my mind throughout the interim,” that this is, in effect, “a continuation of the earlier conversation” (Schegloff 1996a: 451). Referring to oneself by name (“Jack doesn’t like that!”) is another non-default usage that can be used to convey footing shifts of various kinds (see Schegloff  1996a). It is worth noting the clear resonances with structuralism  here. Consider Schegloff ’s analysis of continuers such as “uh huh” presented in his paper “Discourse As Interactional Achievement: Some Uses of ‘Uh Huh’ and Other Things That Come Between Sentences” (Schegloff 1982). In characterizing “discourse as an interactional achievement,” Schegloff  draws attention to the fact that a spate of talk composed of more than one “sentence” (or turn-constructional unit, etc.) is the product of methodic work by speaker and recipient(s) since it necessarily involves the negotiation of places at which transfer to a next speaker might have otherwise been effected. One key resource in such an accomplishment are vocalizations such as “uh huh,” “mm hmm,” “yeah” and the like typically labeled “backchannels” in the literature. Schegloff (1982: 78) notes that in the literature, two related characterizations have been offered to deal with these bits of behavior. According to one, they are evidence of attention, interest, and/or understanding on the listener’s part. (Thus Fries 1952: 49, “signals of this continued attention,” or Kendon  1967: 44, “appears to do no more than signal […] that he is attending and following what is being said”). A second use of such behavior proposed in this literature is that it “keeps the conversation going smoothly” (Dittman and Llewellyn 1967: 342), or “appears to provide the auditor with a means for participating actively in the conversation, thus facilitating the general coordination of action by both participants” (Duncan  and Fiske 1977: 202–203). After noting that while a coherent set of cases can be assembled in which participants use “uh huh” and other such objects to deal with issues of attention (cases in which, e.g., “attention was indeed problematic for the parties” (Schegloff 1982: 79) or in which the “uh huh” responds to the use of a try-marked recognitional  person reference), Schegloff  goes on to point out a number of problems with such characterizations. They are, in the end, of course largely

18

Jack Sidnell

impressionistic. Schegloff  (1982: 81) offers an alternate account, arguing that: perhaps the most common usage of “uh huh,” etc. is to exhibit on the part of its producer an understanding that an extended unit of talk is underway by another, and that it is not yet, or may not yet be (even ought not yet be), complete. It takes the stance  that the speaker of that extended unit should continue talking, and in that continued talking should continue that extended unit. “Uh huh,” etc. exhibit this understanding, and take this stance, precisely by passing an opportunity to produce a full turn at talk. When so used, utterances such as “uh huh” may properly be termed “continuers.”

Within such an account, the notion of signaling attention takes on a different cast. Points of possible completion in a continuing speaker’s talk are structurally relevant positions in which the recipient may properly display their understanding of the current state of the talk and “show their intention to pass the opportunity to take a turn-at-talk that they might otherwise initiate at that point” (Schegloff 1982: 81). So, with “uh huh” etc., a recipient passes the opportunity to do any sort of fuller turn. The analysis begins then with a specification of the particular sequential environment in which “uh huh” and other such tokens are deployed. Schegloff  then goes on to consider what else can occur in that same position noting that were a fuller turn done it would necessarily be a turn of some particular type and would thus be doing some specific action or actions. Thus, the opportunity being passed up by use of “uh huh” is an opportunity “to do whatever might have relevantly been done at that point” (1982: 87). Schegloff  (1982: 87) asks “are there any kinds of actions which have some kind of ‘general relevance’ in conversation,” actions “not made relevant by the particulars of someone’s immediately preceding talk or behavior?” The obvious candidate is other-initiated repair  which Sacks  described in his first transcribed lecture as an “occasionally usable device” meaning that “there doesn’t have to be a particular sort of thing preceding it; it can come at any place in a conversation” ( 1995: I, 6). As Schegloff  (1982: 87) puts it, “it appears that there are no systematic exclusion rules on the possible relevance of nextturn repair initiation in any possible turn position.” “Uh huh,” etc., in passing the opportunity to do a full turn at talk, “can be seen to be passing an opportunity to initiate repair on the immediately preceding talk” thus providing an obvious basis for “the

Introduction

19

ordinary inference that the talk into which they are interpolated is being understood” (Schegloff  1982: 88). Schegloff  (1982: 88) concludes: It is not that there is a direct semantic convention in which “uh huh” equals a claim or signal of understanding. It is rather that devices are available for the repair of problems of understanding the prior talk, and the passing up of those opportunities, which “uh huh” can do, is taken as betokening the absence of such problems.8

Schegloff ’s alternative to the impressionistic account in which “uh huh,” etc. are treated as “displays of attention”9 involves identifying a position, the transition relevance place constituted by a turn’s possible completion, and examining the particular object in relation to the other objects which might have occurred there – and which participants themselves can see could have been done in that position. The result is an interactionally generated, positionally sensitive “paradigm”. 1. Continuer (e.g., “uh huh”). 2. Full-turn (whatever might have otherwise been done). 3. Other-initiation of repair. The sense of “uh huh” (as, variously, conveying attention, understanding, agreement ) is a product of its place within a positionally sensitive yet highly restricted set of alternatives – relations in absentia. Though CA draws, as I hope to have shown, in a basic way on a structural or distributional method , it does not fall into the trap of structuralist analysis which, taken to its logical conclusion, reveals the particularity of any system (e.g., of pronouns or verbal tenses) and the elements of which it is composed. Since each element is defined by its relation to all the others, each element is a unique outcome of the particular system in which it is embedded. The result are accounts that are wholly hermetic and incapable of being compared to one another. The item in the phonemic inventory designated as /p/ for a given language will necessarily be unique to that system of distinctions. The consequences of a strict-structuralism have been noted with respect to diachronic comparision – that is, comparisons of the “same language” at different points in time. As Harris  and Taylor

20

Jack Sidnell

(1997) note, if the consequences of Saussure’s structuralism are accepted, the comparative assumptions of historical linguists , who happily trace the evolution of consonants and vowels from IndoEuropean to the present, are undermined. Accepting structuralist premises, it makes no sense to suppose that the earlier b was the “same consonant” as its later manifestation p, or that “the same word” appeared in Latin as causa but later in French  as chose. For if linguistic units did not exist except as structural units defined within a single linguistic system, it was impossible for any given unit to “survive” from one system A into a different system B at a later point in time.  (Harris and Taylor 1997: 215)

Even where a consonant (e.g., m) appears to have survived and maintained its phonetic character from Latin to French  because at each stage there was a slot that could be filled by a consonant having those phonetic qualities, we should not suppose that it is the same: “For in spite of this apparent identity of phonetic substance, the consonant in question. […] enters into a different set of contrastive relations with other consonants in Latin from those which characterize its position in the French system” (Harris and Taylor  1997: 216–217). CA avoids this problem of particularism through its focus on generic problems of interaction. For example, Schegloff ’s focus in the argument summarized above is the generic problem of coordination and turn-distribution – a problem that has to be solved within any community. We can look across communities and languages to see how local resources  are mobilized to solve the recurrent, generic problems of interaction within human groups. Mobilization of local resources  to solve generic problems of interaction Consider the following example (Sacks et al.   1974: 721): (7)  [T. Labov: Battersea:B:1] 01 Tourist: Has the park cha:nged much, Parky: Oh:: ye:s, 02 03 (1.0) 04 Old man: Th’Funfair changed it’n [ahful lot [didn’it. Parky: [Th- [That05 06 Parky: That changed it,

Introduction

21

At line 05 of this example Parky begins to speak and finds himself in overlap with Old Man. Where does Parky begin to speak? Not after “The,” not after “Fu” or “Fun,” not after “funfair.” Rather, the beginning of Parky’s turn is designed to coincide with the end of “it,” so, after the “The funfair changed it.” Parky has then targeted the first point of possible completion in Old Man’s turn as a place to begin speaking. Notice that Parky starts up here and again at the next two points of possible completion (after “lot” and “it”), not by virtue of any silence (by the time he starts there is no hearable silence) but by virtue of the possible completion of the turn which he has anticipated. The coordination of speakership – i.e. turn-taking – with minimal gap and overlap thus seems to involve, in some kind of essential way, the participants’ ability to project the possible course of the current turn and to thus anticipate when it will reach a point of possible completion. In their 1974 paper, Sacks  et al. (1974: 702) write: There are various unit-types with which a speaker may set out to construct a turn. Unit-types for English include sentential, clausal, phrasal, and lexical constructions. Instances of the unit-types so usable allow a projection of the unit-type under way, and what, roughly, it will take for an instance of that unit-type to be completed. Unit-types lacking the feature of projectability  may not be usable in the same way. [Emphasis added]

In this way, they link the notion of turn constructional units – the building blocks out of which turns-at-talk are composed – to a feature of “projectability.”  In a paper originally delivered in 1973, Schegloff  (1987c: 71) expands on the notion of projectability, writing: One important feature of turn construction […] and the units that turn construction employs (e.g., lexical, phrasal, clausal, sentential constructions) is that they project, from their beginnings, aspects of their planned shape and type. […] There are other sorts of projection that are, or can be involved from the very beginning of a turn. For example, question projection. […] Or: beginnings can project “quotation formats.” Or: a beginning like “I don’t think” can project, in certain sequential environments, “disagreement” as a turn type for its turn. […] Again: turn beginnings are important because they are an important place for turn projection, and, given the importance of turn projection for turn taking, they are important structural places in conversation.

Recent work on Japanese  (beginning with Fox et al.   1996) suggests that the grammatical  properties that make such “early

22

Jack Sidnell

syntactic projections” possible in English may not be shared by all languages. Fox et al.   (1996) and later Hayashi  (2003) argue that various “syntactic practices” of Japanese  result in a “limited projectability.”  These syntactic practices include: 1. a flexible word order  particularly at the beginning of the turn; 2. the use of post- rather than prepositions; 3.  a prevalence of unexpressed constituents (including subject and object); 4. the use of final particles capable of retroactively transforming the constituents to which they attach. The authors (Fox et al. 1996: 208) argue that: the kind of clausal TCU structure these facts commonly lead to typically […] starts with some kind of discourse marker (e.g., ano, nanka), followed by adverbials, or nouns indicating setting of some kind, followed by the verb, and possibly followed by so-called final particles. So what occurs early in the TCU is often only loosely associated structurally with what is to follow. Conversational utterances in Japanese  thus seem not to show tight syntactic organization.

Whereas in English “the subject begins a tightly knit clause structure and hence syntactically is the ‘beginning’ of the clause,” in Japanese “ these is no constituent element that serves as the beginning of a tightly knit syntactic unit – in fact, there is no such tightly knit unit” (Fox  et al. 1996: 209). These syntactic practices then mean that turn-taking  must be managed in a different way in Japanese  than it is in English. The authors (Fox et al. 1996: 213) summarize: English recipients are able to use the beginning of a TCU to project a possible course for that utterance, while Japanese  recipients “wait and see” how the utterance develops. It is thus possible that English recipients are able to predict with such accuracy how the utterance-under-construction will come to an end that they are able to plan their own utterance to start up exactly at the moment the current utterance comes to a possible completion point (this is, in fact, the claim made by Sacks  et al. (1974) with regard to turn-taking in English conversation), with no pause between the end of the current turn and the start-up of their own turn. In Japanese , on the other hand, since recipients are not able to make such detailed predictions about the course of the current utterance, they wait until they hear the last few syllables of the turn (which often contain such “ending

Introduction

23

signals” as final particles, or special completion-relevant verb forms) before starting their own utterance.

The general interactional consequences of these various syntactic practices then include the adoption of a “wait and see”-type approach to turn-taking  (and specifically self-selection), limited mid-turn projectability  of turns-in-progress and an “incremental” or “bit-by-bit” approach to turn construction (Tanaka  2000; Hayashi  1999, 2003). The consequences are manifested in an absence of clause-level recycling in self-repair  (Fox et al.   1996), various kinds of delay in co-participant completions including a prefer ence for terminal item completions (Hayashi 2003), an orientation to the end of the predicate component as transition relevant and a vulnerability of post-predicate components to overlap (Tanaka  2000: 7–8). Japanese  grammar may then help to account for the practices of aizuchi  (Miller 1983; LoCastro  1987). As Schegloff et al.   (1996: 32) remark: The interpolations which in English conversation ordinarily come at the boundaries of larger chunks of extended turns understood to be not yet complete (Schegloff  1982) are produced – and solicited – for smaller chunks of utterance in Japanese , in part to offset otherwise potentially problematic indeterminacies built into Japanese  grammar, indeterminacies which are problematic precisely because of the exigencies of recipient parsing in real time.

As Schegloff  (2000c) notes, “the orderly distribution of opportunities to participate in social interaction is one of the most fundamental preconditions for viable social organization.” In other words, for human groups, turn-taking  is a generic problem of interaction. Like any other such problem, this one must be solved through the mobilization of available local resources . With respect to turn-taking, one particularly important resource appears to be grammar (or “syntactic practices”). In both English and Japanese,  the turn-taking problem gets solved but differently in each case due to the availability of different grammatical  resources. And notice then that differences in the organization of turn-taking appear to have consequences for other practices of talk-in-interaction – ­specifically the deployment of so called “response tokens” (See Hayashi  and Yoon, this volume).

24

Jack Sidnell

Overview of the volume The chapters in this volume are divided into three thematically organized sections. In conclusion here I want to briefly point to some of the issues involved in creating such groupings in the first place. The chapters grouped together in the first section of the book share a focus on repair. Each involves a detailed consideration of the various ways in which particular kinds of interactional trouble are addressed in conversation. At the same time, the reader will find that these chapters deal with a wide range of quite varied phenomena. For instance, Fox  et al.’s analysis of self-repair  initiation in seven typological  and genetically distinct languages leads to a consideration of the influence of word length and syllable structure.  The examination by Egbert et al. of a species of other-initiation leads to a consideration of prosody  and pre-announcements . And though Wu ’s chapter begins with other-initiated  repair, it comes eventually to an examination of epistemic  issues implicated in talk-in-interaction. The point then is that within a given single domain a very wide range of practices are implicated. Indeed, it is quite impossible to deal solely with the phenomenon of repair, for instance, without at the same time considering the organization of turns and the sequencing of actions. So, while the chapters in each section address particular domains of talk-in-interaction, they do not attempt to reduce the phenomena they examine to that domain. The second section of the book considers responses, and, again, although there is clear thematic continuity across the chapters, each raises distinct issues and locates a particular dimension for comparative analysis. Here the prefer ences for contiguity  and agreement  described initially by Sacks  (1987) are particularly important as are issues of epistemic access. The chapters show that although there is much in common across languages in terms of the resources available for constructing responses, the particular grammatical  and lexical characteristics of these languages also result in considerable variability. Hayashi and Yoon compare not only different languages (English, Japanese, Korean) but also different approaches to social interaction. The authors conclude their chapter with a discussion of the ways in which the phenomenon they

Introduction

25

consider in Japanese and Korean has been analyzed in previous studies as an expression of cultural themes (e.g., “mutual dependency”), before offering an alternate account that emphasizes the particular syntactic practices and grammatical characteristics of these languages. Taken together the chapters in this section show that while responding clearly presents similar problems everywhere, interactional solutions are adapted and shaped by the available resources of particular languages. The chapters of the final section maintain a focus on languagespecific resources in the construction and sequencing of action investigating the use of language-specific resources to respond to assessments (Sorjonen  and Hakulinen, Sidnell) and to show relations between utterances that are not adjacent (Bolden). Sorjonen and Hakulinen  show that, in Finnish,  word order  does not code basic grammatical relations (as it does in English and many other languages) and can thus be deployed to index subtle distinctions of epistemic access, speaker stance  and interactional alignment . While there are roughly comparable practices in languages that use word-order to code grammatical  relations, these Finnish  practices of responding are intimately tied to the grammar of this language and thus ultimately unique to it.     The broad theme relating all the chapters, then, is how the generic problems of interaction among humans (coordination of opportunities to participate, fixing problems, responding in such a way as to convey an understanding of what the previous speaker has done, constructing a course of action in a sequence, etc.) are solved through the mobilization of specific local resources . Those resources include not only the grammar and lexicon of a particular language but also prosody  and the body. It is the mutual shaping of these resources, the structures of interaction and the production of coordinated courses of action that these chapters investigate. Acknowledgments I   thank Nick Enfield , Chuck Goodwin , Makoto Hayashi , Manny Schegloff,  and Tanya Stivers  for helpful written comments on an earlier version of this chapter.

26

Jack Sidnell

Notes 1 Variously termed descriptive, functional or cognitive linguistics . For a useful overview of language universals  that contrasts these two approaches see Comrie 1981. Some major landmarks in the descriptive/ functional tradition are, for instance, Li and Thompson (1981), Shopen  (1985). 2 So-called “non-configurational” languages which exhibit free word order provide a striking illustration of the way grammar shapes the practice of repetition. Thus Hale writes of Warlpiri: “sentences containing the same content words in different linear arrangements count as repetitions of one another” (Hale 1983:5). 3 While each language offers unique possibilities for social action, there are also clear points of convergence as well. Indeed in some cases it is possible to identify the very same practice in typologically and genetically unrelated languages. For instance, Stivers  (2004) shows that a practice involving the use of multiple sayings such as “no no no” to halt an ongoing course of action is found not only in English but also in Japanese  and Russian . 4 The chapters by Fox et al. and Rossano et al.   depart from the collection-based, qualitative CA approach described here and develop their analyses through the use of quantitative methods. For discussion of such methods applied to the materials of social interaction see these chapters as well Schegloff ’s commentary. 5 See Goffman  (1964) for an early juxtaposition of two research traditions: one in which features of context are correlated with the features of speech and another in which the focus is on the structures of social interaction themselves, that is, “the neglected situation.” 6 See, however, Egbert (2004) for an examination of some occasions in which just these issues come to the interactional surface. 7 As Makoto Hayashi  points out, CA studies do sometimes begin with action-types: Pomerantz  (1978) on compliments, Goodwin and Goodwin (1987) on assessments, and Drew (1984) on invitations are examples. These studies differ from those in cross-cultural pragmatics by taking the action-implementing turn as part of a larger organization of the turn at talk (i.e. the Goodwins’ study of assessments) or sequence of action (Pomerantz  and Drew). Emanuel Schegloff  notes that although the titles of the papers by Pomerantz  and Drew refer to compliments and invitations, the actual texts are about responses to compliments and invitations. 8 Schegloff  goes on to note “Further, the use of other-initiated  repair as one way of pre-indicating the imminent occurrence of disagreement

Introduction

27

(Schegloff et al.   1977: 380) suggests why uh huh’s and the like can be taken as indications of agreement with the speaker of an ongoing extended unit. For if disagreement were brewing, then opportunities to initiate repair would supply a ready vehicle for the display and potential deflection of that disagreement. Passing the opportunity to raise problems of understanding may be taken as indicating the absence of such problems. It may also be taken as indicating the absence of that which such problems might have portended – disagreement – and thus be taken as indications of agreement.” 9 As Schegloff  notes, “attention signals” is a gross under-specification. First because it is not clear why attention should be signaled on the occasion in which “uh huh” is used and not others in which it is not and second because it is not clear why “uh huh” should be used rather than some other signal such as gaze.

Part II

Repair and beyond

 2 Repetition in the initiation of repair Ruey-Jiuan Regina Wu

Introduction The organization of repair has received sustained interest in the study of language and social interaction over the past decades. Although, until recently, research in this area has based its claims primarily on English materials, there has been a growing interest, especially in the past ten years or so, in exploring how repair operates in languages other than English. This expanding body of research includes studies of German  (e.g., Egbert  1996, 1997b, 2004; Selting 1988, 1992, 1996; Uhmann 2001), Japanese (e.g., Fox  et al. 1996), Korean (e.g., Kim 1999a, 2001), Thai (e.g., Moerman  1977), and Mandarin  Chinese (e.g., Chui 1996; Tao  et al. 1999; Wu 2006; Zhang 1998), among others. Some of these studies have focused on the mechanisms of self- and other-initiation of repair in the languages examined (e.g., Chui 1996; Kim 1999a, 2001; Moerman 1977 ; Zhang 1998), while others have uncovered the linkages between repair and other aspects of interactional practice, such as prosody  (e.g., Selting  1996; Tao  et al. 1999) and bodily conduct (e.g., Egbert 1996), and still others have explored the relation between repair and syntax from a cross-linguistic perspective (e.g., Fox  et al. 1996). As part of this growing effort to understand the organization of repair across languages, this chapter investigates two repeat­formatted other-initiated  repair practices in Mandarin  conversation. Using the methodology of conversation analysis (CA), this study will show that the two Mandarin repair initiations under examination, like other-initiation of repair in English, serve not only to initiate repair but also as vehicles for accomplishing additional

32

Ruey-Jiuan Regina Wu

negatively valenced actions, such as displaying a stance  of disbelief or nonalignment . However, in further explicating the common sequential and activity contexts of these two practices, I will also describe a previously undescribed division of labor between repair initiations and demonstrate how the use of these two Mandarin practices is sensitive to two intertwining axes: the epistemic stance  of the speaker who initiates the repair and the sequential context and positioning of the initiation of repair. The analysis draws upon a corpus of approximately two hours of telephone conversations and seventeen hours of videotaped face-to‑face conversations among family members, friends, and acquaintances. The data were collected in mainland China , Taiwan, and the USA. In what follows, I will first briefly review previous work on ­other-initiation of repair and lay out a few initial observations regarding the use of the two Mandarin  practices under examination in clear-cut contexts of repair. I will then examine occasions in which they are used to implement actions beyond repair initiations and discuss how these uses can be related to the basic properties they exhibit as repair initiators.1 The candidate phenomenon: repeat-formatted other-initiation of repair In general, research on other-initiation of repair has been pursued along two different lines. The first line of inquiry (e.g., Curl 2005; Egbert  1996; Kim 1999a, 2001; Schegloff  2000b; Schegloff  et al. 1977; Zhang 1998) is concerned mainly with the internal organization of repair sequences, exploring issues such as the locus of repair initiation relative to the trouble source  turn as well as the repertoire of practices for the initiation or resolution of repair. The second line of inquiry (e.g., Couper-Kuhlen  1992; Drew  1997; Egbert 1997b, 2004; Kim 1999a; Robinson 2006; Schegloff  1997b; Selting  1988, 1992, 1996) focuses not so much on repair sequences per se, but rather on the larger sequential contexts in which such repair initiations and sequences figure, as well as their interactional uses in these contexts. One prominent theme along this second line of inquiry is the double-barreled nature of other-initiated  repair – that is, its

Repetition in the initiation of repair

33

capacity for, and frequent use in, implementing additional interactional projects while simultaneously serving to initiate repair. Egbert (1997b), for example, demonstrates how initiating repair in multiparty German conversation can be used strategically not only as an entry or exit device to such conversations but also as a display of affiliation among conversational co-participants. Selting  (1988, 1992, 1996) shows that prosodically marked other-initiation of repair in German can carry an emotive overtone of “astonishment” and indicate a problem of expectation on the part of the speaker. Drew  (1997) focuses on what he terms “open class” repair initiators in English (e.g., “Pardon?,” “Sorry?,” “What?”) and shows how their use can embody the speaker’s treatment of the turn being targeted as either topically or sequentially disjunctive or in some way inappropriate. And the strong connection between the use of other-initiated repair and the display of incipient nonalignment on the part of its speaker has also been documented in several studies (e.g., Heritage  1984: 339–344; Kim 1999a; Pomerantz  1984; Schegloff 1997b, 2007a; Schegloff et al. 1977  ). The following example, from Schegloff (2007a: 102–103) , is a case in point: (1)  Schegloff (2007a:102– 103) 1 Bee: a→ =[Why] whhat’sa mattuh with y- Yih sou[nd HA:PPY,] hh 2 Ava: [Nothing. ] 3 Ava: b→ u- I sound ha:p[py?] 4 Bee: c→ [Yee]uh. 5 (0.3) 6 Ava: d→ No:,

Here, Ava initiates repair (Arrow B) of Bee’s assessment (“Yih sound HA:PPY,” Arrow A) of Ava’s emotional state. To this repair initiation, Bee responds in the next turn (Arrow C) with a confirmation . Although this response by Bee suggests that she is treating Ava’s repair initiation, prima facie at least, as involving a hearing or understanding problem, Ava’s subsequent disconfirmation (Arrow D) nonetheless retroactively indicates that this prior repair initiation serves not merely to initiate a repair but to portend disagreement  as well. The present study pursues the second line of research and considers a similar phenomenon in Mandarin  conversation as that illustrated in Extract 1. An initial sense of the presence of a similar

34

Ruey-Jiuan Regina Wu

operation – and a further complication – involved in the use of repeat-formatted other-initiated  repair in Mandarin is evident in the following two excerpts, both paralleling Extract 1 closely: (2)  (CMC01–403A) 1 C:

ta yi xiaoshi:: fu ni duoshao, xianzai. she one hour pay you how:much now “How much does she pay you- for an hour:, now?”

2 W: a→ pianyi de le. sanshi. cheap NOM CRS thirty “Cheap. Thirty dollars.” 3 C: b→ yi ge xiaoshi sanshi? one C hour thirty “Thirty dollars for an hour?” 4 W: c→ (uh [huh) PRT PRT “(uh [huh.]” 5 C: d→ [Bu pianyi. N   cheap [ “Not cheap.” (3)  (CMC 05–06, 00:20, audio 315a) 1 L: a→ hai, hebei jiu mei shenme hao difang. (sigh) (province) just N what good place “((sighs)) There are not many good places in Hebei.” 2 M: b→ hebei a?= (province) PRT “Hebei A?”= 3 L: c→ =uh. PRT = “Yeah.” 4

(0.5)

5 M: d→ langfang hai xing. (place) still okay “Langfang is okay.”

As with Extract 1, in each of these excerpts, a speaker initiates repair (Arrow B) by repeating part of the prior assessment by another (Arrow A); and in each excerpt, following a recipient confirmation (Arrow C ), the repeat speaker moves to launch a

Repetition in the initiation of repair

35

disagreeing  response (Arrow D): In Extract 2, the speaker delivers a straightforward negation of the prior assessment; and in Extract 3 the speaker introduces an exception to the recipient’s just-proffered negative evaluation of the province by claiming that one of its ­cities – i.e., Langfang – is “okay.” Despite the similarity, it can be noted, however, that the repair initiations in these extracts are not in entirely the same format: Whereas the repeat in Extract 3 is suffixed with the particle a, 2 the one in Extract 2 is not. What Extracts 2 and 3 suggest, then, is both a commonality and variation between English and Mandarin . On the one hand, as in English conversation, repeat-formatted other-initiation of repair in Mandarin apparently can serve not only to initiate repair but also as a vehicle for accomplishing negatively valenced actions such as projecting disagreement . On the other hand, however, in initiating a repair through the use of repeat to accomplish such an action, a Mandarin speaker will have to select between at least two alternative practices: a question-intoned repeat3 and a repeat suffixed with the final particle a. The second of these practices appears to have no direct analogue in English.4 As will be shown below, the selection between these two Mandarin  practices in implementing negatively valenced projects is not random; rather, the division of labor appears to be consistent with, and may in fact be carried over from, the basic meanings they index when serving as straightforward repair initiators. To demonstrate such a connection, let us turn first to a brief overview of the use of these practices in clear-cut contexts of repair initiation. Question-intoned repeats and a-suffixed repeats in clear-cut contexts of repair As straightforward repair initiations, question-intoned repeats and repeats suffixed with a commonly adumbrate different problems their speakers are having with respect to the trouble source  turns targeted and frequently engender different types of responses from their recipients. Repeats suffixed with a are commonly heard as confirmation  questions, i.e. the use of this format makes a recipient’s confirmation or disconfirmation relevant in next position. This is illustrated respectively in Extracts 4 and 5, both from a conversation recorded in

36

Ruey-Jiuan Regina Wu

Beijing between two acquaintances. Immediately prior to Extract 4, M has just asserted that life in Beijing is boring. In Extract 5, the topic has turned to a neighborhood where M used to live. (4)  (CMC 05–05, 00:40, audio 243a) 1 L:

suibian ni. as:wish you “Whatever.”

2

(0.3)

3 L: a→ chabuduo zhao ren jia le de le. almost find person marry ASP get CRS “(I’ll) just get married to some guy.” 4

(0.5)

5 M: b→ zhao ren jia le de le a?= find person marry ASP get CRS PRT “Just get married to some guy A?”= 6 L: c→ =uh.= PRT = “Yeah.”= 7 M: d→ =wo zher hen duo ren dasuan qu ne. I here very many person plan marry PRT = “I know lots of guys who are ready to marry.” (5)  (CMC 05–03, 01:45, audio 130a) 1 M:

bu, yuanlai women jia zhu nar? no originally we home live where “No, where did we live before?”

2 M:

we’re gunna< go: to: 21 Spokane ’n another week or 22 so so (.) I’d like tuh start lookin.’ 23 (0.7) 24 S: °(Oh)-° Okay. I:’ll=uh I’ll think about it.

After the son merely acknowledges (but arguably does not agree  with) his mother’s reminder/request (Lindström 1997), “Mm: that’s

Repairing reference

117

right.” (line 4), the mother pursues his agreement  by beginning to account for needing the list (line 6; see the mother’s later pursuit at lines 20–22, especially her repeat of “we’re gunna go to . . .”). The mother’s request contains an implicit potential offer to buy “big” gifts (lines 2–3, “small duh big”), and, at line 7, the son interrupts her by saying that a large gift has already been agreed upon, “Well you got thuh big one.”. Here, the son refers to an as-of-yet-­unnamed “big” present with “the big one” (which turns out to be an airplane ticket). Grammatical ly, “the big one” is the direct object in this trouble-source unit and is a placeholder that is specified in the repair operation with “thee: =uh: ticket.” (line 13). In other words, the reference in the trouble-source unit is a more “granular” reference form (Schegloff  2000a) than the reference form in the repair operation. In this case, “the big one” is treated as being underspecified by the dad, who initiates repair, but not by the mom, who displays that she understands “the big one” by beginning to respond at line 11, “Well-,” which is comparable to how she re-begins at line 17, “We: ll”. At line 10, the dad initiates repair with “Wha:t.” (which is produced with final-falling intonation). The son responds by specifying “the big one” with “thee:=uh: ticket” (line 13). There is evidence that the dad’s “Wha:t.” constitutes repair initiation. At line 16, his “Oh,” which sequentially is a third-positioned action relative to the repair sequence (Schegloff 2007b ), claims a change of state from unknowing to knowing (Heritage 1984). And, at line 18, the dad responds to the son’s “Well you got thuh big one” by partially discounting it with a (delayed and) dismissively produced “°#Yeah.#°”, thereby treating the “Wha:t.” as having delayed the progression of the son’s claim in order to better understand it. (The mom similarly dismisses the son’s claim at lines 11, 17, and 19.) For a second English example, see Extract 8, which is drawn twenty-seven minutes into a telephone call between Sue and her boyfriend Max. Sue has initiated all of the major topics and has done a majority of the talking. (8)  Phone (CF:6507) 01 S: 02 03 M: 04 05→a S:

.hhh Anyway there’s nothing on you(r)- more on your mind is there. (0.4) Uh:not really. h=Oh:.=hh (.) .hhh (and) a:re- (.) are all our

118

Maria Egbert et al.

06→a our conversations gunna end up like this, 07 (.) 08 S: On thu[h phone,-] 09→b M: [Wha:t. ] 10→c S: .hhhhhh (0.2) I mea[n] 11 M: [(O]h./No.) h= 12→c S: =.hhh (I’ll=say) something:you know pretty:=h (.) 13→c >somewhat-< (.) you know pretty good that hap14→c you know happened (to you then you) say something 15 .hhhhhhhhhh[h h-=[or ] 16 M: [.pt=We:l[l >you kno]w< you gotta 17 understand th’t I’m:kind=of a little ti:red,

Although Sue’s action at lines 1–2 solicits additional conversational topics (Button  and Casey  1984), it is grammatical ly designed to prefer  (and thus presume) a “No”-response. Given that Max has not previously initiated any major topic , Sue’s turn can be heard as an interpersonal complaint  regarding Max’s lack of enthusiasm or interest in terms of talking to her (for in-depth research on complaints, cf. Heinemann and Traverso, forthcoming; Yoon 2006). In the wake of Max’s (delayed) agreement, “ U h: not really” (line 4), Sue asks/accuses: “are all our conversations gunna end up like this” (lines 5–6; note the extreme-case formulation “all our conversations”; Pomerantz  1986). Grammatically, Sue’s “like this” is an adjunct (in the form of an adverbial expression of manner) and refers to the immediately prior sequence of action (at lines 1–4). In response, Max initiates repair with “Wha:t.” with final-falling intonation (line 9), to which Sue responds by specifying what she “meant” (see line 10, “I mean”) by “this.”. That is, she specifies her complaint by providing a characterization of how the call has “ended up” (lines 12–14). Max orients to his “Wha:t.” as having initiated repair on Sue’s turn and thus as having delayed its progression to better understand it, by subsequently responding to it by defending himself: “We:ll >you know< you gotta understand th’t I’m: kind=of a little ti:red” (lines 16–17). For a final English example, see Extract 9, which is drawn from an American  telephone call between two college students, Carla and Rich. (9)  Humorous [UTCL:D08] 01 02

C: Peet’s a:lways there. ((i.e. at the apartment)) R: °Y°:eah he is.=

Repairing reference

119

03 C: =B’t your never there. ((i.e. at the apartment)) 04 (1.6) C: (No:pe) 05 06 (1.0) R: Hh hh [hh °hh° ] 07 08 C: [Where are] you. on thuh pho:ne, R: .hhhh ((laugh-relevant inbreath)) 09 C: ↑What’re you↓ sa:ying? 10 R: .huh ((laugh)) 11 12 (0.4) C: Te:ll me:. 13 R: .hhh hh=heh (0.5) .huh (0.2) No:thing I’m 14 15→a j’st- I find it humorous. 16→b C: ↑Wha:t.↓ 17→c R: Just (tw)- listening duh you. C: What I sa:id? 18 R: Yes. 19 C: Peet’s always the:re, 20 R: Yes. an’ I’m n:ever ‘ere. 21 C: So where are you. 22

When Rich fails to respond (at line 4) to Carla’s assertion that Rich is never at his apartment, “B’t your never there.” (line 3), Carla produces a candidate response (for Rich) that aligns with her position: “No:pe” (line 5). After a long gap of silence at line 6, Rich laughs, “Hh hh hh °hh°” (line 7), and continues laughing at lines 9, 11, and 14. Perhaps because of Rich’s displayed lack of attention (i.e. his failure to respond at line 4), or because Rich’s laughter (at line 7) is disjointed from prior talk (i.e. lines 1–5, which neither participant oriented to as being funny), Carla begins to suspect – evidenced by her chastising question “↑What’re you↓ sa:ying?” (line 10) and its pursuit “Te:ll me:.” (line 13) – that Rich might be talking to, and laughing about something with, his roommate Peet. Rich ultimately answers Carla at lines 14–15: “No:thing I’m j’st- I find it humorous.”. Grammatically, Rich’s “it” is the direct object of this trouble-source unit. As a pro-form, “it” is analyzable as being underspecified relative to its full-reference form, which is the source of what Rich finds humorous. Indeed, as the repair operation to Carla’s “↑Wha:t.↓” (line 16), Rich responds by specifying the humor source: “Just (tw)- listening duh you.” (line 17). To review Extracts 4–9, “Was denn”/“Was.”/“What.” are produced with either clearly falling intonation (in the case of “Was.” and “What.”) or with non-strong-rising intonation (in the case of

120

Maria Egbert et al.

“Was denn”), which differentiates them from open-class  forms (Drew  1997). The trouble sources are pro-terms or indexical  expressions (e.g., “it”, “that”, “this”; in the data), adverbial expressions of manner (e.g., “like this”), and placeholder nouns (e.g., “the big one”) that refer to “things”, which include not only realworld physical objects (e.g., photo album, parking token, airplane ticket), but also more abstract objects (e.g., a source of humor, a ­characterization of a telephone call, and a habitual physical action). “Was denn”/“Was.”/“What.” implement repair initiation by halting the forward progression of a prior course of action (regarding progressivity , see Lerner  1996) in order to deal with a particular type of referential trouble, i.e. its claimed underspecification. “Was denn”/“Was.”/“What.” are responded to with specifications of referring expressions. The referring expression in the trouble sources may or may not have been used in sequentially proximate prior environments, including the prior turn.  Further evidence: The case of “doubles ” So far, we have examined formal features of the repair initiators “Was denn”/“Was.”/“What.” and their uptake by co-participants. These results are strengthened by observations of a special kind of context in which “Was denn”/“Was.”/“What.” can occur, and that is when they are produced in the same turn with another repairinitiation action. In such multi-unit turns (called “doubles”  by Schegloff  et al. 1977), the second repair-initiation attempt is typically “stronger” (more specific) than the first in terms of locating the trouble source . In doubles, “ Was denn”/“Was.”/“What.” are placed first, followed by stronger confirmation  requests  of candidate specifications of trouble-source referring expressions. For a German example, see Extract 10, which is drawn from a dinnertime interaction. (10)  WEIN [Fischessen_1.31.10] (simplified) 01→a A: Ptz ptzptzptz ºderº is lecker. der g- der Pts pts pts ºitDEMº is tasty. that- p- that Pts pts pts ºthat oneº’s tasty. that- p- I 02→a schmeckt mir am besten von allen. tastes to me the best of all. like that one best of all of them.

Repairing reference

121

03 B: Ja? Yeah? A: [Mh ºhmmº] 04 [Uh ºhumº] [ ] 05→b D: [Was denn. ] [What PRT. ] [ “What.” ]

[Dieser wein hier? ] [This wine here? ] [This wine here? ] [ ]

06→c A: [((Points at wine)) ] 07→c A: mh ºhmmº uh ºhumº 08 (0.5) ((D starts turning bottle to read label))

As context, two minutes prior to Extract 10, the interactants discussed what they would drink next. Annette had suggested continuing to drink another bottle of the same white wine, whereas Sybille had suggested a bottle of red wine. In Extract 10, at lines 1–2, Annette produces an extreme-positive assessment (Auer  and Uhmann 1982; Pomerantz  1984) of the same bottle of white wine that she had suggested earlier: “Der schmeckt mir am besten von allen.”/“I like that one best of all of them.”. Annette produces a slight head nod toward the bottle of wine during the production of “der”/“that one”. Grammatical ly (in German), Annette’s “der”/“that one” is the subject. Note that the German “der” has the same form as the article “der” in “der Wein”. Annette’s “der”/ “that one” relies on a demonstrative  determiner to refer to a referent indexical ly (Hanks  1992). In spoken German, this form is often used as a pronoun (i.e. instead of “er”) or as a demonstrative (i.e. instead of “dieser”). German grammar thus provides for an indexical  form which is not easily translated directly into English. At line 5, David treats the referring expression as being underspecified by initiating repair with “Was denn” (line 5), but then immediately following it up with a request for confirmation of a candidate referent: “Dieser wein hier?”/“this wine here?”. David’s candidate solution, “Dieser wein hier?” is grammatically congruent with the pronoun “der” in the trouble-source turn in that both are morphologically marked for the grammatical  gender  masculine and case-marked for the nominative. In moving from “Was denn” to a

122

Maria Egbert et al.

candidate solution, David displays an orientation to his own “Was denn” as doing the job of targeting a “thing” referent in need of specification to him. Note that Annette similarly orients to the function of “Was denn”; that is, immediately after its production, and before David produces his candidate solution, she specifies her “der”/“that one” by pointing at a specific bottle of wine. Thus, in doubles with “Was denn”, the second repair-initiator offers a candidate specification of the troublesome indexical . For an English example, see Extract 11, which is drawn from a phone conversation between Joe and his mother. Prior to this fragment, Joe and Marsha have discussed a variety of topics dealing with Marsha. Deep into the call, Joe asserts that “he can’t talk very long” (line 1). With a delayed “by thuh way” (line 3), Joe orients to his assertion as being “misplaced” relative to prior talk (Schegloff  1984, 1987c), and, at lines 6–7, Joe returns to prior talk by producing a summary assessment of it. (11)  Stanford (Marcia and Joe) 01 J: 02 03 04 J: 05 06→a J: 07 08 09→b M: 10→c J: 11 12 J: 13 14 M: 15

Oh we can’t talk very long ‘cause I=‘as jus’ talkin’ tuh gra’ma. (0.4) by thuh way. (0.4) But this sounds like=ay- very interesting situation you’re in you’re in here. (0.2) What. Mi::ne? Ye:ah. (0.5) With Stanford research an’ everything. (1.4) Oh I myself am auditioning for Ca:melot this June.

In the trouble-source unit (lines 6–7), Joe’s “this”, in combination with “a very interesting situation”, alludes to, and thus is analyzable as being underspecified relative to, a particular “situation”. After a slight pause (line 8), Marcia initiates repair with “What.” (line 9), targeting “this” in the trouble-source turn, and then immediately follows it up with a request for confirmation of a candidate domain of “situation”, “Mi::ne?”, which attempts to clarify the nature of the “interesting situation” being referred to. Here, Marcia displays

Repairing reference

123

an orientation to her own “What.” as doing the job of targeting an underspecified referring expression. Joe treats Marcia’s request for confirmation  as such by first confirming, “Ye:ah.”, and then by supplying a particular situation: “With Stanford research an’ everything.” (line 12).  A contrasting practice with a similar form: “Für was.”/“For w hat.” The function of “Was denn”/“Was.”/“What.” can be put into relief by examining a different, but allied, repair-initiation action: “Für was.”/“For what.” The comparison of these two different repair‑initiation actions is presented to shed further light on the unique function of “Was denn”/“Was.”/“What.”. In Extracts 4–9, repair-related trouble was associated with referring expressions contained in trouble-source units themselves. Alternatively, we will now discuss one other way in which repair-related trouble can be associated with referents. In these cases, none of the referring expressions in the trouble-source action are treated as presenting “trouble”, but, rather, the trouble-source action presupposes a referent (e.g., an elaboration of a direct object), and the knowledge / understanding of this referent is necessary to adequately “grasp” the trouble-source action. This alternative type of reference trouble can be dealt with by an alternative repair-initiation practice, “Für was.”/“For what.”, which prompts (Lerner  2004: 169) the prior speaker to elaborate a prior utterance. The trouble-source utterance is syntactically complete, and the repair operation in the form of an increment provides pragmatic completion. For a German example, see Extract 12. The trouble-source unit is Oma’s announcement: “ Ich hab’ jetz’ wieder ‘nen neues rezept gekriecht”/“I have now gotten another prescription again” (line 4). (12)  Für was (Oregon 1A_1:38) 01 M: =.hh Wie oft musst’n zur krankengymnastik. =.hh How often must+you to+the physical therapy. =.hh How often do you have to go to physical therapy. 02 O: Äh dreimal die woche °ne.° Uh three times the week °PRT.° Uh three times per week you know.

124

Maria Egbert et al.

03 (.) 04a→ O: Ich hab’ jetz’ wieder ‘nen neues rezept gekriecht, I have now again a new prescription got, I have now gotten another prescription again, 05b→ M: Für was. For what. (0.5) M: Ach so f[ür die gymnastik. 06 Oh i see f[or the therapy. [ 07c→ O:

[Für die krankengymnastik. [For the physical therapy.

Grammatically, Oma’s “rezept”/“prescription”, is the direct object. Based on the analysis of Extracts 4–9 (above), to initiate repair with “Was denn”/“Was.” would be to claim that “rezept”/“prescription” is itself underspecified. However, Markus initiates repair with “Für was.”/“For what.”, which alternatively claims that the troublesource unit elided a referring expression necessary for his understanding and solicits an increment  (Schegloff  1996b; 2001) to the trouble-source unit. In this case, Markus’s “Für was.”/“For what.” is responded to with an elaboration of the direct object (i.e. the source mandating the prescription). Both participants orient to this as the repairable . Oma responds by producing a grammatical  increment  to her trouble-source question that supplies a new referent: “die krankengymnastik”/“the physical therapy” (line 7). Prior to Oma’s response, Markus himself claims that he has resolved the trouble with “Ach so”/“Oh I see” (Heritage  1984; Golato  and Betz  2008), and then repairs the trouble with “für die gymnastik”/“for the therapy”, which displays his understanding of his “Für was.”/“For what.” as seeking an elided referring expression. For an English example, see Extract 13, in which the troublesource unit is Ada’s question at line 6. (13)  Loan (CH:4365) 01 A: .mtch So:u- (.) I=s- I:was thinking 02 o[f sen]ding mom an’ da:d (.) uh:m about a 03 M: [.hhh] 04 A: hundred bucks. 05 (.) 06a→ A: .hhh Does that sound reasonable to you,=

Repairing reference

125

07b→ M: =For what. 08 (0.9) 09c→ A: Off of what I owe them.

Grammatically, Ada’s “that” (line 6) is the subject. Based on the analysis of Extracts 4–9 (above), to initiate repair with “What.” would be to claim that “that” is itself underspecified and seek its specification, such as a recharacterization of Ada’s proposal at lines 1–4. However, Max initiates repair with “For what.”, which alternatively claims that the trouble-source unit elided a referring expression necessary for his understanding and solicits an increment  to the trouble-source unit. Similar to Extract 12 (above), Max’s “For what.” seeks an elaboration of the subject (i.e. Ada’s reason for sending the money). This is oriented to by Ada, who responds by supplying her reason, “Off of what I owe them.” (line 9). (The) Difference(s) between “Was.” and “Was denn” As mentioned in the introduction, there is an easily describable difference in form between “Was denn” and “Was.”. “Was denn” contains the particle “denn” and is delivered with an intonation contour that ranges from falling to slightly rising. “Was.” does not contain a particle and is always delivered with clear final-falling intonation. Given these form features, German “Was.” is akin to English “What.”. However, distributionally, when used to initiate repair, “Was denn” (75 percent of corpus) occurs much more frequently than “Was.” (25 percent of corpus). Furthermore, and perhaps more importantly, similar to English “What.”, in German only “Was denn” is used to accomplish a specific nonrepair function, namely a “go-ahead” action to pre-sequence-initiating actions (Schegloff 2007), such as a summons, preannouncement , and story/ joke preface (see examples below). In our data, we found no cases of “Was.” used as “go-ahead” action, suggesting that “Was.” is reserved for repair-initiation actions only. Such distributional and action-based differences suggest that “Was denn” (1) is the more generic, common, or default form relative to “Was.”; and (2) may be the more appropriate equivalent of English “What.”. What follows are two examples of pre-announcements (Terasaki  1976) that are relevantly responded to, and progressed, by “Was denn”/“What.”, respectively. As Schegloff  (1997b) argues, there is

126

Maria Egbert et al.

not a one-to-one correspondence between practices and actions, and, in the following cases, “Was denn”/“What.” are not used to implement repair initiation. For a German example, see Extract 14, from an interaction between two adults and two children, with one of the children beginning to tell a story. (14)  Weisstduwas (SandL_T1_0.22.48) 01→a S: Un weißt du (.) wa:as, An you know (.) wha:t, 02→b A: Was denn, What PRT, What. 03→c S: Und die .hh (ni.hhna) .H hat(te) auch And the .hh ((name .hh)) .H has/had also And .hh ni.hhna has also once 04 mal ihre ((2 syll.)).H[HH PRT her ((2 syll.)).H[HH (had) her ((2 syll.)).H[HH [ 05 A:

[Ja? [Yes? [Really?

Extract 15 is drawn from an English telephone conversation between two sisters, Kathy and Janet, the former of whom is planning a vacation to Cabo San Lucas and has been searching for reasonably priced airline tickets. (15)  Cabo (Sister Calls.2) 01→a K: An’ then uhm guess what. 02→b J: What. 03→c K: ’Cause I called ‘er because 04 .hh Kaytlen fo:und uh::m (0.2) 05 things for a hundred an’ fifty. 06 to go duh Cab[o.] 07 J: [.H]HHH She di:d?

In both Extracts 14–15 (above), line 1 involves the first part of a preannouncement  sequence, which initiates a course of action and solicits a response. Similar to many preannouncement  sequences, these (see “Arrow A”) project the presence of announcables without

Repairing reference

127

explicating them. Here, they are alluded to by “Was denn,” and “What.”, respectively, which refer to actions or events. These preactions function to frame their projected announcements as “news” for recipients (Terasaki 1976) and make use of underspecified referents (actions or events in the lives of the interactants) to which the prefer red response “Was denn,”/“What.” then engenders the actual announcements. At line 2 in both examples, “Was denn,”/“What.” functions as a second-part “go-ahead”, which is oriented to as such by pre-announcers when they ­proceed to deliver their announcements (lines 3–4 and 3–6, respectively), thereby explicating the referring expression. Rather than halting the progressivity  of the  pre-announcement  sequence, “Was denn,”/“What.” propels the course of action. Extracts 14–15 were presented for two reasons. First, the interactional achievement of “Was denn”/“What.” as second pair parts of a pre-announcement sequence is different from initiating repair, while they share the feature that they request more specification. In response to both functions, a reference is “unpacked”. Second, our preceding comparison of pre-announcement  go‑aheads and repair initiators sheds light on a hypothesis of how “Was denn”/“Was.”/“What.” may differ in their repair initiation function. In German, relative to “Was.”, “Was denn” is the more default repair-initiation form, and it may be the case that “Was.” is doing something unique. This aligns  with Thurmair’s (1991: 382) observation that German w-Fragen (wh-questions) typically contain the modal particle “denn” and that such questions without “denn” are marked. All of this, though, begs the question of what “denn”’s uniqueness might be in the context of repair-initiation. Because we have too little data to pursue this hypothesis further, we offer some speculative possibilities. An initial way of getting at differences between “Was denn” and “Was.” is to examine aspects in which they do not differ. Our limited data suggest no differences in terms of idiolectal and dialectal factors, nor medium (i.e. telephone vs. co-present interaction), nor number of participants (i.e. two or more), nor distance between the trouble-source referring expression and the reference form to which it (sometimes) connects back. There also seems to be no difference in prefer ence organization, i.e. “Was denn” and “Was.” can target trouble-source units which are preferred or dispreferred .

128

Maria Egbert et al.

Regarding this last observation, however, relative to “Was denn” both native-speaker authors of this chapter (Maria Egbert and Andrea Golato) perceive “Was.” as sounding more “terse”. Their vernacular understanding is also expressed by research in linguistics , which reports that “denn” makes utterances friendlier and possibly less blunt (Durrell 1992; Helbig 1988: 107; Hentschel and Weydt 1983: 266; Schwitalla 2003: 153). One subtle difference might exist at the sequential level. In each of our six cases of “Was”., the trouble source is always the unit immediately prior to “Was.”. In our fifteen cases of “Was denn”, this is so for thirteen cases, but there are two cases in which the trouble-source unit is farther removed. Thus, it is possible that the effectiveness of “Was.” relies on a more proximate relationship between itself and the trouble-source unit, whereas “Was denn” may remain effective given a more distant sequential relationship. We speculate that the particle “denn” serves to claim that repairinitiators have a general sense of intersubjectivity amid the need for reference specification. Such a sign of general intersubjectivity would not be necessary with “Was.” because the trouble source  is immediately prior. Although more research on actual conversation is necessary, the aforementioned speculation is in line with linguistic descriptions of the function of the particle “denn”. When “denn” is used in questions, linguists  have argued that it indicates that such questions refer to immediately prior talk (Dittmann 1980; Franck  1980: 222; Helbig 1988: 107; Thurmair 1989: 164, 166), and thus are not asked “aus heiterem Himmel” (“out of the blue”) (Zifonun et al. 1997: 1219). The particle “denn” is also said to target things that are known, or assumed to be known, by hearers (Helbig 1988: 107). Along these lines, Franck (1980: 225) argued that “denn” is frequently used in utterances that request further explanation before its speakers are able to respond to the content of the utterance (i.e. in repairs). Discussion Given the important relevance of referencing to intersubjectivity, it makes sense that the organization of repair provides conversationalists with specific practices for repairing reference-related

Repairing reference

129

troubles. Lerner (2004: 180–181, footnote 25) noted that there are multiple practices with which speakers can deal with “inadequate or insufficiently complete or a not yet response-ready turn that has reached possible completion.” Our small collection of thirty-eight instances from a corpus of ninety-five hours suggests that there is a specific category  of repair initiation designed to request specification of “thing”-referring expressions which the repair-initiating speaker claims to be underspecified and thus inhibiting full understanding. In German and English, this can be realized with “Was denn”/“Was.”/“What.” produced with nonrising intonation. This chapter extends prior research on German (Selting  1987a, 1987b, 1987c, 1988, 1992) and American English  (Schegloff  et al. 1977; Schegloff  1997b) by comparatively analyzing one particular type of repair initiation that requests  the specification of “thing”referents that are somehow embodied in the components of ­trouble-source units themselves. This practice was implemented by “Was denn”/“Was.” in German and “What.” in English, produced with either final-falling or slight (but not strong)-rising final intonation. Future research needs to examine other types of troublesource specific repair initiators, including those that target person, place, and time-referring expressions (such as “Wo.”/“Where.”, and “Wann.”/“When.”) (See Schegloff  et al. 1977, for comments on English “Where”, see also Lerner  2004). For now, we consider the term “thing” (as in “thing”-referent) to be an analytic one. In the data, “things” include not only realworld physical objects (e.g., photo album, parking token, airplane ticket), but also more abstract objects (e.g., a source of humor, a characterization of a telephone call, a habitual physical action, and an action/assertion). Future research needs to explore in more detail how members orient to persons, places, times, and “things”. As argued most forcefully by Selting  (1987a, 1987b, 1987c, 1988, 1992), our analysis reiterates that intonation is an important, constitutive feature of practices of other-initiated  repair. As suggested by our analysis of “Was denn”, when languages provide particles, they too can be constitutive features of repair initiation. Relative to the distribution of previously analyzed classes of other-­initiated  repair, such as open-class  initiators (Drew  1997), rising‑intoned interrogative words (e.g., “Who?”, “When?”, “Where?”), and various types of repeats and reformulation of trouble

130

Maria Egbert et al.

sources (Sacks  and Schegloff  1979), “Was denn”/“Was.”/“What.” repair initiations seem to be rare. This rarity might be explained by the presence of a set of first-order conversational practices and prefer ences regarding reference that obviate the relevance of “Was denn”/“Was.”/“What.”. For example, the preference for self­correction  (Schegloff  et al. 1977) and the associated practice of “try-marking” (Sacks  and Schegloff  1979) tend to preclude the need for “others” to initiate repair, for example, through iterations of trouble-source unit completion, recipient silence, and same-speaker continuance (Heritage  2007). Furthermore, there is a preference for using recognitional  reference forms (at least when referring to persons; Sacks and Schegloff 1979), which, in our data, are not targeted by “Was denn”/“Was.”/“What.”. Finally, there are practices regarding the use of locally initial and subsequent reference forms that promote intersubjectivity (Schegloff  1996a). Thus, “Was denn”/“Was.”/“What.” are only “needed” when other conversational safeguards have “failed”. Our findings add to the growing evidence that at least the context-free, structural organization of repair outlined first for English (Schegloff  et al. 1977; see also Schegloff  1992a), and replicated for Thai (Moerman  1977) and for German (Egbert  1996), seems to be universal . Our findings further suggest that different languages (in our case, German and English) may contain similar linguistic and intonational resources (in our case, the lexical item “Was.”/“What.” produced with falling intonation) to accomplish similar repair-initiation actions. Method ological issues in conducting a cross-linguistic analysis One conversation analytic goal is to analyze how linguistic resources are used in talk-in-interaction to achieve social actions, thus contributing to a theory  of social order. Since languages provide different resources, cross-linguistic research contributes to the theoretical question of what dimensions of human sociality are universal  and what are specific to particular linguistic communities (see the introduction to this volume, see also the collection in Enfield  and Stivers  2007, and here in particular, Hanks  2007 and Levinson  2007). Given this theoretical relevance, we would like to make transparent how method ological decisions and procedures in conducting a comparative analysis may influence the analysis

Repairing reference

131

and thus potentially impact the theoretical development. We are addressing how the role of English as data and as a lingua franca in the conversation-analysis community may shape our findings. Similar to many other comparative studies, the present chapter analyzes the commonalities and differences of one phenomenon across two languages. All cross-linguistic studies on repair have taken English as the basis for comparison, for example, Moerman ’s (1977) analysis of repair in Thai, Egbert ’s (1996, 2002) work on repair in German, Kim’s (1993) study of generalized other-initiated  repair in Korean , or Sidnell ’s (2007a) analysis of repairing person reference in a small Caribbean community. When doing so, it may be tempting to look for those formats that have been found for English, yet it is vital that the researcher remain open for all kinds of formats, including ones not attested in English. 5 Similarly, it may be tempting to compare items across languages based on form congruities (e.g., downward intoned German “Was.” and downward intoned English “What.”). However, the present chapter has shown that more can be gained by comparing functions of utterances, i.e. actions across languages. Similarly, our analysis cautions against translating utterances based on (perceived) phonological or lexical correspondences. For example, based on having similar lexical forms, linguistic meanings, and prosodic shapes, one might be biased toward translating the German “Was.” as the English “What.”. In contrast, our analysis suggests that the German “Was denn” is a more appropriate translation because it and the English “What.” are more similar in terms of the actions they accomplish. That is, unlike the German “Was.”, the German “Was denn” and the English “What.” accomplish both repair initiation and goahead responses to pre-sequences. In sum, we suggest that translations be based on an analysis of action, sequential placement, and function rather than on perceived form congruities (for a similar argument, see Betz  and Golato  2008). Many comparative studies show that it is difficult or impossible to accurately translate interactional data into English. Even when a morpheme-by-morpheme gloss is provided with the transcript, the original language is not fully represented. One way that a reader of English can be made aware of this dilemma is by leaving words such as particles in the original language within the English rendering (Sorjonen  2001). This works well for highlighting the focal phenomenon, yet it would render a transcript illegible if used for all

132

Maria Egbert et al.

items which lack an appropriate translation. Descriptions of what the English translation does not capture can be cumbersome for the reader, and the researcher is forced to find a balance between focusing on the relevant features of the phenomenon and to what level the nature of the data can be sustained.   Acknowledgments All three authors contributed equally to this research. We thank Emma Betz for helping with transcription and bibliographic work and Christopher Stewart for creating the PRAAT files. Additionally, we thank Emma Betz, Trine Heinemann, Anna Lindström, MarjaLena Sorjonen, and Jakob Steensig for their feedback on individual data segments. We are grateful to Emanuel A. Schegloff for feedback on an earlier draft and to Jack Sidnell, the Editor, for both feedback and support. All remaining errors are of course our own. Notes 1 “Repair” is defined as the mechanism by which trouble in speaking, hearing, and understanding is claimed and resolved (Schegloff et al.   1977). 2 We intentionally avoid the terms “locally subsequent” (vs. locally initial) and “non-recognitional” (vs. recognitional  ) because they have only been empirically verified with regard to person-reference, and this chapter deals with thing-reference (but see Golato  2005 for a discussion of locally initial and locally subsequent forms for thing-references in German). 3 Here and in other data samples we chose to summarize stretches of the interaction which would have been too long to display. 4 We would like to thank Emanuel A. Schegloff  for drawing our attention to designedly underspecified references. 5 The idea that our knowledge of talk-in-interaction in most languages may be shaped by what is already described for English is due to the history of the field, whose first-generation researchers have reported profound results mainly on English. Therefore, taking English as a basis of comparison is more a decision due to the history of science than to scientific reasoning.

Part III

Aspects of response

5  Projecting nonalignment in conversation Anna Lindström

Projecting nonalignment Early conversation-analysis (CA) research established that ­nonaligning actions are typically delayed and that turn-initial  objects can project nonalignment. These features provide a resource for recipients, allowing them to anticipate nonaligning responses and to revise their prior action before such a response is realized in this way avoiding the occurrence of rejection or disagreement . Sacks  (1987) showed that in conversation there is an association between agreement  and contiguity  on the one hand, and disagreement  and noncontiguity on the other. Agreeing  utterances tend to be contiguous while disagreeing  utterances are noncontiguous. Consider Extract 1. (1)  (Sacks  1987: 58) 01 A: 02 B: 03 04

Yuh comin down early? Well, I got a lot of things to do before gettin cleared up tomorrow. I don’t know. I w- probably won’t be too early.

As Sacks noted, B’s disagreement – “ probably won’t be too early” – is not only weak, additionally, and more important for this study, it is not produced until the end of B’s turn. The disagreement  is prefaced by “Well,” an account “ I got a lot of things to do before gettin cleared up tomorrow,” a hedge “I don’t know,” and a selfinterruption “I w-” all of which serve to delay its production. The distinctive features of agreeing and disagreeing turns were explored

136

Anna Lindström

further by Pomerantz  (1975, 1978, 1984). Focusing on how predisagreements  shape sequence trajectories she noted that: a gap after a first pair part to which agreement /disagreement is relevant may function as an “unstated” or “as yet unstated” disagreement . That is, if a recipient delays in initiating his talk post the completion of a turn to which agreement /disagreement is relevant, that delay-gap may be disagreement implicative for his [sic] co-participant. As such, that ­co-participant may resume talk post a gap with talk which displays a sensitivity toward a discrepancy between the parties.  (Pomerantz  1975: 79)

This is illustrated in the following Extract 2. (2)  (SBL:3.1.-8). (Pomerantz  1975:79, line numbers added) 01 B: 02 03 B: 04 A:

. . an’ that’s not an awful lotta fruitcake. (1.0) Course it is. A little piece goes a long way. Well that’s right.

B makes an assertion in line 1 “an’ that’s not an awful lotta fruitcake.” Faced with no uptake from A (line 2), she revises her assertion in line 3 “Course it is. A little piece goes a long way.” B’s revision potentiates A’s agreement  in the ensuing turn “Well that’s right.” In Pomerantz’s  terms, B’s revision shows her sensitivity toward the discrepancy between A and B’s displayed points of view. In this example, a silence was treated as disagreement  implicative. Other devices for projecting nonaligning responses include prefaces, partial repeats, anticipatory accounts , and “pro-forma” agreements  (Heritage  1984b, 1988; Levinson  1983; Pomerantz  1975; Schegloff  1995b, 2007b). Pomerantz (1975) argued that modifications and revisions such as B’s “Course it is. A little piece goes a long way.” coupled with the tendency to delay or withhold actual disagreement  has “a minimizing effect on the occurrences of disagreements  and a maximizing effect for agreements . In such instances, it is not only what would be a disagreement may not get said, it is also that what comes to be said may be said as an agreement” (Pomerantz  1975: 81). In line with this work, alignment can be understood as the product of finely tuned negotiation between speaker and recipient. The present study focuses on how this negotiation is sequentially realized over the course of each emerging TCU by demonstrating how a prosodic variant of the affirmative response

Projecting nonalignment in conversation

137

token “ja” can be used to project a nonaligning responsive action in Swedish conversation. The dictionary-assigned meaning for “ja” is “yes.” There are several prosodic variants of the Swedish “ja”. I use the term “curled” to draw attention to two prosodic features: a lengthened vowel and a slight rise in pitch toward the end of the syllable (represented by colons and underlining in the transcript). The collection of curled “ja” turns include tokens that begin with an open /a/ as well as ones that begin with a soft /j/. The arrowed lines in Extract 3 show a curled “ja” response to a proposal. (3)  Visit (MOL:1:A). In line 10, A is referring to the fact that she has to help her children get accustomed to institutional day care by spending time with them at the day care facility. 01 A: → Hörru ja tänkte höra ifall ni ville ha Listenyou I thought hear in case you wanted have Listen I was going to ask if you wanted to be 02

en liten påhälsning¿< a little visit paid a visit

03 V:

↑Ja: de vill vi gärna?↑ Yes that want we gladly Yes we would love to

04 A:

Utav mej å ba:rnen¿ By me and the children

05 V:

Ja::: jättegä:rna? Yes very gladly Oh yes of course

06 A: I veckan¿ In the week This week 07 V: → .hh Ja:: (.) de e lite kärvt ((smilevoice)) men .hh Ja:: (.) ‘t’s little difficult but .hh Ja:: (.) it’s a little tricky but 08

de- vi ska se vicken da hade ‘ru tänkt då, ‘t- we will see which day had “you thought then ‘t- let’s see which day had you thought of then

09 A: Nej ja har inte tänkt nånting för nästa vecka No I have not thought anything cause next week No I hadn’t thought anything it’s just that I begin

138

Anna Lindström

10

börjar “ja” sko:la in dom, begin I school in them schooling them next week



[a discussion about Vera’s new job omitted]

50 A: >Är det kärvt den här veckan sa du< Is it difficult this here week said you You said it’s tricky this week

Anita asks Vera whether she would like a visit in lines 1–2, “Hörru ja tänkte höra ifall ni ville ha en liten påhälsning”/“Listen I was going to ask if you wanted to be paid a visit.” Anita is addressing her proposal to Vera and family with the second person plural form of the pronoun “ni” in line 1. Vera enthusiastically accepts the proposal on her family’s behalf with a two-unit turn in line 3, “↑Ja: de vill vi gärna?↑”/“Yes we would love to.” Upon hearing Vera’s acceptance, Anita specifies the proposal by stating who the prospective visitors will be. This is done with an increment that is syntactically parasitic on Anita’s prior turn in lines 1–2 (Lerner  1987, 1989). Vera responds with heightened enthusiasm in line 5, “Ja::: jättegä:rna”/“Oh yes of course.” Anita then adds another increment  that proposes a time for the visit, “I veckan”/“this week.” After an inbreath, Vera responds with a curled “ja” turn. The curled “ja” projects but does not implement a nonaligning stance . This analysis is supported by the fact that turn-transfer does not occur during the micropause after the curled “ja”. In refraining from claiming the floor ­during this pause, Anita shows an orientation to the relevance of an extended turn. The type of nonalignment that is being instantiated is specified within the ensuing parts of Vera’s turn. Instead of accepting Anita’s proposal, Vera states “de e lite kärvt”/“it’s a little tricky.” Vera begins the next turn constructional unit (TCU) with a contrastive conjunction “men”/“but.” This syntactic linkage to the prior TCU indicates that the emerging TCU will contrast with its predecessor. Thus it projects that although the time Anita has proposed is problematic for Vera she might still be willing to work something out. However, Vera does not bring this TCU to completion (note the self-interruption of de- in line 8). Instead she initiates an insertion sequence, “ vi ska se vicken da hade ‘ru tänkt då”/“so which day had you thought of” that indicates that her acceptance is contingent on the time frame for the visit, and the turn is thereby

Projecting nonalignment in conversation

139

hearable as a possible pre-rejection  of the proposal. The past tense formulation “had you thought of” coupled with the inference marker “då” conveys that Vera has inferred that Anita already had a particular day in mind when she made the proposal. Anita rejects Vera’s assumption that she had called with a pre-planned agenda in her next turn, “Nej ja har inte tänkt nånting”/“No I have not thought anything.” That she is addressing this assumption is partially conveyed through the reuse of the word “thought” from line 8. Anita then gives an account “ för nästa vecka börjar ja sko:la in dom”/“it’s just that I begin schooling them next week” that emphasizes that from her ­perspective it is urgent that the visit takes place in the near future. This is an ability account  (Labov and  Fanshel  1977) that has a no‑fault quality (Heritage  1984a, 1988) in that it “does not implicate a lack of willingness” (Heritage  1988: 18). Rather, it references a circumstance that is beyond Anita’s control (i.e. the requirements posed by her children’s day-care provider) as the reason why the visit needs to be scheduled sooner rather than later. Vera then pursues a discussion about Anita’s new job (transcript not shown). Anita returns to the issue of scheduling the visit in line 50 by formulating the upshot of Vera’s curled “ja” turn, “Är det kärvt den här veckan sa du”/“You said it’s tricky this week.” This formulation does not merely show that Anita has understood Vera to convey that the proposed time frame for the visit is problematic but also that Anita may be willing to back down from her original proposal. The analysis of this example illustrates that the curled “ja” projects nonalignment and that the recipient, in this case Anita, can use this as a resource for repositioning her actions in a way that promotes alignment. The results of this study are in line with CA work that has shown that “close and systematic attention to phonetic detail can open the way for a better understanding of the organization of conversational interaction” (Local and  Kelly 1986: 204, cf. Couper-Kuhlen and  S elting  1996; Egbert  et al.). A case in point is Sorjonen ’s analysis of the Finnish response particle “nii” (Sorjonen  1997, 2001). Sorjonen  suggested that intonational variants of this response particle embody different epistemic  claims. When “nii” is spoken with a falling intonation contour in an environment where a display of recognition of a referent is relevant it claims

140

Anna Lindström

recognition. The “nii” in Extract 4 occurs after Asta has referred to a third (nonpresent) party by first and last name. (4)  (Tuire/Danish  Visitor) (Sorjonen  1997:431–432). Simplified transcript, syntactic glossing line omitted. 01 T:

Haloo?, Hello?,

02 A:

.mth ö Tuire Haimakainen?,=

03 T:

=.mt >Puhelime-ssa,< =.tch >Speaking,
↑Hei,< >Hi:,
Nii.< 11 A:

on otta-nu yhteyt-tä (.) semmose-s asia-ssa --has contacted me (.) by reference to ---

Sorjonen  argued that in withholding talk during the silence in line 9, Asta is showing an orientation to the relevance of a display of recognition by Tuire. Tuire’s “Nii” in the next line (line 10) is produced with falling intonation. It claims recognition of Marja Niemelä (Sacks and Schegloff 1979). This analysis is supported by Asta’s next turn where she continues without further describing Marja Niemelä. When “nii” is used as a response to utterances that contain a referent that is treated as unfamiliar to the recipient by contrast it is delivered with a nonfalling terminal contour. In these cases, “nii” receipts the entire prior utterance as part of a larger incomplete unit of talk and invites the recipient to continue with that unit. This “nii” thus embodies a weaker epistemic claim than the “nii” that is produced with falling intonation. Extract 5 shows a “nii” that is spoken with a nonfalling terminal contour.

Projecting nonalignment in conversation

141

(5)  (Sorjonen  1997: 438). Simplified transcript, syntactic glossing line omitted. 01 T: – ja sit so-i puhelin ja .hhh jah tota (.) siel and then the phone rang and .hhh and well (.) there 02 ol-i semmonen (.) tyttö joka me ol-la-an tava-ttu sillo was this (.) girl whom we met 03 neljä vuo-tta sitte Rooma-ssa. four years ago in Rome 04 A: Nii?, 05 T: .mh Ja tota uusseelanti-lainen tyttö. .hh ni se (0.2) .mh And um a New Zealand girl. .hh she (0.2) 06 ol-i tulo-ssa Suome-en pyöräl-ile-mä-än, was coming to Finland to bike, 07 A: Äö:?, Ea:?

Tiina is telling Arto about her midsummer celebration. In her turn in lines 1–3 she introduces a character who will feature centrally in her telling with a nonrecognitional reference  (Sacks and  Schegloff  1979): “this (.) girl whom we met four years ago in Rome.” Arto responds with “Nii” that is produced with a slightly rising terminal contour. Sorjonen  argued that this response token treats Tiina’s talk as incomplete and invites her to continue. This also entails an agreement with the epistemic  assumption of the prior utterance, i.e. that Arto does not know the person who Tiina introduced in her prior turn. This claim is supported by Tiina’s next turn where she continues her story by characterizing the person she introduced in lines 1–3. A comparison of recipient uptake after “nii” in these two examples demonstrate that variations in prosody  alter how otherwise identical lexical items are understood. Beginning with Sacks  (1987), I reviewed early CA research that has established a link between contiguity  and agreement in conversation. Sacks (1987) and Pomerantz (1975, 1978, 1984) showed that agreement can be a negotiated outcome by describing how speakers revise the action of a prior turn to accommodate the displayed stance  of their recipient. Another set of CA studies have explored the link between prosody  and interaction by showing that the prosodic features of response tokens are sequentially and interactionally significant. Sorjonen (1997, 2001) argued that intonational variants of the

142

Anna Lindström

Finnish  response particle “nii” embody different epistemic  claims. When produced with a falling intonation, “nii” can be heard as a claim to recognition of a referent. When delivered with nonfalling terminal intonation by contrast, it invites the recipient of the “nii” turn to continue with the talk that was begun in her prior turn. The following section is devoted to establishing the central claim of this chapter, namely that the curled “ja” projects nonalignment. I do this by first exploring a range of cases in which the curled “ja” initiates a turn where a nonaligning action is implemented. Second, I examine deviant cases where the recipient claims the floor interruptively immediately after the curled “ja” has been produced. By using this position to revise the stance enacted in the first pair part of the curled “ja” sequence, recipients display that the curled “ja” was understood as projecting a nonaligning responsive stance . Data The data consists of audio recordings of naturally occurring ­telephone conversation which were made in three Swedish families during 1991. The conversations that figure in this corpus were made between landline telephones, and none of the families that participated in the study used telephones with automatic caller identification. The corpus that forms the basis for this chapter consists of twenty-five sequences where the curled “ja” occurs in turn-initial position  of a turn in which a responsive action is due. Initiative actions in the curled “ja” sequence include information requests , action requests, advice , assessments, offers  , and proposals. These actions are accomplished through a variety of linguistic devices such as interrogatives, statements, and follow-up questions. Unlike syntactic interrogatives, follow-up questions do not include a subject. They are constructed with a clausal or phrasal TCU and are syntactically and semantically parasitic of the sequential context (cf. Extract 3, lines 4 and 6). All first pair parts (FPPs) in the curled “ja” collection are built to take a “yes”/“no”-type response. Analysis Prima facie evidence for the argument that the curled “ja” projects nonalignment is that the curled “ja” can frequently be found in

Projecting nonalignment in conversation

143

the initial position of turns which implement a nonaligning action. The curled “ja” prefaces three types of nonaligning actions: hedges, pre-rejection s, and rejections. In the following I will explicate how nonalignment is done with the curled “ja” turn in each of these categories . I have used the term “nonaligning” rather than “disaligning” in this chapter to capture the most prevalent action initiated with curled “ja” turns namely the hedge. Twelve of the curled “ja” turns in the corpus either preface an actual hedge or are oriented to as projecting a hedging response. A hedging response does not fully align with the prior turn nor does it reject it. This is shown in Extract 6. (6)  Wallpaper [GRU:7:B]. Tore and Cajsa are building a house together with Tore doing most of the construction. He is calling Cajsa at work to give her an update on the work. Parts of the interior walls of the house are being covered with panelling. 39 T:

=.hhhh Få- få hålla på å sätta upp den här .hhhh Have- have hold on and set up this here .hhhh Will- will have to put up this

40

panelen i da ‘rå paneling in day then paneling today then

41 C:

Ja:? Yes

42 T:

Mm:,

43

(.)

44 C:

Då: blir de panelat färdit i helgen då, Then becomes it panelled ready in weekend then Then it’ll be all panelled this weekend then

45 T: → Aa: de vete tusan, ((creaky voice)) Aa: that know thousand Aa: lord knows 46 C:

.hhh För ja börja fundera me Leif också de .hhh For I began wonder with Leif also this here .hhh Cause I started to wonder with Leif and the

47

här me: tapetserarbordet å hhh here with the wallpapering table and hhh stuff about the wallpapering table and hhh

48

allting, everything

144

Anna Lindström

Here Tore describes the construction work that he will do in his turn at lines 39–40, “Få- få hålla på å sätta upp den här panelen i da rå”/“Will have to put up this paneling today then.” Cajsa receipts this with an acknowledgment (line 41). Tore’s subsequent third-position acknowledgment may propose topic  closure (see Hayashi  and Yoon, this volume). After a micropause, Cajsa reopens the previous topic by drawing an inference from the prior talk “Då: blir de panelat färdit i helgen då”/“Then it’ll be all panelled this weekend then.” Tore responds with a curled “ja” followed by the idiomatic expression “de vete tusan”/“Lord knows.” This hedging response casts doubt on Cajsa’s inference without actually disagreeing  with it. Cajsa orients to Tore’s curled “ja” response as nonaligning by accounting  for why it is relevant for her to propose that the panelling should be finished by the weekend (line 46–48). Extract 7 is from a conversation between six-year-old Henrik and his father, Torkel. The curled “ja” turn is produced as a response to the father’s question whether Henrik will return home (lines 2 and 4). (7)  Come home some time [VAT:11: B]. Henrik is calling home from a friend’s house. 01 H:

H[EJ’RÅ, Hi’then Bye then

02 T: [Kom- kommer’ru hem sen då¿ Com- come’you home later then Are- are you coming home later then 03

(0.3)

04 T:

Nån gång, Some time At some point

05 H: → Ja:hh ja vet inte, Ja:hh I know not Ja:hh I don’t know 06 T:

Ja:¿ Yes

Henrik proposes a conversational closing  in line 1 by initiating a good-bye sequence. Torkel claims the floor interruptively to ask

Projecting nonalignment in conversation

145

Henrik, “Kom- kommer’ru hem sen då”/“Are- are you coming home later then.” Faced with no uptake from Henrik (line 3), Torkel ­recompletes his prior turn with the adverbial phrase “Nån gång”/ “At some point.” While the question in line 2 made the specification of a time possible but not necessarily relevant, this question merely requests  a “yes”/“no”- type response. The recompletion can thus be hearable as adjusting to Henrik’s lack of uptake in line 3. Henrik responds in line 5 with a curled “ja” turn. The curled “ja” is separated from the second TCU by an audible outbreath. Torkel treats the curled “ja” as incomplete by refraining from claiming the floor at this point. Henrik continues with a disclaimer, “ja vet inte”/ “I don’t know” that neither affirms nor denies Torkel’s question. Another hedging curled “ja” response is shown in Extract 8. Torkel is calling his wife Liv at work to ask her what clothes their son should wear for soccer practice. (8)  Pants and trousers (VAT:11:B). 01 T:

.h Va heter’e va skulle Henrik ha på sej What call’t what should Henrik have on himself What was I gonna say what should Henrik wear

02

de grö:na¿ the green the green stuff

03 L:

Ja:: fö[r han lär väl ha långa byxor, Yes for he must väl have long pants Yes because he should probably wear pants

04 T:

[VicketWhichWhat-



Short spate of talk omitted where Torkel and Liv dispute the colors of the different items of clothing.

15 T:

Bå:de byxorna å tröjan? Both the pants and the sweater

16 L: → Ja: han kan väl ta tröjan runt mi:djan Ja: he can väl take the sweater around the waist Ja: he can at least put the sweater around the waist 17 → åtminstone om de bir kallt °får han väl at least if it gets cold gets he väl if it gets cold he can

146

Anna Lindström

18 → sätta på° sen har han väl nån svarta put on then has he väl some black put it on then he has some black 19 → te:shirt eller fotbollste:shirtar t-shirt or soccer t-shirts 20 → ligger i garderoben, lying in the closet 21 T:

°Ja°¿ Yes

Torkel requests  confirmation  in lines 1–2 by asking whether Henrik should wear “the green stuff”: “Va heter’e va skulle Henrik ha på sej de grö:na”/“What was I gonna say what should Henrik wear the green stuff.” Liv confirms in the next turn noting the need to wear pants. In Swedish, “långbyxor”/“long pants” are contrastive with “kortbyxor”/“short pants or shorts” (see the literal translation of line 3). A side sequence ensues with Torkel disputing Liv’s characterization of the color of the clothing (lines 4–14). Torkel then asks with a follow-up question in line 15 whether Henrik should wear both the pants and the sweater: “Bå:de byxorna å tröjan”/“Both the pants and the sweater.” Liv gives a curled “ja” response that neither completely aligns with nor disconfirms Torkel’s question by stating that Henrik should bring the sweater but not put it on: “han kan väl ta tröjan runt mi:djan”/“he can at least put the sweater around the waist.” In Extracts 6–8, the curled “ja” is followed by a hedge that neither aligns with the FPP nor completely rejects it. The next category of curled “ja” responses that I will discuss, pre-rejection s, are more strongly disaffiliative. There are eight pre-rejection  relevant instances of curled “ja” in the corpus. In Extract 9, Jan is calling his sister-in-law Vera in Sweden from a London airport. (9)  Landvetter [MOL:3:B]. Vera and her family live within two hours driving distance of Landvetter airport. Ulla, mentioned in line 24, is Vera’s sister and Jan’s fiancee. 16 J:

Ja ja e: i London allså, Yes I ‘m in London see/in other words Yes I’m in London you see

Projecting nonalignment in conversation 17 V:

↑E:: ’RU:¿ Are you You are

18 J:

Så ja tänkte fråga (.) pt om eh- (0.4) om ni So I thought ask (.) pt if eh- (0.4) if you So I was going to ask (.) pt if eh (0.4) if you

19

kunde- (.) ja skulle behöva liksom ha: could- (.) I would need kind of have could- (.) I would sort of need

20

h skjuss från Landvetter h ride from Landvetter h a ride from Landvetter

147

21 V: → .hh Aa: när då ’rå? .hh Aa: when then then .hh Aa: so what time would that be then Ja kommer klockan- vid midnatt kommer ja22 J: I come clock- at midnight come I I arrive at- around midnight I come- at 23

klockan tolv, clock twelve twelve o’clock

24 V: → pt heh heh heh ˙hh Aa:: ˙hh eh kanske U:lla kommer? pt heh heh heh ˙hh Aa:: ˙hh eh maybe Ulla comes pt heh heh heh ˙hh Aa::˙hh eh Ulla might come 25 J:

Ja just de, Yes right that Yes right

Vera treats Jan’s announcement  that he is in London as news in line 17. Instead of responding to the newsmark, Jan begins to make a request for a ride from the airport beginning with a ­preliminary “Så ja tänkte fråga”/“So I was going to ask,” which marks the upcoming action as delicate (Schegloff  1980). Jan’s request for a ride from the airport imposes on Vera in at least two ways. First, it is made late (i.e. on the day of Jan’s arrival). Second, compliance with the request would require Vera to make a two-hour drive to the airport in the middle of the night. Jan’s orientation to the first of these issues is evident in that later in the conversation he indeed states that he had been trying to reach Vera earlier. The request

148

Anna Lindström

includes two self-interruptions “pt om eh-”/“pt if eh-” and “om ni kunde-”/“if you could-.” The second of the self-interruptions could be heard to be going to a direct request “if you could pick me up at the airport.” This formulation is dropped for an indirect one that states a need, “ja skulle behöva liksom ha:h skjuss från Landvetter”/“I would sort of need a ride from Landvetter.” Vera responds to the request with a curled “ja” turn, “Aa: när då rå,” literally translated as “.hh Aa: when then then.” The duplication of the inference marker emphasizes that Vera’s willingness to grant the request is contingent. Like the request in lines 18–20, Jan’s answer to Vera’s question is hesitant and replete with selfrepairs, “Ja kommer klockan- vid midnatt kommer ja- klockan tolv”/“I arrive at‑  around midnight I  come- at twelve o’clock.” This may indicate an orientation to the time as problematic as does Vera’s laughter in line 24. Rather than granting the request, Vera states equivocally with another curled “ja” turn that “kanske U:lla kommer”/“Ulla might come” in line 24. Jan’s neutral uptake to this turn “Yes right” is compatible with the analysis that the request has not been granted. The curled “ja” is followed by a rejection in the next two examples (10 and 11). (10)  Dagens Nyheter (MOL:2:A). From a telephone call to a street vendor who mans a newsstand where papers and fast food is sold. The caller (C) has just asked the street vendor (S) whether a copy of last Sunday’s newspaper is still available (he is calling after Sunday). 08 S: [De kan no hända men iså:fall finns de It can probably happen but in that case is there It may be possible but in that case there won’t be 09

ingen e:xtrabila:ga till den utan då e’re no extraenclosure to it but then ‘t’s any magazine section then it’s

10

bara Dagens Ny:heter¿ only Dagens Nyheter just Dagens Nyheter

11 C: Ja bara Dagens Ny:heter ja¿= Yes just Dagens Ny:heter yes 12 S:

=Jo men de [tror ja att vi ha:r,] =Jo but it think I that we have =Jo but I think we have that

Projecting nonalignment in conversation 13 C: 14

[Ka- kan Ca- can Ca- can

du you you

149 ] titt(h)a:, look go check

(.)

15 S: → Ja:: de fi:nns Ja:: it exists Ja:: it’s there 16 C:

KH[HH

17 S: [ja ve:t de. I know that I know it 18 C:

.hh Eh nu e’re så här att ja: sitter i- i Orvenbo .hh Eh now “s’t so here that I sit in- in Orvenbo .hh Eh now I am in- in Orvenbo now

19

här å kommer väl inte iväg förrn om en timma here and come probably not away until about one hour and I probably won’t leave for about an hour

20

när stänger’u? when close you when do you close

The street vendor’s response in lines 8–10 suggests that parts of the Sunday paper are still available. When the caller replies that he is only interested in the paper itself (without the magazine section), the street vendor states that he believes he has it, Jo men de tror ja att vi ha:r”/“Jo but I think we have that.” The caller interrupts the street vendor’s turn in progress to request that he go look for the paper, “Ka- kan du] titt(h)a:”/“Ca- can you go check.” The street vendor’s response to the request is slightly delayed. Instead of granting the request, the street vendor asserts with a curled “ja” prefaced turn that he has the paper, “Ja:: de fi:nns”/“Ja:: it’s there.” Upon hearing this, the caller lets out a big sigh (line 16). This is treated as a display of irritation by the street vendor who retorts with an upgraded epistemic  assertion, “ja ve:t de”/“I know that.” A different kind of rejection is shown in Extract 11. (11)  Tax money (MOL:4:A). Runar has just asked his nephew Allan whether Allan’s father (Runar’s brother), Orvar has returned home. Orvar is a medical doctor employed by the Swedish government who apparently has been out on business travels. Allan responds that he does

150

Anna Lindström

not know whether Orvar has returned. This response is acknowledged and the sequence continues as follows. 34 R:

Han e bara ute å (lurvar)¿ He is just out and putzes He is out (all the time) putzing around

35 A:

pt Aa: han e bara ute å förlustar se:j, pt Yes he is just out and covorts himself pt yes he is out covorting (all the time)

36 R:

Ja*. ((*=glottal stop)) Yes

37 A:

Gör av me våra skattepengar. Makes off with our taxmoney Spends our tax money

38 R: → Ja::= 39 A:

=˙hA,= =Yes=

˙h ja har (precis en) känsla 40 R: → =Hörrudu- =Listenyouyou- ˙h I have (exactly a) feeling =Listen up ˙h I have a certain feeling 41 R:

av att [du också gör de, of that you also do that that you do that as well

42 A: 43 A:

[˙hh

Att ja: också hh g(h)ö- [heh heh heh˙heh heh heh That I also hh d(h)o- heh heh heh˙heh heh heh That I also spend-

44 R:

[HAH HAH HAH HAH

45 A:

J(h)a:(h) de kanske e nåt släktdra:g J(h)a:(h) it perhaps is some familytrait (Yes) maybe it is a family trait

46

hörru[du¿ listenyouyou (don’t you think)

In line 34, Runar jokingly evaluates Orvar’s travels, “Han e bara ute å (lurvar)”/“He is out (all the time) putzing around.” The Swedish word “lurvar” denotes walking or moving about without a sense of purpose. Allan responds to Runar’s turn with an upgraded assessment that treats Orvar’s travels as more illicit, “han e bara

Projecting nonalignment in conversation

151

ute å förlustar se:j”/“he is out covorting (all the time).” After an acknowledgment by Runar in line 36, Allan reclaims the floor to make another assessment that is stronger and more negative than the other two as it underscores that Orvar’s travels have a negative impact on others: “Gör av me våra skattepengar”/“spends our tax money.” The assessment makes relevant a second assessment (Pomerantz  1975, 1984). To merely discuss this spate of talk in terms of the relevance of a second assessment, however, does not fully capture the social intricacies involved. In a study of American  family dinner conversations, Ochs et al.  (1989, 1992) argued that one resource for building solidarity and cohesiveness within the family was to align the members of the family against a third nonpresent party. The possessive pronoun in line 37, “våra”/“our” could be heard to refer to taxpayers in general as well as Runar and Allan in particular. Allan thereby builds an alliance with Runar against Orvar (who is Runar’s brother and Allan’s father). With his assessment, Allan invites Runar to express solidarity with him (via a prefer ence for agreement ). However, in response to this, Runar produces a curled “ja” in line 38. On the immediate completion of the curled “ja,” Allan reclaims the floor with “.hA.” Runar orients to Allan’s turn as an interruption. The turn-initial summons in line 40, “Hörrudu-“/“listen up” is doubly contextual (Heritage  1984b). It links back to the prior turn by admonishing Allan for claiming floor prematurely. It also points forward by marking the upcoming action as problematic (Schegloff  1970, 1995b, 2007b). Runar agrees  with Allan’s assessment but rejects the alliance by suggesting that Allan is just as bad as Orvar in his next TCU, “ja har (precis en) känsla av att du också gör de,” “I have a certain feeling that you do that as well.” Although Runar produces this without laughter, Allan treats it as humorous by laughing in his next turn and then attempts to turn the tables on Runar by suggesting that this is a family trait (lines 45–46). When the curled “ja” turn is brought to completion, it typically implements a nonaligning action. Most commonly it does this with a hedge that teeters between alignment and nonalignment. My collection also includes curled “ja” turns that are prerejections. Only two curled “ja”s in my database are followed by rejection. This distribution suggests that the curled “ja” is more closely associated with minor nonalignment than with rejection

152

Anna Lindström

or disagreement  per se. That the curled “ja” typically precedes a nonaligning action does not prove that the curled “ja” response token projects nonalignment. To establish the latter point, we must isolate the recipient’s orientation toward the curled “ja” in and of itself. In the next set of examples, the recipient claims the floor after the curled “ja” is produced but before the remainder of the turn it prefaces has been articulated. The actions they engage in show that they orient to the curled “ja” as foreshadowing a nonaligning response. The analysis presented in the following builds on and contributes to CA studies that have established that interactants orient to the emergent and dynamic quality of each turn at talk, parsing it for the action(s) that are being accomplished (cf. Goodwin and Goodwin   1992; Heritage  1988; Jefferson  1973; Lerner  1987; Sacks  et al. 1974; Schegloff et al.   1996). Schegloff  (1996c) observed that any utterance in conversation may be understood to go through three phases: as (incipient) next, as current, and as prior. That is, as a currentrecipient-of some-talk/potential-next-speaker parses it in the course of its progressive articulation, potential response types and lines are engendered, subject to revision and replacement  as the current talk is further produced bit by bit.  (Schegloff  1996c: 97)

Recipients’ orientation to the bit-by-bit construction of talk becomes particularly clear when one examines talk produced in overlap. As Jefferson (1973, 1983) has shown, overlap onset is precisely placed with respect to the turn that is already in progress, and its resolution is delicately managed. Discussing overlap, Lerner  (1996) argued: Just as each next turn in a conversation is ordinarily designed to follow the just prior turn, so a recipient’s utterance begun internal to a turn’s talk (i.e. in the midst of the projected turn space of the current speaker) is shaped by, and understood by reference to, its placement. (Lerner  1996: 251)

In Extract 12, the recipient of the curled “ja” turn enters the turn space of her interlocutor. The placement of her turn relative to the turn-in-progress coupled with the kind of action she performs lends support to my argument that curled “ja” is understood to project nonalignment. The example is from a conversation between two young mothers, Nina and Liv. Nina’s young children are in

Projecting nonalignment in conversation

153

Liv’s home-based day care. Nina is a nursing aide who works on a shift schedule that involves night work. Her husband (Janne) is a teacher. In the excerpt below they are discussing Nina and Janne’s prospective work schedules. (12)  Work schedule [VAT:4:B]. Nina is in the midst of describing days when her and her husband’s working hours are so spread out that they do not need to leave their young children in day care. 01 N: [>Å likadant som tillexempel<  >And same as for example<  >And it’s the same if say< 02 om ja börjar- när ja börjar fy:ra när ja har if I begin- when I begin four when I have if I begin- when I begin at four when I have 03

natten¿ the night the night shift

04 L:

Ja:? Yes

05

(0.4)

06 N: (Ja) behöver ju inte åka före  (0.2) tjugi i fyra,= (I) need ju not go before twenty to four I would not need to go earlier then twenty to four right 07 i fyra,= to four to four right 08 L:

=Å [Janne kommer ha:lv fem, =And Janne comes half five =And Janne comes home four thirty

09 N:

[(De e ju-) (It’s ju-)

10 N:

[Ja seYes laYes at the la-

10 L:

[Före halv fem, Before half five Before four thirty

154 11 N:

Anna Lindström Senast [alltså, Latest that is At the latest that is

12 L: 13 L:

[Ja:, Yes .hh Ja han [slutar fyra då, .hh Yes he quits four then .hh Yes he gets off at four then

14 N:

[(Då hinner ja) (Then manage I) (Then I’ll have the time)

Ne::j när slutar’om slutar före 15 L: No when quit  they quit before four what No when do they get off they get off before four don’t they 16

fyra va? four what four don’t they

olika )] 17 N: → Ja::: d[e (e lite Ja::: i t (‘s little different) Ja::: it (depends they have-) 18 L: → 19 N:

[De bero:r lite på- a::¿] It depends little on yes It depends yes

Dom har inte riktit (.) samma ti:der, They have not really (.) same times They don’t have exactly (.) the same times

Nä:e, 20 L: No

In line 13, Liv states that Nina’s husband’s workday ends at four o’clock. This is offered as a concessive inference across a series of turns beginning in line 8. Although her turn in line 13 is formatted as a statement, it could be heard as a question because it focuses on a B-event  (Labov and  Fanshel  1977). Instead of answering Liv, Nina reclaims the floor interruptively to continue her own previous turn by beginning to formulate its upshot, “(Då hinner ja)”/“Then I’ll have the time.” She yields the floor to Liv before her turn has been brought to completion. In lines 15–16, Liv self-corrects the time reference she gave in her previous turn, “Ne::j när slutar’om

Projecting nonalignment in conversation

155

slutar före fyra va”/“No when do they get off they get off before four don’t they.” Here, Liv uses a locally subsequent reference form “they,” in locally initial position (Schegloff 1996a). Presumably she is referring to Janne and his colleagues at the school. The turn is done as a self-correction  through the turn-initial “ Ne::j”/“No,” and the stress on före/“before.” In contrast with her turn in line 13, which also focused on Janne’s work schedule, this turn is formatted as a syntactic interrogative and includes a turn-final tag, “va,” and is thus hearable as in pursuit of a response which Nina indeed begins to provide in line 17 with a turned prefaced by a curled “ja”. Liv reclaims the floor just after the articulation of the first consonant of Nina’s second TCU, i.e. after “de e lite olika” (literally translated as “it’s little different”). The “early” placement of Liv’s turn is not haphazard but sequentially motivated. Specifically, this is an instance of a recognition point entry. Lerner  (1996) proposed that one feature of many recognition point entries is that they are done not merely as responses to a forecasted TCU, but are used in circumstances where it would be felicitous to pre-empt the current speaker’s utterance before completion – i.e. to forestall the action that is currently underway and recognizable as part of the TCU’s projectability.  (Lerner  1996: 252)

The positioning of Liv’s turn in line 18 is thus not early but precisely placed to pre-empt the nonaligning stance that was projected with the curled “ja.” While the position of Liv’s turn claims an understanding of the action projected by the curled “ja” it does not demonstrate that understanding. The demonstration is achieved through the utterance itself, “De bero:r lite på- a::”/“it depends a little on the circumstances yes,” which is spoken in overlap with and aligns with the second TCU of Nina’s turn, “de e lite olika”/“it’s little different.” By capitalizing on the nonalignment projected by the curled “ja,” Liv has thus been able to maneuver herself from a position of projected nonalignment to actual alignment with her co-participant. Extract 13 is from a conversation between Malena and Pihlander. Malena has just bought a used car from Pihlander. Pihlander is calling Malena to remind her to register the car in her name as he

156

Anna Lindström

otherwise would be responsible for any traffic violations that she may incur. (13)  Gasoline (GRU:4:B). After having underscored the importance of changing the registration, Pihlander informs Malena that he has an extra set of keys for the car that she can pick up at her own convenience. Once these two issues have been settled, Malena raises a new topic  in line 42 by asking Pihlander what grade of gasoline he has been using in the car. “Ninety-six” and “ninety-eight” refer to two grades of gasoline with ninety-eight being the more expensive grade. 39 P:

[>Får hämta dom nån gång då sen< >Can fetch them some time then later< (You) can pick them up at some later time then

40 M:

Ja? Yes

41 P:

A[a:,

42 M: [˙hh Du “ja” tänkte på bensi:n (eh*) visst ˙hh You I thought on gas (eh) no doubt ˙hh Hey I was thinking about gas you usually do 43

tankar du s- nittisex? fill you s- ninetysix fill up with s- ninetysix I suspect

44

(0.6)

45 P: → Ja::[(h) 46 M: → 47 P:

[Eller tanka’ru nittiåtta?= Or fill you ninetyeight Or do you fill up with ninetyeight

=Ne:j nittisex, =No ninetysix

48 M: Nittisex (bör ja väl [kunna tanka’rå,) Ninetysix ought I probably able to gas’then I ought to be able to fill up with ninetysix then

Malena’s question in lines 42–43 is built to take an affirmative response through the adverb “visst.” Pihlander begins to respond in the arrowed turn with a curled “ja.” Like the recipient of the curled “ja” turn in Extract 12), Malena revises her prior turn upon hearing the curled “ja.” She does this by giving a contrastive alternative to the one she had offered in line 43, “Eller tanka’ru nittiåtta”/“Or

Projecting nonalignment in conversation

157

do you fill up with ninetyeight.” Her turn is built as a revised query through its placement (immediately after Pihlander’s curled “ja”) and through the use of the turn-initial “ Eller”/“Or.” In revising her query, Malena is trying to head off the nonaligning response that she heard to be underway at line 45. Pihlander then goes on to assert with an other-correction in line 47 that he indeed uses 96-octane gasoline. In this example, then, the curled “ja” was not produced (or “intended”) to project nonalignment. That Malena nonetheless heard it that way underscores that for the negotiation of agreement  and disagreement , and indeed for interactional practices generally, it is not “the intention” but the interactional effect that matters. Discussion When used in turn-initial position  of a responsive turn where a display of alignment is relevant, curled “ja” projects nonalignment with the action of the prior turn. Since it projects rather than accomplishes nonalignment, the curled “ja” furnishes a resource for recipients as they can recast their prior action upon hearing the curled “ja” to change the sequence trajectory from projected nonalignment to actual alignment. This was illustrated by Extract 12, where the recipient reclaimed the floor upon hearing the curled “ja” to revise the position she had assumed in her prior turn. As my analysis showed, the placement of her turn was crucial in that it allowed her to maneuver herself from a position of projected nonalignment to actual alignment with her co-participant. The curled “ja” is therefore an important tool for the achievement of alignment and affiliation  as a negotiated outcome. This is the first empirical study of curled ja. Future studies could examine the syntactic form of the FPP to see whether it shapes the way alignment and nonalignment are done within the responsive turn. For the purposes of the analysis presented in this chapter I have not found any reason for treating “ja:” and “aa:” as interactionally distinct. However, this should be examined in future work. I suggested that the curled “ja” may be more closely associated with slight nonalignment than outright disagreement  or rejection. This should also be explored in future studies.

158

Anna Lindström

Acknowledgments The study draws on analyses originally presented in my doctoral dissertation (Lindström 1997, 1999). I would like to thank Emanuel Schegloff, John Heritage, Steve Clayman, Trine Heinemann, and Jack Sidnell for valuable comments on earlier versions of this manuscript. This study is one of several investigations in the project Language and Social Action: A Comparative Study of Affiliation and Disaffiliation Across National Communities and Institutional Contexts (financed by the Swedish Council for Working Life and Social Research and the Swedish Research Council).

6  Two answers to inapposite inquiries Trine Heinemann

Introduction In recent years, one major focus of conversation analysis (CA) has been to investigate how knowledge  is distributed and negotiated among participants in interaction. Research has demonstrated that for conversationalists what each of them knows and how they know it are accountable  matters (Heritage  1984a, 1998; Drew  1991; Schegloff  1996b; Roth 2002; Heritage and  Raymond  2005; Raymond and  Heritage  2006; Stivers  2005; Clift  2006a, 2006b; Golato and  Fagyal 2008 ). As is the case for most CA research, the majority of these studies have been based on the analysis of English data. However, as Schegloff  (2006) argues, human beings as a species face the same problems and issues in everyday life independently of language or culture. As “the primary, fundamental embodiment of sociality” (Schegloff  2006: 70), interaction is designed to address these problems and issues, though the way in which this gets done varies across languages or cultures (cf. Schegloff  2002a). It thus stands to reason that practices parallel or analogous to those described for English exist in other languages. Once identified, such practices may provide a basis for cross-­linguistic comparison – allowing us to see the way in which the local resources  of a particular language are mobilized to solve generic problems of interaction. In this chapter, I explore how Danish speakers orient to knowledge  in interaction. Specifically, I identify two Danish practices, through which a recipient can index that a question is inapposite, because it inquires into something that should already be known to the questioner, and compare these with one previously described for English.

160

Trine Heinemann

In what follows, I first discuss the use of the English changeof‑state token “oh” (Heritage  1984a, 1998). I then identify the Danish practices used for accomplishing equivalent actions. Having established that Danish, like English, has a practice for conveying that a question targets something that should already be known to the questioner, I proceed to identify another Danish practice used for the same purpose. This second practice constitutes a stronger implementation of inappositeness than the English “oh” because speakers here explicitly reveal why a question ought not to have been asked. The chapter concludes with a discussion of how the two practices reflect the different ways in which a question can be understood to be inapposite and provides some suggestions as to how this difference may be reflected in other languages. Answers that index a question as inapposite The English change-of-state token “oh” and its use in question–answer sequences  Question–answer sequences are concerned in a basic way with the exchange of information and, as such, invoke the relevance of who knows what. In the following extract, both questioner and answerer display themselves as being more or less knowledgeable in relation to each other. First, J asks a question about the future activities of a person called Susan. J thus portrays herself as less knowledgeable about this matter, than the recipient, M. In turn, M aligns with the expectation that she is more knowledgeable than J by providing an answer to the question. Finally, by producing the change-of-state token “oh” as a receipt of the answer, J claims to have been informed on a matter where she was previously uninformed (Heritage 1984a). (1)  (WPC:1:MJ(1):1) (Heritage 1984a, ex. 20, pp. 308) 01 02 03 04 05 06

J: When d’z Sus’n g[o back.= M: [.hhhh J: =[( ) M: =[u-She:goes back on Satida:y= J: =O[h:. M: [A:n’ Stev’n w’z here (.) all las’week…

In Extract 1, the distribution of knowledge  is clear-cut, and the participants’ roles as questioner and answerer are in alignment 

Answers to inapposite inquiries

161

with their display of who is the least and the most knowledgeable. This is not always the case. Interaction is abundant, with instances in which a speaker asks a question to which she already knows the answer. In ordinary, everyday conversations, such questions are commonly used as vehicles for actions such as challenging the recipient (Heritage  2002; Heinemann, 2008 ) or making a direct complaint  (Monzoni, in press ).1 Even when they do not serve directly as vehicles for disprefer red actions, answerers often treat such questions as inapposite and problematic. One way in which recipients mark this, in English, is by prefacing the answer with “oh”. In the following example, Steven’s description of Robert Maxwell’s parents as having been killed by the Nazis (→1) causes Lesley to make the inference that Maxwell is of Jewish origin (→2). Steven confirms this with an “oh”-prefaced response (→3), which marks the answer as self-evident and the question as one that need not have been asked (Heritage  1998). (2)  (Field:2:3:9) (Drew and Holt  1998, ex. 6, pp. 500)2 01 Ste: →1 Well he didn’t either ‘ee had a bad start I mean ‘ee had 02 ‘iz (0.3) .t.k.hh father shot by the Nazis ‘nd ‘is uh .hh 03 mother died in:Auschwitz yih kno:w [so 04 Les: [Oh really:?= 05 Ste: =So ‘eez [had the:( ] )06 Les: →2 [Oh ‘z a Je:w] is he Je:w? 07 (.) 08 Ste: →3 Oh yeah. 09 (.) 10 Ste: He’s had k- eez a Czechoslovakian Jew so so [eez had 11 Les: [Yes 12 Ste: k- eez had quite a- checkered career already=

“Oh” is used in different sequential positions in Extracts 1 and 2 to implement different actions. In Extract 1, it occurs in third position, after a response to a question, and is used to claim that the speaker has been informed about something she was previously uninformed about. In Extract 2, by contrast, “oh” occurs in second position, as a preface to the answer to a question, and it indicates that the question, by inquiring into something that was already available to the questioner in the preceding sequence of talk, is inapposite. Heritage  (1998) argues that “oh” has an overarching function as a change-of-state token, where variations such as those between

162

Trine Heinemann

Extracts 1 and 2, “are managed through sequential and contextual particularizations of the particle’s placement” (Heritage 1998: 327). Thus, the “oh” in Extract 1 marks a shift in the state of the speaker’s knowledge , whereas the “oh” in Extract 2 marks a shift in the speaker’s state of attention (Heritage  1984a, 1998). “Oh” provides a solution to two different interactional problems in these extracts. In Extract 1, the problem that “oh”-prefacing addresses is that of how a participant can show that an answer to her question has been informative. In Extract 2, by contrast, “oh”-prefacing displays to a questioner that her question was inapposite. In the following section, I explore how these interactional problems are solved in Danish. The Danish change-of-state token “nå” The two functions that “oh”-prefacing is used for in English are accomplished through two different expressions in Danish. This language, like English, has what may be described as a changeof‑state token (see Nielsen 2002), i.e. one that can be used to mark that an answer to a question has informed the speaker where she was previously uninformed, as in the following extract. Here a question is asked (→1), an answer given (→2), and, in third position, the answer is receipted (→3). The extract comes from a face-to-face conversation between two female students, Birthe and Carina. In lines 1–3, Birthe inquires whether Carina has found any flats advertised in the newspaper. Initially, Carina, who is chewing some food, disconfirms this question with a headshake and a negative minimal response token “Mm mm” (line 5). She then readdresses the question with a confirming “Jo”/“Yes”, followed by a specification of the flat’s location. In line 11, Birthe receipts the information with the particle “nå”, registering that she has now been informed. (3)  (DLAU:KC “Flat hunting”)3 01 Bir: →1 Angående den lejlighed d[er. ] Var der andre About that flat. [ ] Were there any others 02 Car: 03 Bir: →1 i avisen, in the newspaper, 04

(0.4)

05 Car:

Mm mm.

[Mm, ]

Answers to inapposite inquiries 06

163

(0.4)

07 Bir: Ne[j. ] N[o. ] 08 Car: →2 09

[Jo ]. Ude ve” Bilka, [Yes]. Out by Bilka, (0.4)

10 Car: →2 eller i Store Husby. or in Store Husby. 11 Bir: →3 Nå. Oh.

As this extract attests, “nå” works as a change-of-state token similar to “oh” in English (see Extract 1). In contrast to English “oh,” however, “nå” is never used as a response to an inquiry, to mark that the answer to the inquiry was already known.4 Instead, this is accomplished through the production of multiple response particles produced within the same prosodic unit, as demonstrated in the following section. Answers to inapposite questions in German  and Danish It seems safe to assume that it is possible to ask a question to which the answer is already known in all languages, and prior research has established that this is certainly the case for Danish (Heinemann 2008 ). Consequently, it can be expected that Danish speakers have ways to index the inappositeness of such questions, just as English speakers have, even if this cannot be accomplished by using the change-of-state token “nå”. Heritage  (1998: 294) describes the function of “oh”, when used in response to inquiry, as proposing “that the question questioned something which could (or should) be taken for granted.” A similar description of a Danish practice is found in Heinemann  (2003). Here I note that multiple productions of the negative response particle “nej”/“no” within the same prosodic contour marks a previous turn “as in some way conveying something so obvious that it need not have been stated in the first place” (Heinemann  2003: 380). Likewise, Golato and Fagyal (2008  ) suggest that in German, certain multiple response particles are used to index a question as “requesting (obvious) information that was already presented or implied [ … ] in prior turns.”5 

164

Trine Heinemann

Extracts 4 and 5 are instances of multiple response particles used to answer questions in German and Danish. When compared with the English Extract 2, we can see that in all three cases a question (→2) questions something that is implied in prior turns (→1), and that this inappositeness is indexed in the recipient’s answer (→3). As earlier noted for the English example (Extract 2), the fact that Maxwell had a father who was shot by the Nazis and a mother who died in Auschwitz strongly implies that he is of Jewish origin, and the question “Oh ‘z a Je:w is he Je:w?” questions something which is self-evident.6 Similarly, in the German example below, M, in lines 14–15, asserts that his dentist subjected him to unnecessary dental work. For M to know and assert this, he has to have been informed of this after the dental work was done. Thus, with the inquiry in line 18, “hat sich das nachher rausgestellt?”/“did that come to light later?”, G questions information that is already conveyed through M’s assertion. Here, M indexes the inappositeness of the question by producing two instances of the positive response particle “ja”/“yes” within the same prosodic contour.7 (4)  (OregonT1A_514_offenerBiss) (Golato and Fagyal 2008  , ex. 8, pp. 258–259)8 05 M:

[.hh denn ähm denn nen zahnarzt, ich (g-) der [.hh cause uhm cause a dentist, i (g-) who

06

damals als ich bei mei- bei de- bei unserm back then when I was at my- at the- at our

07

zahnarzt war, .hh dentist, .hh

08 G:

mhm. uhum.

09 M:

zu dem meine ganze familie immer hingegangen is. to whom my entire family always went.

10

>und auch der kieferorthopäde=die ham mir ja da >and also the orthodontist=they quacked

11

im mu:nd< rumgedoktert, .hh around< in my mouth, .hh

13 G:

mhm. uhum.

14 M: →1 und ähm:ha:m sachen gemacht die ich gar nich and uhm:did stuff that I actually did not

Answers to inapposite inquiries

165

15 →1 ge↑braucht hab. .hh ↑need at all. 16 G:

ach so:? oh rea:lly?

17 M: ich hab äh:[m, I have uh:[m, 18 G: →2

[hat sich das nachher rausgestellt?= [did that come to light later?=

19 M: →3 =ja^ja. =yeah^yeah. =that’s ^right. 20

(.)

21 M:

.h der zahnarzt hat mich zum kieferorthopäden .h the dentist sent me to the orthodontist,

22

geschickt, und hat ä:hm hat gesacht ich bräucht ne a:nd said that I needed

23 M:

feste spange, braces,

In the following Danish example, Kirsten is describing a trip she took several years back. During this description, she mentions a church she visited and expresses her belief that this is the church where a friend’s husband is buried. This leads Gunnar to make the inference – somewhat delayed – that the friend’s husband is dead. Being buried clearly entails being dead. Hence, Gunnar’s inquiry “A’ han død”/“Is he dead” questions something that is conveyed in Kirsten’s prior turns at talk. As in the German example, this inappositeness of the question is indexed through the production of two positive response particles within the same prosodic contour. (5)  (MITHVDP31Aug0606 “Dead and buried”) 01 Kir:

Å’ der lå nemlig en kirk’ som lå °så’n° And you see there was a church situated °like°

02

>meget meget meget< højt oppe.→Å’ der tror jeg< >very very very< high up.→And there I think
al’så børnenes far< ligger °there he e-° >you know the children’s father< is

166

Trine Heinemann

05 →1 begravet, buried, 06

(1.1)

07 Gun: →2 A’ han død, Is he dead, 08 Kir: →3 Jah=jah. Jahm’ han døde .hh (2.0) Han var Yes=yes. Yes but he died .hh (2.0) He bloody 09

sgu ikk’ død da vi var herovre.= well wasn’t dead when we were over here.=

Extracts 2, 4, and 5 show how three languages – English, German, and Danish – allow speakers to index that they are answering a question the answer to which should already be known to the questioner. Whereas this is accomplished in English through prefacing the answer turn with the change-of-state token “oh”, in Danish (and German) this is done through the production of multiple response particles within the same prosodic unit. Generic features of inapposite questions and their answers Heritage  (1998) demonstrates that the question–answer sequences  in which “oh”-prefaced answers occur have a number of features in common beyond the fact that the question being asked is one to which the answer is already known. Here, I will briefly discuss how some of these features are also present in the Danish (and German) question–answer sequences . First, questions that receive an “oh”-prefaced answer are typically “mispositioned”, or “off the sequential track” (Heritage 1998: 307). This means that such questions are oblique, in that they target ancillary matters and have the potential to disrupt the progressivity  of an ongoing activity (Koenig 2007). In Extract 2 above, for instance, Steve is describing the “bad start” to life that Robert Maxwell had, exemplifying this by providing the information that his parents were killed during World War II. Though the reason for their being killed was that they were Jewish, this is not central to the description of Maxwell’s “bad start”. When Leslie inquires into this matter, she suspends the progressivity of Steve’s extended description. This is emphasized by the position of the inquiry as being in overlap with – and interruptive of – Steve’s turn in line 5.

Answers to inapposite inquiries

167

Inapposite questions in German and Danish are similarly mispositioned. The inquiry in the German example is produced in overlap and suspends M’s description of the excessive dental work he has endured. In the Danish example, there is no overlap, and the inquiry is produced after a pause of 1.1 seconds. However, the matter of “Tine’s former husband” is specifically introduced as ancillary, through Kirsten’s use of the adverb “faktisk”/“actually” (see Clift  1999). When Gunnar, through his inquiry, focuses on this issue, Kirsten’s telling about her trip is suspended. That the questions in Extracts 2, 4, and 5 are oblique or mispositioned is evidenced by the fact that recipients return to the activity they were engaged in after providing an answer. In the English example, Steve resumes the description of Maxwell’s life with a repeat of the conjunction “so” then closes the description with the figurative expression “checkered career” (Drew and  Holt  1998). Likewise, in the German  example, M continues his description of the unnecessary dental work he endured, after answering G’s oblique question. In the Danish case, this type of evidence is not available, as Kirsten self-interrupts her answer when she realizes that though her friend’s husband is dead he may not have been buried (or dead) at the time at which she went to the church. However, in the following case, the answerer’s treatment of the question as mispositioned is evident. Extract 6 comes from a telephone call between the local mayor, JO and a council member, Jens. JO has called Jens to let him know that a political meeting, taking place the day after, has been moved forward so that Jens will be able to participate before going on to a committee meeting. As will become evident, however, Jens was under the impression that the political meeting was scheduled on the day of the phone call and is, in fact, ready to go to that meeting when JO calls. (6)  (TH/S2/78 “Changing clothes”) 01 JO:

Ø::h Hvis vi så ku’ holde mødet der E::h If we could then have the meeting there

02

lidt i seks::, just before six,

03

(0.5)

04 Jen:

Jerh,= Yeah,=

168

Trine Heinemann

05 JO:

=ne’e ve’ Kurt, =at Kurt’s place,

06 Jen:

Jerh, Yeah,

07 JO:

Så ka’ du nå å’ komme te:eh:te’ bå’e det ene Then you can manage to get to both one

08

å’ d[et andet ikke¿] and t[he other TAG¿] and [the other right¿ ]

09 Jen:

[ . h h h h h [ . h h h h h [.hhhhh

J]Ahm:så si’r vi da det, Y]Es-but then say we then that, O]kay then that’s what we’ll do,

10 JO: Ska’ vi ikk’ si’:det¿= Shall we not say that¿= Isn’t that what we’ll do¿= 11 Jen: →1 =Johm’ jeg var ellers klædt om¿ =Yes-but I was otherwise dressed about¿ =Yes but I had actually changed my clothes¿ 12

(0.4)

13 JO: →2 >.hh< Var du klædt om? >.hh< Were you dressed about? >.hh< Had you changed your clothes? 14 Jen: →3 Jah=jah=ja:h=å’:: f [orb]eredt.=.hhh [Det var Yes=yes=yes=and p[rep] ared.=.hhh [Those were 15 JO: 16 Jen:

[( )]

[(

fan]d’me hunderd’tredive hårde sider. blo] ody well a hundred and thirty tough pages.

17 JO: )] 18 JO:

Ja det a’ hårdthh. Det a’ hårdt ahah= Yes it’s tough. It’s tough ahah=

In lines 1–5, JO proposes when and where the political meeting will be held. His turn in lines 7–8, “Så ka’ du nå å’ komme te: eh: te’ bå’e det ene å’ det andet ikke¿”/“Then you can manage to get to both one and the other right” implies that the meeting has been coordinated to accommodate Jens. Jens’ response does not align  with this implication as, for instance, an expression of gratification would have done. His acceptance is delayed by a loud and long inbreath, potentially projecting a disprefer red answer

Answers to inapposite inquiries

169

(Heritage  1984b; Pomerantz  1984). The inbreath in combination with the way in which Jens’ acceptance is formulated, “JAhm: så si’r vi da det,”/“OK then that’s what we’ll do,” comes across as resigned concession. This lack of enthusiasm makes JO pursue a more adequate response, with the inquiry “Ska’ vi ikk’ si’ det”/“Isn’t that what we’ll do”. As a negative interrogative, this inquiry is strongly biased toward a positive, confirming answer (Heritage  2002; Heinemann  2006). Jens, however, produces only a minimally accepting “Jo”/“Yes”, then goes on to explain that he had already changed his clothes, i.e. gotten ready for the meeting (line 11).9 This turn provides a possible explanation for Jens’ unenthusiastic acceptance of the rescheduling of the meeting: He believed the political meeting to be taking place on the day of the telephone call and had thus already incorporated the meeting into his schedule. With the rescheduling, he now finds himself in a position where on one day (the day of the telephone call) he is ready to go to a meeting that will not take place, whereas the next day he will be forced to go to two meetings in close succession. In line 13, JO reformulates Jens’ statement as an inquiry, requesting Jens to reconfirm that he had “changed his clothes”. This question entirely fails to take the implications of Jens’ statement into consideration and is, in that sense, off the sequential track. Jens’ response to the inquiry orients to this misplacement. First, he treats the question as inapposite through the production of multiple response particles (“jah=jah=jah”). Second, he continues on a track that focuses on his expectation that the meeting is taking place on the day of the phone call: Not only had he changed his clothes, he had also prepared by reading “130 tough pages” (presumably a report to be discussed at the meeting). He does so, through latching an “and”-prefaced turn (Heritage and  Sorjonen  1994) to the last of the response particles, thus signaling that this turn is part of the overall sequence of which the question–answer  sequence was a suspension. A second observation made by Heritage  (1998) about “oh”prefaced answers to inquiry is, that though these answers indicate “some inappositeness or difficulty about a prior inquiry”, they only do so implicitly and leave “the source of the difficulty as something to be searched for and resolved by the questioner” (Heritage  1998: 295). In Extract 2, for instance, Steve does not overtly state

170

Trine Heinemann

that it follows from the information provided by him (that Nazis shot Maxwell’s father and that his mother died in Auschwitz), that Maxwell is of Jewish origin. The Danish and German  extracts above suggest that multiple response particles are similarly inexplicit as answers to inapposite inquiries. In Extract 4, Kirsten does not state that the fact that a person is buried entails that that person is dead. Nor does Jens, in Extract 6, remark on the fact that he has already explicitly stated that he had changed his clothes at the point at which JO inquires about this. Rather, as Golato and  Fagyal (2008 ) note, the answerer, through the production of multiple response particles merely indicates the misaligned placement of the turn to which she is responding. The inexplicit nature of “oh”-prefaced responses and their functional equivalents in Danish and German  seems to be a direct consequence of the fact that the questions to which these practices are used as answers are oblique. Should answerers deal with the inappositeness of such questions in a more explicit manner, this would only serve to further suspend the progressivity of the activity that the participants were engaged in before the question was asked. Moreover, by explicating exactly why a question should not have been asked, the answerer would be challenging the questioner’s inapposite behavior in an overt manner, making a reaction relevant from the challenged  party (see, for instance, Dersley and Wootton  2000; Koshik, forthcoming ; Monzoni 2008 ). Such a reaction would, in turn, lead the participants further away from the suspended activity and perhaps even result in a complete abandonment of this activity. As practices used for answering questions to which the answer is already known, “oh”-prefaced responses and multiple-response particles are thus perfectly designed to handle the answerer’s conflicting interests so that she can index the inappositeness of the question without further diverting attention from her overall interactional project. Explicit targeting of the inappositeness of a question: The use of the modal adverb “da” The previous examples have shown how recipients can index that the question to which they are responding is inapposite without explicating exactly what was problematic about the question. On

Answers to inapposite inquiries

171

occasion, however, recipients do explicate why a question need not – indeed, ought not – have been asked. In the following, I identify one practice for doing so in Danish. Extract 7 is a first instance of this. Here, a multiple “jaja” indexes the prior question as inapposite and is then followed by additional material which explicates what knowledge the recipient possessed that should have prevented her from asking the question. Kirsten and Gunnar have been married for about twenty years. Prior to this extract, they have been talking about Carl Peter, who is married to Gunnar’s sister. Throughout their discussion, Kirsten has treated Gunnar as being the most knowledgeable about all things concerning Carl Peter. She does so again in lines 4–5, where she inquires whether Gunnar has ever met Carl Peter’s father. (7)  (MITHVDP31Aug0606 “Carl Peter’s father”) 02 Gun:

KR HH

03

(2.5)

04 Kir:

>M’ har du< no’ensinde mødt:e:h hans far >But have you< some-time met:e:h his father >But did you< ever meet eh his father

05

egentli’ >al’så fordi det var< no’et me’ actually >ADV because it was< something with actually >you know because there was< something

06

(1.1)

07 Gun: → Jah=jah ham kender du da ås.’ Yes=Yes him know you adv also. Yes=Yes you da know him as well. 08

(0.5)

09 Kir:

Carl Peters far? Carl Peters father?

Kirsten’s inquiry is positively framed. With the inclusion of the negative polarity item “nogensinde”/“ever” and the adverb “egentlig”/“actually”, she displays her belief that Gunnar has never met Carl Peter’s father (Quirk et al. 1985). Because of the distribution of knowledge , the inquiry implies that Kirsten herself has never met Carl Peter’s father. If Kirsten had met him, then presumably Gunnar would have as well – and, more importantly, Kirsten would know this.

172

Trine Heinemann

The kind of knowledge that is at stake here is not available in prior turns. Prior to this extract, Gunnar and Kirsten have discussed only Carl Peter, not his father. Any knowledge that Kirsten has about Carl Peter’s father is thus dependent on who she is in relation to Gunnar, rather than on what he has said in prior turns. In addition to this, Kirsten’s inquiry is produced in a sequential environment different from the ones discussed in prior sections. It follows a lapse in the conversation (see lines 1–3) and does not suspend an ongoing activity. Nevertheless, Gunnar treats the inquiry as inapposite by producing multiple instances of the response particle “ja”/“yes”. Furthermore, and in contrast to the extracts discussed previously, he goes on to explicate exactly what it was that Kirsten knew and should have taken into consideration before ­asking the question. Thus, in line 7, Gunnar states that Kirsten has herself met Carl Peter’s father. The statement concerns a “B-event” (Labov and Fanshel 1977) in that it makes a claim about the recipient, Kirsten’s experiences. In addition, the use of the verb “kender”/“to know” is an upgrade of the verb used by Kirsten, “mødt”/“met”, and implies that Kirsten has had more than one encounter with Carl Peter’s father. Finally, the addition of the term “ås”/“as well” links Kirsten’s experience directly to Gunnar’s. This statement is thus exquisitely designed to convey exactly what was problematic about Kirsten’s question, namely the presupposition that she herself had not met Carl Peter’s father. It is, however, only through the presence of the modal adverb “da”, that Gunnar explicitly rebukes Kirsten for failing to consider already known information in designing her question. Though this adverb can be translated into English as “surely” or “really”, translation fails to capture its interactional import. To the best of my knowledge, “da” has not been described in any systematic fashion. However, the similar adverbs, “doch” in German  and “toch” in Dutch , have.10 Though these descriptions are based on invented examples, some initial observations about them can serve as a point of departure for the analysis of “da”. For the German  adverb “doch”, Blass (2000) argues that this is a marker of mutual manifestness (Sperber and  Wilson 1986), which serves to strengthen what the speaker is saying, in order to weaken or strengthen a previous assumption (stated or implicit). Blass (2000) provides the following case as an example of how “doch” is used and how it should be interpreted.

Answers to inapposite inquiries

173

(8)  (Blass 2000, ex. 14, pp. 48) Komm doch rein! Come as-it-is-manifest-to-do in contrary-to-what-you may-be-thinking “Why don’t you come in?”

For this example, Blass (2000: 48) notes, that doch indicates that the speaker and hearer have manifest a cultural norm, which is “to just come in.” Doch also indicates that the speaker is negating any assumption that made the hearer act against the manifest norm. To paraphrase, “It is the norm to enter without being told and any other assumption has to be eliminated.” In English the question form has the same implications: “It is normal to come in, so why don’t you?”

In a similar fashion, Foolen (2006) argues that the Dutch “toch” is used in contexts where the speaker had assumed that the recipient shared the same knowledge and attitude toward a particular matter but where there is now reason to be uncertain about this. He provides the following example, from a novel by the Dutch  writer Maarten ‘t Hart. (9)  (Foolen 2006:67, ex. 12)

Maar ik kan toch thuis koffie zetten But I can make coffee at home, can’t I?

For both “doch” and “toch”, there are indications, then, that speakers use these adverbs to imply that the recipient has failed to behave in a manner that is congruent with some knowledge, attitude or normative expectation that is shared by both parties. “Doch” and “toch” serve to emphasize the existence of this shared knowledge , attitude or normative expectation by forcefully rejecting any suggestions to the contrary. Gunnar’s use of “da” in the extract above suggests that this Danish modal adverb functions in a similar manner. The presence of “da” emphasizes that the fact that Kirsten has met Carl Peter’s father is shared knowledge and that any assumption to the contrary is incorrect. Gunnar thus rebukes Kirsten for asking a question that presupposes the opposite to what she already knows, and the question is in this way treated as inapposite. In the following, I will discuss in more detail how “da”, when used in turns that are responsive to a question, serves to explicate

174

Trine Heinemann

why that question was inapposite and thus to imply that the questioner has behaved in a morally problematic manner. First, I demonstrate how the presence of “da” in answer turns serves to target and reject a position taken (implicitly) by the questioner in asking a particular question. Second, I demonstrate how speakers use “da” to indicate that the recipient has committed a moral breach by failing to follow normative expectations. Finally, I show that recipients of “da” turns orient in their responses to these aspects of “da” by conceding that they did have access to the knowledge that should have prevented them from asking the question and that they have consequently behaved in a morally problematic manner. “I don’t know” answers One way in which to illuminate how “da” serves to target and reject a position taken (implicitly) by the questioner is to contrast cases where this adverb is present in the answer, with similar cases where it is absent. The following two extracts allow for such a comparison. These are instances of “I don’t know” answers, i.e. answers where a speaker claims to lack the knowledge required to answer a question. These answers are particularly interesting because they themselves reject an implication – or expectation – of the question: that the person to whom the question is directed will be able to answer. As the contrast between the two extracts attests, the presence of “da” in such answers upgrades the rejection and marks the answerer’s lack of knowledge as self-evident and the question, consequently, as highly inappropriate. In the first extract, an “I don’t know” answer without “da” is delivered. Gunnar and Kirsten are having a break during their bicycle vacation on an island. Both of them have been on the island (separately) before, but Kirsten has repeatedly treated Gunnar as the one most knowledgeable about the island and their whereabouts. In line 2, she does so by inquiring about the distance to a small village, Søby. (10)  (MITHVDP31Aug0606 “Distance to Søby”) 02 Kir: .mlhh Hvor langt a’ der egentli’ te’ Søby .mlhh How far is it actually to Søby

Answers to inapposite inquiries

175

03 herfra, from here, 04 (0.8) 05 Gun: Det ve’ jeg ikk.’ ↑ti kilometer, That know I not. ↑ten kilometers, I don’t know. ↑ten kilometers,

In response to Kirsten’s inquiry, Gunnar first claims a lack of knowledge with “Det ve’ jeg ikk’”/“I don’t know”. He then proceeds to attempt to answer the question nevertheless, by providing a possible answer, “ti kilometer”/“ten kilometers”. With this “best guess”, he displays his willingness to answer the question – and does not in any way indicate that he finds it problematic that Kirsten has asked him this question. By comparison, in Extract 11, the answerer adds “da” to her “I don’t know” answer, thus treating as incorrect and inappropriate the expectation that she is more knowledgeable on the matter questioned than is the person asking the question. KM has called to talk to Jens. Jens’ wife, Fie, answers the telephone. After a greeting and a self-identification (in line 2), KM inquires about Jens’ availability by asking “Har du Jens hjemme”/“Do you have Jens home”.11 Fie disconfirms this (in line 5), and the sequence is completed through KM’s third position receipt “Nåh”/“Oh” in line 6. Fie then asks whether she was supposed to have Jens home, suggesting that KM expected this to be the case and that she thus knows more about Jens’ whereabouts than his own wife (in line 8). (11)  (TH/M2/9 “Missing husband”) 02 KM:

Jah. Go’dag An’Sophie KristaMerete, Yes. Hello An’Sophie KristaMerete,

03 Fie:

Jah hej, Yes hi,

04 KM:

Har du Jens hjemme¿ Have you Jens home¿ Do you have Jens home¿

05 Fie: Nejh. No. 06 KM:

Nåh. Oh.

176

Trine Heinemann

07

(0.3)

08 Fie:

Sku’ jeg ha’ det? Should I have that? Should I?/Was I supposed to?

09

(0.1)

10 KM: → De::tehhh V(h)e’ jeg da i(h)kk’(hh) Tha::thehh K(h)now I adv n(h)ot(hh) I do da n(h)ot(hh) kn(h)ow Nejh, de:t det tror jeg ikk’. 11 Fie: No, tha:t that think I not. No I don’t think so. 12 KM:

>Nåh.< >Oh.
11 B: Si’. °Vabbe’° Yes Alright Yes. °Alright° 12

A

(0.8)

B

Gaze, questioning, and culture A

195

B

13 A: ↑Anzi niente. Actually nothing ↑Actually nothing. 14

A

B

(0.2) A

B

15 →B: Ni[ente. Nothing No[thing. A

B

16 A: [Perche’ sono andato su’ a piedi. Because be.1s gone up by feet [Because I went up on foot. 17

A

B

(0.2) A

B

A

B

18 A: Se pr[endi] l’ ascensore paghi. If take.2s the elevator pay.2s If you t[ake] the elevator you pay. 19 →B: 20 21 →B:

A

B

[eh?] eh [eh?] A

B

(.) A

B

Si’. Yes Yes.

196

22 A:

Federico Rossano et al. A

B

Si’ Yes Yes A

B

A

B

A

23 B: Allora ci vado anch’ io. Then cl. go.1s also I Then I will go too.

B

At line 2, B asks a question about the cost of entering the building represented in the picture. The speaker (B) turns toward the recipient (A) at the beginning of the question while the latter keeps looking at the picture. After a delay in responding, B starts a candidate answer quantifying the possible cost (ten), but at this point A looks toward B and provides an answer. At line 13, A modifies the answer provided at line 5 by claiming that to enter that place he did not pay anything. Even though A starts the turn with a pitch reset and emphasizes the word “nothing,” A does not look at B. At line 15, B produces an other-initiation of repair by asking for confirmation  (and displaying disbelief at the claim that A did not pay anything). B starts turning toward A during the silence at line 14 and therefore starts the initiation of repair already ­looking at the addressee. B holds his gaze toward A until the end of line 19, where he is not initiating repair but rather pushing for a “yes”/“no” answer to his question at line 15 (“eh” can be used as a tag in Italian  questions). At line 21, B asks for confirmation  of the last claim about the free entrance to the building, but this time he is looking at the pictures and looks toward B only at line 23, when he produces a closing remark that accounts for the occurrence of the question at line 2 (his possible interest in visiting the place represented in the picture). With Extract 1, we have illustrated, among other things, that even in situations where objects relevant for activities at hand are present, speakers tend to look at recipients while asking questions. With the following extract we want to show what happens when two participants are just talking about future plans and are not dealing with relevant objects in the surrounding environment. Extract 2 shows:

Gaze, questioning, and culture

197

1. Speaker and recipient tend to look up mid-question during the question that initiates a new course of action (see also line 2 in Extract 1). 2. During the course of action, speakers tend to look from beginning to end of each question. 3. Recipients sometimes look away before completion of the question (presumably to plan the answer to the question), though not during every question. In Extract 2, the participants are two female friends sitting at a table (at a 90-degree angle), and the conversation has just started. We focus on the gaze behavior during lines 1, 3, 6–7, 12, 16, and 20. (2) Italian  2GGOSS-stasera 00:15 A

A

B

B

A

B

01 →B: .hh A(h)l l o r a stas(h)era cos’ e’ che fate S(h)o tonight what is that do.2s .hh S(h)o t(h)onight what are you doing A

02 A: Eh Eh Eh

B

A



B

andiamo a Villa Chiara = go.1p to Villa Chiara we go to Villa Chiara =

A

B

A

B

03 →B: = Ma a che ora vi incontrate But at which hour you meet.2p = But at what time do you meet 04 A: 05

A

B

A

B

Vado alle nove e mezza dalla Gloria. Go.1s at nine and half to Gloria I go to Gloria’s (house) at nine thirty. ((long side sequence about Gloria’s recent guests)) A

B

A

B

A

B

06 →B: Nove e mezza ma andate subito a Villa Chiara Nine and half but go.2p immediately to Villa Chiara Nine thirty but do you go immediately to Villa Chiara

198

Federico Rossano et al. A

B

07 →B: alle nove e mezza? = At nine and half at nine thirty? = A

B

A

B

08 A: = Con- le  dieci con  gli altri. With the ten  with the others = With- at ten with the others. 09

A

B

(0.4) A

B

10 A: Ci incontriamo. Cl. meet.1p We meet. 11

A

B

(0.6) A

A

B

B

12 →A: Ci [v i e n i? Cl. come.1p Do you come there? ((to Villa Chiara)) A

B

A

B

A

B

A

B

13 B: [Io esco alle nove_ (0.5)  ah io pero’ se arrivo I get.1s out at nine   oh I but if arrive.1s [I get out at nine_ (0.5) oh but if I come 14 15

A

B

arrivo a mezzanotte eh arrive.1s at midnight eh I arrive at midnight eh A

B

(0.6)

((A displays shocked facial expression))

Gaze, questioning, and culture A

199

B

16 →A: Alle nove esci? At nine get.1s out At nine you get out? A

B

A

B

17 B: E[h Eh Y[es ((confirming something already said)) 18 A: 19

[Cazzo Dick [Shit ((and then A nods, looking down)) A

B



(0.6)

A

B

(1.2) (1.8)

A

B

A

B

20 →A: Domani sera che fai? Tomorrow evening what do.2s Tomorrow evening what do you do? 21

A

B

(0.5)

At line 1, the recipient starts looking at the speaker before the latter turns toward her; by the time the question at line 3 starts, both participants are looking at each other, and only the recipient withdraws from mutual gaze before completion of the question. The same happens during the question at lines 6–7. All three questions can be taken to be requests  for information by B about A’s plans for the night, probably meant to get an invitation  by A. At line 12, A does indeed invite B to go out with her and her friends, and, again, we see that the speaker is looking at the recipient from the beginning to the end of the question, while the recipient (B) looks toward the speaker mid-turn when she has already started producing the turn at lines 13–14. The question at line 16 is not only a repair initiation but also a way of displaying surprise and disbelief (as can be seen from the

200

Federico Rossano et al.

co-occurring facial expression) and in this case both participants are looking at each other from beginning to end of the turn. The course of action reaches its completion at line 18 and both participants look down toward the table. At line 20, however, A starts a new sequence and asks B about her plans for the following day. Here we see that the speaker starts looking at the addressee only after the beginning of the question and the recipient does not look at the questioner at all. Also in this case (see questions in Extract 1), no remedial action is taken by the speaker to obtain the recipient’s gaze  back and the recipient is not looking at any other relevant objects on the table. These two extracts, which represent patterns observed throughout ten hours of dyadic interactions in Italian , suggest that many of the claims about the organization of gaze  in interaction previously proposed do not apply to Italian data. In particular, if we focus on questions, it appears that there is much less gaze mobility  throughout the question than compared to previous general claims about gaze in any utterance,10 and there is no particular sensitivity to the lack of recipient gaze (differently from C.  Goodwin 1980, 1981). Moreover, speaker gaze  appears to be much more predictable and relevant during questions than recipient gaze . If we then look at two typical patterns of gaze  during questions in Tzeltal  and Yélî Dnye  , we can see that while in the former case participants can produce entire sequences without ever looking at each other (see Extract 3), in the latter case participants can be seen to sustain gaze toward each other during questions without any withdrawal (see Extract 4). Extract 3 is taken from an interaction between two middleaged Tzeltal  sisters. Both participants are sitting, side by side on a bench, with their backs against a wall, approximately two meters away from each other. They are looking straight ahead, away from each other’s faces. B has been telling A about a relative of theirs, who had come supposedly to the village for a meeting but had gone straight back to town without waiting for the meeting. Lines 2 and 4 are our focus. (3)  Tzeltal 2005_v5A_Q19 11:53 01

A

(1.3)

B

Gaze, questioning, and culture A

201

B

02 →A: ma’yuk li’ y-ak’ junta tz’i bi not.at.all here 3E-give meeting PT TAG He didn’t hold a meeting at all here then eh? A

B

03 B: ma’yuk laj not.at.all QUOT Not at all, reportedly 04 →A:

A

B

ju’uk no No? A

B

05 B: ma’yuk laj jalaj ix not.at.all QUOT last.long ACS He did not stay long (enough) at all, reportedly.

During both questions, both participants keep looking away from each other and do not turn to each other even during the production of the answers. This contrasts quite clearly with what can be observed in interactions in our dataset for Yélî Dnye, where, in dyadic interactions,   participants seem to look at each other most of the time. Extract 4 is taken from an interaction between two Rossel  men who are sitting face-to-face, less than one meter apart. A is complaining  about the fact that some individuals who owe him money have not repaid him yet. The gaze behavior at lines 4, 6, and 7 is our focus. (4)  Yélî r03_v19_s2 00:13:53 01 A: awêde nga today here 02 A:

anî tóó, 1sFUT.PUNCT sitting

u pyinê 3POSS quest

d:a ngmêê, ngmepe 1sIMMPAST.PUNCT exchange.PROX paying.back I am here today, to search for it ((the shell money)), to pay it back law nkwodo até nî kmungo. law on.top just 1sIMMPAST.PUNCT put.inside I took it up to the law (told the Peace Officer)

202 03

Federico Rossano et al. (0.4) A

B

A

B

04 →B: n:uu ye ngmepe Who that.near.Addr paying.back Who is paying it back (to you) 05 06 →A:

A

B

(0.4) A

B

:êê uh uh?

07 →B: n:uu ye ngmepe Who that.near.Addr paying.back Who is paying it back (to you) 08

A

B

(0.4) A

B

09 A: kî pini dy:eemi knî that.distal man.spec man.and.brother-in-law AUGM That man and his brothers-in-law

In this extract, we can see that participants sustain mutual gaze  throughout the three questions: a request for information, its repetition,  and a repair initiation. Altogether, the participants sustain mutual gaze  across silences and into the answer at line 9, showing the opposite pattern of what was shown in Extract 3. It is noticeable that not only in Italian  but also in Yélî Dnye   a speaker can sustain gaze toward  the addressee from the beginning to the end of the question, without looking away before uttering it or looking toward the addressee only while approaching the completion of the turn. These extracts raise multiple questions. How representative are they of the common gaze patterns during question–answer  sequences in each culture? Are the clear differences from prior

Gaze, questioning, and culture

203

claims in the literature indicative of a different type of ­organization of gaze in interaction, at least in these three cultures? Are there cultures where participants tend not to look at each other at all or tend to look all the time? And if we believe Simmel’s claim (1969) about the importance of mutual gaze  for human sociality, what would be changed in the interactions and social relationships of human beings living in a culture that appears to minimize the occurrence of mutual gaze (e.g., Tzeltal )? All these issues invite the development of a proper coding scheme and the collection of a larger sample to assess how many of the patterns here observed are still systematic in a larger dataset. For this purpose, we used extensive video recordings of natural conversation from these three cultures and developed a comparative coding scheme, as described below. Data and method  The data we used for our comparative analysis of gaze behavior in interaction comes from a total of twenty-nine dyadic interactions11 in three different languages: Italian , Tzeltal,  and Yélî Dnye  .12 Italian is a Romance language spoken primarily in Italy, Tzeltal is a Mayan  language spoken in Chiapas, Mexico, and Yélî Dnye   is an isolate, presumably of Papuan origin, spoken on Rossel  Island, Papua New Guinea. More precisely, each dataset has been collected in a specific part of the area where each language is spoken, and we believe the data are representative of the three different cultures, which are relatively homogeneous within each dataset.13 Figure 7.1 shows the location of the three cultures – as the geographical dispersion suggests, there has been no cultural contact between any of them, and thus any similarities in gaze pattern can be presumed to derive from universal  tendencies or from systematic functions within an underlying shared interaction system. The ten dyadic interactions in Italian  were recorded with two cameras, one on each participant, while the ten interactions in Tzeltal  and the nine in Yélî Dnye   were recorded with one camera, but with an angle that allowed for a good view of both participants’ faces. Twenty different individuals are included in the Italian dataset, seventeen in the Tzeltal one and eighteen in the Yélî  Dnye one. The conversations recorded were ordinary and casual and not

Figure 7.1  Location of three cultures

Italian (Rossano)

Yélî Dnye (Levinson)

Tzeltal (Brown)

Gaze, questioning, and culture

205

solicited by the researchers. The participants of each interaction knew each other, being either relatives, friends or acquaintances. The interactions analyzed were dyadic because the presence of just two participants simplifies issues such as next-speaker selection (see e.g.,  L erner 2003) and who has been addressed by the speaker ( Goodwin 1979). Specifically, in a dyadic interaction if one person is talking, the other person is the one addressed. This means that if the speaker looks at the addressee during a turn, this is not done in the service of disambiguation to show who is being addressed and who is being selected to speak next. This allows gaze, freed from an address function, to serve a range of other interactional functions. This extensive database of interactions in the three cultures was sampled in a systematic way. Within each cultural sample, we selected roughly ten dyads and searched for question–response sequences until we had 300 such sequences for each cultural ­sample. By “question” we understood any utterance that functioned as an information soliciting action, regardless of whether it was in interrogative form or otherwise marked morphosyntactically, lexically or prosodically. For example, many “yes”/“no” questions in Tzeltal  and Yélî  Dnye are delivered in declarative format with falling intonation – they are recognizable as questions just because they appear to be statements about facts that are privileged information of the recipient’s, of the kind “You have a stomach ache” (cf. Labov and Fanshel’s  1977 “B-event statements” ).14 Note also that we excluded requests for things other than information: We were interested only in utterances that required a verbal or visible communicative response, which could thus be coded as an “answer,” “nonanswer response” or missing response. The question–response sequences collected in this way are not homogenous in function. Schegloff ’s (1984) point is well taken: An utterance can be a question and, at the same time, much more besides. Still, it is by virtue of being the first pair part of an adjacency pair  (cf. e.g.  Sacks 1987;  Schegloff 2007b; Schegloff  and  Sacks 1973), calling for a specific type of second pair part, that questions do the multifarious things they do. For this study we have focused on just three gross functions: requests  for new information, requests for repair  (e.g., for repetition  or clarification ), requests for confirmation  (of previously established information or

206

Federico Rossano et al.

presumed facts). The reason for distinguishing these is the different degree to which they push the ongoing business forward (i.e., different degrees of “progressivity”;  ( see Schegloff 2007b; Stivers  and  Robinson 2006), as opposed to invoking prior utterances or sequences. An other-initiated  repair, for example, is a momentary hitch, occasioning the insertion  of a sequence into whatever business is at hand – it is hoped and expected that completion of the repair sequence is immediate upon response, and work on the interrupted sequence can resume. Question–response sequences have a number of advantages for this study focusing on gaze behavior. First, the category  is “emic,” a category for participants – an unanswered question is a potential matter for complaint. Second , like in all adjacency pairs, there is turn-transition  relevance (cf. Sacks  et al. 1974) immediately after the first pair part; moreover, normally a question is delivered as a simplex unit, a single turn-constructional unit in the great majority of cases (see Extracts 1–4). Third, the two parts, question and response, potentially form a complete sequence, and again in the majority of cases actually do so. Thus over the course of a question–response sequence, two turns have been traversed, the former speaker has become a ratified recipient, and the former recipient a ratified speaker, and, in addition to speaker change occurring, a whole sequence has been jointly engendered. These are all important loci where gaze has been thought, or might be thought, to play a crucial role in organizing interaction. We should add that most if not all the prior claims and rules about gaze in interaction did not specify whether they would apply only within a specific culture or to specific actions. They are usually presented as applying generally to every single turn at talk whenever two people are talking to each other. If we find that those claims do not hold for at least one type of action (questions) and for one or more cultures, the generality and universality of such claims can be considered disproved. This would further confirm the importance of restricting claims in terms of the domain under investigation and, even more importantly, the necessity of moving beyond the “turn level” and including the details of the actions performed and the ­sequential organization of participants’ talk as important variables that affect gaze  in interaction, as proposed by  Rossano (2005; 2006a; 2009).

Gaze, questioning, and culture

207

We coded for the participant role (e.g., speaker vs. recipient), the primary action being implemented with the question, the question type, and the occurrence or not of a visible and/or verbal response. We then developed a refined coding  for gaze for both speaker and recipient. We wanted to capture, inter alia, absence of gaze at interlocutor and its possible cause, full gaze, unilateral vs. mutual gaze , and gaze at crucial loci within the first pair part. Details about the coding  can be found in the appendix. The inter-reliability coding performed on 40 percent of the gaze coding  for the Tzeltal  data showed a Cohen Kappa statistic k = .875, S.E. = 0.03, p < 0.0001 (Cohen 1960) which is considered an almost perfect agreement (Landis and Koch 1977).15 Results In this section, we report the main outcomes of the statistical analyses, commenting briefly on the import of each finding but leaving aside extended discussion until the next section. First, let us compare speaker gaze  and recipient gaze  during the question utterances, asking whether speakers or recipients gaze at their interlocutor at least once during the question (from here on we refer to these as Q-speakers and Q-recipients, to remind the reader that we are referring to speakers and recipients of questions only). A first surprise is that, as made clear in Table 7.1 and Figure 7.2, in each language Q-speakers look at Q-recipients more often than vice versa. In particular, speakers gaze  at addressees in 65.7 percent of the questions in Tzeltal , in 73 percent in Italian,  and in 79.7  percent in Yélî Dnye . Recipients gaze  at speakers in 42.3 percent of questions in Tzeltal, in 63.3 percent in Italian, and in 67.3 percent in Yélî Dnye . This is unexpected because it contrasts with all the above-mentioned works that found that participants Table 7.1   Instances of Q-speaker and Q-recipient gaze toward other participant, by language

Speaker Recipient

Tzeltal

Italian

Yélî Dnye

197 (65.7%) 127 (42.3%)

219 (73%) 190 (63.3%)

239 (79.7%) 202 (67.3%)

208

Federico Rossano et al. 90% 80%

Percentage

70% 60% 50%

Speaker gaze up

40%

Recipient gaze up

30% 20% 10% 0% Tzeltal

Italian

Yélî Dnye

Languages

Figure 7.2  Gaze toward other participant, by language

generally look more while listening than while speaking (see ­section “Earlier Work on Gaze in Interaction”), while it confirms what was shown in Extracts 1 and 2 in Italian. We therefore carefully checked the statistical reliability of more speaker than recipient gaze in each language through logistic regression analysis.16 Table 7.2 shows the results of the logistic regressions run for each language in which the standard error has been corrected for the clustering of the data by interaction.17 The difference between speaker and recipient gaze  behavior is statistically significant for each language. Table 7.2 shows that when a Tzeltal  participant acts as a Q-speaker, he or she is 2.61 times more likely to gaze at the addressee than when he or she acts as a Q-recipient. It shows that when an Italian  participant acts as a Q-speaker, he or she is 1.57 times more likely to gaze at the addressee than when he or she acts as a Q-recipient. It finally shows that when a Yélî Dnye participant acts as a Q-speaker, he or she is 1.90 times more likely to gaze at the addressee than when he or she acts as a Q-recipient. We turn now to consider possible cross-cultural  differences in gaze behavior. Figure 7.3 compares the gaze behavior of Q-speakers and recipients across the three languages. It shows that, in both, Q-speaker and recipient role participants look at each other more

Gaze, questioning, and culture

209

Table 7.2.  Logistic regression analysis predicting gaze toward other participant in relation to role (Q-recipient as reference group), by language Predictor

 β

S.E. β

p

Odds ratio

95% C.I.

Role Tzeltal (speaker) Constant

  0.96 −0.31

0.17 0.28

0.000 0.023

2.61

1.86, 3.66

Note: Model Evaluation: Wald χ2(1) = 30.53, p < 0.001 

Predictor

β

S.E. β

p

Odds ratio

95% C.I.

Role Italian (speaker) Constant

0.44 0.55

0.15 0.30

0.002 0.073

1.57

1.17, 2.09

Note: Model evaluation: Wald χ2(1) = 9.37, p < 0.01

Predictor

β

S.E. β

p

Odds ratio

95% C.I.

Role Yélî (speaker) Constant

0.64 0.72

0.23 0.32

0.006 0.276

1.90

1.21, 2.99

Note: Model Evaluation: Wald χ2(1) = 7.68, p < 0.01

90% 80%

Percentage

70% 60% Tzeltal Italian Yélî Dnye

50% 40% 30% 20% 10% 0%

Speaker gaze up Recipient gaze up Participation role

Figure 7.3  Gaze behavior across the three cultures clustered by participation role 

210

Federico Rossano et al.

Table 7.3.  Logistic regression analysis predicting Q-speaker gaze toward Q-recipient in relation to language (Italian  and Tzeltal  as reference group) Predictor

 β

S.E. β

p

Odds ratio

95% C.I.

Tzeltal (speaker) Yélî (speaker) Constant

−0.35   0.37    0.99

0.40 0.29 0.25

0.385 0.212 0.000

0.71 1.45

0.32, 1.54 0.81, 2.59

Note: Model evaluation: Wald χ2(2) = 4.85, p = 0.09 

Predictor

β

S.E. β

p

Odds ratio

95% C.I.

Italian (speaker) Yélî (speaker) Constant

0.35 0.72 0.65

0.40 0.35 0.30

0.385 0.039 0.036

1.41 2.05

0.65, 3.09 1.04, 4.05

Note: Model evaluation: Wald χ2(2) = 4.85, p = 0.09

in Yélî  Dnye than in Italian  and more in Italian than in Tzeltal , although the significance of these differences in frequencies will be shown to vary depending on the participation role . Let us first take the question of speaker gaze: Are  Q -speakers more likely to gaze at their recipients during questioning in one of the languages compared to the others? If we compare Italian  and Yélî Dnye , the answer is no: Taking Italian as the comparison language, there is no significant difference between Q-speakers’ gaze  behavior in Italian and Yélî  Dnye (see Table 7.3, top half) and equally no significant difference between Q-speakers’ gaze  behavior in Italian and Tzeltal . However, when we take Tzeltal as the comparison language, there is a significant difference in Q-speaker gaze  between Tzeltal and Yélî  Dnye (see Table 7.3, bottom half). Tzeltal Q-speakers are significantly more economical with their gaze to recipients than Yélî Dnye Q-speakers. These latter results, however, should be taken with extreme caution, as the model evaluation shows that the Wald statistic is not significant. This means that the predictor (language) should be rejected as a good predictor of the outcome (Q-speaker gazing at Q-recipient). Language in general does not significantly predict speaker gaze , at least for these three languages.

Gaze, questioning, and culture

211

Table 7.4  . Logistic regression analysis predicting Q-recipient gaze toward Q-speaker in relation to language (Italian  and Tzeltal   as reference group) Predictor

 β

S.E. β

p

Odds ratio

95% C.I.

Tzeltal (recipient) Yélî (recipient) Constant

−0.85   0.18    0.55

0.40 0.42 0.29

0.035 0.677 0.062

0.43 1.19

0.19, 0.94 0.52, 2.74

Note: Model evaluation: Wald χ2(2) = 7.36, p < 0.05 

Predictor

 β

S.E. β

p

Odds ratio

95% C.I.

Italian (recipient) Yélî (recipient) Constant

  0.85   1.03 −0.30

0.40 0.41 0.27

0.035 0.013 0.273

2.33 2.79

1.06, 5.16 1.24, 6.27

Note: Model evaluation: Wald χ2(2) = 7.36, p < 0.05

In short, Table 7.3 shows that Italian  Q-speakers’ gaze behavior  does not differ significantly from Tzeltal  and Yélî  Q-speakers’ gaze  behavior, while Tzeltal and Yélî  differ significantly but in a model that suggests language does not properly predict differences in Q-speaker gaze behavior . Let us now turn to the gaze behavior of the recipient during questioning. Once again, there is no significant difference between Italian  and Yélî  gaze behavior, now considering Q-recipients (Table 7.4, top half). But both Italian and Yélî are strikingly different from Tzeltal in this regard: The difference between Q-recipient gaze behavior in Tzeltal and the other two languages is significant (see Table 7.4, bottom half) and so is the model used to test these differences (see model evaluation of Table 7.4). Table 7.4 shows that Yélî  Dnye Q-recipients’ gaze behavior is not significantly different from Italian  Q-recipients’ gaze, whereas Tzeltal  Q-recipients’ gaze behavior is significantly different from Italian and Yélî  Q-recipients’ gaze. In particular, Italian and Yélî  Q-recipients are respectively 2.33 times and 2.79 times more likely to look at the speaker during a question than a Tzeltal Q-recipient is. In sum, Q-speakers’ gaze behavior is fundamentally similar across the three languages.18 Although there is a significant

212

Federico Rossano et al.

difference between the two extremes (Tzeltal  and Yélî  Dnye), Italian  stands in the middle and is not significantly distinct from either. Moreover, the evaluation of the model shows that language is not a good predictor of differences in speaker gaze. All three languages show a significantly greater proportion of speaker gaze compared to recipient gaze during questioning. On the other hand, gaze behavior by the Q-recipient shows more fundamental differences in the case of Tzeltal: Now Italian and Yélî Dnye  show similar, relatively high patterns of gaze by the Q-recipient to the speaker, while Tzeltal shows significantly less. These results suggest a strong tendency to uniformity in speaker gaze behavior, despite the literature (reviewed above) that suggests that speaker gaze in general varies radically across cultures. In contrast, there is a clearer cultural difference in the way in which gaze is deployed by recipients, and the implication is of course that gaze plays a differential role across cultures as a signal of active recipiency . If we now look at the occurrence of mutual gaze  between Q-speaker and Q-recipient, that is, whether mutual gaze is engaged sometime during the course of asking a question, we find that mutual gaze occurs in just over 50 percent of questions in Italian  and Yélî  Dnye, but in only 33.7 percent of questions in Tzeltal  (see Table 7.5). Once again, this aligns Italian and Yélî  Dnye as highgaze cultures, with Tzeltal as a low-gaze culture, along the lines shown in Brown   and Levinson ’s previous comparative work on Tzeltal and Yélî  Dnye (2005; Levinson and Brown 2004). Table 7.6 shows that Tzeltal is significantly different from Italian  and Yélî  Dnye in terms of the occurrence of mutual gaze  during a question, while Italian and Yélî  Dnye do not differ significantly. We noted earlier that some of the gaze literature (e.g., C.  Goodwin 1980; Goodwin 1981) suggests that as a rule speakers check that the recipient is attending, expecting to find a recipient gazing back at that point. This predicts of course that mutual gaze should occur in nearly all cases in which the speaker looks at the recipient, and where not, remedial action (e.g., getting attention by restarting an utterance) might be undertaken. The numbers for our data do not support any such rule, at least for question–answer sequences. Consider that in approximately 20 percent of questions in Italian  and Yélî Dnye , and in more than 30 percent of questions in Tzeltal,19  the

Gaze, questioning, and culture

213

Table 7.5.  Instances of mutual gaze   by language

Mutual gaze

Tzeltal

Italian

Yélî Dnye

101 (33.7%)

159 (53%)

173 (57.7%)

Table 7.6.  Logistic regression analysis predicting mutual gaze  in relation to language (Italian  and Tzeltal   as reference group) Predictor

 β

S.E. β

p

Odds ratio

95% C.I.

Tzeltal (MG) Yélî (MG)

−0.83   0.18

0.41 0.42

0.042 0.672

0.44 1.19

0.20, 0.97 0.53, 2.69

Note: Model evaluation: Wald χ2(2) = 6.37, p < 0.05 

Predictor

β

S.E. β

p

Odds ratio

95% C.I.

Italian (MG) Yélî (MG)

0.83 1.00

0.41 0.43

0.042 0.020

2.29 2.72

1.03, 5.06 1.17, 6.35

Note: Model Evaluation: Wald χ2(2) = 6.37, p < 0.05

speaker is gazing at the recipient and the recipient does not gaze back before completion of the turn (see Table 7.7), and no remedial action is taken (see Extracts 1 and 2 for further qualitative evidence). We turn now to consider the distribution of gaze across finer coding  categories. As mentioned, questions can implement rather different actions, and our coding scheme characterized these accor­ ding to their “progressivity,”  distinguishing information-seeking questions from repair initiation questions and from confirmation seeking questions. Table 7.8 and Figure 7.4 show the distribution by language of these main classes of action. The residual category “Other” includes different actions such as offers , invitations , requests  for further clarification , topic  profferings, etc., that were implemented through the questions – they did not amount to a large number of instances for any given action for each language. There are slight differences in the occurrence of these different types. For example, Yélî  Dnye has more requests  for information

214

Federico Rossano et al.

Table 7.7.   Questions in which speakers look at recipients and recipients do not look back, by language

Speaker gaze, no recipient gaze

Tzeltal

Italian

Yélî Dnye

96 (32%)

59 (19.7%)

66 (22%)

Table 7.8.   Distribution of actions performed by questions, by language

Request for information Other-initiated repair Request for confirmation Other

Tzeltal

Italian

Yélî Dnye

107 (35.7%) 34 (11.3%) 115 (38.3%) 44 (14.7%)

126 (42%) 43 (14.3%) 68 (22.7%) 63 (21%)

143 (47.7%) 53 (17.7%) 84 (28%) 20 (6.6%)

60.0% 50.0% 40.0% Tzeltal Italian Yélî Dnye

30.0% 20.0% 10.0% 0.0%

Req. for info

O.I. repair

Req. for conf.

Other

Figure 7.4  Distribution of actions by languages

and other initiation of repair than the other two languages, while Tzeltal  has more requests for confirmation  . Again, the possibility arises that these differences in the frequencies of action types could account for some of the cross-cultural  differences in gaze. If, for example, gaze is especially associated with requests for information and repair-initiating questions, given

Gaze, questioning, and culture

215

Table 7.9.   Q -speaker looking at Q-recipient, by action and by language

Request for information Other-initiated repair Request for confirmation

90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0%

Tzeltal

Italian

Yélî Dnye

71 (66.4%) 26 (76.5%) 81 (70.4%)

87 (69%) 33 (76.7%) 50 (73.5%)

110 (76.9%) 43 (81.1%) 72 (85.7%)

Tzeltal Italian Yélî Dnye

Req. for info.

O.I. repair Actions

Req. for conf.

Figure 7.5  Distribution of percentages of each action in which the Q-speaker looks at the addressee, by language

that there are more of these in Yélî Dnye than in Tzeltal , this could explain the distribution of gaze shown in Figure 7.3. However, Table 7.9 and Figure 7.5 show that this is not the case. For each action, Q-speakers look more in Yélî  Dnye than in Italian  and more in Italian than in Tzeltal, although the difference between Q-speaker’s gaze behavior in Italian and Tzeltal is minimal, exactly as shown in the general distribution. It is noticeable here that, of the three actions, in each language requests for information are produced less often with speaker gaze than are other-initiations of repair and requests for confirmation  . Considering Rossano ’s (2005, 2009) claim that gaze behavior is mainly organized around larger structures than turns (sequences and courses of action) and that most of the shifts in gaze direction should occur at the beginning and end of these structures and around self-repair and speech disfluencies (on this point cf.  B eattie, 1979), the differences in the amount of gaze per action

216

Federico Rossano et al.

could be accounted for by their sequential position. The fact that other-initiations of repair do not occur at the beginning of a sequence or course of action, and the same usually holds for requests  for confirmation, whereas  requests for information can occur in first position, may account for this difference. Another general fact emerging from all three cultures is that questions initiating repair are especially likely to involve both speaker and recipient gaze. One possibility is that gaze here reinforces that the Q-speaker, though speaking, is committing to a more attentive engagement as a listener in the conversation. The speaker of the repair initiation is asking the recipient to repair his or her prior talk because of some problem, and the speaker of the repair question can ask to delay the progressivity  of the talk by projecting a full recipiency  once the repair is produced. Moreover, if speaker gaze can exert pressure for a response, this pressure can be particularly valuable given the social cost of asking a prior speaker to repair his or her own prior talk and the additional social cost of not being able to provide an appropriate response in case of troubles in hearing or understanding. Alternatively, one may take other-initiation of repair to convey disaffiliation, non-alignment  or some other problematic stance  toward the prior utterance; it may therefore be a kind of interactionally “loaded” moment, comparable to challenges , which are usually delivered while looking at the addressee. Independently of which is the favored account, it is clear that delaying the progressivity of the conversation and asking for repair is an interactionally problematic action, and the occurrence of more gaze by both speakers and recipients might be related to this potential problematicity. We turn now to consider Q-recipient gaze within each action type. Table 7.10 and Figure 7.6 show that, for each action, the distribution of Q-recipient gaze follows the general distribution of recipient gaze as shown in Figure 7.3. This means that Yélî  Dnye and Italian Q-recipients look more during each action than Tzeltal  speakers do, and the difference between Italian and Yélî  Dnye is less marked than the difference between these two languages and Tzeltal. It is noticeable that in each language other-initiation of repair tends to get gaze more often from the Q-recipient than the other two actions. Moreover, in Italian and Yélî Dnye, requests for confirmation   are the actions that are least likely to elicit gaze from

Gaze, questioning, and culture

217

Table 7.10.   Q -recipient looking at Q-speaker, by question type and by language

Request for information Other-initiated repair Request for confirmation

Tzeltal

Italian

Yélî Dnye

40 (37.4%) 18 (52.9%) 55 (47.8%)

80 (63.5%) 28 (65.1%) 38 (55.9%)

100 (69.9%) 40 (75.5%) 51 (60.7%)

80.0% 70.0% 60.0% 50.0%

Tzeltal

40.0%

Italian Yélî Dnye

30.0% 20.0% 10.0% 0.0% Req. for info

O.I. repair

Req. for conf.

Action

Figure 7.6  Distribution of percentages of each action in which the Q-recipient looks at the Q-speaker, by language

recipients, while in Tzeltal it is requests for information that are the actions least likely to get gaze from the recipient. As we noted at the outset, questions are perceived as first pair parts of adjacency pairs, making a response by the recipient the relevant next action. This is the case independently of whether they are initiating a sequence, they are inserted in it, or they are working as post-expansion . Table 7.11 shows how many of our questions got proper answers, responses that would not count as proper answers (e.g., “I don’t know, maybe”) or no response at all. The results show that in each language proper answers are produced in response to questions two-thirds of the time or more. Moreover, the number of questions that receive no response at all is minimal (on average, in 12 percent of cases).20 It is clear that overwhelmingly questions get responded to (on average, in 88 percent of cases; see Table 7.12).

218

Federico Rossano et al.

Table 7.11.  Distribution of response types by language

No response Non-answer responses Answers

Tzeltal

Italian

Yélî Dnye

15 (5%) 41 (13.7%) 245 (81.7%)

45 (15%) 71 (23.7%) 184 (61.3%)

50 (16.7%) 52 (17.3%) 198 (66%)

Table 7.12.  Distribution of responses to questions, by language

No response Response

Tzeltal

Italian

Yélî Dnye

14 (4.7%) 286 (95.3%)

45 (15%) 255 (85%)

50 (16.7%) 250 (83.3%)

Gaze per se does not, nor did we expect it to, account for the occurrence of an answer or a nonanswer response, as action types, question format, and many other factors can come into play here. There is, however, an interesting correlation between lack of Q-recipient gaze (and of mutual gaze  as a consequence) and the fact that an answer does not get a response. If we take Q-recipient gaze to be a means for displaying recipiency , we should expect that when Q-recipient gaze is lacking there should be some interactional consequence. Given that here we are looking at questions, the most notable consequence would be problems in responding to a question. We are here considering only the most macro display of “trouble” in responding: i.e., not responding at all. It is clear that other micro-consequences can occur and would be worth examining in future work. Notice that we are not expecting that every time the recipient is not looking there will be no response but rather that absence of Q-recipient gaze can be a good predictor of upcoming trouble in responding, or, more specifically, of no response. Extract 5 (as well as Extract 1) provides a qualitative example of cases when the Q-recipient is not looking at all during the ­question and there seems to be a problem and a response is lacking. In Extract 5, two male Italian  undergraduate students are preparing for an exam. B is asking A questions and is holding the notes they have used to prepare for the exam. B has already passed the exam while A has to take it the following day. The topic being  discussed

Gaze, questioning, and culture

219

is the kinds of diseases that one should report to the officials while working as a veterinarian. We focus on the questions at line 1, 3, and 5. (5) ITALIAN 2PEXAM-sapere (42:49) A

B

A

B

01 →A: Ma lei vuol sapere ste cose qui? but she want.3s to know these things here Does she want (us) to know these things here? A

02

B

(1.6) A

03 →A:

B

Eh? eh Eh? A

B

A

04 (1.0)

B

A



(0.2) (1.6)

A

B

(0.4)

B

A

B

A

B

05 →A: Lei son ste cose qua [che vuole sapere?] she are these things here that want.3s to know Are these the things [that she wants (us) to know?] A

B

A

B

A

B

06 B: [Lei vuole sape]re quelle she want.3s to know those [She wants (us) to know those A

B

07 ‘segnalazioni immediate’ = si’ questo e’ importante signaling immediate yes this is important ‘inform immediately’ = yes this is important

In all three questions the recipient is not looking at the speaker but at the notes (although the answer to the questions is not in the notes). The question at line 1 is not answered, and A pursues it at lines 3 and 5, until he finally gets an answer at lines 6–7 (and

220

Federico Rossano et al.

Table 7.13.  Distribution of questions that do not get responded to in which the recipient does not look at the speaker

No recipient gaze

Tzeltal

Italian

Yélî Dnye

9/14 (64.3%)

23/45 (51.1%)

32/50 (64%)

B looks toward A while producing it). Notice that even though B answers, the formulation of the answer is not straightforward (the “yes” comes only mid line 7). B was not looking at A during the questions and it is clear that he had problems dealing with the question(s). See also Extract 1 for similar evidence. If we focus purely on the numbers from our dataset, we see that in Italian  and Yélî  Dnye, the proportion of gaze absence by the recipients of the questions that are not responded to is higher than in the total of the questions (see Table 7.13). 21 We can test whether lack of recipient gaze is a good predictor of lack of response by dividing the response possibilities into response vs. no response, as in Table 7.12. We tested this hypothesis by creating a model that takes into account other variables such as speaker gaze, recipient gaze, and language spoken (Table 7.14). Table 7.14 shows that, in general, lack of Q-recipient gaze is a good predictor of lack of response after a question, while speaker gaze and language spoken are not (at least for this dataset). Table 7.15 shows that lack of recipient gaze works differently in each language. It is a significant predictor of lack of response in Italian  and Yélî Dnye , but not for Tzeltal . This means that while the proportion of no responses preceded by lack of Q-recipient gaze is significantly higher than for all questions in Yélî  Dnye and Italian, this is not the case for Tzeltal. These results are particularly relevant because they show that, in Italian and Yélî Dnye , Q-recipient gaze appears to be related to the doing of recipiency,  and, in this respect, the lack of it can predict possible upcoming trouble in terms of the occurrence of a response to the question. However, in Tzeltal,  recipient gaze does not seem to have this function, and this fits with the general tendency for Tzeltal recipients to gaze less at the speaker than in the other two

Gaze, questioning, and culture

221

cultures. This raises important issues about the general function of gaze as a display of recipiency: It suggests that this role for gaze is not universal , and there must be other possible ways in which good recipiency can be signaled. All the results just outlined distinguish questions in which the participants looked at any time toward the other participant from those in which the participant never looks at the other. The coding  scheme we had developed, however, was more specific, as it distinguished thirteen possible scenarios (six distinct locations for gaze for each of speaker vs. recipient, and the possibility of mutual gaze ). Once these data are taken into account, the results are again surprising, as they contrast with claims by Kendon  (1967) for a wider range of actions about when speakers typically look away and when they typically look at the recipient within the turn. Tables 7.16 and 7.17 show that, in each language, only in a small minority of questions do participants look up during the question if they were not already looking at its beginning. In particular, for Tzeltal  Q-speakers, this happens in 10.7 percent of the questions, in Italian  in 16.4 percent, and in Yélî  Dnye in 12.3 percent (these percents are the sums of rows 4, 5, and 6 in Table 7.16). More specifically, if  looking toward the recipient near the completion of an utterance invites speaker transition (see Kendon 1967), row 5 in Table 7.16 shows that in our question–answer  data this happens very infrequently. For Tzeltal this happens in 7.7 percent of the questions, for Italian in 5 percent, and for Yélî  Dnye in 5 percent. Table 7.17 shows percentages for Q-recipients of gaze ­“immobility”  within a question similar to those in Table 7.16 for

Table 7.14.   Logistic regression analysis predicting no response Predictor

β

S.E. β

p

Odds ratio

95% C.I.

No recipient gaze Speaker gaze Tzeltal Yélî Constant

  0.61 −0.31 −1.00 −0.29 −1.78

0.25 0.23 0.54 0.41 0.35

0.015 0.191 0.063 0.481 0.000

1.84 0.73 0.37 0.75

1.12, 3.02 0.46, 1.17 0.13, 1.06 0.33, 1.68

Note: Model evaluation: Wald χ2(6) = 84.25, p < 0.00001

222

Federico Rossano et al.

Table 7.15.  Logistic regression analysis predicting no response if Q-recipient is not gazing, by language Predictor

 β

S.E. β

p

Odds ratio

95% C.I.

No recipient gaze (Tzeltal) Constant

   0.11

0.54

0.842

1.11

0.39, 3.21

−3.00

0.42

0.000

Note: Model evaluation: LR χ2(2) = 0.40, p = 0.84 

Predictor

 β

S.E. β

p

Odds ratio

95% C.I.

No recipient gaze (Italian) Constant

  0.70

0.33

0.031

2.02

1.07, 3.83

−2.03

0.23

0.000

Note: Model evaluation: LR χ2(2) = 4.61, p < 0.05

Predictor

 β

S.E. β

p

Odds Ratio

95% C.I.

No recipient gaze (Yélî) Constant

   1.59

0.33

0.000

4.92

2.58, 9.37

−2.29

0.39

0.000

Note: Model evaluation: LR χ2(2) = 25.22, p < 0.001

Table 7.16.   Q -speaker gaze, by language

1. No gaze 2. Gaze up all the time 3. Gaze up at beginning, away before completion 4. Gaze up on new referent 5. Gaze up approaching completion 6. Gaze up any other time

Tzeltal

Italian

Yélî Dnye

103 (34.3%) 146 (48.7%) 19 (6.3%)

81 (27%) 162 (54%) 8 (2.7%)

61 (20.3%) 187 (62.3%) 15 (5%)

2 (0.7%) 23 (7.7%)

8 (2.7%) 15 (5%)

7 (2.3%) 15 (5%)

7 (2.3%)

26 (8.7%)

15 (5%)

Q-speakers. In particular, a Q-recipient looks toward the speaker’s face after the beginning of a question in 15.4 percent of questions in Tzeltal , in 16 percent in Italian,  and in 7.7 percent in Yélî  Dnye (the sums of rows 4, 5, and 6 in Table 7.16).

Gaze, questioning, and culture

223

Table 7.17.   Q -recipient gaze, by language

1. No gaze 2. Gaze up all the time 3. G aze up at beginning, away before completion 4. Gaze up on new referent 5. G aze up approaching completion 6. Gaze up any other time

Tzeltal

Italian

Yélî Dnye

173 (57.7%) 63 (21%) 18 (6%)

110 (36.7%) 120 (40%) 22 (7.3%)

98 (32.7%) 162 (53.3%) 17 (5.7%)

0 (0%) 29 (9.7%)

4 (1.3%) 14 (4.7%)

1 (0.3%) 11 (3.7%)

17 (5.7%)

30 (10%)

11 (3.7%)

These results contrast with the patterns claimed by  Kendon (1967) to occur in English and suggest, as argued by Rossano  for  Italian (2005; 2009), that the utterance per se is not an adequate level for describing gaze behavior in interaction, as very little seems to happen to gaze direction in dyadic interactions during a single utterance. 22 Finally, we noticed that looking away, or better our coding  “no gaze toward addressee” could actually refer to three different situations: 1. The participant is just looking down or away but at nothing in particular. 2. The participant is looking away because of the content or the delivery of the question (e.g., someone is pointing and looks in the direction of the point, or is talking about an object present in the surrounding environment and looks at it). 3. The participant is involved in a competing activity (e.g., ­eating, weaving, drawing on the sand). We therefore monitored whether the lack of gaze toward the addressee could be considered “motivated” or not and whether this alternative focus was related to the talk or to a competing activity. Tables 7.18 and 7.19 show the great cross-cultural difference in terms of what else is going on while talking and how often people’s eyes get redirected toward different semiotic entities. Tables 7.18 and 7.19 show that, while in our Italian  interactions Q-speakers and Q-recipients can be seen to be looking away for a possible reason around 80 percent of the time, this happens around 50 percent of the time in Yélî  Dnye interactions and

224

Federico Rossano et al.

Table 7.18.   Questions in which lack of Q-speaker gaze toward Q-recipient motivated by question content or competing activity.

Total motivated Q. delivery/content Other activity

Tzeltal

Italian

Yélî Dnye

6/103 (5.8%) 5/103 4/103

70/81 (86.4%) 16/81 62/81

33/61 (54.1%) 20/61 21/61

Note: The sum of “gaze orientation motivated by question delivery/content” and “gaze orientation motivated by competing activity” does not correspond to the “Total motivated gaze orientation” because in some instances participants are conflating the two possible motivations, e.g., if someone points to an object in a picture while the competing activity is looking at pictures or if someone is drawing on the sand and uses deictics in his or her speech to refer to the object just drawn.

Table 7.19.   Questions in which lack of Q-recipient gaze toward Q-speaker motivated by question content or competing activity

Total motivated Q. delivery/content Other activity

Tzeltal

Italian

Yélî Dnye

8/173 (4.6%) 4/173 4/173

85/110 (77.3%) 20/110 77/110

42/98 (42.9%) 10/98 39/98

only around 5  percent of the time in Tzeltal  interactions. This means that in the culture where most of the time participants do not look at each other (Tzeltal) this is not because they are looking in the direction of a pointing finger or because they are usually involved in competing activities. Rather, they tend to look down, toward their hands or legs or in the middle distance but not at a specific object or in a recognizable direction. As will be explained in the following section, speakers of Tzeltal and speakers of Italian inhabit very different material worlds, and the interactions were recorded mainly at home in Italian, and mainly outside the home in Tzeltal and outside in Yélî. It  is remarkable that the “nonlooking” at the other participant is mainly a clear looking at something else in the Italian interactions, and half of the time in the Yélî  interactions, but almost always looking down or mid-distance in Tzeltal. This might be another cultural difference in terms of displaying attentive recipiency. If, in an Italian conversation, one is expected to be looking at the speaker most of the time, then not looking

Gaze, questioning, and culture

225

at the speaker is accountable and should be “motivated”. In contrast, if the default of recipiency is not looking at the speaker, then ostensive focus toward other objects or activities might mean that the recipient is not listening to the speaker at all. So by not looking at anything else, Tzeltal recipients might simply be displaying full commitment to attending to the conversation, while the same behavior enacted by an Italian participant would appear to display complete uninterest toward the conversation. In other words, speakers in these two cultures are dealing with the same problems (full recipiency and commitment to listening) using opposite visible displays. These behaviors follow directly from the nature of preferred gaze behavior in conversation within their respective cultures. To sum up this section, the following general patterns of gaze in question–answer  sequences across the three cultures emerged: 1. Q-speakers gaze at recipients in about two-thirds of the questions, but Q-recipients’ gaze is more variable and less frequent. 2. Q-speakers gaze least on the most progressive actions. 3. Q-recipients gaze most on the least progressive actions. 4. Mutual gaze  is far from assured during such sequences and never occurs in much more than half of the cases. There are, however, some striking cross-cultural  differences, most clear in the Tzeltal  vs. Yélî  Dnye contrast: Here, both Q-speakers and Q-recipients gaze significantly more in Yelî Dnye than in Tzeltal. In the Tzeltal case, gaze is not a good indicator of active recipiency,  and it plays no role as a predictor of whether a response will be forthcoming. Moreover, nonlooking at the other participant usually means very different types of orientation in the three cultures. In particular, participants do very different things with their eyes if they do not orient them toward the other person: They are mainly oriented toward competing activities in Italian  interactions, they are oriented toward competing activities or to locations motivated by question delivery only half of the time in Yelî Dnye interactions, and they are mainly oriented down or middistance in Tzeltal interactions. These differences cannot be accounted for in terms of different patterns in the uses of questions within the three cultures.

226

Federico Rossano et al.

Reasons for, and consequences of, cultural variability in gaze behavior in interaction Brown  and  Levinson (2005; Levinson and Brown 2004) have noted, based on many years of observation, that differences in gaze behavior between Yélî Dnye and Tzeltal speakers correlate with differences in observed seating patterns. Broadly speaking, Yélî  Dnye speakers of Rossel  Island prefer to sit face-to-face within reach of each other, while Tzeltal speakers of Tenejapa prefer to sit side by side or at an angle.23 In our samples of interaction, these patterns are indeed replicated. There are, of course, additional cultural constraints here – Rossel Islanders do not invite people of other families into their homes, and they have no chairs, consequently all interactions were filmed standing, squatting or seated on the ground in the open air. The Tenejapan home is also relatively private, and visitors would normally be seated on benches or small chairs on a patio or enclosed public area: The recordings are therefore filmed either outside or in the less private parts of the home. The Italian interactions were filmed inside the home, where seating is normally arranged around a table for face-to-face interaction. Our recordings show that participants speaking Italian  and Yélî Dnye  prefer sitting face-to-face during conversation, while participants speaking Tzeltal prefer sitting side by side. 24 However, as can be seen in Figures 7.7–7.9, in all three cultures participants can be seen sitting side by side or face-to-face and engaging mutual gaze  independently of their body position. The fact that in each culture we can observe both side by side and face-to-face configurations, and in both cases participants can be seen to engage mutual gaze , suggests a certain caution in distinguishing these cultures in terms of typical or normative seating configurations. Nevertheless, in our sample in Italian  and Yélî  Dnye, in most of the dyadic interactions participants sit face-toface, whereas in Tzeltal  they sit side by side. 25 These different seating configurations would suggest that in Tzeltal looking at each other is a more marked act (see, e.g.,  Schegloff 1998c on body torque), given that it is often necessary to turn one’s head to look at the other participant (Brown and  L evinson, 2005; Levinson and Brown, 2004), and this, of course, is consonant with the less frequent gaze in interaction. On the other hand, we should bear in mind

Gaze, questioning, and culture

227

Figure 7.7  Italian  dyads

that in our dataset Tzeltal speaker gaze is not significantly ­different from Italian speaker gaze, and this shows that there are distinct limitations on the ecological determinism of gaze behavior. The early literature on human kinesics  also throws considerable doubt on any theory  of ecological determinism of gaze behavior .

228

Figure 7.8  Tzeltal  dyads

Federico Rossano et al.

Gaze, questioning, and culture

229

Figure 7.9  Yélî Dnye  dyads

These studies showed that participants tend to arrange themselves according to the interactions they plan to have (Ekman and Friesen 1974;  Kendon 1977; Sommer 1959, 1962). While it is true that the physical structure of the environment constrains interactional positioning ( Goffman 1971;  Goodwin 1981), these

230

Federico Rossano et al.

physical arrangements tend to directly reflect the preferred cultural patterns. In short, we would argue that the Tzeltal  seating patterns, for example, directly reflect the preference for more limited, more controlled gaze behavior, in just the same way that the unfurnished and thus more unconstrained sitting arrangements of Yélî  Dnye speakers favor the eyeball-to-eyeball interaction that is the ­dominant mode. So what does account for the cultural differences? Brown  and  L evinson (2005; Levinson and Brown 2004) note that the contrast between Tzeltal and Yélî Dnye is partly based on different practices for the display of recipiency . For example, during extended turns at talk such as a telling, Tzeltal recipients are expected to respond at regular intervals with significant verbal material, repeating parts of the immediately prior utterance (Brown 1979: Chapter 4, 1998, 2007). In contrast, Yélî  Dnye speakers have an extended inventory of visual feedback signals, including facial gesture s specialized for assent, surprise, continuer function, and so forth ( L evinson 2007). The Yélî  system presupposes the likelihood of gaze (or at least close peripheral vision), while the Tzeltal system seems built to assume its absence. The low-key role that gaze plays in Tzeltal interaction implies, as mentioned, other effective displays of recipiency. Extract 6  shows how repetitions can signal recipiency and understanding of what has been asked in a much more precise way than is observable in Italian  and Yélî Dnye. 26  It shows a “yes”/“no” (polar) question– answer  sequence in Tzeltal. This kind of question particularly allows an answer to be constructed by repetition  with minimal modification – a repeat rather than “yes” is in fact the default for an affirmative response to a “yes”/“no” question (unlike what was shown by  Raymond [2003] for conversations in American English ); the Tzeltal words for “yes” have much more restricted functions ( Brown 1998, in press). (6)  Tzeltal T002020, 13:13 01 AMT: kuxul-0 to wan s-tukel i (.) muk’ul jme’tik i? living-3A still perhaps 3E-self DEIC big Mrs. DEIC? Is “big Mrs.” ((honorific for “your mother”)) still alive? 02 CH: kuxul-0 living-3A She’s alive.

Gaze, questioning, and culture

231

Extract 6, together with Extract 3, illustrate how repeating part of the polar question in the answer is an appropriate and typical way of responding in Tzeltal . 27 If one probes further for why Tenejapan Tzeltal  speakers are relatively gaze aversive, there is little doubt that it has to do with politeness  and decorum: Tzeltal interactors will even turn their backs, or hide behind a structure, if the conversational matter is in any way face-threatening (e.g., in the case of substantial requests  – see  Brown 1979 for details).  Darwin (1872) was particularly interested in gaze aversion  as a symptom of self-conscious shyness and shame, associated especially with the appropriate demeanor of Victorian women. From this, it seems to acquire the semiotics of self-denigration, which figures in many honorific systems (see Brown and  L evinson 1987). Such an account predicts gaze aversion, especially in disprefer red responses, which needs further checking in the data. If the semiotics of gaze aversion has a “natural” source, we may expect this to hold in all three cultures. Implications for general patterns of gaze in interaction As already mentioned, one of our main cross-cultural ly stable findings is that Q-speaker gaze is significantly more expectable than Q-recipient gaze in each culture. This was not predicted by the prior literature, where, on the contrary, the assumption has been that generally recipients are expected to gaze, and speakers may well only glance at recipients somewhere in the course of the utterance, e.g., toward its end (e.g., Duncan  and Niederehe 1974;  Goodwin 1981;  Kendon 1967). Two further points are in order. Some of these earlier claims are phrased in terms of rates of gaze, i.e., gaze over time, while our data has been coded as gaze over turns (question or response). Nevertheless, since we have found that if there is gaze it tends to be more or less constant over these units (see Tables 7.16 and 7.17), our findings too imply a temporal preponderance of speaker gaze. This constancy of gaze during the asking of a question also contrasts with another general claim, made most forcefully by  Kendon (1967), according to which speakers tend to give quick glances at recipients while recipients tend to hold their gaze toward the speaker. Our results show that only rarely will a participant start

232

Federico Rossano et al.

glancing at the other and then look away and maybe glance back again during a question. 28 Where speaker gaze is momentarily averted this seems to be occasioned by gaze pointing for the other participant, or, in the case of recipient gaze, by the need to redirect the eyes toward the object the other participant is talking about. Gaze redirection in these instances – looking away from the recipient – is not related to the management of turn-taking . The general lack of mobility in gaze direction within a question has implications for the proper level of organization for gaze in interaction. A question typically consists of a single turn constructional unit (TCU). This suggests a general equivalence between our unit “question” and a turn at talk, and, indeed, 90 percent of the questions of the Italian  corpus, and 77 percent of those in the Tzeltal  and in the Yélî  Dnye corpora, constitute a turn by themselves. It follows straightforwardly that gaze cannot be playing an internal function within the turn, e.g., by warning of turn transition (as Duncan and Fiske 1977; Duncan and Niederehe 1974;  Kendon, 1967 had supposed). For the same reasons, C. Goodwin’s  (1980; 1981) “recipiency”  rules for American English  interactions do not seem to hold for question turns in these three languages. Recollect that, according to these rules, the recipient’s gaze should be toward the speaker by the time the speaker looks at the recipient, failing which the speaker can solicit such attention through, for example, self-repair . Table 7.7 has already shown that in 20–30 percent of questions in our samples, the speaker looks but the recipient fails to look back, and clearly the speaker has not extended the turn to successfully achieve mutual gaze . In Italian, for example, in only four out of fifty-nine (or 7 percent) questions of this kind do sound stretches, cut-offs or pauses occur.29 If we look at the thirty-eight Italian questions in which the speaker is gazing from the beginning but the recipient looks up only later, we find a cut-off or a sound stretch in only seven (and only three seem to have any timing correlation to the return of recipient’s gaze toward the speaker). The difference between Goodwin’s   findings and ours may be due to a number of factors, in particular related to the sampling of the data and the details of the coding . For example, we looked only at question sequences, while Goodwin’s data was coded intact and no particular attention was paid to the specific actions that the participants were implementing with their turns at talk

Gaze, questioning, and culture

233

(C. Goodwin 1980). Still, it is possible that Goodwin’s account in terms of speakers’ restarts eliciting recipient gaze has captured a practice that is especially prominent in American English .30 In any case, what is clear is that neither the importance of obtaining the recipient’s gaze before the completion of the turn nor the general normativity of this behavior can be observed in our samples. What role could the restriction to question sequences play in our data? Clearly, by being first-pair parts of adjacency pairs, questions imply turn transition, and in a dyadic situation they also imply to whom the next turn belongs. If gaze here plays no addressee-selection role, it may especially play the role of exerting pressure for a response ( Rossano 2006b, 2009; Stivers and  Rossano in press). It is possible too, that gaze in questions is especially devoid of turn-transition functions, since recipients will know, as soon as they detect a question, that on completion an answer is immediately relevant. Despite finding little evidence for a role for gaze  in regulating turn-taking  in our three cultures, there does seem to be a correlation between the variation in gaze behavior within a question and the sequential position of the question. This trend can be seen reflected in Table 7.9, where requests  for information, which are mostly sequence initial, are accompanied by the least amount of speaker gaze. Requests for repair and confirmation , which deal with prior business, are associated with more speaker gaze. This same pattern is reflected in recipient gaze, at least as regards repair (Table 7.10): There is more recipient gaze during repair initiation, a class of action that is by definition not in first position. These trends have been further explored in the Italian  data, which has been coded also in terms of the sequential position of the question, by which we mean, for example, whether it initiates a new course of action or not (see  Schegloff 2007b). Fifty-nine of the Italian questions were starting a course of action, which is to say less than 20 percent.31 The question here is whether in this sequential position the gaze behavior is comparable to the one observed for the total of the questions, when their sequential position was not taken into account. We want to focus in particular on whether participants tend to start courses of action already looking at each other or not, which is to say whether they tend to look up only during the sequence-initiating  question. The Italian Q-speakers look at Q-recipients before the beginning of turns that start new courses

234

Federico Rossano et al.

of action only in 27.1 percent of the cases, while Q-recipients do so only in 25.4 percent of the cases. In 42 percent of these fifty-nine questions the Q-speaker directs his or her gaze toward the recipient only during the course of the question and not before the beginning. In 35.6 percent of these fifty-nine questions initiating a course of action, the recipient directs his or her gaze toward the speaker only during the course of the question, not before the beginning. If these latter percentages are compared with the general proportion of gaze shifts toward addressee during the question shown in Tables 7.16 and 7.17 (16.3 percent for speakers and 16 percent for recipients), it is clear that there is more “mobility” in terms of gaze when initiating a course of action than when questioning in the middle of a course of action (Rossano  2009). These numbers therefore confirm what was already shown in Extracts 1 and 2. Moreover, these numbers suggest that Italian speakers do not regularly start new courses of action only after they have secured gaze from the recipient. Actually, they look at the addressee before starting speaking only a fourth of the time in this specific sequential position. We have here claimed that participants will tend to look up during the question rather than before the beginning of it mainly during questions that initiate new courses of action. However, if we focus, for example, on Q-recipients in Italian  conversations, we can see that out of the forty-eight questions in which Q-recipients look up during the question and not before, twenty-seven do not occur at the beginning of a course of action. Why in these cases is the recipient not looking at the speaker from before the beginning of the question (as appears to be the case in most of the non-initial questions) but instead looks up only during the question? Two clear interactional patterns can be observed in the Italian  data. The first one unfolds as follows. Some of these questions are initiating a post-expansion  of the sequence at a point when the sequence has been treated as complete by the other participant. In other words, the speaker who initiates the post-expansion re‑opens a finished sequence. The recipient of the post-expansion is the person who had brought the sequence to possible completion. As mentioned earlier, sequence completion typically involves withdrawal of gaze ( Rossano 2005), and the recipient will therefore have withdrawn his or her gaze before the occurrence of the question. When the other participant launches the post-­expansion, the recipient

Gaze, questioning, and culture

235

abandons any other activity and reorients gaze toward the current speaker before completion of the question. Extract 7 shows this pattern. A and B are a couple and they are having lunch. A has announced that he is going to borrow a motorbike from his grandfather but he is worried about all the additional expenses. B reassures him about the expenses and promises to lend him her helmet for few months so that he can delay the expense for it. The target is the gaze behavior during the question at line 8. (7) Italian:2PLUNCH1-casco 14:14 01 B: Io il casco te lo presto (0.6) e poi:: I the helmet you it lend.1s and then I will lend you the helmet (0.6) and the::n 02 prima o poi te lo compri before or after yourself it buy.2s sooner or later you will buy one 03 (0.5) 04 B: Perche’ per questa estate ti presto quello integrale, Because for this summer you lend.1s that full face Because for this summer I will lend you the full face one, 05 quando arriva l’inverno che l’integrale when arrive.3s the winter that the full face when winter comes the full face one A

B

A

A

B

B

A

B

06 (0.9) serve a me te lo c o m p r i. Serve.3s to me yourself it buy.2s (0.9) is needed by me you buy one for yourself. A

B

07 (0.9) A

A

B

(0.7) (2.0)

B

A



A

B

(0.4) B

08 →A: Quanto costa un casco?   POST-EXPANSION   How much cost.3s a helmet How much does a helmet cost? 09 B: Non lo so Not it know.1s I don’t know

236

Federico Rossano et al.

B’s turn between lines 1 and 6 is closing  the topic  of general expenses for the motorbike and the issue of the cost of the helmet by promising A to lend him her helmet until the following winter (the interaction occurs at the beginning of April). At the end of line 6, both participants are not looking at each other (see  Rossano 2005), and for two seconds there is silence. Both participants are oriented toward the dishes in front of them on the table. At line 8, A asks a question about the cost of the helmet, and, by doing so, he reopens the sequence, starting a post expansion, to which B responds with a non-answer response. The sequence is then further expanded. The second pattern of shift in recipient gaze during a question unfolds as follows. The person that has been coded as a recipient for the question was actually speaking in overlap or displaying an intention to initiate a turn at talk by producing a long in-breath when the speaker of the question starts his or her turn. The beginning of the question initiates in overlap and the recipient of the question orients toward the speaker only at the moment in which he or she abandons his or her overlapping turn and acts as a recipient. Extract 8 shows this pattern (see also line 12 of Extract 2 for additional evidence). A and B have met to prepare for an exam, and B is supposed to ask A questions to check whether he is ready for the exam. Our target is the gaze behavior at line 6. (8) Italian:2PEXAM-zonose 40:35 01 B: Cioe’ cosa hai letto della Marelli I mean what have.2s read by Marelli I mean what did you read by Marelli 02 A: Ho letto gli appunti della:dell’ Ilaria che erano have.1s read the notes of of Ilaria that were.3p I read the notes by:by Ilaria that were 03 fra fotocopie e:.hh e articoli di legge among Xeroxes an:d and articles of law among Xeroxes an:d .hh and law articles 04

(1.8) A

B

A

B

05 A: Cioe’ so:: [so le classificazioni,] I mean know.1s know.1s the classes, I mean I kno::w [I know the classes,]

Gaze, questioning, and culture A

B

A

237 B

A

B

A

06 → B: [Cioe’ le zoonosi in Italia principali quali sono I mean the zoonoses in Italy main which are [I mean which are the main zoonoses32 in Italy 07

B

(0.5)

A: Le zoonosi in Italia? 08 The zoonoses in Italy The zoonoses in Italy? 09

(0.4)

B: Eh 10 Eh B: Eh

B’s request for information at line 1 is responded to by A at lines 2 and 3. After a silence of almost two seconds, A starts talking again, providing further specification of what he knows and therefore what B can ask him questions about. However, immediately after the beginning of line 5, B, who is looking at some notes he is holding in his hands, starts talking as well and asks A a possible exam question. This is followed by a repetition  of part of line 6 that initiates repair at line 8, which is confirmed by B, and the sequence continues with A’s tentative answer to B’s question. If we look at the gaze of the participants, we can see that while B keeps looking at his notes throughout the entire question at line 6, A, who was looking away toward his right side, turns toward B as soon as the first two words in B’s turn are produced. A, therefore, turns toward B as soon as he hears B talking in overlap and abandons the turn that he was producing soon after. This shows that once the sequential environment of the question is taken into account, some further order is observable. In particular, it shows that the beginning and ending of sequences and courses of action are the most vulnerable environments – in Italian at least – in terms of sudden shifts in gaze direction, and it also accounts for those instances in which the shift in gaze is not due to the sequential position of the question but rather to a parallel shift from speakership to recipiency that had to occur because of overlapping speech.

238

Federico Rossano et al.

Conclusions It is clear that gaze plays a delicate and complex role in social interaction. In this paper we have explored this conversational role mainly through the crude instrument of descriptive statistics. We have done this primarily because we have here been pursuing similarities and differences between conversational practices in unrelated cultures. A statistical approach can firmly establish the existence of distinct practices in the utilization of gaze, practices that have only partly been observed in qualitative analysis. If, for example, we focus just on Extracts 3 and 4, we can immediately notice the importance of having grounded their representativeness in a larger corpus. While being representative of a large number of questions in Tzeltal  and Yélî Dnye, Extract s 3 and 4 only represent a minority of cases in each language. Indeed, mutual gaze  (as shown in Extract 4) occurs in only half of the questions in Yélî  Dnye and is sustained from beginning to end of the question in even fewer cases. In more than 65.7 percent of questions in Tzeltal, the speaker looks at the addressee, and the recipient looks in 42.3 percent of the cases. This means that the number of instances in which neither participant looks during a question (as shown in Extract 3) would necessarily be a minority. Exposing the extremes, even if recurrent and not isolated cases, could risk distracting us from a proper reconstruction of what empirically appears to be the most general pattern in a culture. The use of large corpora, in this sense, is necessary if the goal is to develop a comparative analysis of the systematicity of specific practices across different cultures. By proceeding in this way, we have indeed found distinct cultural practices, especially in the relative lack of use of gaze as an indicator of engaged recipiency  in Tzeltal . We have also noticed alternative “home positions” for the eyes when a participant is not looking at the other’s face: down, toward hands or legs or middistance for Tzeltal participants, mainly toward objects or places motivated by question delivery for Italian  participants, and a bit of both for Yélî  Dnye participants. We know from observation that there are also other distinct practices that could be culturally specific, for example the use of gaze to “point” to unseen, even distant, entities in Yélî  Dnye ( L evinson 2007). All these practices

Gaze, questioning, and culture

239

clearly affect the “accountability” and semiotic relevance of specific gaze orientations in conversation and merit further examination to establish their interactional consequences. But of at least equal importance are the general findings that hold across all three cultures and thus are candidates for general interactional properties. First, during questions, it is the Q-speaker rather than the Q-recipient who is more likely to be gazing. As pointed out repeatedly, this is at odds with prior generalizations about the role of gaze in conversational practice in general, where its functions have been presumed to be largely a display of recipiency  and attentiveness. Second, the dominant pattern is with gaze being fixed throughout the turn, and thus it cannot play a systematic role (in the question context at least) as a turn-transition  cue, or as a cue that participant role is switching. Third, the findings in all three languages are consistent with a detailed qualitative analysis of gaze in Italian  (Rossano  2009), which shows that gaze is tied into sequence initiation and sequence completion rather than into the turn-taking  system directly. We therefore think it likely that this observation too has general application across cultures. While our study has distinct limitations, especially because of its restriction to question sequences and the relative lack of detailed sequential analysis (at least that we can report on here), it is nevertheless clear that the standard accounts for the role of gaze need modifying either in general or at least with regard to question sequences in particular. Whichever account is preferred, it is clear that our understanding of the systematicity of gaze behavior in interaction has been affected. The most notable piece of news is probably the fact that Q-speaker gaze behavior seems to be much more similar across languages than Q-recipient gaze behavior. In this sense, ethnographic descriptions that describe lack of eye contact and very brief glances could well be partially correct but perhaps only as descriptions of typical recipient behavior. The difference in gaze behavior observed between Tzeltal , Italian,  and Yélî  Dnye strongly challenges the view that looking at the speaker’s face would always be the default gaze behavior for a recipient. Nonetheless, the nonuse of gaze as a reliable indicator of recipiency in Tzeltal can be shown to require some substitute system, which in Tzeltal seems to be

240

Federico Rossano et al.

provided by the repetition  response system (Brown   and  L evinson 2005; Levinson  and Brown 2004). This study therefore confirms that generic conversational functions reliably invoke machinery that will handle them ( Schegloff 2006). There are, however, a series of open questions that emerge from our first empirical pass through the data. What are the functions of speaker gaze in dyadic interaction? How is recipiency  displayed and are recipiency cues somehow systematically organized and ranked? How do we account for a system in which Q-recipients look less than Q-speakers and yet in more than 40 percent of the cases? Is there something special about “questions” that affects gaze deployment in a very specific way? It is clear that much more fine-grained analysis of gaze behavior in interaction is needed. Action types, at least at the level of the description that we chose, do not really account for the differences in recipient gaze behavior. It seems reasonable to invoke as an account the material world the participants are inhabiting, but we feel obliged to emphasize that Tzeltal  participants are similar to the Italian  when they speak but not when they listen to questions. The Tzeltal participants are inhabiting the same environment throughout the interaction and yet behave differently according to their participation role . We are aware of the risks involved in isolating a specific practice (gaze in interaction) and trying to understand how this changes the interactional environment. However, the fact that gaze toward addressee is deployed very often during questions but not “all the time” or “never” in all three cultures here examined (and in all cultures examined in previous studies) suggests that its occurrence is systematic and interactionally managed and probably affects the interaction in significant and specific ways. As the quotation by Lord Chesterfield at the beginning of this chapter suggests, gaze in interaction is a potent thing. And, as Darwin  insisted, it has deep ethological roots and so is inevitably deployed in interaction in all cultures as a valenced indicator of visual attention – although, as we have shown, it is not always deployed in the same ways. Yet gaze in interaction has received relatively little attention in the past twenty years. We hope this chapter will encourage further investigations of how gaze is deployed

Gaze, questioning, and culture

241

in ­different settings, different cultures, different participant roles, and in performing different actions.   

Appendix A:  Coding  For numerical purposes, we developed the following coding scheme. For the verbal component, we used the following five ­categories  (bear in mind that all the interactions were dyadic, so the effective recipient is by default the nonspeaker). A. B. C.

D. E.

SPEAKER/RECIPIENT: who is the speaker of each question. QUESTION TYPE: “yes”/“no” (polar) question, ­wh-question or alternative question. ACTION: e.g., request for information, request for confirmation , other initiation of repair, offer, invitation, seeking agreement , etc. RESPONSE: answer (interactionally appropriate), nonanswer response (e.g., “I don’t know, maybe”), no response.33 SEQUENTIAL POSITION: initial position or further in the sequence (only in Italian ).

We coded for thirteen different possible gaze configurations: 1. (a) The speaker does not look at the recipient at all during the question. (b) The recipient does not look at the speaker at all during the question. 2 (a) Speaker looks at recipient from beginning to end and was already looking before the beginning of the question. (b) Recipient looks at speaker from beginning to end, and was already looking before the beginning of the question. 3. (a) Speaker is looking at recipient before the beginning of the question but looks away during the question. (b) Recipient is looking at speaker before the beginning of the question but looks away during the question. 4. (a) Speaker looks toward recipient during the introduction of a new referent in the question. (b) Recipient looks toward speaker during the introduction of a new referent in the question.

242

Federico Rossano et al.

5. (a) Speaker looks toward recipient on nearing completion of the question (last two words or last two syllables if question made of two words). (b) Recipient looks toward speaker on nearing completion of the question (last two words or last two syllables if question made of two words). 6. (a) Speaker looks toward recipient in any other position during the question. (b) Recipient looks toward speaker in any other position during the question. 7. The occurrence of mutual gaze  at any time during the question turn. (We define mutual gaze as both participants looking at each other’s face or eyes simultaneously.) Finally, we coded for how often the lack of speaker or recipient gaze was motivated by: 1. the content of the question and its delivery (e.g., presence of deictic s, pointing, talking about an object present in the surrounding environment); 2. a competing activity such as eating, weaving, drawing on the sand, etc. Appendix B:  Symbols for gaze orientation A A

A

A

B

: Mutual gaze. 

B

B

: A looks away and B looks away. : A looks down oriented toward B. B looks away.

B

A

B

A

B

A

B

A

B

: A looks away. B looks down oriented toward A. : A and B are looking down in front of them. : A looks at B. B looks down. : A looks at B. B looks away.

: B looks at A. A looks down.

Gaze, questioning, and culture A

B

A

B

A

B

A

B

A

B

A

B

A

: B looks at A. A looks away.



: A looks down. B eyes closed.



: A eyes closed. B looks down.



: A eyes closed. B eyes closed.



: A looks at B. B eyes closed.



: A eyes closed. B looks at A.

B

A

B

A

B

A

B

A

B

: A away. B eyes closed. : A starts turning toward B who is looking down. : A starts turning toward B who is already looking at A.



B

A

B B

A

B

A

B

A

B

A

B

A

B

A

: A starts raising the gaze toward B who is looking down.

: A starts raising the gaze toward B who is already ­looking at A.

A

A

243

: A looks away B looks mid-distance up left. : A looks away B looks mid-distance up right. : A looks down B looks mid-distance up left.

: A looks down B looks mid-distance up right.

: B starts turning toward A who is looking down. : A starts raising the gaze toward B who is looking down.

: A looks mid-distance up left. B is looking away.



: A looks mid-distance up left. B is looking down.

B

: A looks mid-distance up right. B is looking away.

244

Federico Rossano et al.

A

B

A

B

A

B

: A starts turning toward B who is looking away.

A

B

A

B

A

B

A

B

A

B

: B starts turning toward A who is looking away. : B starts raising the gaze toward A who is already ­looking at B.



: B starts raising the gaze toward A who is looking down.



: A looking at B. B looks mid-distance up right.

B

A

B

B

A

B

A

B

A

B

A

B

: B starts turning toward A who is already looking at B. : B starts raising the gaze toward A who is looking away.

A

A

: A looks mid-distance up right. B is looking down.

: A mid-distance right. B eyes closed. : A mid-distance right. B looking at A.



: A mid-distance left. B looking at A.



: A looking at B. B mid-distance right. : A looking at B. B mid-distance left.



: A eyes closed. B looks away. : A looking mid-distance up left. B looking up.

Notes 1 Cited in Goodwin (1981: 62), after Goffman  (1953). 2 “Participants utilize both their bodies and a variety of vocal phenomena to show each other the type of attention they are giving to the events of the moment, and, reciprocally, the type of orientation they expect from others. [ . . . ][Engagement displays] permit those present to display

Gaze, questioning, and culture

245

to each other not just speakership and hearership but ­differentiated ­attention to, and participation in, the talk of the moment” (Goodwin 1981: 124–125). 3 “To extract robust outcome-based conclusions about how physicians (or patients) should conduct themselves in specific moments in the flow of the medical encounter, it is important to find a meeting point between the two methodologies of coding and microanalysis [ . . . ]. In other words, beyond the intrinsic worth of an analytical framework responsive to very granular, individual moments in the physician–­ patient encounter, we need one that simultaneously supports coding at a broader level of granularity sufficient to reach beyond individual cases to generate findings at a statistical evidential standard” (Heritage and Maynard 2006: 8). 4 “Children at a very early age do not blush; nor do they show those other signs of self-consciousness which generally accompany blushing; and it is one of their chief charms that they think nothing about what others think of them. At this early age they will stare at a stranger with a fixed gaze and unblinking eyes, as on an inanimate object, in a manner which we elders cannot imitate” (Darwin 1872: 328). 5 Duncan and colleagues actually refer to shift in head direction as a turn-yielding cue but specify that this should be taken as a proxy for “eye direction” (Duncan and Fiske 1977: 211). 6 In a direct response to Beattie’s paper, Kendon (1978) argues that the data used by Beattie (dyadic conversations between students and their supervisors) was not comparable to the one used in his study (ordinary dyadic conversation between Oxford undergraduates), because of the asymmetric status of the participants and the formality of the situation, further indicating that the kind of interactional situation participants are dealing with may well affect the deployment of gaze. 7 There are, however, works (e.g., Erickson 1979; LaFrance 1974; LaFrance and Mayo 1976) that claim racial differences in this respect: Black Americans look more while speaking than while listening, while white Americans follow the opposite pattern. 8 In a footnote, (Goodwin 1981: 57) admits that even though he proposes this as a rule applying to turns in general, this pattern is not found in every turn at talk. 9 With the terms “courses of action accomplished through one or more sequences of talk,” Rossano  refers to the fact that, to be considered completed, most actions require at least the occurrence of some sort of response or reaction by the other participant, and, therefore, an initiating action usually starts the development of a course of action produced by more than one participant. For example, the gist of a request

246

10

11 12

13

14

Federico Rossano et al.

for information can be considered accomplished only if the other participant provides the information, and an offer is completed only if it is accepted or rejected and the thing offered is provided to the person to whom it had been offered. This means that the occurrence of an utterance and the action(s) that it implements opens the possibility (and sometimes it normatively expects) the occurrence of another set of utterances or actions that would allow the gist of the initiating action to reach a socially appropriate completion. This can be achieved in two turns or may require larger structures, though the participants’ orientation toward completing the gist of the initial action remains the same. Kendon is one of the few who distinguish between speaker gaze during “short questions” and speaker gaze during “long utterances”; he suggests that during short questions “[the speaker] will look steadily at [the recipient] while he asks his question, and where [the speaker] is asking a series of questions, unless he has to pause in thinking of the question, or unless he has to pause in formulating it.” (Kendon 1967: 47). Ten interactions in Italian, ten  in Tzeltal,  and nine in Yélî  Dnye. In some interactions in Tzeltal  and Yélî Dnye,  some children and/or other adults can be seen in the videos that capture the interactions. The interaction remains dyadic for current purposes as these bystanders are not addressed, they never intervene in the conversation, nor they are looked at. As we are not interested in the content of their conversation but rather in the gaze behavior of the participants, the presence of neglected bystanders does not have any significant impact on the gaze behavior of the dyad. The Italian  speakers were all natives of the region Emilia-Romagna, in the north of Italy, and all the Italian data were recorded in Bologna, Italy. The Tzeltal speakers all belong to the ethnic group of Tenejapans (c. 25,000 strong), and all were recorded in situ within their territory. The Yélî Dnye speakers are the exclusive inhabitants of Rossel Island, and form a small population of 4,000 souls, nevertheless divided into two dialect groups – the recordings are wholly of eastern dialect speakers. From each culture we had participants of both sexes; ages ranged from twenty to fifty years old for Italian and Yélî Dnye speakers and from thirty to sixty years old for Tzeltal speakers. The three authors are either native speakers (Rossano ) or long-term ­fieldworkers (Brown , Levinson ) in their respective communities reported on here. Some of these could be understood as noticings, but often they address already-raised issues so are functioning as requests  for confirmation or prods for expansion.

Gaze, questioning, and culture

247

15 Rossano  coded the linguistic/interactional aspects of questions for Italian , Brown for Tzeltal and Levinson for Yélî Dnye . Rossano coded for gaze the Italian and Yélî  Dnye data (as well as 40 percent of the Tzeltal data, checking consistency of coding), while Brown  coded for gaze the Tzeltal data. 16 Logistic regression is a statistical model used to predict outcome variables that are categorical (e.g., yes or no). In this specific case, if a specific participation role  (e.g., being a speaker) predicts significantly better than the other (being a recipient) whether an individual will be looking at the other participant, then we can say that being a speaker rather than a recipient significantly affects the likelihood of the occurrence of gaze toward the other participant. In other words, participants look at each other differently depending on which participation role they have during a question. We are grateful to Tanya Stivers  for help with the statistical analyses. 17 From now on, all logistic regression analyses reported here have to be considered corrected for the clustering of the data by interaction. This is meant to take into account the fact that different interactions (and in particular what the participants are doing within a specific interaction) could affect the general pattern more than others and therefore the general distribution could be the product of one or two biased interactions rather than a systematic pattern observable across multiple interactions. 18 Interestingly, in a small corpus of questions in not experimentally elicited dyadic English interactions, Beattie (1978: 13) reports that the speaker gazes at the addressee at least at some point during the question in thirty out of thirty-nine questions, which corresponds to 76.9 percent of the questions. Similarly, Kendon (1967: 45) reports that speakers look toward addressees in 31/41 (75.6 percent) of the questions of his corpus of British English dyadic interactions. While both are very small corpora, probably not fully representative of dyadic interactions in British English, it is remarkable that the percentage of speaker gaze during questions would be in between the ones found for Italian and Yélî Dnye, which are not significantly different from each other. 19 These percentages can be obtained by subtracting the instances of mutual gaze from the instances of speaker gaze. 20 Most of the questions that receive no response are directly followed by the same speaker elaborating or clarifying the question and therefore still pursuing a response. The relevance of a response is still present, and interactional contingencies can account for the lack of responses (Rossano  in press).

248

Federico Rossano et al.

21 For Tzeltal the recipient is not gazing in 57.7 percent of the 300 questions here considered, in Italian in 36.7 percent, and in Yélî Dnye in 32.7 percent. 22 There is no doubt that in dyadic conversations there are instances in which gaze direction can vary within one utterance, and this is usually related to the activities at hand and what is going on in the interaction (moreover, specific actions can require a very specific gaze deployment). What we are saying is that during questions in these three cultures we do not see much gaze mobility, and, therefore, there is at least one domain and three cultures in which prior general claims about gaze behavior in any turn at talk do not hold. 23 We should emphasize here once again that there are a number of ­distinct social groups who speak Tzeltal, with distinct customs; our generalizations here are about Tenejapans. 24 This matches previous claims by Brown and  L evinson (2005; Levinson and Brown 2004). However, the fact that alternative sitting configurations are observable within the same samples as well as the rather small number of dyadic interactions for each culture analyzed here suggest caution in terms of generalizations. 25 It is not always clear whether participants have chosen one configuration because of a general preference for that seating configuration or because of external factors such as presence of benches, preference for sitting in the shade, surrounding noise, etc. 26 See for a comparison the following example in American English, recorded in a medical interaction (Boyd and Heritage 2006: 158): 01 Doc: Is your mother alive, 02 Pat: No: .

27 This repetition -as-affirmation default has been shown to apply in a number of languages, not only within the Mayan  family but also in completely unrelated languages (e.g., Irish, Welsh, Finnish, Estonian). 28 See row 3 of Tables 7.16 and 7.17 for instances in which a participant is looking at the addressee before the beginning of the question and looks away before completion of it. It never happens more than 6–7 percent of the time in any of the three languages, and in some configurations (e.g., speaker gaze in Italian) it is even rarer. 29 Moreover, the fact that there is a sound stretch, a cut-off or a pause in the question does not mean per se that they have been produced to elicit gaze back, as issues such as self-initiation of repair could be in play here. 30 There are a few cases in our data where a self-repair  may play a similar role to that in American English.

Gaze, questioning, and culture

249

31 As expected, most of them were requests for information (44/59, 74.6  percent) with the remaining ones performing different actions such as offering or doing a topic proffer, but clearly not initiating repair. 32 A zoonosis is a disease that can be transmitted to humans from other vertebrate animals. 33 We coded as responses both visible and verbal responses. A head nod or a head shake in response to a polar question was coded as an answer, exactly like pointing to an object on the table in response to a question such as “where did I put my notes?” Shoulder shrugs or similar gestures whose meaning could not be clearly established were instead coded as non-answer responses. “No response” means that no verbal or visible response was provided.

8 Negotiating boundaries in talk Makoto Hayashi and Kyung-eun Yoon 

Introduction Across languages, we find minimal vocalizations, such as “mhm” in English and “un” in Japanese , that are used by participants in interaction to respond to co-participants’ prior or ongoing talk. These “response tokens” have drawn a sustained interest in the study of language and social interaction over the past several decades (see Sorjonen  2001 for a useful review). One major line of research in this area has been concerned with distributional analysis, exploring statistical correlations between types of response tokens and turn-taking  (e.g., Duncan  1972, 1974; Duncan and Fiske 1977) or correlations between the frequency of response tokens and such parameters as culture and gender (e.g., Maynard  1986, 1990; White 1989; Fishman 1983). Because their primary interest is in discerning patterns for the aggregate occurrences of response tokens across situations, these quantitative studies tend to isolate response tokens from their local contexts of use, glossing the interactional contingencies to which participants orient when producing response tokens in any given context. By contrast, a line of research on response tokens that emerged within conversation analysis (CA) pays more serious attention to the sequential/interactional contexts for the deployment of response tokens (e.g., Jefferson  1984, 1993; Schegloff  1982; Heritage  1984, 1998, 2002; Goodwin  1986; Mazeland  1990; Beach  1993; Sorjonen  1996, 2001, 2002; Gardner  2001; Mori  2006; among others). These studies investigate the kinds of interactional activities the participants engage in that give rise to the deployment of particular response tokens, how the co-participants orient to them, and what consequences their deployment has for the organization of

Negotiating boundaries in talk

251

subsequent talk. This research has shown that the “meanings” of response tokens depend crucially on their placement within the ongoing sequence of actions and activity, and that it is therefore necessary to examine the details of the sequential/activity contexts as part of the “semantics” of response tokens. In the present study, we explore the situated meanings of minimal response tokens in Japanese and Korean (“ un” and its variants and “um” and its variants, respectively; both roughly equivalent to “yeah” or “mhm” in English) produced in a specific sequential context – i.e. in a slot after one’s talk has been responded to by a recipient with a minimal “yeah”/“mhm”-like acknowledgment token.1 To describe this sequential context schematically: →

A: B: A:

turn’s talk minimal yeah/mhm-like acknowledgment token minimal yeah/mhm-like acknowledgment token

Position 1 Position 2 Position 3

The practice to be examined is thus characterized as the “thirdposition deployment”  of a minimal response token. 2 It should be noted that the minimal response tokens in Position 3 are produced by the primary speaker-so-far (i.e. the speaker of substantive talk in Position 1) rather than the recipient-so-far. Previous studies have paid almost exclusive attention to response tokens as “recipient activities,” and little has been investigated about primary speakers’ deployment of response tokens. Our study thus aims to shed light on the roles of response tokens as primary speakers’ activities in organizing the unfolding courses of talk in interaction. This study is also intended as a contribution to cross-linguistic analysis of response tokens (cf. Beach  and Lindström  1992). Most of the past studies focused either on analyzing multiple functions of a single response form in a given language (e.g., Heritage  1984; Beach  1993; Sorjonen  2002; Mori  2006) or on comparing functions of different response forms within a single language (Jefferson  1984, 1993; Goodwin  1986; Gardner  2001; Sorjonen  1996, 2001). Our study, by contrast, examines comparable response tokens used in comparable sequential/activity contexts in two different languages. Through this examination, we demonstrate how the practice of deploying third-position  minimal response tokens provides a resource for speakers of both Japanese  and Korean  to accomplish similar interactional ends under similar interactional contingencies.

252

Makoto Hayashi and Kyung-eun Yoon

In what follows, we analyze two related yet distinct types of usage of third-position minimal response observed in Japanese and Korean  conversations. We will show that in both types of usage, the deployment of third-position minimal response tokens is “closure implicative”; that is, it proposes and projects termination of what has been going on prior to it. In the first type of usage we examine, it is the topic /activity-so-far that is projected to be terminated. In other words, a speaker-so-far deploys third-position minimal response to indicate that the topic talk or activity in which the participants have been engaged so far is now put into a “state of attrition” (Jefferson  1993) and that a new topic or activity can be relevantly initiated. A third-position minimal response is thus used as a resource to negotiate topic/activity boundaries and achieve topic/activity transition. In the second type of usage, it is the turn-so-far that is projected to be terminated. That is, third-position  minimal response is used as a “turn-exit device”  (Jefferson  1981). As will be discussed below, speakers of Japanese  and Korean  sometimes produce syntactically incomplete turns that may nonetheless be regarded as “pragmatically complete” (i.e. complete from the perspective of executing an action). Such turns can at times cause a problem for turn transfer because recipients may not treat them as complete turns due to their syntactic incompleteness, while the speaker-sofar may be ready to give up the floor. Thus, when a syntactically incomplete, yet pragmatically complete turn is produced and is only met with minimal response from the recipient, the speaker may deploy third-position minimal response to indicate that his or her turn is over now and that it is the recipient’s turn to do some extended talk. Negotiating topic /activity boundaries The first type of usage of third-position  minimal response examined here is its deployment in the context of negotiating topic / activity boundaries.3 As previous studies have shown, bringing an ongoing sequence of topic talk or activity to closure is an interactive achievement – an achievement that commonly involves turn-by-turn negotiation among the parties, rather than unilateral execution by a single speaker (Schegloff and Sacks 1973; Maynard 

Negotiating boundaries in talk

253

1980; Jefferson  1984, 1993; Drew  and Holt  1998). Deploying third-position minimal response, then, is one method that is ­available for the participants to propose the relevance of imminent topic/­activity closure. By responding to a co-participant’s minimal acknowledgment of the prior talk with another minimal acknowledgment token, the speaker-so-far conveys that there is nothing more to add to the prior topic/activity and therefore that it is an apt moment for topic/activity transition to take place. The following fragment presents an example from a Japanese  conversation. This fragment is taken from a phone conversation between two adult friends who used to be on the same track and field team in high school. Prior to this segment, one of the participants, Mika, announced her recent visit to a stadium (called Hattori Stadium) where they often had district track meets when they were in high school. This announcement  touches off a reminiscing activity (lines 1–5), which then leads to Eiji’s “my-side telling” (from line 7 onward) about his visits to stadiums they used to compete in. In this telling, he mentions that, though he sometimes visits another stadium (called Nagai Stadium), he has not visited Hattori Stadium for quite a while. Examine in the following fragment how Eiji brings his “my-side telling” to closure with a thirdposition  minimal token in line 36. (1)  [Japanese: K MI 9 (telephone:Eiji and Mika)] 01 E: sabu torakku mitaina n mo::natsukashii na:::. sub track like N also nostalgic FP I feel nostalgic about the practice track, too. 02 °.hhh° (.) demo anmari::hh anm(h)ar(h)i but not.much not.much °.hhh° (.) But I don’t::hh I don’t 03

ob(h)oet(h)enai na:: [.hhhhhhh] u:::[::n. hhh] don’t.remember FP yeah remember m(h)uch::. .hhhhhhh Yeah::::. hhh

04 M: 05

[huhuhuhu] huhuhuhu

[.hhh moo] already .hhh It’s

haruka mukashi no ohanashi desu ka[ra ne:::] by.much long.ago LK story CP because FP already such a long time ago, so:::

254

Makoto Hayashi and Kyung-eun Yoon

06 E: 07

[ soo desu] so CP That’s

ne:::. nanka:NAGAI::::no hoo WA:: FP like Nagai LK side TP right. As for Nagai::::,

08 M: u:n Mhm 09 E: ano:: yoko ni amehuto joo ga dekite::: uhm next.to PT American. field SP is.built football uhm, they built an American football staduim right next to it, so … : :((14 lines omitted in which Eiji tells Mika :that he visits Nagai Stadium every so often :because he goes to American football games :played in the new stadium next to it.)) : 24 E: . . . nagai no kyoogijoo o [ne::] Nagai LK stadium O FP 25 M:

[u:::]n. Mhm

26

(.)

27 E:

nagameteta n desu ke[do.] was.looking N CP but ((Lines 24 and 27)) So I was looking at Nagai Stadium.

28 M: 29

no hun ni mamireta [( LK dropping by stained stained by pigeons- pigeons’ droppings

30 E: 31

[ano] hato ni:- (.) hato that pigeon by pigeon That ((filthy stadium)) )]

[s(h)oo HHsoo soo soo soo] R(h)ight HHright right

soo. right right right.

32 E: .hhhhhh demo ne:nanka ano::::acchi wa:: but FP like uhm there TP .hhhhhh But uh::::m, the other one,

Negotiating boundaries in talk

255

33

hattori no hoo wa moo::: zen:::zen:: Hattori LK side TP EMPH not.at.all Hattori Stadium, I don’t go there at all

34

ja nai ka na:::.= CP NEG Q FP anymore, I guess.

– Position 1

35 M:

=u::n. Mhm

– Position 2

36 E: →

u::n. Mhm

– Position 3

37

(0.2)5

.hhhh soo ka:::::iya ii desu ne::::: 38 E: so Q well good CP FP .hhhh I see. Well, it’s good ((that you went to Hattori))

After explaining why he still sometimes goes to Nagai (lines 7–27, most of which is not included here), Eiji brings the focus of his talk back to Hattori Stadium and states that he does not go there at all (lines 32–4). Eiji’s utterance here connects back beyond the adjacently prior talk/details (i.e. talk about his visits to Nagai) to an earlier part of the ongoing talk (cf. lines 1–3). It also presents an upshot of his “my-side telling” and its connection to Mika’s original announcement  about her recent visit to Hattori; that is, that in contrast to his occasional visit to Nagai, he has not visited Hattori for quite a while. Utterances of this kind, connecting back to an earlier part of an ongoing telling and presenting an upshot of the telling-so-far, often serve to indicate that the teller is becoming disengaged from an item-by-item development of topic  talk, and they are therefore commonly associated with topic closure (Drew  and Holt  1998). In response to this closure-relevant statement, Mika produces a minimal acknowledgment token, “u::n”/“mhm” (line 35), and continues to display her alignment as a recipient to Eiji’s telling. Eiji then responds to Mika’s display of “passive recipiency” (Jefferson 1984) with a minimal response token of his own (line 36), and, by doing so, indicates that he is ready to disengage from the current line of “my-side telling” about his visits to old stadiums. Indeed, after a brief silence (line 37), Eiji makes a transition from “my-side” talk to “your-side” talk by producing a response to (and a positive assessment about) Mika’s original announcement  of her recent visit to Hattori Stadium.

256

Makoto Hayashi and Kyung-eun Yoon

The next fragment from a Korean  conversation shows a case in which a parenthetical  activity embedded within another activity is brought to closure with third-position  minimal response. Prior to the following segment, Gina mentioned a tasty noodle product, and Sumi asked for its brand name. Just as Sumi brings out a paper to write down the name of the product, Gina begins to voice a concern about its availability in stores because it is a seasonal product. Examine how this parenthetical activity of “(potential) problem announcement” is brought to closure with a ­third-position ­minimal response in line 13 before the participants resume the activity of giving/writing down the product name. (2)  (Korean : Lunch Talk [face-to-face: Sumi and Gina]) 01 S:

[((brings out a paper))

02 G: [yocumey kacyeo-nunci moll -a.= these.days bring -if not.know-INT I am not sure if they bring the noodles ((from the warehouse)) these days. 03 S:

=[ ney:.] Mhm.

04 G:

=[waynya]myen yelum -ey ha:ncham nemwu manhi because summer-in a.long.time very a.lot because they bring a lot of those noodles

05

kacyeo-nuntey, bring -CIRCUM all summer long, but

06 S:

u u:m. m Mhm.

07 G:

kyewul-i[-nikka.] winter-be-CONN because it’s winter now.

08 S:

[ u :m ] Mhm.

09 G:

han pen tto- (0.3) one time again Once again- (0.3)

10 S:

nacwungey (ye- pom -ina yelum later sum- spring-or summer Later (when sum- spring or summer

Negotiating boundaries in talk

257

11

toy -myen) become-when comes,)

– Position 1

12 G:

°um° °Mhm°

– Position 2

13 S: → ey. Mhm 14

– Position 3

(1.2) ((Sumi gets ready to take a memo.))

tongchimi mwul-nayngmyen-i –ntey -yo, pwukkyeng 15 G: Tongchimi mwul-nayngmyen-be-CIRCUM-POL Peking It’s Tongchimi Mwul Nayngmyen, but I guess it’s 16

tongchimi mwul-nayngmyen tongchimi mwul-nayngmyen Peking Tongchimi Mwul Nayngmyen.”

In lines 2, 4–5, and 7, Gina develops a line of talk in which she suggests a potential problem with the product’s availability in stores in the season in which this conversation takes place. This “problem announcement” is responded to by Sumi in lines 10–11 with a suggestion of a “problem solution,” i.e. that one only needs to wait a few months to see the product available again in stores.4 In the context of discussing a problem, presenting a solution can be implicative of the closure of that discussion. Gina responds to this statement of problem solution with “um”/“mhm” produced in soft voice (line 12) and displays minimal acknowledgment/­agreement . To this, Sumi produces a third-position  minimal response token of her own (line 13) and displays her stance  that she is not going to pursue the current line of talk about the problem of the noodle product’s availability any further.5 This proposal of topic / activity closure is further reinforced by Sumi’s nonvocal conduct in the ensuing silence (line 14), in which she moves to prepare to write down the name of the product and proposes the relevance of resuming the activity that the participants were about to engage in before the parenthetical sequence began. Gina concurs with Sumi’s proposal by moving on to mention the brand-name of the noodle product (lines 15–16). Thus, just as in the previous instance, thirdposition minimal response is used in the juncture where topic/ activity closure and transition are jointly accomplished. The two fragments we examined demonstrated that third­position  minimal response can be used by speakers of Japanese 

258

Makoto Hayashi and Kyung-eun Yoon

and Korean as a means to effect imminent closure of the preceding line of talk and transition to a new topic /activity. In both cases, we observed the following sequence of actions around the deployment of third-position minimal tokens. First, the utterance produced in what is labeled as Position 1 above presents some kind of “closurerelevant” statement, such as a reference to what was said earlier, an upshot of what has been discussed, etc. This is receipted by a co‑participant with a minimal acknowledgment in Position 2, which registers no more than minimal receipt of (and/or minimal agreement  with) the prior talk. This, then, is responded to by a reciprocal minimal response produced by the speaker of the utterance in Position 1, which serves to decline to take the opportunity in that slot to say anything more about the previous topic/activity. This exchange of reciprocal minimal response tokens displays both speakers’ orientation to withholding any further topical development or elaboration and thereby paves the way for a transition to a next topic/activity. Schematically:   Position 1 Position 2

A: Closure-relevant statement B: Minimal response (registers minimal receipt/ agreement) Position 3 A: Minimal response (declines further development of topic/activity) Position 4 B/A: New topic/activity initiation

We should note here that this sequence of actions at the juncture of topic /activity transition is not exclusively observed in Japanese and Korean  conversations. For instance, Drew  and Holt  (1998), who examine the use of figurative expressions in English conversation, describe a similar course of actions observed at topical junctures (which they term the “topic transition sequence”) and explore how figurative expressions regularly occur as topical summaries in the slot equivalent to Position 1 in the schema above (see Extract 3). Jefferson  (1993) also describes how a summative commentary is followed by reciprocal exchanges of minimal responses that put the topic talk-so-far into the “state of attrition” and lead to topic shift (see Extract 4); see also Schegloff  and Sacks  1973: 318). (3)  (English: Drew  and Holt  1998: 501–2; slightly modified) 01 R: B’t take this with a dollop’v salt you kno::w I’m02 I’m ba:sic’ly quite happy b’t quite relieved it’s

Negotiating boundaries in talk

259

03 the sheer organization ‘n getting all, everything – Position 1 04 done in th’ da:y. – Position 2 05 L: Yes:that’s ri:ght,= – Position 3 06 R: =Ye[s. 07 L: [Yes. Ye[h .tch .hhhhhh Wuh[O k a y, W’ddiyou wanna talk t’ me 08 R: – Position 4 09 abou(h)t (4)  (English: Jefferson  1993: 20; slightly modified) 01 M: .hh Well you never kno:w do you someti:mes 02 yo[u feel as if you don’t want to stay in the= 03 J: [No:. 04 M: =sa:me pla[:ce, .hh tha[t where you’ve been= [(pla:ce.) [Ye:s. 05 J: – Position 1 06 M: =with your pa:ren[ts:.hh [Ye:s. – Position 2 07 J: 08 (.) – Position 3 09 M: Mm[:. .hh [But uh::anyway, 10 J: 11 (0.3) 12 M: .mptlk By the wa:y Janet did you get . . .  – Position 4 13

Thus, the practice described in this section is by no means unique to Japanese and Korean . There is, however, a potentially distinctive feature in the way third-position  minimal response can operate in Japanese and Korean that has not been observed in English. That is, at least in some instances of third-position minimal response in Japanese and Korean, a speaker uses the minimal token to close down the topic  talk/activity-so-far even though his or her immediately prior turn is still grammatical ly incomplete. An example of this is found in Extract 2, examined above, which is partially reproduced in Extract 5 below. (5)  (Korean : Lunch Talk [face-to-face: Sumi and Gina]) 10 S:

nacwungey (ye- pom -ina yelum later sum- spring-or summer Later (when sum- spring or summer

11

toy -myen) become-when comes,)

– Position 1

12 G:

°um° °Mhm°

– Position 2

260

Makoto Hayashi and Kyung-eun Yoon

13 S: → ey. Mhm 14

– Position 3

(1.2) ((Sumi gets ready to take a memo.))

15 G: tongchimi mwul-nayngmyen-i –ntey -yo, … Tongchimi mwul-nayngmyen-be-CIRCUM-POL It’s Tongchimi Mwul Nayngmyen, but …

In her “problem solution” statement in lines 10–11, Sumi produces a subordinate clause that ends with the subordinate-clause marker “-myen”/“when.” In more interactional terms, Sumi’s utterance here is the “preliminary component” of a “compound turn constructional unit (TCU)” (Lerner 1991) in the form of [X-myen + Y] ([When/If X + then Y]).6 At the component boundary within this compound TCU, Gina provides a weak acknowledgment/agreement (line 12). Rather than continuing to produce the main clause, or the “final component” of the compound TCU, Sumi then produces a third-position  minimal token (line 13) and indicates that she is not going to go on with the turn-so-far even though it has not been brought to syntactic completion.7 Thus, third-position  minimal response can be used not only as a device to exit from the topic /activity-so-far, but also from the turn-so-far. While this feature is not observed in all cases of thirdposition minimal response in Japanese  and Korean , as evidenced by Extract (1) above, it is observed regularly enough to merit analytic attention in its own right. We thus turn now to the usage of third-position minimal response as a means to exit from a notyet‑complete turn. Negotiating turn closure This section examines how a third-position  minimal response token is used as a “turn-exit device,”  i.e. a device to propose the closure-relevance of the turn-so-far by conveying that the speaker is not going to add anything more to the current turn. The deployment of this practice is often observed when speakers of Japanese  and Korean  produce utterances that are incomplete from a syntactic point of view, yet which may nonetheless be regarded as complete from the perspective of executing an action. Such “syntactically incomplete yet pragmatically possibly complete turns”

Negotiating boundaries in talk

261

can sometimes engender a problem in coordinating speakership transfer because recipients may not treat them as complete turns due to their syntactic incompleteness, while the speaker-so-far may regard his or her turns to be complete within the particulars of the given contexts in which they are produced. When this type of incongruence of interpretation arises, the speaker-so-far may use a third-position minimal response token to mark the closure of his or her turn and propose the relevance of speakership transfer. Now, what motivates speakers to leave their utterances syntactically incomplete in the first place? In our database, we observe several specifiable environments in which “syntactically incomplete yet pragmatically possibly complete turns” are recurrently observed. One is where a speaker regards the rest of his or her utterance to be easily inferable from the context and thus avoids stating the obvious. An example is seen in Extracts 2 (and 5) above, in which the main clause or the “final component” that would follow Sumi’s utterance in lines 10–11, “nacwungey ye- pom-ina yelum toy-myen”/“Later when sum- spring or summer comes,” is easily inferable by Gina, given that it alludes to what Gina herself stated in the immediately prior talk, i.e. that people at grocery stores lay in a stock of the noodle product in the summer (lines 4–5). Thus, in terms of executing the action of “stating a solution to a possible problem with obtaining the noodle product,” Sumi’s utterance in lines 10–11 can be regarded as complete for all practical purposes. Sumi then deploys a third-position  minimal response token to exit from the turn and move on to resume the activity that has been put on hold. One common environment in which stating the obvious is avoided is where the speaker is voicing something interactionally delicate, such as a disagreement  or bad news. In this type of environment, third-position  minimal response can be used as a way to exit from a turn while leaving the “core” of the delicate matter unsaid (and inviting the co-participants to infer it). The following fragment from a Japanese conversation provides a case in point. In this fragment, Seiji, who often takes a sauna at a public bathhouse, is telling the recipients, Akira and Harumi, about the kind of conversation he would have with other people while taking a sauna. He claims that, because people want to make sure they can get out when it gets too hot, they would deliberately choose “safe” topics

262

Makoto Hayashi and Kyung-eun Yoon

such as weather, etc., on which they would not disagree  with one another or get into an argument (lines 1–2, 4, 7–8). Examine how Akira responds to Seiji’s telling (lines 10–12) and how Seiji in turn responds to Akira’s response (lines 14–15). (6)  (Japanese : RKK 24 [face-to-face: Seiji, Akira, and Harumi]) 01 S:

(yoo suru ni) iya sonna koto nai desu yo::: in.short no such thing NEG CP FP

02

toka yuu yoona [koto yuu] to::giron ni= QT say like thing say if argument PT

03 A: 04

[a: : : : : : : .] Oh: : : : : : .

=natteshimatte::= become

((Lines 01–02, 04)) In short, if you say something like, “I don’t think so,” you get into an argument and, 05 H:

=u:(h):(h):(h):(h):n.= Y(h)e(h)a(h)h:::.

06 A:

=[demo::] but But

07 S:

=[nanka:]:moo atsui detai to omotta like already hot want.to.get.out QT think Like, when you are hot and want to get out,

08

toki ni [derenai [heh heh heh heh .hhhhhhhh when PT can’t.get.out you can’t. heh heh heh heh .hhhhhhhhh

09 H:

[heh    heh    heh    heh    heh

10 A: [shotaimen no 1st.encounter LK I guess it’s more or 11

kaiwa tte daitai sonnan janai conversation QT more.or.less like.that CP:NEG less like that whenever you talk to people who

12

ka na::. FP FP you meet for the first time.

13

(0.7)

Negotiating boundaries in talk

263

14 S: u::::n .hh demo hora:: warito:s- (.) sauna nakama but y’know sort.of sauna friend Hmm:::.hh But y’know, since people become, like, 15

mitaini nacchau kara sa::, like become because FP “sauna buddies,”

16 H:

hhehh=

17 A:

=u:::n. ((with a slight head nod)) Mhm

18 S: → u:::n. Mhm

((with a slight head nod))

19

(0.4)

20 H:

(nani) sauna suki na no what sauna like CP FP (What), do you like sauna?

– Position 1

– Position 2 – Position 3

In response to Seiji’s characterization of “conversations in a sauna,” Akira comments that the same holds more or less true whenever people meet for the first time and have a conversation, whether in a sauna or not (line 10–12). As it points out the potential generalizability of Seiji’s characterization to just about any conversation among people meeting for the first time, Akira’s comment can undermine the very point that Seiji is trying to make, i.e. that the choice of conversational topics in a sauna is intimately related to the external constraints unique to the setting of sauna-taking. In lines 14–15, then, Seiji responds by stating that the people he converses with in a sauna are his “sauna nakama”/“sauna buddies” that he sees repeatedly, thereby suggesting that what Akira said does not apply to the conversations that he is describing. Note that Seiji’s utterance in lines 14–15 is constructed as a clause marked with “kara”/“because,” a preliminary component of a potential compound TCU in the form of (X-kara + Y) ([Because X + Y]).8 Seiji could go on to produce the final component, such as shotaimen demo nai n da yo ne “it’s not really that we meet for the first time,” and thereby make explicit his point of disagreement  with Akira’s prior comment. However, he opts to leave it unsaid; in this context, what would be said in the final component is easily inferable, yet saying it overtly may cause some interpersonal tension. Thus, by proposing to exit from the turn with a third-position  minimal

264

Makoto Hayashi and Kyung-eun Yoon

response in line 18, Seiji avoids stating the obvious and potentially delicate. This proposal of exiting from the turn is accepted and confirmed by Harumi, who initiates a new sequence of talk in line 20. In the previous Extract and also in Extract (2), the syntactically incomplete turn produced in Position 1 embodied a responsive action to what was said prior to it, and the third-position  minimal response served not only the work of exiting from the turn-so‑far while leaving the obvious and/or delicate unsaid, but also the work of exiting from the topic  talk/activity-so-far. Now, when a syntactically incomplete turn in Position 1 embodies a sequence-initiating  action that makes a recipient’s response relevant next, third-position minimal response can be used to accomplish the work of pursuing a more-than-minimal response from the recipient. That is, when the speaker considers the recipient’s production of a mere acknowledgment token in Position 2 to be inappropriate as a response to his or her prior talk in Position 1, the speaker may deploy a minimal response token in Position 3 in order to prompt the recipient to revise his or her response and produce a more sequentially/interactionally appropriate response. A case in point is seen in the following Extract. In Extract (7), Young, the host of a dinner gathering at her apartment, mentions her idea of purchasing some kind of rug to put on the living room floor, and she solicits others’ ­opinions about it (lines 1–3, 5–6). (7)  (Korean : Dinner talk among five friends [face-to-face: Young, Joo, and Hoon]) 01 Y: wuli-twu::, yeki-taka mwe -lul ilehkey we -also here-to something-AC like.this Like others, we have to- we are thinking about 02 com ↑kkal↑ -aya- ↑kkal↑ -kka-pwa:, i a.little put.down-must- put.down-what.if this putting something here like this, on here, ((what 03 wi-ey:, on-to do you think?)) 04 H: i wi-ey-ta[ka? this on-to-to On here?

Negotiating boundaries in talk

265

05 Y: [nemwu com ssellenghan ke-kath too a.little chilly thing-seem It looks too chilly, you know? (.) 06 -ay:, (.) ikhey cip-i:, -INT like.this home-NM Like, ((I mean)) this place, you know? 07 (0.5) 08 J: mwe sikthak? what dining.table “What, ((you mean)) the dining table?” : :((5 lines omitted where Young clarifies that :she is referring to the floor.)) : 14 J: a pata:k¿ oh floor Oh ((you mean)) the floor! 15 Y: u:ng. Ye:s. 16 (1.0) 17 Y: nemwu com ssellenghan ke-kath-ay:, too a.little chilly seem.like It looks too chilly, you know? 18 (2.2) 19 Y: °com° ttattushan ke-llwu:, a.little warm thing-with With °a little° warm thi:ng,

– Position 1

20 J: °um° °Mhm°

– Position 2

21 Y: u:m.

– Position 3

22 (.) 23 H: °patak-ey iss -nun ke° [kuntey] floor-on exist-RL thing DM °But the floor material° 24 J: 25 H: (

[kuntey]:(.) DM Bu:t (.) )[(

)]

266 26 J:

Makoto Hayashi and Kyung-eun Yoon [nam -tul]-i sayngkakha-n[u:n]= others-PL -NM think -RL

27 Y: [ e?]= Huh? 28 J: =kule:n↓ khapheythu-kathun ke-malkwu:, such carpet -like thing-not ((Lines 26 and 28)) Not the kind of carpet that others would think,”

After clearing up Joo’s misunderstanding about the referent of yeki (“here”; line 1) and ikhey cip (“this place”; line 6) in an other-­initiated  repair sequence (lines 8–15; mostly omitted), Young resumes the original activity and pursues Joo’s response to her idea about putting a rug on the living room floor. In line 17, she first repeats a part of what she said in lines 5–6 (“nemwu com ssellenghan ke-kath-ay:, “It looks too chilly, you know?”) and thereby re-proffers an opportunity for Joo to respond to her idea about carpeting. Though this utterance is produced while Young looks down wiping the dining table, she makes it clear that she is pursuing Joo’s response when she subsequently looks up and brings her gaze  to Joo while producing her utterance in line 19. In this utterance, she provides a further detail about her idea of carpeting, thereby providing more material for the recipient to respond to. It finally succeeds in eliciting a response from Joo, but only in the form of a minimal acknowledgment token (line 20). It is at this moment that Young deploys a third-position  minimal response (line 21). By indicating that she is not going to continue her syntactically incomplete utterance in line 19, Young proposes to exit from the turn and hands the floor back to Joo, thereby offering her another opportunity to respond to her prior statement. This move then succeeds in eliciting more elaborate responses from the recipients: Hoon, another participant who has been attending to Young’s talk, starts to present his opinion about Young’s idea (line 23), joined slightly later by Joo (line 24), who makes a suggestion that Young should not buy the kind of carpet that other people often buy (lines 26 and 28). In this section, we examined the use of third-position  minimal response as a turn-exit device . We observed that speakers of Japanese  and Korean  sometimes leave their utterances syntactically incomplete in certain interactional environments, such as those

Negotiating boundaries in talk

267

where the rest of the utterance has already been clearly adumbrated in the prior context, and/or where saying the rest of the utterance would put the speaker (and the recipient) in an interactionally delicate situation. In those contexts, a speaker may use third-position minimal response to propose the closure of, and an exit from, his or her prior (syntactically not yet complete) turn.9 When the syntactically incomplete utterance in Position 1 embodies an action that makes a recipient’s response relevant next, third-position minimal response can be used to accomplish the work of pursuing a more-than-minimal response from the recipient. That is, when a speaker expects the recipient to produce a more-than-minimal response, yet only receives a minimal acknowledgment token, the speaker can deploy third-position minimal response to reproffer an opportunity for the recipient to revise his or her earlier response and provide a more relevant and appropriate response. Summary and discussion In this study we have explored the practice we termed the “thirdposition  deployment of minimal response tokens” observed in Japanese  and Korean  conversations. The guiding perspective for our exploration is that, in order to understand the meanings of response tokens, it is crucial to examine the details of the sequential/activity contexts in which response tokens are embedded, such as what activities the participants engage in that give rise to the deployment of given response tokens, how the co-participants orient to them, and what consequences their deployment has for the organization of subsequent talk. We identified two broad types of actions that are accomplished through the deployment of third­position minimal response. One is to bring closure to the topic / activity-so-far by proposing the speaker’s imminent disengagement from it. The other is to bring closure to a syntactically incomplete turn by indicating that the speaker is not going to continue with the turn-so-far. Thus, we observed that third-position minimal response is used as a resource to regulate boundaries both at the level of turn organization and at the level of topic/activity organization. By way of concluding the present study, we will discuss some implications of our findings.

268

Makoto Hayashi and Kyung-eun Yoon

Response tokens and cultural style In the literature on Japanese communication style, response tokens have been one of the most common objects of study (e.g., Mizutani 1982, 1988; Maynard  1986, 1989; LoCastro  1987; White 1989; to name a few).10 The focus of many, if not all, of these studies has been to claim a direct linkage between response tokens (especially their frequency in Japanese conversation) and cultural traits and values of the Japanese. Thus, it has often been asserted that the behavior of sending frequent “back-channel cues” (or “aizuchi,”  the Japanese term for listener response often used in these studies) in Japanese communication is a quintessential manifestation of the Japanese cultural values, such as “omoiyari”/“consideration/ empathy,” “ wa”/“harmony,” “sasshi”/“empathetic understanding,” etc., as seen in the following excerpts: [T]he emphasis on maintaining group harmony and on having smooth relations would probably cause speaker-listeners to use more aizuchi  to show willingness to co-operate in the conversation and to show support of and attach value to the speaker on the part of the listener. This could be the underlying factor behind the notion that it is impolite in Japanese  not to use back-channel cues.  (LoCastro  1987: 110) The cultural value most relevant to the use of backchannels concerns the Japanese  concept omoiyari, which, according to Lebra (1976), is a key concept for understanding Japanese people. [. . .] [T]he concept generally refers to the creation and maintenance of smooth and pleasant human interactions. [. . .] To maintain harmony, unanimity, or mutual understanding, people must be most sensitive to the recipient’s point of view and feelings. Being empathetic with others’ ideas and wishes may require going beyond indirectness and politeness [. . .]  and involve compliance with the other’s ideas, even if they are opposed to one’s own (Lebra 1976). [. . .] In conversation, the fear of deviating from the speaker’s viewpoint and the eagerness to anticipate, understand, and accommodate the other’s idea may, in part, be demonstrated by the frequency with which the Japanese listener interjects with a backchannel.  (White 1989: 67)

Along similar lines, Iwasaki ’s (1997) study of what he calls the “loop sequence,” which is very similar to the practice examined in the present study, presents the following conclusion.11 I propose that the frequent use of the loop sequence is a consequence of the Japanese conversationalist’s preference toward “ mutual dependency,” a concept held to be important in Japanese interaction. [. . .] The producer of the loop-tail [roughly equivalent to third-position minimal response in

Negotiating boundaries in talk

269

our study] reveals his helpless and submissive position (“I cannot contribute to the floor now. Help me out.”) and at the same time his trust in, or dependency (or amae) on, the other who he believes can provide a help. [. . .] The recipient of the loop-tail may either take up his responsibility and become the floor holder, or he may show his own helpless stance and counters the other’s loop-tail with his own loop-tail. That is, taking a floor in conversation is sometimes more of an obligation that a more powerful party must fulfill at a given moment in interaction rather than a right that he is entitled to.  (Iwasaki  1997: 688–689)

From the point of view of the present study, however, there are some serious drawbacks in this type of claim for a direct linkage between response tokens and cultural values. That is, these accounts are invoked on behalf of the aggregated occurrence of response tokens (or the loop sequence), yet they do not show in any way whether the participants themselves in fact orient to such notions as empathy , mutual dependency, etc., when they produce response tokens or the loop sequence in the particular interactional contexts in which they are situated. For example, if we look again at the Extract examined above, we do not find any evidence that the producers of third-position  minimal token (or what Iwasaki  1997 calls the loop-tail ) display their “helpless and submissive position” or their “trust in, or dependency [. . .] on” their interlocutors. The problem lies in the fact that, because these accounts are formulated in a manner far removed from the details of the actual contexts in which response tokens occur, it is unclear whether they provide any useful explanation of the meanings to which the participants themselves orient (Schegloff  1982, 1993; Zimmerman  1993). Of course, this is not to say that response tokens (or the loop sequence) are never used to display cooperation, empathy , dependency, etc. If one can show that displaying cooperation, empathy , dependency, etc., with a response token is a relevant action for the participants in a particular moment within the trajectory of the particular interactional activity that they engage in, accounts based on such notions may be viable for those particular cases. However, unless one can establish that displaying cooperation, empathy , etc., is invariably relevant for the participants when producing response tokens across different situations, such accounts as those cited above, however intuitively appealing they may be, remain equivocal as general explanations for the “meanings” of response tokens (Schegloff  1982, 1993; Sorjonen  2001).

270

Makoto Hayashi and Kyung-eun Yoon

As our analysis of third-position  minimal response has shown, the meanings of response tokens depend crucially on the local sequential/interactional context of their usage and the kinds of tasks the participants orient to accomplishing by using response tokens in that context. Thus, rather than assuming a direct linkage between response tokens and cultural values and proposing premature generalizations about the meanings of response tokens for a particular cultural group, we should pay close attention to the immediate interactional environment within which a response token is precisely placed, and with respect to which it gets its meaning within the moment-by-moment contingencies of interaction. By doing so, we can gain an understanding of the detailed workings of response tokens in situ, not as an undifferentiated class of expressions taken out of context, but rather as deployable “devices” with which to accomplish different interactional ends in different contexts (Jefferson  1984, 1993; Sorjonen  2001). We hope that our study serves as a reminder of the importance of grounding our claims about response tokens (or any bit of human conduct, for that matter) in the careful observation of empirical details of their actual context of production. Cross-cultural /cross-linguistic comparisons While the predominant focus of cross-cultural/cross-linguistic studies is on substantiating differences rather than evidencing similarities (Schegloff  1987a; Beach  and Lindström  1992), the present study was not motivated by a concern with demonstrating crosscultural/cross-linguistic differences. Rather, our interest has been to explore the operation of systematic solutions to certain general organizational problems of interaction and to evidence that those systematic solutions are utilized by speakers of different languages under similar interactional circumstances. Regulating the completion (or expansion) of turns, topics, activities, encounters, etc., is an omnipresent problem that requires continual resolution by the participants in interaction in any language community. Our study focused its inquiry on one specific form of conduct (minimal response) produced in one specific sequential environment (after a recipient’s production of a minimal response to one’s prior talk) and demonstrated how it is mobilized in both Japanese and Korean 

Negotiating boundaries in talk

271

language communities as a systematic solution to the problem of regulating turn completion and topic /activity closure in that specific sequential environment. As discussed above, Japanese  and Korean  are not the only languages in which third-position  minimal response can be deployed to regulate topic /activity boundaries; it has been reported that English speakers utilize a similar practice to accomplish topic/ activity transition (Jefferson  1993; Drew  and Holt  1998). As for the usage of third-position minimal response as a device to exit from a syntactically incomplete turn, we are not aware of any study that has reported on this practice in other languages. Given the limited amount of empirical work done on response tokens in different languages, we cannot claim anything conclusive about whether the usage of third-position minimal response as a turnexit device  is unique to Japanese and Korean. We certainly need further research to verify this point. That said, we have some tentative basis for speculating about the possibility that third-position minimal response as a turn-exit device may indeed be more likely to occur in Japanese and Korean because of certain distinctive features of turn constructional practices observed in these two languages, which may not be shared by some other languages, such as English. Our speculation is based on recent work on turn­constructional practices in Japanese and Korean and their relation to turn-taking (Maynard 1989; Fox et al.   1996; Kim 1999b, 2001; Tanaka  1999, 2000; Hayashi  2003, 2004; Morita 2005; Young and Lee 2004; Iwasaki  2007). It has been pointed out that, in conversational Japanese  and Korean , speakers often adopt an “incremental” or “bit-by-bit” approach to turn construction. That is, rather than producing a whole turn at once, Japanese and Korean speakers tend to ­produce smaller units of talk (often phrasal units) during the course of a turn and leave a space at the end of each such unit for the recipients to produce a minimal response to acknowledge the just-produced unit (Maynard  1989; Kim 1999b, 2001; Tanaka  2000; Iwasaki  2007). As they receive such acknowledgments from the recipients at intra-turn phrasal unit boundaries, the speakers build up the whole turn by piecing together these smaller units. This type of incremental turn construction, it has been argued, relates to several characteristics of the “syntactic practices” (Fox  et al. 1996)

272

Makoto Hayashi and Kyung-eun Yoon

employed by both Japanese and Korean speakers in conversational interaction. These syntactic practices include the following: In conversational Japanese and Korean, (1) word order  within a clause is quite flexible, particularly toward the beginning of the clause; (2) many syntactic constituents (including core arguments) can be, and often are, left unexpressed (i.e. the prevalence of so-called “ellipsis”); and (3) elements placed toward the end of a clause (e.g., clause-conjunctive markers, predicate negation, quotative markers, etc.) can retroactively transform the grammatical  status of the clause-so-far (Tanaka  1999; Hayashi  2003). If word order is flexible and many syntactic constituents can be left unexpressed, and if clause-final elements can retroactively transform the grammatical  status of what has been produced so far, then the presence of some constituent at the beginning of a given TCU provides relatively little information as to how the rest of the TCU will eventually develop. This observation has led some researchers to claim that Japanese (and Korean) recipients tend to adopt a “wait and see” approach to turn-taking , which manifests itself in the frequent interpolation of acknowledgment tokens at TCU-internal phrasal unit boundaries discussed above. Thus, Fox  et al. (1996) argue that, if turn-­beginnings do not provide the recipient with much information about how the utterance is going to proceed, then it makes sense for speakers to produce relatively short [units] whose interactional implications the recipient can acknowledge or question as the speaker works on a larger turn (exactly what happens with multi-TCU turns in English). This allows the recipient to acknowledge small pieces without having to know exactly where the speaker is going with the full turn.  (Fox et al. 1996: 212)

The discussion presented so far seems to provide some basis for arguing that the usage of third-position  minimal response as a turn‑exit device  is particularly fitted to the type of turn-­constructional ­practices described for Japanese  and Korean . That is, as discussed above, due to the limited mid-turn projectability of how the turn will eventually develop, recipients in Japanese and Korean conversations are systematically motivated to provide acknowledgment tokens at turn-internal junctures where the speaker’s turn is still not yet complete. When, as it happens, the speaker decides that the syntactically incomplete bits of talk that the recipient has just acknowledged are already sufficient from the perspective of executing an action, he or

Negotiating boundaries in talk

273

she can take advantage of the intra-turn acknowledgment token produced by the recipient as an opportunity to counter it with another minimal token and thereby indicate that the speaker is not going to continue with the turn any more. In other words, the very operation of third-position minimal response as a turn-exit device appears to rest on the systematic provision of intra-turn opportunities for recipients to insert acknowledgment tokens during the course of the speaker’s turn in Japanese and Korean. Thus, we would like to suggest that turn constructional practices in Japanese and Korean offer the kind of interactional environment that seems to be more amenable to the deployment of third-position minimal response as a turn-exit device than turn-constructional practices in some other languages, where there is no such systematic provision of intra-turn opportunities for inserting acknowledgment tokens by recipients. Of course, what has been discussed above remains speculative until further research is done on response tokens in languages other than those discussed here. We hope that our study serves as encouragement for others to explore the detailed workings of response tokens in a much wider range of different languages. Acknowledgments Earlier versions of this article were presented at the International Conference on Conversation Analysis (Helsinki, May 10–14, 2006) and the 92nd Annual Convention of National Communication Association (San Antonio, November 15–19, 2006). We are grateful for invaluable feedback provided by audience members at both meetings. We are also indebted to Hiromi Aoki, Kaoru Hayano, Kiyomi Kawakami, Mi-Suk Seo, and Jack Sidnell for their helpful comments on earlier drafts of this paper. Appendix AC CIRCUM CP DM EMPH FP

accusative particle circumstantial copula discourse marker emphasis marker final particle

274 INT LK N NEG NM O PL POL PT Q QT RL SP TP

Makoto Hayashi and Kyung-eun Yoon intimate speech level nominal linking particle nominalizer negative nominative particle object particle plural marker polite speech level particle question particle/suffix quotative particle relativizer suffix subject particle topic particle

Notes 1 We realize that “yeah” and “mhm” in English can work quite differently, especially with regard to whether they function to hand the floor back to the immediately prior speaker (i.e. the continuer usage; Schegloff ) or whether they serve to project their producer’s incipient shift to speakership (Jefferson 1984; Drummond and Hopper 1993a, 1993b). It is not clear to us where “un” in Japanese and “um” in Korean stand in this regard, and it is for this reason that we use the term “minimal acknowledgment” in a somewhat loose manner to capture this equivocality. 2 We exclude from our analysis those cases in which the deployment of a response in Position 3 is made relevant by the type of minimal response produced by a recipient in Position 2. For instance, such minimal responses as “Really?” produced by a recipient in Position 2 make relevant a subsequent response by the speaker-so-far in Position 3, as seen in the following fragment: (a)  [Heritage 1984: 302] R: And I got athletic award C: REALLY? R: Uh huh.

Position 1 Position 2 Position 3

The sequential import of a minimal response in Position 3 in cases like this is very different from that of a minimal response in Position 3 when it follows a “yeah”/“mhm”-like acknowledgment token in Position 2, which does not make a subsequent response relevant. It is therefore important to make an analytic distinction between cases like (a) and those examined in this study. The practice we examine in our study,

Negotiating boundaries in talk

275

then, is not simply a third-position  deployment of minimal response but is more appropriately characterized as the “post-minimal-response deployment” of a minimal response token. For convenience, however, we will use the label “third-position minimal response” in the rest of this article to refer to the practice of deploying a minimal response after another minimal response in Position 2. 3 We juxtapose the terms “topic/activity”  here in order to capture the fact that the practice being examined is used to negotiate not only the boundaries of “what is being talked about” (i.e. topics) but also a shift in “the course of action that participants engage in” (for example, a shift from discussing the availability of a product to writing down its name; see Extract 2). 4 Gina’s self-interrupted utterance in line 9 could be an initiation of a similar statement, though it is not entirely clear. 5 “Ey” is a polite version of “um.” 6 Lerner  (1991) uses the term the “preliminary component” to refer to a non-final sub-unit within a compound turn-constructional unit (e.g., [If X] of [If X + then Y]). The term “final component” is used to designate the final sub-unit that completes the whole compound unit (e.g., [then Y] of [If X + then Y]). 7 Compare this with more common, “unmarked” cases in which the speaker continues after the recipient’s acknowledgment token and brings the turn to syntactic completion: [Korean : Two Friends] 01 K: hankwuk ay–tul wena: k chongmyenghay–kacikwu:, Korea kid–PL very smart -because Because Korean  kids are very smart, 02 J:

u: m= Mhm

03 K: → =kumpang kumpang ttalaha–nta. quick quick catch.up–DC they catch up so fast.

Below is a similar case from Japanese:   [OBS 9] ((Middle-aged mothers are discussing how their daughters now wear their old clothes.)) 01 C:

tenkin no tanbi ni sutechatta kara sa::,= transfer LK every.time threw.away because FP Because I threw away ((my old clothes)) every time we moved,

02 B:

=[u:: n. Mhm

03 ?:

=[u:: n. Mhm

276

Makoto Hayashi and Kyung-eun Yoon

04 C: → nai kedo, attara kirareta not.exist but exist.if could.wear we don’t have any left, but if we did, 05 →

(kamo ne, °soo yuu no.°) maybe FP so say FP maybe ((my daughter)) could wear such clothes.

8 It has been pointed out in the literature that so-called conjunctive particles such as kara (“because”) and kedo (“although”) have come to be used somewhat like final particles in conversational Japanese  and that they do not always indicate that the preceding clause is a “subordinate” clause to be linked with a forthcoming (or already-produced) “main” clause (Kokuritsu Kokugo Kenkyuujo 1960; Nakayama and IchihashiNakayama 1997; Iwasaki and Ono 2002). While this is true, examination of our data concurs with Tanaka ’s (1999: 195–199) observation that, when conjunctive particles are used turn-finally (as opposed to turn-internally), they are typically delivered in a turn-final, falling intonation. Furthermore, even when a conjunctive particle is produced in a falling intonation, it can be retroactively recontextualized as a “subordinate” clause-marking by producing a “main” clause after it. Thus, a third-position minimal token such as that observed in Extract 6 can be seen as a way to “seal off” the possibility of structural expansion of the turn-so-far. 9 In a study of some uses of “yeah” by Korean  learners of English in spontaneous interaction, Park (2004) reports on a usage of “yeah” by those learners that is strikingly similar to the usage of third-position  minimal response as a turn-exit device  described in the present study. According to Park (2004: 91): “[N]onnative speakers often end their grammatically complete unit with continuing intonation, rising intonation, or simply non-ending intonation, and thereby, not properly ending their turn intonation-wise. As a result, they get native speakers’ ­continuer and miss an opportunity to leave their turn properly. They then need to use some extra marker or device to show their interlocutor that they are actually finished and need to leave their turn explicitly.” The “extra marker or device” that Park refers to is the minimal token yeah, as illustrated by the following fragment. [Park, 2004: 91] ((The instructor (T) has asked the student (D) which chapter of the textbook was the most difficult to study.)) 01 D: A lot and in case of:: 11, 02 T: mhm, 03 D: Word is very (0.2) difficult=hhh[h 04 T: [Right. Right. 05 There were a lot of animal name[s and so:: okay 06 D: [Yeah

Negotiating boundaries in talk 07 D: 08 09 T: 10 D: 11 T:

277

um: and:: (2.8) But I like to have a chance with(0.4) with other people and listen to other opinion::, mhm, °Yeah Okay. That’s good. All right. That’s basically that.



D’s utterances in lines 1, 3, and 7–8 constitute her answer to the instructor’s question about which chapter of the textbook was the most difficult for her to study. Notice that D produces the sentential unit in lines 7–8 with continuing intonation at its end (represented by a comma). This appears to prompt T to produce a continuer (“mhm” in line 9), which displays her understanding that D’s answer is not yet complete. Faced with this continuer, D produces “yeah” to indicate that she was in fact finished with her answer before the instructor’s continuer and that she is not going to continue her turn any more.   Park (2004) does not address the native-speaker practice of deploying third-position minimal response in Korean, and she appears to regard the use of “yeah” shown above as a practice creatively devised by nonnative speakers in order to cope with their less-than-fully-­developed linguistic and interactional competence. However, it seems clear to us that there is a strong connection between what Korean learners of English do with “yeah” to exit from a turn and what native Korean speakers do with “um” (and its variants) in similar interactional contexts. While, of course, nonnative speakers’ conduct cannot always be accounted for in terms of L1 transfer, careful examination of littleknown L1 practices may sometimes help understand what appears to be a peculiarly nonnative practice in a second language. 10 Though response tokens in Korean  have not been studied as extensively as the Japanese  counterpart, there are a growing number of recent studies that explore their functions in conversation; see K. Kim 1999b; H. Kim 2004; Young and Lee 2004. 11 What Iwasaki (1997) refers to by the “loop sequence” subsumes the type of sequence of utterances examined in this study, but it is not limited to it. Though the schematic representation of the loop sequence Iwasaki provides (see below; slightly modified from Iwasaki 1997: 662) corresponds fairly closely with what we examine in our study, he includes those cases where the second-position utterance by speaker B is an utterance that makes a response relevant next, such as a soo nan desu ka “Oh, is that right?” SPEAKER A (Some utterance) ee (“Yeah”)

SPEAKER B ee (“Yeah”) (Some utterance)

278

Makoto Hayashi and Kyung-eun Yoon

As stated in Note 2, we believe that the sequential import of a minimal response in third position following an utterance like “a soo nan desu ka”/“Oh, is that right?” is very different from that of a minimal response following another minimal acknowledgment token. For that reason, we made an analytic distinction between these two types of sequences and focused only on the cases where a minimal response follows another minimal acknowledgment token.

Part I V

Action formation and sequencing

9  Alternative responses to assessments   Marja-Leena Sorjonen and Auli Hakulinen

Introduction In this chapter we examine part of the paradigm of utterance types used to agree  with a prior assessment in Finnish. We use the term “assessment” in the same sense as Pomerantz (1984) and Goodwin  and Goodwin  (1992), for example, to refer to an evaluative act, typically performed by an utterance that contains a negative or positive predication of a referent or a state of affairs expressed by the subject or the object of the sentence. Alternatively, the assessable is something that can be inferred from the context. We focus on cases in which agreement  is accomplished by presenting “the same evaluation ” as that of a previous assessment (cf. Pomerantz  1984: 66–68). In each of them, the recipient repeats part of the preceding assessment by her co-participant, leaving out the assessment term.1 Figure 9.1 shows the range of alternatives; those we will discuss are marked in boldface. The schema is a simplification in which we have, for the purposes of illustration, used as the first assessment a prototypical clause-type for assessing something: a predicate nominal clause with an evaluating adjective complement. In practice, there is variation in the clause type, and the evaluating element, “X” in our schema, is often a more complex phrase. Equally there are other verbs besides the copula  “on” used in this context, but “on” is by far the most frequent one. What is striking in this schema is that of the six alternative response types, only two begin with an element other than the verb. This reflects the facts that in Finnish the subject is not an obligatory clausal element, nor is the order between the subject and the verb

282

Marja-Leena Sorjonen and Auli Hakulinen S

1st assessment :

V

se mekko on that dress

is

X

| V

S

X

hieno | on(pa) se mekko hieno great | is(CLI)i that dress great ‘that dress is great’

response :

ADV + V niin on

V+S

on se

V (+ V) on on on

S+V

V + PRT on joo

se on

i -pA

is a clitic particle which here adds to the emphatic function of the sentence (see Hakulinen et al 2004: 800).

Figure 9.1  Responses to assessments with a verb repeat

grammatical ly constrained (see, e.g., Heinämäki 1976; Vilkuna  1989; Helasvuo  2001). In contrast with a language like English, for example, in Finnish, the word order verb + subject  does not, by itself, convey interrogativity. The crucial element in a “yes”/“no” interrogative is the question particle -ko (in colloquial Finnish also -ks), which is added to the initially positioned verb: “on-ko se totta?”/“is-q it true?” Accordingly, verb-initial word order can be deployed for indicating that a declarative sentence is either emphatic, as in the schema above: “on-pa se mekko hieno”/“that dress is really great” (literally, “is-clitic that dress great”), or responsive  – for example, that it is an answer to a question, an assent or an agreement to an assessment (Hakulinen et al. 2004: 1316, 1325). So, one possible answer to the question “on-ko se totta”/“is it true” is a verb-initial sentence “on se”/“it is” (lit. “is it”). In our previous work (Hakulinen  and Sorjonen,  forthcoming), we examined four alternative assessment responses, “on,” “on on,” “on se,” and “on joo.” We found that a mere verb repeat  (V) asserts agreement and is usable in a wide range of sequential contexts. The reduplicated verb repeat (VV), in comparison, occurs in a more narrow range of sequential contexts. The design of the prior assessment is especially relevant for the understanding of its use. Specifically, we found reduplicated verbs used where prior assessments expressed incontestable truths such as culturally conventionalized proverbs and the like. Finally, the (V + joo) format was found to be implicative of topic  closure and routinely occurring in sequential environments where closure is a relevant possibility. In this chapter we will focus on three of the alternatives: “on se,” “se on” and “niin on.” In these alternative assessment response

Alternative responses to assessments

283

formats there is, in addition to the verb, a deictic element: either the third-person anaphoric  pronoun “se”/“it” (in plural “ne”/“they”) or the adverb “niin.” The adverb “niin” is a derivative of the pronoun “se”/“it”; it is its plural form in the instrumental (instructive) case, meaning approximately “so; in that way.” It is even used as a response particle; in this usage, as well, the anaphoric meaning is still discernible (see Sorjonen 2001: 281–282, 287). Our initial aim was to explore the minimal pair “se on” vs. “on se.” However, it turned out that there were very few instances of “se on” in our data. Instead, the response type “niin on” proved to be extremely frequent. For that reason we examine all three alternatives. We show that the detailed structure of the response, including its word order  and the types of elements it contains, specifies the way in which a “same evaluation ” should be understood in a given sequential context. We argue that a key parameter these alternatives index is the extent to which the participants approach the object of evaluation from the same point of view. Our database consists of approximately thirty interactions including both face-to-face conversations and telephone calls. Most of our data come from conversations among friends and family members, but we have also included a number of videotaped institutional interactions, for example from a social-insurance office and from a hair salon. VS response implying difference in perspective We first consider cases where, in her response, the recipient repeats the finite  verb, followed by a pronominal subject (or by a pronominal object as in Extract 2 below). We refer to this as a VS response. By using the VS response, which sustains through ellipsis the evaluative term in the prior assessment, the recipient agrees  with her co‑participant’s assessment. We argue further that, at the same time, she implies that their experience of the object of evaluation and its relevance for them is of a different kind. The nature of the difference may but need not be specified in the subsequent talk. Especially in conversations between friends, the VS response may be found in sequences where the topic  is equally accessible to the participants. These cases thus appear to run against the meaning of the VS response as one that implies a difference

284

Marja-Leena Sorjonen and Auli Hakulinen

in the  perspectives. For instance, in the following example the ­participants are ­discussing the weather. (1)  (SKK/Sg01 B03, 8–9) 01 L:

Minkäslainen sää teillä nyt on.= what.kind weather you.ade now is What kind of weather do you have now.=

02 A:

=.hh Tänään on kaunista. Ihan kirkas taivas näkyy olevan =.hh today is beautiful quite bright sky seems be.prpc =.hh Today it is beautiful. There is quite a clear sky

03

tuol[la.hh there over[ there

04 L: 05

[Niin- e niin täälläki ja niin komee ruska että [so er so here.too and such great autumn.coloring that [So it is here as well and such a great autumn coloring

.hh[h

06 A:

[JOO:.hh [YEA:.hh

07

(.)

08 L:

että nyt:hän siellä [o( )( )( ). that now.cli there [w( )( )( ). that now it (would) ]

09 A:

] ]

[Nyt kelpais olla]k[i. Mut kyllä [now worth.would be.cli but sure [Now it would be something. But it is really

10 L:

[Ai( ) ]

11 A:

sie]l on ihanaa heti kun ei sada. there is lovely immediately when neg rain lovely there as soon as it’s not raining.

12 L:

↑Nii, [Joo, ↑Right, Yeah,

13 A: → 14 →

[.hh Kyl se on: syksy on niin mahdottoman [ prt it is autumn is so impossibly [.hh It really is: the autumn is so extremely

kaunis.[h beautiful.h

Alternative responses to assessments 15 L: →

285

[On se.= [is it 2 .

16 A: → =.Jo[o .Y[ea 17 L:

[>Kyllä mä vi- ei viikonloppuna menen< .hhh mä meen [> sure I we- neg weekend go.1 I go.1 [>I’m surely wee- at the weekend I’m going< .hhh I’m also

18 kans t- kääntää ↑maat ja .hh laittamaan kuntoon varmuuden also turn lands and put shape.ill safety’s going to dig the land over and .hh put everything in shape to be 19 vuoks kaikki jos (.) jos sitte ei tuu enää °mennyks° sake everything if if then Ø neg comes anymore going on the safe side if (.) if one doesn’t get to go there °anymore°.

Anna’s (A) assessment in lines 13–14 is located in a larger segment of weather talk. It was preceded by talk about Leila’s (L) summer cottage which Anna had checked out while visiting hers during the weekend. The assessment in lines 13–14 moves the talk into autumn weather in general. This generalization can be heard as closing  implicative. However, it receives a VS response, which, we argue, foreshadows further talk by suggesting that the participants have different perspectives on the topic  being discussed. The ensuing talk explicates the difference in the participants’ perspectives . In overlap with a further closing-implicative turn by Anna (an inhaled response token “.joo,” cf. Hakulinen 1993) in line 16, Leila, continuing at the pitch level of her preceding response, tells of her plan to visit her cottage the following weekend. This leads into a troubles-telling; it turns out that Leila is considering selling her cottage. The VS formatted response is thus deployed by the recipient, at a potential place for moving into the closing  of the call, as a way of constructing a bridge from the joint talk about the weather to her own, related talk about giving up her summer place. In the next example from a social-insurance office, the official (O) responds with a VS-formatted turn to an assessment by the client (C) in line 10. The client has come to ask whether he should pay back some of his study grant, as he has earned too much in his part-time work to be eligible for the full amount that he received.

286

Marja-Leena Sorjonen and Auli Hakulinen

(2)  (Kotus, T1000: 2–3, Social insurance office, Helsinki) 01 O:

Paljokos sulla on ollu niitä tuloja. how.q.cli you.ade is be.ppc those.par income.pl.par How much income have you had.

02 C:

No mull on olluk kaheksantoisttuhatta et well I.ade is had eighteen.thousand prt Well I have had eighteen thousand so

03

seittemäntoisttuhatta seittemänsataa. seventeen.thousand seven.hundred.

04

(.)

05 O:

Joo katotaa (hetki). prt look.pass moment Yeah let’s have a look (a moment).

06

(2.0) Official taking a printout from the printer

Mä otan tom päätöksen tosta samalla mä 07 O: I take.1 that decision there.ela same.ade I I’ll take the decision from there while I’m 08

[Takes a paper from the printer [haen sen. ] [fetch.1 it ] [fetching it ]

09 C: → [Se ov vä]hä hankala arvioida etukätee. [it is some difficult estimate Ø beforehand [It is a bi]t difficult to estimate ((them)) in advance. 10 O: → O:n ne. joo; se o ihan totta. is they prt it is quite true O: n ne. yes; it is quite true. Tuntilaisena teen. 11 C: hour.ag.ess do.1 I do ((the work)) as a timeworker ((“paid by the hour”)) 12 O:

clears throath krhrr

13

(2.0) Official reading the text on the computer screen

14 C:

°Joo° °Yeah°

15

(0.2) Official reading the text on the computer screen

16 O:

.mt Eli kakstuhatta↑kaks: on tulluv valvonta, prt two.thousand.two has come control .tch So two thousand and ↑two: there’s been a control,

Alternative responses to assessments

287

The assessing utterance in line 9 is an account: The  client provides his reason for having to pay back some of the allowance. The utterance does not contain any explicit person reference forms referring to the client himself but is formed with a zero-person  format (Hakulinen  1987; Laitinen  2006). Through it, the account  is constructed as generic (i.e. applicable to anyone), and this invites the official to display affiliation  with the client and in that way her acceptance of the account  (cf. Sorjonen  2001: 131–140). The official treats the turn as one that sought agreement  by responding with a repeat of the ­copula verb “o:n,” followed by the subject pronoun “ne”/“they” (line 10). The copula “o:n” is here in its singular form since in colloquial Finnish there is no number agreement in the third-person ­plural (see also Extract 4, line 6). The antecedent to the plural pronoun “ne” is found in line 1, “niitä tuloja”/“the income,” which is plural in Finnish. In line 9, the reference is achieved with a zero anaphora object, marked with a Ø-sign in the English gloss. In Finnish, the conditions for zero anaphora are more flexible than in English; anaphoric  subject and object pronouns can alternate with a zero both across sentence boundaries and across speaker change (see Laitinen  and Hakulinen,  2008). By using this response format the official implies that the participants’ perspectives  on the issue are different. While the client presented his evaluation as someone who is expected to have correctly estimated his income, the official’s response is understood to be based on the experience she has gained in her job, dealing with other clients in a similar position. This difference in perspective is treated as readily understood – it need not be spelled out. From the assertion of agreement , the official proceeds to upgrade her agreement: “It  is quite true.” With her turn then the official accepts the client’s account  for having earned too much money. The repeated verb is accented, and also lengthened. Our intuition is that the prosody  contributes to the reassuring flavor of the response. What we have seen so far is that the VS response, while asserting agreement , conveys a difference in the participants’ perspectives  on the assessable item (autumn weather, difficulty in anticipating earnings). Even though the prosody  may work to strengthen the agreement (as in Extract 2), the verbal construction opens up a possibility for further talk from a different perspective. The ­dif­ference hinted at by the design may but need not be specified in the ensuing talk.

288

Marja-Leena Sorjonen and Auli Hakulinen

SV response implying independence of speaker’s stance  Most of the alternative responses in the schema presented above are verb-initial – word-order being a means of indicating that the turn is a response. As there are also two alternatives that do not begin with a verb, the question we face is what speakers are doing when using these formats. We shall first examine the S + V format where the recipient starts the response with an anaphoric  subject pronoun, followed by a repeat of the verb in the prior assessment. Our intuition is that by starting the response with the subject, the recipient, while agreeing  with the evaluation presented by the first assessment speaker, implies having already held the same stance  (see also Tainio 1993: 189–193). In the few cases we have, the following is the only one that features an exact repetition  of both the subject and the verb in the co-participant’s prior assessment. The example comes from our field notes, which is why some details of the sequential environment are not available. We participated in a conversation in which a colleague was praised for having given a successful job talk. After a characterization by B of the kind of talk the candidate had just given, speaker A, who did not participate in the job talk, presents an assessment which consists of an evaluation of the person as a teacher in general. It is formatted as a predicate nominal clause, with an SV word order , and it evaluates the person in extreme terms. (3)  (Field note, spring 2007) 01 A:

Se on ihan hirveen hyvä opettaja. it is quite terribly good teacher He is just an extremely good teacher.

02 B: → Se on. it is He3 is

By responding  with the format “se on,” B both asserts agreement  with A and implies the independence of her stance . In the prior talk, B was the one who was telling the others about the specific event, portraying the success of the colleague’s teaching. Now that A presents her assessment, B’s turn can be heard as agreeing  but simultaneously confirming A’s stance . The function of the SV response comes close to that of the partial modified repeats in

Alternative responses to assessments

289

English investigated by Stivers  (2005: 138–143). Stivers  examines response turns consisting of a repetition of the subject and the verb of the prior assessment (e.g., “it is”), as in our Extract 3 above. She shows that by producing the repeat verb with marked prosodic emphasis (stress) the second speaker claims primary rights to the claim. In Extract 3 above, the copula  verb in line 2 carries an accent too. However this is not an obligatory feature of SV responses as illustrated by Extract 4 below. Thus it seems that an action that in Finnish is performed with the word order  of the response is expressed through prosodic means in English. In Extract 3, the fact that the response begins with the subject, instead of the finite  verb, is essential for its function. However, unlike in Extract 3, in our other cases of SV-formatted responses, the subject pronoun differs grammatical ly from that of the previous assessment. The following example from a hair salon contains a modification – specifically, a change from singular to a plural in the subject position (lines 3 and 6). (4)  (Kotus, T1208: 61, Eastern Finland, hairdresser’s) 01 C:

Joo. Yeah.

02

(33.7) H cutting Client’s hair

03 H: → .mt Voi mahoto mite o itsepäine hius. oh impossible how is obstinate hair .tch Oh my god what an obstinate hair. 04 C:

Mm-m.

05

(0.6)

06 C: → .nf Ne on. they is. 07

(2.0)

08 H:

Mite sie sitä aina kotona ite laitat. how you it.par always home.ess yourself make.2 How do you set it at home by yourself.

09 C: No geelillähän sitä pittää °mh°, (0.7) muottoilla. well gel.ade.cli it.par Ø must shape Well you have to use gel °mh°, (0.7) to shape it. 10

(5.0)

290

Marja-Leena Sorjonen and Auli Hakulinen

11 C: °Kyl°lä se aina asettuu ku s’tä aikasa, surely it always settles when Ø it.par time.poss It does settle alright when one just ((treats)) it 12

aikasa vaa °et[tä° (men]ee). time.poss prt so.that (goes) for awhile so that it (goes ((down)))

13 H:

[Mm-m. ]

At line 3, the hairdresser evaluates the client’s hair. By using the nominative singular “hius”/“hair” instead of the plural form “hiukset,” the variant used in everyday speech, she treats the client’s hair as material for professional designing. The client agrees  with the evaluation of his hair, but the subject initial format (line 6) indexes his independent knowledge of the matter at hand. By choosing the plural form of the anaphoric pronoun “ne”/“they,” he brings forth his own experience, that of doing his hair every day. This is then elaborated in the subsequent talk (lines 9–12). This difference in the speakers’ perspectives resembles what we discussed above, in connection with Examples 1–2, with “on se” (VS) format. However, in our present example, the difference in the perspective is conveyed by the change in the grammatical  number of the subject (from singular to plural), whereas the SV order is doing another job – that of indicating independent access to, and prior stance toward , the object of the assessment (the speaker’s own problematic hair). Thus the two formats differ in the following way: The VS format conveys a difference in the perspectives of first assessor and recipient, and the SV format conveys the prior existence of the assessment recipient’s stance . In the SV format, expressing agreement  may be intertwined with a function of������� confirming the co-participant’s stance . “Niin on” response displaying unmodified agreement In the introduction to this chapter we mentioned that while the subject initial “se on”/“it is” response is infrequent in our current database, there is another, much more frequent response type in which the verb occupies the second position within the turn. This response format begins with the adverb “niin”/“so; in that way” which is then followed by a repetition of the finite verb of the first assessment

Alternative responses to assessments

291

(“niin on”/“so is”). With this response type, the recipient asserts unmodified agreement  with the prior speaker, without the further implications characteristic of SV and VS responses. In the following example the speakers discuss the reservation they need to make for a badminton field where they have just started to play. In lines 8–10 they compare this field with the one where they used to play. (5)  (SKK/Sg06 A06) 01 S: .mt Et tota niin .hhhhh hh mä [tilaan viel toisen ja sit prt prt prt I [order.1 yet another and then .mt So uhm then .hhhhh hh I’ll [reserve yet another ((time)) and then 02 V: [hhhhhhhhh 03 S: katotaa vähän millanen m:millane se kenttä et onks se look.pass a.bit what.kind what.kind the field prt is.q it we’ll see a bit what kind m:what kind of a field it is and if it is 04

sama järjestely ku viimeks. same arrangement as last the same arrangement as last time.

05 V:

°Jo[o°. °Yeah°.

06 S: 07

[>Sehän ei ollu mikää jättihyvä mut kylhä siin [ it.cli neg was any super.good but sure. cli it.ess Ø [It wasn’t y’know particularly great but one can surely

pelaa.= plays play on it.=

08 V: → =Kyl siin pelaa joo ei se tos >et ne oli sure it.ess Ø plays yeah neg it there prt they was =Sure one can play on it yeah it won’t there >so those were 09 →

kyl kivat kentät oikeestaan tuolla (1.2) .mt Olarissa. sure nice fields actually there Olari.ine real nice fields there in fact (1.2) .tch in Olari.

10 S: → Niin on. Nehän on tosi hyvät. so is they.cli is true good.pl Niin on. They are real good. 11

(1.0)

12 V:

Et kylhän tos tietyst voi sitte (.) harkita. prt sure.cli there of.course Ø can then consider. So one can indeed think about it then of course.

292

Marja-Leena Sorjonen and Auli Hakulinen

13 S:

Välillähän sitä voi harkita sometimes it-par can consider Every now and then one can think about it

14

[(tietyst sitte) [of.course then [(of course)

15 V:

[Täytyy sit vaa kattoo tosiaanki [must then just look indeed [One just must see to it indeed

16

et e(h)i tost r(h)ytmist tipahda ku heh se o that neg that.ela rhythm.ela drop.out when it is n(h)ot to drop out from the rh(h)ythm as it heh seems to be

17

vitunmoinen työ [näköjää päästä toho ryt(h)mii näi fucking job [apparently get that.ill rhythm.ill thus a fucking hard job to get into the rh(h)ythm like this

18 S: 19 V: 20 S: 21 V:

[h(h)h he he he

p(h)uhelim(h)itse ainaki. [.hhhh et mä en sit phone.through at.least prt I neg.1 then over the pho(h)ne at least. [.hhhh so I don’t [hihihihi .mh(h)h £mielellään nyt niinku tipu siitä£. preferably now prt drop.out it.ela £like to drop out it sort of£.

After agreeing with Sami (S) that the present field is manageable, Ville (V) moves to give a positive evaluation of the field where they used to play (lines 8–9). Sami agrees with this using the “niin on” format response in line 10. As was mentioned in the introduction, the adverb “niin” is derived from the anaphoric pronoun “se”; its meaning is not quite equivalent to the English adverb “so,” but could be approximately glossed as “in that way.” “Niin” refers to the evaluative term in the co-participant’s preceding assessment (to “kivat”/“nice” in line 9 in this example). “Niin on” is the only one of the response formats that has an evaluative element as one of its components: After “niin on,” Sami explicates his agreement  by upgrading the assessment, ­further demonstrating his agreement with the evaluation. In the “niin on” response above, there is a change in tense: While Ville presented his evaluation in the past tense (“oli”/“were,” line 8), Sami uses the present tense and thereby highlights the current relevance of the assessment. In this sequential context, this choice of

Alternative responses to assessments

293

tense implies that there is still a possibility of returning to the old field. In the subsequent talk, this option is acknowledged by Ville (line 12). The use of the unmodified, strong agreement presented by “niin on” is not restricted to affiliative contexts. Our next example comes from a situation in which there is disagreement  between two speakers. A group of men are discussing what a boss ought to do with an employee who is found drunk at the worksite. What is at issue here is whether drunkenness is a sufficient reason for being fired or whether some additional troubles are required, such as a fight, to warrant firing. In lines 15–16, VM expresses his opinion that fighting is worse than using alcohol. To this, MI responds by saying that fighting should be seen as an aggravating circumstance. In this way he implies that drunkenness should be seen as the primary reason for being fired. The previous speaker VM receives this with a “niin on” response (line 19). (6)  (SKK/Sg003 A, 12) 01 V:

=Mut oliks (.) oliks se tää Kivine ku aiheutti sen but was.Q was.Q it this name who caused the =But was it (.) was it this Kivine who caused the

02

luunmurtuman tai [siis (.) tappe[lun.= bone.break or [prt fight fracture or [I mean (.) the[ fight.=

03 M: 04

[No ei: [well neg [Well no:

[tässä sitä ole:kyllä muuta [here it is in.fact other [it doesn’t say so here4 other

ku että. than that than

05 (1.0) 06 V: 07 M: 08 ?:

Sen yhteydes[sä vaan, its connect [ion.ine only Just in connection with it, [Nii nii että (.) [ol ryyppäämine ja [ prt prt that [was drinking and [Yes yes that (.) [((there)) was drinking and [O[W-

294

Marja-Leena Sorjonen and Auli Hakulinen

09 M:

[sitte, [then,

10 V:

[°Joo joo (.) mm°,= [° Yeah yeah (.) mm°,=

11 M:

=Mutta tuota noin ni joka tappauksessa ni, but prt prt prt any case.ine prt =But um er so in any case that,

12

(1.0)

13 M:

↑joko hänen ois kylymästi< (.) potkassa either he be.cond coldly should sack ↑either he should coldly (.) should sack this (.)

14

tämä (.) (vi[hollinen). this (.) (en[emy). 5

15 ?:

[(Tää vih[ollinen.) [(This enemy.)

16 V:

[>Minä katson ainaki siis< että ku [I look.1 at.least prt that when [>I think at least that is< that when

17

tulee tappelu ni se on .hh siis pahe- se on melkeen comes fight so it is therefore worse it is almost there is a fight then it is .hh in fact wor- it is almost

18

pahempi ku: (.) .= worse than alcohol’s use worse than (.) .=

19 R:

[Nii;= [Yeah;=

20 M: → =Se on ras:kauttava asian[. ] it is aggravating circumstance ] =It is an agg:ravating circum[.] 21 V: →

[Niin on. ] [so is ]

22 →

°se on:° (.) sillon se on (.) pahoinpitely. it is then it is assault °it is:° (.) then it is (.) an assault.

23 M:

Mm-m,

24 V:

Ja (.) vaikka kats- ajattelee asiaa siis ihan tollaviisii and though Ø look thinks the.matter prt quite that.way And (.) even if one look- thinks about the matter just like

Alternative responses to assessments

295

25 noin vaan ö:.mt ei mitenkää hh (.) juridisesti niinku sanotaa like.that neg any.way legally as say.pass that er .tch not in any way hh (.) legally as they say 26

[hianost. .hh [refinely. hh

27 M:

[Mm-m,

28 V:

vaan >tota noin< ihan noin kaverien kesken et jos tulee (.) but prt prt quite prt mates.gen among that if comes but >sort of< just among mates that if there is (.)

29

ryypätessä tappelu nin minusta sillon ni (.) saa kyl drinking.ine fight so I.ela then prt Ø may surely a fight amidst drinking so in my opinion then (.) one can

30

antaa lem:put. give sack give the sack.

31 R:

Niin kyllä °se on°, so sure it is So sure °it is°,

Ei siis pelkkä alkoholinkäyttö nii sitä mä en sano 32 V: neg therefore mere alcohol.use prt that I neg.1 say I mean not just the use of alcohol so I am not saying 33

et saa suoranaisesti [sitä voi varottaa ensin. ] that Ø may simply [prt Ø can warn first ] that one may downright [one can give a warning first.]

From the strong agreement  with “niin on” (line 21), V moves to present his own interpretation of the situation. Note that at the beginning of line 22 what looks like an instance of the format “se on” is in fact an instance of self-repair: Speaker V abandons his utterance beginning and restarts it with an adverb “sillon”/“then.” He defines a fight in legal terms, as his co-participant had also done, but he uses the technical term “pahoinpitely”/“assault.” In this way he ignores the implication that the primary reason for getting fired is being drunk, conveyed in the first assessment, thereby holding to his own stance  expressed in lines 16–18. In the subsequent talk, he further explicates his stance  (see lines 24–33). In this example the recipient thus uses “niin on” strategically: He asserts agreement  as a way to get back to enforce his own stance .

296

Marja-Leena Sorjonen and Auli Hakulinen

In the next case, the “niin on” response is also followed by a disaffiliative utterance by the same speaker (see Sorjonen  2001: 174–176 for an analysis of the same response). Iia has called to ask whether Mia would join her to see a play in a summer theatre. Instead of responding to the inquiry, Mia has delayed the response by asking about the price of the tickets. Iia answers by saying that she does not know the price and then goes on to tell Mia what she knows about the availability of tickets. Her turn in line 1 is part of that telling: There are more tickets available for weekdays than for weekends. (7)  (SKK/Sg08 A01, 2) 01 I:

J’ >mut sit arki-iltoina on ihan (.) tost and but then weekday-evenings.ess is right that.ela An’ >but then for weekday evenings there are still tickets

02

heinäkuun:viidennest päiväst lähtien ni °lippuja.° july.gen fifth.ela day.ela starting prt tickets right (.) from July:fifth onwards.

03

(1.1)

04 M: Onkohan se hyvä.=Eihän sitä tiedä ennen kun näkee. is.Q.cli it good neg.cli it Ø knows before Ø sees I wonder if it’s good.=One doesn’t know before seeing. 05 I:

Mm:.

06 M: → M’t kyl kuus tuntii on ai:ka kau°hee°. but indeed six hours is quite awful B’t six hours is really qui:te aw°ful°. 07 I: → Niin on; Siin on kaks väl°iaikaa°. so is it.ess is two intermissions Niin on; It has two inter°missions°. 08

(1.2)

09 M:

°Jeoo°. °Yeah°

10

(0.8)

Mut kai se on semmone et se pitäs 11 M: but maybe it is that.kind that it should Ø But I guess it’s the kind of thing that one should 12 13 I:

nähä [°kuitenki°. see [°anyway° [°Mm:.°

Alternative responses to assessments

297

14 (1.8) 15 I: °nJoo.° .hh et jos mä nyt ostan johonki< (0.6) päivään prt so if I prt buy.1 some.ill day.ill °Yeah.° .hh so if I buy tickets for some< (0.6) day, 16 liput, ni ostanks mä sulle °kanssa°. tickets so buy.1.Q I you.all also shall I buy for you °too°. 17 M: ↑Nii sä tarttet sillo varmaan mun opiske- aijaa mut ei prt you need then certainly my student prt but neg ↑Right you probably need then my stu- oh but there 18 ei siell_oo, onks siellä mitää opiskelija neg there is is.Q there any student isn’t, is there any student ((discount))

In line 6, Mia produces a negative assessment of the length of the play. In this context, the assessment implies a possibly untoward stance  toward going to the theatre. The recipient responds with “niin on” by which she asserts agreement  with the co-participant (line 7). She thus focuses on the general stance -taking character of her co-participant’s utterance. After that, she addresses the implication that the length of the play might discourage them from seeing it. She does this by indicating that the play will include two ­“intermissions.” This continuation thus disaffiliates with the ­co-participant in terms of the possible consequences of her assessment. In lines 11–12, following trouble-foreshadowing silences (lines 8 and 10) and her minimal acknowledgment, the first assessment speaker produces a turn with which she implies the possibility of an acceptance of the invitation . In this case, a participant asserts agreement  with a “niin on” response in the midst of a pending response by her co-participant who had produced evading and rejection-implicative responses to her invitation . With the “niin on” response she affiliates with her co-participant’s assessment as a general one of the state of affairs. She then, however, moves to express disaffiliation  with the kinds of consequences the assessment could be heard to have for the ongoing larger activity. The “niin on” response can also be used as a means to treat an utterance as an assessment although it was not primarily designed as one. In the following extract the recipient responds with “niin on” to an utterance that is hearable as a piece of advice . Tiina

298

Marja-Leena Sorjonen and Auli Hakulinen

is talking over the phone with Ripa and Anni. Both Tiina with her husband and Ripa with Anni are renovating their homes. As part of promising Ripa and Anni that she will visit them, Tiina complain s about the hard work she and her husband have been doing and mentions their intention to take some time off (lines 3–5, 8). Ripa responds with an utterance that is constructed as a positive evaluation of taking a vacation but whose primary function, due to its sequential position and the attention getting particle “kuule” (“listen,” see Hakulinen et al. 2003), is to give advice . The utterance contains the zero person  that leaves it unspecified for whom it is good to take a vacation. This utterance gets a “niin on” response by Tiina (line 11). (8)  (SKK/Sg06 A01 20) 01 T: .hh Kyl me sit tullaan kuha me täst vähä sure we then come when we here.ela a.bit .hh We’ll definitely come when we have ((sorted out)) a bit here 02 A:

[Joo::. [Yea::h.

03 T:

[Me ollaan nyt ajate]ltu et sit ku Kalle lopettaa [we have now thought that then when 1nameM finishes [We have been thinking that after Kalle finishes

04

hommat ni sen jälkeen me otetaan ii:sisti et ei chores prt it.gen after we take easy prt neg his job then we are going to take it easy as we don’t have

05

[me nyt jakseta tällai [ku me ollaa nyt kak]s [we now have.energy this.way [as we have now two [the energy ((to work)) in this way “cos we have

06 A:

[Joo:. [Yea:.

07 R: 08 T:

] ]

[Joo:. [Yea:.

viik[koo oltu tuolla .hhh wee [ks been there have been there now for two weeks .hhh

09 R: →

] ] ] ] ]

[Ja sit on k- sit on kuule he:le]vetin hyvä [and then is listen then Ø is listen hell.gen good [And then it’s l- then it’s listen da:mn good

Alternative responses to assessments

299

10 → ottaa välillä lomaa?, take now.and.then vacation to take a break every now and then 11 T: → Niin o[n. so is 12 R:

[(-) siit [ju()(-) [(-) from it (-)

] ]

13 T: →

[Se on erittäin tärkeetä. ] [it is extremely important ] [It is extremely important. ]

14 A:

[(Nii. Sit siin on) [prt then there is [(Yea. Then there is)

] se et ] it that ] the fact that

15

vaik me ollaan tehty vaa viikko tätä although we have done just week this although we’ve been doing this for a week only

16

mut siis me ollaan nukahdettu joka ilta kello but prt we have fallen.asleep every night clock but I mean we have fallen asleep every night at

17

kymmenen ([)(-) ten ten o’clock ([) (-)

18 T:

] ] ]

[.mhh Joo ] sama meilläki et ku – – [ prt ] same we.ade.cli that when [.mhh Yeah] the same with us too that when – – -

By responding with “niin on,” Tiina disattends the advice  giving function of Ripa’s turn by making use of the alternate ­interpretation that the zero person  reference allows: She treats Ripa’s utterance as one that presented a generic assessment on what would be a good way of conducting one’s life and asserts agreement with it. In this way she transforms the ongoing activity from advice - giving to one in which equal partners assess a state of affairs. Subsequently (line 13), she explicates her stance  with a strong positive assessment. By using here the assessment term “tärkeetä”/“important,” she echoes the voice of lifestyle advice , enforcing her personal competence in the matter. In this case, with the “niin on” response and its continuation Tiina emphasizes the joint character of the activity and in so doing resists the advice  offered by Ripa.

300

Marja-Leena Sorjonen and Auli Hakulinen

Conclusion In this chapter we have discussed three alternative ways of constructing an agreeing  response to an assessment, and the interactional functions of these responses. All the responses we have analyzed are clausal, that is, they contain a finite verb. Two of these – the VS and SV formats – include the anaphoric  subject but leave out the evaluative term of the first assessment altogether. By contrast, the subjectless variant “niin on” with the anaphoric adverb “niin”/“so; in that way,” is composed of a predication only: The object of evaluation expressed with the subject-NP in the first assessment is in this format treated as fully shared. The “niin on” response, being a predicating expression, foregrounds the evaluation by having the adverb in the initial position. These two aspects make “niin on” a construction for expressing strong agreement. The three response types we have analyzed show that participants in interaction do not merely express agreement in terms of its strength. We have seen that the detailed structure of the response, including its word order  and constituents (subject or object pronoun or the adverb “niin”), conveys the way in which a “same evaluation ” should be understood in some particular sequential context. What seems to be at issue is the extent to which the participants approach the object of evaluation from the same point of view. Thus the “on se”/“is it” response carries an implication that the perspective of the response speaker differs from that of her co‑participant. This difference in perspective may be explicated in the subsequent talk. The subject-initial response “se on”/“it is,” by contrast, implies that the response speaker already held the stance  earlier and in so doing underscores the independence of the agreement . This is perhaps most clearly seen in cases where the “se on” speaker’s primary or independent access to the object of evaluation is evident in one way or another from the context. In these cases, the “se on” utterance may perform a double task: While claiming agreement with the prior speaker it also functions as a confirmation of the validity of the co-participant’s assessment due to its conveying the priority of access to the matter at hand. The response types that contain the subject pronoun (“on se”/“is it” and “se on”/“it is”) carry an implication that the response

Alternative responses to assessments

301

speaker qualifies the agreement: Either she implies a difference in the perspective (VS) or conveys that she already held the same stance  before the current interaction. In comparison, the subjectless response “niin on”/“so is” focuses on the sharedness of the current assessing activity: It merely asserts agreement. However, we have seen that this type of agreement can be put to various uses. It strikes us as remarkable that in our data, it often functions as a means of asserting agreement that is then deployed for moving to one’s own, opposing stance . Because the three response types are constructed as clauses they make available to the speaker grammatical  resources other than word order  and the presence or absence of the subject to further specify their point of view on the evaluation. For instance, we have shown cases in which the response speaker changes the tense of the verb or the number of the subject NP so as to modulate or qualify agreement . As we have shown above, a function similar to the one expressed in Finnish by word order  may have to be expressed through prosodic means in English (Stivers  2005; see p. 286 above). In addition, the meanings of the Finnish VS and SV responses resemble what Heritage  (2002) has described as “oh”-prefaced responses  to assessments in English. According to Heritage, “ oh”-prefacing is “a method persons use to index the independence of their access and/or judgment in relation to the state of affairs under evaluation” (2002: 204). This function of “oh” rests on its more general change of state semantics. Even though the issue of epistemic access can be and often is associated with the VS and SV responses, that is not the key issue with them – the difference in perspective (VS) or the independence of the assessment (SV) can get specified in the ensuing talk in several different ways. Furthermore, as compared to “oh,” the VS and SV responses, by being formatted as clauses, have a different range of sequential contexts where they can occur: as responses to assessments and as answers to questions but – unlike the English “oh” – not as responses to informings. The range of morpho-syntactic means, on the other hand, is not available in English given the role that word-order plays in syntactic processes; verb-initial formats are used for interrogatives and imperatives only. We have shown that a number of basic morpho-syntactic characteristics of Finnish are reflected in the organization of assessment

302

Marja-Leena Sorjonen and Auli Hakulinen

sequences. These include, most notably, the nonconfigurationality of Finnish word order , in other words the possibility of using word order to make discoursal and not grammatical  distinctions. In addition, there is the possibility of forming a sentence without a subject: with an anaphoric  zero, a response is tied to the utterance of the prior speaker. Finnish is not unique in terms of its morphosyntactic characteristics. However, languages that share some of the same typological  features may not deploy them for the same interactional purposes. For example, Japanese  (Hayano 2007) appears to have an assessment response format coming close to “niin on” but functioning differently both with respect to its structural relation to the prior assessment and with respect to its interactional functions. On the other hand, Estonian, a language closely related to Finnish, does not deploy the same types of response formats: The only verb initial response is (verb + particle) (Renate Pajusalu, personal communication).   Acknowledgments We would like to thank Jack Sidnell for most helpful comments on our article. An earlier version of this article was delivered at the Annual Meeting of Finnish Linguists , at the Annual Meeting of Finnish Conversation Analysts and at ICCA-06 in Helsinki, and we are grateful to participants in these events for comments. Appendix: Glossing symbols used ADE AG ALL CLI COND ELA ESS GEN ILL INE NEG PAR

adessive case (“on”) agentive morpheme allative case (“(on)to”) clitic conditional mood elative case (“from”) essive case (“at”; “as something”) genitive case illative case (“(in)to”) inessive case (“in”) negative auxiliary or particle partitive case

Alternative responses to assessments PASS PL POSS PPC PRPC PRT Q 1 2 1nameM S V X

303

passive plural possessive suffix past participle present participle particle question particle 1st person singular 2nd person singular male first name subject verb any complement

Notes 1 It is of course also possible to present a same evaluation  by repeating the entire prior assessment, for example, “Se on hieno”/“It is great.” These cases fall out of the scope of our present focus. 2 We will not provide an idiomatic version of the target lines as we do not know what the exact functional equivalents of English in these cases would be. 3 The pronoun se is used for both human and non-human referents in most of the colloquial varieties. 4 The problem that the men were to solve was presented to them in ­written form. 5 The speaker indeed refers to the problematic worker as an “enemy”.

10 Language-specific resources in repair and assessments   Jack Sidnell

Introduction  For humans, social action is, by its very nature, embedded in the rich semiotic structures of the life-world. Such structure not only provides the background against which particular actions are recognizable but also the raw materials out of which they are constructed in the first place. In talk-in-interaction, social action is built up out of the particular prosodic, lexical, and grammatical  resources of a given language and is thus necessarily endowed with a partially language-specific  character. In this chapter, I show that the use of “if” as a preface to next-turn repeats provides speakers of Caribbean English Creoles with apparently unique possibilities for social action with no clear analogue in other languages. In their basic, canonical use, “if”-prefaced repeats are used to initiate repair of a prior turn formatted as a “yes”/“no” question (YNQ). In this sequential context, the “if”-preface acknowledges the ongoing activity of questioning even while it suspends it in order to initiate repair. This use of “if”-prefacing appears to be a reflex of grammar in language varieties which, unlike many other varieties of English, do not use inversion to form YNQs. We can see this use of “if”-prefaced repeats then as the mobilization of local grammatical resources to solve a generic interactional problem: the problem of conveying that one heard a prior trouble-source turn to be a “yes”/“no” formatted question. This basic usage makes possible a number of other distinct practices quite independent of initiating repair. In the second section of the chapter I turn to consider two of these. First, I discuss the use of “if”-prefaced repeats as second assessments. I argue that, as second assessments, “if”-prefaced repeats treat a prior assessment as if it were a question and thus as

Language-specific resources

305

something said with some uncertainty. Second assessors use such “if”-prefaced repeats to not only agree with a prior assessment but also and simultaneously to claim epistemic  priority from second position (Heritage  and Raymond  2005; Stivers  2005). Second, I discuss the use of “if” as a stand-alone item. In the particular case I examine, stand-alone “if” apparently treats a prior turn as stating the obvious and thereby casts the emerging agreement  and alignment of the moment as an artifact of a general consensus. An examination of “if”-prefacing in these language varieties opens up a number of issues for comparative conversation analysis. First, it shows that differences in the way a distinction is coded can have significant, domino-like, effects which link together apparently unrelated practices in talk-in-interaction. Second, it illustrates the way in which practices of speaking that solve generic interactional problems are shaped in part by the specific character of locally available resources such as, for instance, language-specific  grammatical  resources. Other-intiated repair Other-initiated repair ( OIR) is a subset of practices which participants in conversation employ to address and potentially resolve problems of speaking, hearing or understanding (Schegloff et al.   1977). The following example presents an instance from American English . Here Jon and Guy are discussing where they might be able to play golf later in the afternoon. (1) NB 1.1. 01 J: Well I’m s:↑ure we c’get on et San Juan ↑Hi:lls ↑that’s 02 ni: ce course ah only played it ↑o:nce. 03 G: °Uh huh?° 04 (0.6) 05 G: a→ .hhh °↑It’s not↑ too bad,° 06 (0.4) 07 J: b→ Hu:h? 08 G: c→ ‘S not too ba:d, 09 (.) 10 J: No:.

In extract 1, repair is initiated (arrow B) in the turn directly following the trouble source (arrow A ) and the repair itself follows in the turn after that (arrow C). The repair as a whole constitutes

306

Jack Sidnell

a sequence inserted between the second assessment “It’s not too bad” at line 5 and the confirmation  of it at line 10. It is important to see that although repair is used here to fix an apparent problem of hearing – an orientation to which is revealed, in part, through the use of repetition in line 8 – it is implicated, at the same time, in the larger course of action being pursued. That is, at lines 1–2 Jon has offered up San Juan Hills as a place he and Guy might play golf and has recommended it on the basis that it is a “nice course” and that he has “only played it once.” Guy’s “It’s not too bad” responds specifically to the assessment “nice course,” refusing to endorse it with an equivalent or upgraded evaluation (e.g., “yeah, real nice,” etc.). It is within this context of emerging disalignment that Jon initiates repair. In doing so he provides Guy with an opportunity to revise, modify, or withdraw his nonaligning assessment. And notice that when Guy does not do this, Jon treats it as if it were a positive assessment agreeing  with “no:.” Jon might have agreed with “yeah” but this would have been to accept the negative assessment and the rejection of Jon’s proposal this embodies. The point is, then, that while repair is often used to fix problems of hearing, speaking, and understanding, it is not simply or solely a piece of the “systems engineering” as Goffman  (1976) might have suggested. Rather, when we look at particular instances we find that whatever else repair may be doing it is also a vehicle for social action. A number of comparative studies suggest that other-initiated  repair is, in terms of its organization, essentially generic and, in terms of its distribution, universal (see, e.g., Egbert 1996; Moerman  1977; Ochs 1984; Schegloff 1987a; Sidnell 2006; Wu 2006). At the same time, these studies indicate that there are also some important and striking differences across communities. This variation within a clearly identifiable domain makes repair an attractive topic  for comparative study as many of the chapters in this book demonstrate (for instance, Fox  et al., Egbert  et al., Wu).  Other-initiated repair  in two Caribbean communities The data for this study comes from long-term fieldwork in two Caribbean communities. Between 1994 and 1996, I spent twelve months in a rural Indo-Guyanese village of approximately 700 people that I refer to as Callander (see Sidnell  2005a for an ethnographic 

Language-specific resources

307

analysis of talk and knowledge in this village). I recorded about eighty-five hours of talk in the village, the bulk of which consists of ordinary conversation in people’s homes, at gathering places along the road and in the rumshop. Due to the character of the phenomenon examined in this chapter I also draw from a number of interviews I conducted with villagers. Between 1999 and 2004, I spent approximately seven months conducting fieldwork in Bequia  – the largest of the Grenadine islands with a population of about 5,000. Although this corpus includes a large number of audio recordings, for the present study I have drawn exclusively on a smaller corpus of approximately ten hours of video-recorded data. The video-recordings capture ordinary conversation in the yard of an extended family and in the homes of island residents. An examination of the recordings collected in these two communities reveals that the organization of other initiated-repair in the Caribbean data is remarkably similar to that in American English as described in Schegloff et al.  ’s well-known study (1977).1 Several different strands of evidence may be cited in support of this claim. First, the various formats used to initiate repair are almost ­identical. Thus, in the Caribbean data the major formats for initiation include open-class “ what?” and “huh?” as well as questionintoned repeats and class-specific question words such as “who?,” “where?,” and combinations thereof (e.g., “They well have who?”). Second, a robust prefer ence for self-correction is observable in that others typically await the first possible completion of a turn before initiating repair; indeed, as Schegloff et al.   (1977) observed for American English, others typically delay initiation even further allowing a slight gap to develop between the completion of the turn containing the trouble source  and the other-initiation of repair. Moreover, others typically only initiate repair, leaving it up to the speaker who produced the trouble source  to provide a correction. Third, other initiated repair is used in a remarkably similar way not only to fix troubles of speaking, hearing, and understanding but also to ­convey ­disagreement , disbelief, and other problems of interactional alignment .2 This suggests a base of generic principles organizing repair across differences of language, culture, and social arrangement (see Moerman  1977, Schegloff  1987a, 2006; Sidnell  2001 and 2007a). There are, however, some differences. Elsewhere I’ve discussed one

308

Jack Sidnell

such difference – a repair initiation format apparently specific to Bequia  that is particularly fitted to deal with problems encountered in the making of recognitional  reference to persons (Sidnell  2007a). In what follows, I discuss another difference having to do with the way in which a participant can, in the course of otherinitiating repair, convey that she heard that the trouble-source turn was a YNQ.  “If”-prefaced repeats in the initiation of repair The collection of other-initiated repairs includes a small number of instances in which a partial repeat or wh + partial repeat formatted initiation is prefaced by “if”. Pat’s turn at line 2 of Extract 2 is an instance. Here Pat and Benson are sitting side by side in the yard that adjoins Benson’s small house. It is a week after Carnival and Pat has stopped by for a visit with Benson’s neighbor. (2) #187_Q2 qt 51:50 01 B: yu biin hii fu kanival (.) Pat? were you here for Carnival Pat? 02

(.)

03 P: → if mi bin wa? if I was what? 04 B:

Bekwe fu kanival? Bequia Carnival?

05 P:

yeah: yeah

When Benson asks Pat in line 1 if she was in Bequia for Carnival, Pat responds by initiating repair with “if mi bin wa?”/“If I was what?’ – an “if”-prefaced repeat. The preframing “mi bin”/“I was” combined with the question word “wa”/“what” isolates “hii fu kanival”/“here for Carnival” as the trouble source  to be repaired. By prefacing the turn with “if,” Pat also shows that she heard the turn addressed to her as a YNQ. At line 4, when Benson repairs the reference, he preserves the status of his turn as a question by producing it with rising intonation. At line 5, Pat answers the question with “yeah.” The function of this “if”-preface then is to show that the one initiating repair heard that the turn containing the trouble source  was produced as a YNQ.

Language-specific resources

309

In most varieties of English, one standard way of forming YNQs is to invert the ordering of subject and auxiliary verb in simple declarative constructions (Quirk et al. 1985). So “You’re going for a nap” becomes “Are you going for a nap?”. Such inverted syntax can be preserved in repeats which initiate repair so that, for instance, repair is initiated with forms such as “Am I GOing?” or “Am I going for a what?”. The fragment given as Extract 3 provides an illustration from American English  conversation: (3) TC II(b): #28 6 P: D’yknow where Mister Bowdwin is. 7 (0.2) 8 B: Wha:t? 9 (·) 10 P: hhuh-hhuh-°hu-° [˙hhh 11 B: → [Do I know where who? 12 P: Leo is. 13 B: No. 14 P: Oh. Okay.

Here, by preserving the inverted syntax of the trouble source  turn (“Do you know . . . ?” becomes “Do I know . . . ?”), Bush shows that he heard the turn to be a YNQ. Notice then that Pyatt’s repair in line 12 not only fixes a problem of person reference by substitut­ing “Leo” for “Mr Bowdwin,” it also completes the question that Bush began in his turn at line 11. Notice moreover that Bush displays an orientation to the ongoing activity of questioning by providing an answer at line 13. In the creoles of the Caribbean there is no auxiliary-subject inversion in YNQs; indeed, there is no syntactic category  of auxiliary verb for such an inversion to operate upon (see Winford 1993). Instead, in these varieties, a turn’s status as a question is constituted through a range of features of design and context. In the following case, Naomi asks her nephew Ozan, who is looking at a newspaper, whether he read a story about “the Brewster boy.” (4) Q1.12.55–14.02 02 N: → Oozan yu riid ting bou i Bruusta bo::y?  Ozan did you read the thing about the Brewster boy? 03

(.)

04

hii kripl an ii ina jeel, he’s crippled and in jail

310 05 O:

Jack Sidnell oo gad, mi keer abou hii?= Oh God, do I care about him?

A number of factors intersect to help the turn at line 2 come off as a question. First, it is produced with slightly rising intonation and a sound stretch on “boy.” Second, this is a clear B-event  in the sense that whether Ozan has read the story is clearly something Ozan knows better than Naomi. And, third, at the completion of this unit, Naomi stops speaking momentarily thereby providing Ozan with an opportunity to respond. When he doesn’t, Naomi builds a next component on to the turn. In the following case (Extract 5), Emmanuel and Viv are explaining to Baga that the anticipated time of his arrival prevented Emmanuel from going to town earlier in the day. At line 7, Baga formulates the upshot of what has been said with a “so”-prefaced turn. In this case then the B-event  character of the turn is established by the immediately subsequent talk. Although there is no discernable rise in the intonation of the turn the recipients immediately recognize this as a question and respond accordingly. (5) Garden Scene qt. 2: 15 03 E: ai woz goin tong tu bo- wii [gain ai gaing I was going to town too bu- we’re going I’m going 04 V: [ya bika hii bin fu yea because he was to 05 go tong ai se wait bika de see di man go to town I said wait because they said the man was 06

koming tudee. coming today

07 B: → so hi kudn go tong bika wen mi comin So he couldn’t go to town because of when I was coming 08 E:

[yeah

09 V:

[yeah

A final case illustrates (Extract 6) another important resource in the design of questions. Here Cat remarks to Roger that the mother of a passerby is from Hamilton (their natal village). Winnie agrees  with this and subsequently elaborates, asking if Roger knows “Joanne the daughter of Dorothy Quashie.” As she says this, she adjusts her body so as to achieve mutual gaze  with Roger. This is

Language-specific resources

311

sustained throughout the production of the turn. As Stivers  and Rossano (2009 ) have recently shown, gaze like this is a resource for mobilizing response and here it appears to contribute to the constitution of this turn as a question. (6) OFB2.27.00–27.24 him mada biilongz tuu Hamilton. 01 C: his mother is from Hamilton. 02

(0.6)

03 W: ya yu noo ahm Kwi- Joo An  Ya you know uhm Qui- Joanne 04

fo Daretii Kwashii? the daughter of Dorothy Quashie?

05 C:

ya:. Daretii Kwashii. yeah. Dorothy Quashie?

06 R:

Oh::.

These examples illustrate some of the practices such as intonation and directed gaze  that speakers use in designing a turn to be a recognizable question. With respect to the argument being developed here, it can be noted that these features cannot be carried over to turns that other-initiate repair in such a way as to show that the trouble source -turn was heard to be a YNQ. In turns that other-initiate repair, intonation is typically deployed in the service of repair-relevant tasks – e.g., marking the position of a trouble source  within a repeat. The use of gaze  to convey that a question is being asked is wholly indexical  and thus similarly not transferable to turns that initiate repair. So these important resources for showing that a question is being asked are not available to the one initiating repair as resources for showing that the trouble-source turn was heard to contain a YNQ. Notice, moreover, that in these varieties a simple repeat of the trouble source  turn will not be grammatical ly marked as a question and will be potentially hearable as confirming what has been questioned. In the Caribbean creoles then there is no subject-auxiliary inversion in YNQs, and the intonation of the trouble-source  turn is not preserved in turns that other-initiate repair. “If”-prefacing fills the resultant gap, conveying that the recipient heard the trouble source  turn to be a question. This is the basic interactional environment

312

Jack Sidnell

for “if”-prefacing, and the practice appears to be related in a basic way to the grammar of questions in these varieties. Because they are organized as a series of questions and answers, interviews provide a rich source of data. The examples below are from interviews I conducted with villagers in Guyana. (7) Callander_Interview_T41a (13: 17) 01 J: arait om:(1.2) yuu eva kos, alright, uhm do you ever curse 02

(0.8)

03 G: → if mi eva kos? if I ever curse? 04 J:

o

mhmo=

05 G: =yes mii a kos Yes I curse (8) Callander_Interview_T42b (30:10) 01 J: uhm:(.) yuu eva kos an ting? Do you ever curse and that? 02

(0.2)

03 Sh: → if mii eva kos? if I ever curse? 04 J:

ye::s. yes

05 Sh:

ye:[s yes

06 J: [eh he he ha huh (9) Callander_Interview_ T42b (31:02) 01 J: alrait uh:m (.) yu eva hiir ma:n kos? Have you ever heard men cursing? 02

(0.4)

03 Sh: → if mi ev hiir man kos? if I ever heard men curse? 04 J:

yes= yes

05 Sh: =yea:h, yeah

In each case, the recipient of a YNQ responds initially with an “if”-prefaced repeat and thereby initiates repair of the question.

Language-specific resources

313

The “if”-prefaced repeat begins an insertion  sequence that temporarily suspends the activity of questioning underway. These examples suggest that creole “if” functions as an “interrogative particle” (Sadock and Zwicky 1985 ) that marks whatever it prefaces as a question. Further evidence for such an analysis is provided by instances of reported speech in which, instead of using a verb such as “ask,” creole speakers often use a “say” + “if” construction. In the following example (Extract 10), for instance, Seeta asks if my wife Allison is still planning to go to Georgetown the next day. When I confirm this, Brammie asks, “Shi noo tong?”/“Does she know her way around town?”3 With “mos bi”/“most be,” Seeta responds in a way that shows she is less than completely certain and is basing her answer on inference. The reported speech at line 8 is apparently addressed to me. (10) Guyana_T5a.pm3_29: 10 01 S: alison go tong tum[ara sti:l Is Allison going to town tomorrow still? 02 J:

[yea:h yeah

03

yeah. yeah.

04 B:

Shi noo tong? Does she know her way around town?

05

(0.2)

06 S:

mos bi Most be.

07

(0.6)

08 → hi see if shi noo tong He asked whether she knows her way around town. 09 J:

na relii bonot really bu-

I suggested earlier that “if”-prefacing was connected to the grammar of YNQs in a basic way; specifically, given that in these varieties there is no subject-auxiliary inversion in YNQs, a repairinitiating repeat cannot convey that its speaker heard the turn containing the trouble-source to be a YNQ. Reporting speech presents similar issues. If Seeta had reported this without the “if”-prefacing

314

Jack Sidnell

she might have been heard as reporting an assertion. She could perhaps have conveyed that what was being reported was a question by preserving the intonation of Brammie’s original utterance, but this would have been to invite inferences about what she was doing in reporting the talk in this way (mimicking for instance). “If”-prefacing then is a practice for marking a turn as a question in the absence of grammatical  inversion. As we’ve seen, it is used in just this way in both the other-initiation of repair and reported speech .4 Some of this has been noted before. Thus, Cassidy and Le Page’s Dictionary of Jamaican English (1980: 234) describes “if/ef” as an introductory interrogative word, “the abbreviation of some such phrase as I wonder (if, etc.), or Did you ask me (if, etc.).” They report two uses of “if”: “introducing a question” and “introducing the repetition  of a question which (having not been fully understood) is repeated to make sure that that was the question” (Cassidy and LePage 1980: 234). In fact, there is no reason to assume any process of abbreviation or reduction. As Sadock  and Zwicky  (1985: 183) write, “it is not infrequently the case that the formal indications of yes-no questions resemble those of the antecedent of conditional sentences (the clause in a conditional that is expressed in English with if).” They cite biblical Hebrew and German  as examples of languages that mark YNQs and antecedent clauses in conditional constructions in the same way. Having described the use of “if”-prefaced repeats in the otherinitiation of repair, I now turn to consider an apparently derivative use of “if”-prefacing in assessment sequences.  “If”-prefaced second assessments In their basic, canonical environment, “if”-prefaced repeats initiate repair of a prior question turn and, in so doing, begin an insertion sequence that breaks the contiguity of first and second parts of an adjacency pair (Schegloff 2007b). “ If”-prefaced repeats are also deployed in a quite different sequential context – specifically in response to assessments. In this environment, “if”-prefaced repeats are second assessments that agree  with a prior. Extract 11 provides a first illustration of the practice. Here Shanka and Kiki are sitting in their family yard with three young children

Language-specific resources

315

close by. At line 1, Kiki directs one these children  – Zaria – to move and, in close succession, to “pick up that thing” an object on the ground in front of her. She completes the turn with a question about the whereabouts of Zaria’s cousin, Roxanne. Zaria responds only to the final part of the turn – ­waving her arms around in what might be several pointing gesture s. Kiki then directs her attention to the slightly older girl Naksin. As Kiki produces this turn, Zaria runs away. This occasions Shanka’s evaluative assessment of Zaria at line 6 – “Wailnes Zaria a kom wid”/“Wildness Zaria comes with.” At line 7, Kiki responds with “if Zaria wail?”/“if Zaria is wild.” (11) #134_Q1 qt 25: 38 01 K: Zaria muv from de.=tek a ting.  Zaria move from there. Take that thing. 02 K: wapa Rakzan.  Where is Roxanne. 03 

(3.0) ((Zaria is waving arms in possible pointing gestures))

04 K: Naksin lii shii:. tek da- an ting- Jak ting.  Naksin leave her, take that and thing, Jack’s thing 05

(0.2) ((Zaria runs away))

06 S:

Wailnes Zaria a kom wid. Wildness Zaria comes with

06 K: → if Zaria wail? if Zaria is wild 07

(0.4)

08 Naksin kom. Naksin come

We can see then that this assessment sequence is occasioned by a complex set of visible behaviors and witnessable actions: the failure to comply with the directives, the flailing hand gesture s, the running away. We can also observe that although the “if”prefaced turn recycles parts of the earlier talk it is not a verbatim repeat of it. This kind of loose repetition  seems to be doing several things. First, the incorporation of earlier talk marks the “if”prefaced turn as operating on something that came before. Second, the looseness of the repetition suggests that something other than repair initiation is being done. And, third, the particular words

316

Jack Sidnell

reused formulate the “kernel” of something that someone else has said. Notice also that the “if”-prefaced repeat does not elicit any response from the ­recipients – indeed, this seems to close the assessment sequence and the talk turns to other matters. Consider next Extract 12. This begins with Donna calling to her nephew (who is off camera). Although he appears to respond, he does not comply with the request to “come here.” After Kitana beckons the same boy again, Benson turns to Donna and remarks, “he’s rude you know.” This initiates a string of assessments culminating in an “if”-prefaced repeat. (12) #139_Q1 qt 39: 25 01 D: Gusnel kom bai hee Gushnell come over here. 02 (Gushnell):

(for yu)

03 K:

( ) kom he. come here.

04 B:

hii ruud yunoo he’s rude you know o

05 D: ai noo hi ruud I know he’s rude 06

(1.2)

07 B:

riil ruud. real rude

08

(0.4)

09 E:

huu ruud? Who’s rude?

10 B:

da boi [de. That boy there.

11 D: →

[if hi ruud? if he’s rude?

Notice that the “if”-prefaced turn in line 11 once again closes this, quite extended, sequence of assessments. With it, Donna seems to have “the last word” on the matter and the talk turns to other concerns. It is worth noting that in both this and the previous example the assessment is the vehicle for a complain t about a nonparticipant, third party.

Language-specific resources

317

In the final example (Extract 13) of this practice to be considered here, the participants are talking about an upcoming party to be held at a local park. After Manuel notes that people are expected to come from many nearby islands for the event, Cheryl augurs the consequences for the park, saying “da pak a mash op”/“That park is going to be a mess.” This can be heard as a way of saying that the party will be particularly lively and thus as an assessment of the anticipated event. At line 4, Manuel remarks, “if da pak a mash op agen”/“if that park is going to be a mess?” Notice that this “if”-prefaced repeat does not elicit any immediate response from the participants. Rather, after a pause, Cheryl raises the possibility that one of the expected musicians may not show up for the event. Cheryl seems to be backing off her earlier positive assessment here – in saying that Poser might not come to the event she anticipates a possible problem. In overlap with this, Roxanne adds “Sunday is the queen show.” By noting another aspect of the event not so far considered, this is supportive of the positive assessment that Manuel seconded with his “if”-prefaced repeat. Notice also that while the “if”-prefaced turn comes close to a verbatim repeat of what has just been said in line 2, the speaker adds the word “again” and in this way marks it as something other than a word-for-word repeat. (13) #31_AC1 qt 33: 00 01 C: dee kyaan cheenj it. They can’t change it. 02

da pak a mash op.= That park is going to be a mess.

03

=biisaidz besides I

04 M: →

[mi

()

[i- if da pak a mash op agen. if that park is going to be a mess?

05

(0.6)

06 C:

if Pooza na [kom an piipol if Poser doesn’t come and people

07 R: 08 M:

()

[oan Sundee iz di Kwiin Shooo and Sunday if the Queen Show hou ma fu pee fu goo in=I tink iz How much is it to get in= I think it’s

318

Jack Sidnell

This instance differs from those considered above. Here the first assessment – Cheryl’s “da pak a mash op”/“That park is going to be a mess” – while hearable as an assessment, is not clearly designed by its speaker to be one. Rather, Manuel’s “if”-prefaced turn, somewhat creatively, locates in what Cheryl has said something that could be heard as an assessment and responds to it as such. And notice that in this case the participants continue to assess the event. While here again the “if”-prefaced turn does not elicit an immediate response and the talk that follows it is marked by a subtle topical shift, Manuel is not able to achieve “the last word” as Kiki and Donna were in the previous examples. This then looks like an instance in which the practice is stretched to deal with a slightly different turn and, as a result, is not completely successful. In this collection of cases, “if”-prefaced turns are used to do second assessments (Pomerantz 1984). It is notable that in all these cases, the one who produced the initial assessment does not respond to the “if”-prefaced turn. This suggests that the second assessors are not being heard as requesting clarification  or as ­indicating a problem of hearing. What, then, do these “if”-prefaced second assessments do? In their basic interactional environment, “if”-prefaced repeats initiate repair on a YNQ. In other words, “if”-prefaced repeats convey that their speaker has heard a previous turn containing the trouble-source to be a YNQ. In these last three examples, the practice appears to be used to treat a prior assessment as if it were a YNQ. With these “if”-prefaced second assessments, second assessors are doing agreement . One piece of evidence for this is seen in the fact that, like other agreements (and prefer red actions more generally), these turns are closing -implicative – they help to bring topics or sequences to a close. By contrast, disprefer red actions and disagreements  in particular tend to be sequence-­elaborative. So notice that in Extract 11 after the assessment sequence at lines 5–6 the talk turns to other matters (Kiki beckons Naxine). And, in Extract 12, a long string of assessments concludes with an “if”prefaced repeat at which point the participants again turn to other matters.

Language-specific resources

319

But these “if”-prefaced turns go beyond just agreeing . By treating the first assessment as if it were a YNQ, and thus as something said with some uncertainty, an “if”-prefaced second assessment suggests that a first assessment was not strong enough. There is then an epistemic  dimension to this practice. As a number of recent studies have demonstrated, even where participants agree  on some state of affairs, they not uncommonly negotiate the terms of that agreement. As Heritage  and Raymond  (2005) show, an important dimension of this negotiation involves epistemic  rights to assess a given state of affairs. These authors have argued that a first assessment carries an implied claim to epistemic  priority but that this can be managed through various practices of speaking; specifi­cally, there are practices for downgrading claims to epistemic priority in first position (such as tag questions) and, inversely, practices for upgrading claims in second position (such as “oh”-prefacing). “if”-prefaced second-position assessments work in part by treating a first assessment as a question and thus as downgraded relative to a declaratively formatted assertion. With this practice then the participants can be seen to orient to the participants’ differential epistemic  rights to assess. In Extract 13, for instance, Manuel is responding to what he hears as an assessment from his aunt. By virtue of their ages and the social networks in which they participate, Manuel may well feel he has greater rights to assess the party (moreover, it is Manuel who raised the topic  in the first place). Indeed, Manuel has just been telling the other participants about all the people coming for the party from other islands, thereby displaying relatively more extensive knowledge about at least this aspect of the event. 5 In Extract 11, the first assessment is produced by Shanka, who is Zaria’s cousin, and the second by Zaria’s aunt (Kiki), who is partially responsible for her. The assessment here is a complaint and thus the “if”-prefaced format of the second assessment may be selected to deal with a situation in which Kiki feels she needs to reassert her greater rights to assess the child (see Raymond  and Heritage  2006). And, finally, in Extract 12, the participants are again assessing a child in the vicinity. Here Benson is assessing Donna’s nephew. The design of the initial exchange is sensitive to Donna’s greater epistemic  rights here. Specifically, Benson designs his first assessment as a

320

Jack Sidnell

question, in this way inviting Donna to confirm it, which she does in a particularly explicit way (“I know he’s rude”). The sequence continues with Ezekiel initiating repair (“huu ruud?”/“Who is rude?”). After Benson repairs the reference with a demonstrative  referring expression, Donna responds to the initial assessment again, now with an “if”-prefaced turn. Here again, then, relative rights to assess are at issue: This is, after all, Donna’s nephew that Benson and Ezekiel (a family friend) are assessing. Like the modified repeats described by Stivers  (2005) and discussed briefly in the introduction to this volume, ­“if”-prefaced second assessments claim epistemic priority from second position. While these look like quite different practices on the surface, the underlying logic is actually quite similar. Stivers  shows that modified repeats work by confirming a turn that did not invite such confirmation . In this way, modified repeats impute to the first-position speaker an uncertainty that they did not express. By treating a prior assessment as if it were YNQ, “if”-prefaced repeats similarly impute uncertainty to the first ­assessment speaker. So the larger point here is that in this interactional environment, “if”-prefacing treats a prior assessment as if it were a question. “If”-prefacing takes its sense and import in this environment from its canonical use in the other-initiation of repair of a prior YNQ. Its use in second assessments is apparently derivative of a more basic use in initiating repair and this in turn appears to be a reflex of grammar in language varieties that do not use syntactic inversion to form YNQs. Stand-alone if as an interjection “If” is also used as a stand-alone item in a way that suggests a relation to the uses already described. Because it occurs unattached to any other turn components and in a particular sequential position, I’ve described this an interjection. The following case (Extract 14) is illustrative. Here, in a fragment I’ve discussed elsewhere (see Sidnell 2005a: Chapter 8), the participants are a group of men who form what is described in the ethnographic  literature as a “crew” (see Wilson 1973) – adult men who grew up together in the village and consider themselves peers or “age-mates.” They are drinking together at their local rumshop and have been talking about

Language-specific resources

321

deceased persons in their parents’ generation. This is an environment of intense epistemic  competition, and, here, relative rights to assess the person being talked about appear to be at issue. In line 1, John links a new character “Buddy’s father” to the discussion up to this point. This negative assessment (“fucker”) adumbrates a story about Buddy’s father but first elicits a round of strongly aligning  responses.6 (14) Callander_Rumshop_T66 (04.40) 01 John: >Soo bodi dadii, ee da f:uka noo fu po[nish waif.= Just like Buddy’s Father. That fucker loved to beat his wife 02 Jaj:

[da rait. That’s right

03 Rohan: =iif? If? 04 Jaj:

da ii kaal neem. That was his nickname

05 John:

tek a leedi . . . grabbed that lady . . . ((Story continues))

At line 2, Jaj agrees , saying “That’s right” and, moments later, “That was his call-name,” meaning, it appears, that this was such a well-known fact about “Buddy’s Father” (i.e. that he was a “wife-beater”) that people used it to name him. At the interstices of Jaj’s talk lies Rohan’s contribution: “iif?” I originally glossed this as “You’re telling me!” (Sidnell  2000: 79–80), and, in a certain sense, that gloss is perfectly adequate, even if it seems to get the illocutionary value exactly backwards. Notice then that in this context of emerging collective agreement “ if” is not treated as initiating repair. Rather than occupying first position in an insertion  sequence, this stand-alone “if” is structurally similar to the “if”-prefaced repeats in assessment sequences. Here, stand alone “if” does not elicit a response and, indeed, works in conjunction with the expression “da ii kaal neem”/“That was his nickname” to close down this initial assessment sequence and thus forward the talk toward the telling that the turn in line 1 projected. It is likely this use of “if” that the author of the Dictionary of Caribbean English Usage has in mind when he characterizes it as

322

Jack Sidnell

“expressing certainty or strong agreement” and noting that, used in this way, it is typically “uttered with high rising pitch and strongly stressed, sometimes lengthened, vowel” (Allsopp 1996: 303). He provides the following example from a popular Barbadian magazine (The Pelican): “Hah!” Daphne laughed. “You ain’t put he in he place . . . ?” “If?” Claire exclaimed. “Who you think I is? I had to ask he if he think he did talking to one o’ he pissy li’l friends!”

Conclusion Returning to my larger theme, I want to suggest then that the use of “if”-prefaced repeats as second assessments is derivative of and, as such, dependent upon a more basic use of “if”-prefaced repeats to initiate repair of YNQs and that this more basic use of ­“if”-prefaced repeats is, in turn, related to the grammar of YNQs in this language variety. This local instantiation of a generic interactional practice then takes its local character from the grammatical  features of the variety. Tracking “if”-prefaced repeats across their various contexts of occurrence reveals an apparently language-specific web of connections between quite distinct practices. Of course, similar though not identical practices can be observed in other languages. In French , for instance, turns that other-­initiate repair of YNQs may be prefaced by “si”/“if.” In the following case, for instance, N and O are next-door neighbors. At line 1, N compares her living room to O’s, saying that his is bigger. Notice that although the neighbors should have comparable if not equivalent epistemic  access to the objects of comparison, N conveys a degree of uncertainty by formulating her remark as “an impression.” When there is no uptake, N pursues a response with “non?” which clearly makes confirmation  a relevant next action. O then initiates repair of this YNQ with a turn prefaced by “si”/“if” – a repeat of the talk in line 2. (15) Origami (Ex. 2, from Maheux-Pelletier and Golato 2008 ) 01 N: j’ai l’impression que vot’ salon i have the impression that your (sing.) living room i have the impression that your living room

Language-specific resources 02

il est plus grand it (masc. sing.) is more big (masc. sing.) it is bigger

03

(1.5)

323

04 N: non? no? isn’t it? 05

(2.0)

06 O: s’il est plus grand? if it (masc. sing.) is more big (masc. sing)? if it’s bigger? 07 N: qu’ le nôtre ouais than the (masc. sing.) ours (masc. sing.) yeah than ours yeah

It is interesting to note that while YNQs in French  are frequently formed through morpho-syntactic inversion (or the periphrastic construction “Est-ce que . . .”), here the “si”-prefaced turn initiates repair of a question turn not formed in this way. It is not clear at this point how widespread this use of “si” in fact is or whether it can be deployed in other environments such as in response to assessments. In English, as discussed earlier, repair initiations that target YNQs may preserve the inverted syntax of the trouble source  turn and in this way convey that it was heard as a question. This of course obviates the need for a particle like “if.” And, indeed, it appears that treating a prior assessment as if it were a YNQ is not a practice available to speakers of English. While it is possible to initiate repair in response to assessments (e.g., A: “That was good.” B: “Good?”), this is a quite distinct practice from the one I have described in this chapter. Although, with a question-intoned repeat (“good?”), a second assessor may agree  with a first via an upgraded assessment, there is no sense in which this works by casting that prior assessment as a question and thus as something said with uncertainty. Practices like this in English do not do the same work as the “if”-prefaced responses to assessments that I’ve examined in this chapter. This goes to my argument that within the essentially generic organization of other-initiated  repair, language-specific 

324

Jack Sidnell

grammatical  patterns shape and constrain the relevant practices. Moreover, such language-specific patterns and the practices that they provide for may spill into other areas of talk-in-interaction and provide speakers with unique resources for dealing with a range of interactional contingencies .  Acknowledgments Versions of the analysis presented in this chapter were presented at the Meetings of the American Anthropological Association, 2005, Sociolinguistics  Symposium XVII, and the Center for Language Interaction and Culture at UCLA, January 23, 2008. For comments on those occasions, I’d like to thank Mary Bucholtz, Ignasi Clemente, Chuck and Candy Goodwin  , Kira Hall, John Haviland , Keith Murphy, John Heritage, Emanuel A.  Schegloff, and  Tanya Stivers . For providing me with examples and insight based on their native intuitions, I thank Joseph Farquarson, Peter Patrick, and Donald Winford. Notes 1 There are no distributional patterns available for American or other forms of English. However, based on several years of collection, I can attest to the greater prevalence of open-class  type initiators in adult talk in American, British , and Canadian English. 2 These points are substantiated in various publications including Sidnell  2000, 2005a, 2005b, and 2007a. 3 It is not clear from what is available to whom this question is addressed. 4 The use of “if” to mark what follows as a reported question is noted by Holm and Shilling in The Dictionary of Bahamian English. They include the example: “Mama say if you go?” which they gloss as “She is asking whether you went?” (Holm and Shilling 1982: 109). 5 That is, the size of the party. In other respects Manuel defers to the other participants. Consider, for instance, that, in the final line of this extract, he is asking the others how much it will cost to go in. This suggests that rights to assess are just that and may be tied to specific particulars of the assessable event and not others (i.e. having to do with the cost, the anticipated profits and so on).

Language-specific resources

325

6 Domestic violence – primarily violence perpetrated by men against women and children – is a serious problem in Guyana as in other societies. It is important to note that these men consider such behavior to be reprehensible and cause for complaint. This assessment is, in fact, embedded in the preface to a negatively valenced story about “Buddy’s Father.”

11  Implementing delayed actions Galina B. Bolden

Introduction This chapter describes a language-specific solution to a generic and likely universal  interactional issue – how to show that the current utterance is occasioned by and should be understood by reference to something other than the immediately preceding talk. While the default way of connecting an utterance to a prior one is by placing it immediately after the targeted turn (Sacks  1987, 1995; Sacks   et al. 1974), on occasion interlocutors find themselves in need of showing that their current turn is noncontiguous with what came just before. One common interactional task then is to connect the current utterance to some early talk, what Sacks refers to as “skipconnecting” (Sacks  1995: II, 349–351, 356–357). Using the methodology of conversation analysis (CA) to examine Russian  language conversations, this chapter focuses on one linguistic resource interlocutors can use to manage this interactional problem: the Russian discourse particle -to. An investigation of over sixty hours of recorded interactions between native Russian speakers demonstrates that this particle is deployed in order to index the delayed placement of the action implemented by the current turn-at-talk. The particle is typically placed after a word repeat that helps locate the target of the displaced action in prior talk. The chapter starts with a cross-linguistic overview of several currently documented solutions to the problem of contiguity breaks in talk-in-interaction. Turning attention to Russian,  I examine some contexts in which the particle -to is used, including delayed clarification requests and resumptions of previously closed or abandoned

Implementing delayed actions

327

courses of action. Further, I discuss whether the deployment of the particle relates to the topical organization of the unfolding talk or to the activities that are being accomplished through interaction. A number of interactional implications of marking a particular action as delayed are discussed, focusing on action implicativeness and topicality. Overall, the presented analysis contributes to our understanding of how actions are formed to reflect their positioning in ongoing interactional projects. Ways of dealing with contiguity  breaks While ordinarily each turn at talk is understood as emerging from the immediately preceding context and as part of the ongoing activity (Sacks et al. 1974), on occasion, this contiguity  may be broken. A break in contiguity  may involve an interruption of the ongoing activity with a “side sequence” (Jefferson  1972) or a parenthetical  comment, a misplaced introduction of a new interactional project (e.g., Schegloff  and Sacks  1973), or a need to return to or redo some early part of the activity (e.g., Sacks  1995, vol. II). In these contexts, special techniques are used to convey to the interlocutors how the current turn is to be understood in relation to what has occurred prior to it. An overview of the existing literature suggests that discourse markers, prosodic features, and lexical tie-ins are among the most common devices that do this job crosslinguistically.1 Discourse markers conveying to the addressees that an utterance should not be understood as emerging from an immediately preceding context are commonly referred to as misplacement or disjunction markers. English misplacement markers include “by the way” (Schegloff  and Sacks  1973), “hey” (Levinson  1983: 313–315), “listen” (Jefferson  1972: 319; Sidnell  2007b), “look” (Sidnell  2007b), “now” (Aijmer  2002), “okay” (Beach  1993; Filipi and Wales 2003), and “oh” (Bolden  2006; Jefferson 1978), among others. Additionally, several discourse markers can be used specifically to convey the turn’s affinity to some earlier talk:2 such as, English “anyway” (Ferrara 1997; Lenk 1995, 1998; Owen 1985; Takahara 1998), “so” (Bolden 2005, 2009; Howe 1991: 90–93; Jefferson , 1972: 319), “and” (Heritage  and Sorjonen  1994; Local  2004), and “but” (Park 1997). A large array of pragmatic markers

328

Galina B. Bolden

that fulfill these functions has been identified in other languages as well (e.g., Mazeland  and Huiskes  2001; Mori  1999; Park 1997, 1999; Sadler 2006). There is significant cross-linguistic variation in the placement of these pragmatic markers within the turn (or, more accurately, the “turn constructional unit,” see Sacks  et al. 1974). In English (and a number of other languages), misplacement is typically indicated at the beginning of the turn, which allows for an early projection of the turn’s role within the ongoing activity (Heritage 2002; Schegloff  1987b; Sidnell 2007b). Some other languages (such as Japanese and Korean ), however, employ turn-medial and turn-­fi nal markers as well. This late placement allows for recasting of the turn’s function in the course of the turn’s construction (cf. Hayashi 2004). Russian  pragmatic markers may occur in a variety of turn positions, with the delay marker -to always a post-­positioned particle (as discussed below) occurring either turn-­medially or turn-finally. Aside from discourse markers, intonation appears to play a role in marking the turn’s place in the ongoing activity. For instance, variations in pitch, loudness, and tempo as well as a number of articulatory features may be used to show that a new course of action is being initiated (Couper-Kuhlen  2004; Levinson  1983: 313–315) or that some earlier line of talk is being continued (Local  2004). Additionally, lexical tie-ins, such as repetition or recycling of prior utterances, are robust ways of showing a connection between a particular turn and some earlier talk. Repetition may be used as a way of restarting a course of action (e.g., Jefferson 1972; Sacks 1995) as well as resuming a line of talk after an interruption (Jefferson 1972) or after a parenthetical insert (Wong 2000). All of these cues – discourse connectives, prosody , and lexical tie-ins – may be deployed in combination to clearly demarcate the structure of the unfolding talk (e.g., Local  2004). The Russian  discourse particle -to This chapter describes one way in which delayed actions may be implemented in Russian talk-in-interaction: via the particle -to (pronounced as ta/[tә]). This particle has been the focus of considerable attention from Russian-language scholars due to its somewhat puzzling properties. One of the very few post-positioned particles in

Implementing delayed actions

329

Russian, it can be attached, it would appear, to almost any word.3 Prior studies have, almost exclusively, examined the particle from an information-processing perspective and focused on its cognitive rather than interactional functions.4 In these studies, the particle has been described as a marker of emphasis, contrast or theme. Thus, it has been found that the particle can be used to stress the importance of a particular word, phrase or statement (Rathmayr 1989; Vasilyeva 1972), to indicate presence of contrasting elements in discourse (Bonnot 1987; McCoy 2001; Vasilyeva 1972), or, alternatively, to mark old or known information or referent (i.e. “theme”) (Bitextin 1994; Bonnot 1987, 1990; McCoy 2001; Rathmayr 1989; Vasilyeva 1972). The last two descriptors have been argued to relate to the particle’s etymological roots in the Russian indexical pronoun tot “that” (Bonnot 1987, 1990; McCoy 2001; Vasilyeva 1972). While this research has significantly advanced our understanding of the particle’s functions, several factors have limited its scope. The researchers have almost exclusively relied on literary or invented examples in their analysis, and when natural speech was used, sentences (or very short segments) were usually examined in isolation from their interactional context. As a result, little could be concluded about the interactional functions of the particle aside from the general descriptors mentioned above. Specifically, even if we know that -to marks old information (or theme), the question remains why, in a particular context, the speaker would choose to mark something as “old information” given that such marking is optional (and rarely used). What interactional ends does this marking achieve? What are the constraints on the particle’s use? These questions can only be answered when interlocutors’ own orientations and understandings of the unfolding interaction are examined, and for that actual recorded talk-in-interaction is needed. My research into this particle’s use in casual conversation has suggested that it may be deployed strategically to achieve a variety of interactional ends that go beyond information-processing issues, such as to show the speaker’s concern for or interest in the addressee (Bolden  2003, 2005, 2008) or to propose a particular, interactionally driven, understanding of the relationship between the current utterance and prior talk (Bolden 2005). In this chapter, I focus on the discourse connective function of the particle (i.e. its

330

Galina B. Bolden

role in implementing delayed actions) and aim to specify the contexts of its deployment (and nondeployment). Comparison of the environments where the particle is used with those where it could have been used but wasn’t allows for a clearer characterization of the particle’s functions. Contexts of -to use The Russian  particle -to can be deployed in a variety of environments. Of particular interest here are utterances that accomplish sequence-initial actions (such as questions, requests  or invitation s) that require the addressee to respond in a particular way (with an answer, acceptance or rejection, for example). In CA, these are known as first-pair parts of adjacency pair  sequences (Sacks  et al. 1974). As we will see, when the particle -to is deployed in such sequence-initiating turns, it suggests that the action implemented by the turn is, in some way, delayed. In other words, -to-marking indicates that the action launched by the current turn constructional unit is not, in the first instance, connected to the action embodied in the immediately preceding utterance, but to an earlier one. The location of the proper target of the current turn is typically indicated via a word repeat to which the particle attaches. The distance between the -to-marked utterance and the course of action to which it returns (and which it advances) varies. On one end of the spectrum are -to-marked first pair parts that are only minimally removed from the utterance they return to (for example, less than one turn constructional unit away from the target). On the other end of the spectrum are -to-marked turns that return to activities that are several sequences or topics away. 5 Several of these possibilities are discussed below. Delayed clarification requests  in storytelling  episodes One common environment for delayed actions is in storytellings. A story recipient may request a clarification of a prior part of the telling after the narrator has shifted to a new episode in the telling. These delayed clarification  requests , formatted as interrogatives, typically contain a repeat of some element of the targeted part of the telling followed by the particle -to, thereby locating the

Implementing delayed actions

331

t­ arget of the clarification request. In this environment, the return to the prior narrative episode is typically short-lived: As soon as the question is addressed, the narrator returns to the topic  that was suspended. In Extract 1, a -to-marked utterance (at line 13) functions as a clarification question that invites the narrator to explain something about an earlier part of the telling. After the question is answered, the narrator reintroduces the topic  that was temporarily abandoned. (See Appendix for Russian -specific transcription conventions.) (1)  Music Hall (from Kitaigorodskaia and Rosanova 1994).6 Tamara, a former ballerina, is apparently leafing through a photo-album and talking about the shows she performed in. Roza is an eight-year-old child. Two other people are also present. (6:30) 01 R:

A eta shto t^o?zhe spektakl’/ PRT that what also play This is also a play?

02 G:

Da/= yes

03 T:

=Da: >detachka/=eta b- ((squeek)) etat< nazyvalsja yes child this this called Yes child/ this- this one was called

04

“Leningr^ad pad dazhdë:m.”/ Leningrad under rain “Leningrad in the rain”

05 T: .h >Ty panima,eshi prosta lilsja dosch./ and simply poured rain and it was simply pouring/

12

=My< ta byli [vse v klewe PRT were all in pla(stic) We were all wearing pla(stic)

13 R: →

[A kak zhe dosch ta tak/ PRT how PRT rain PRT so How did they make rain?

14 T:

↑A vot tak vot/ vada./ PRT PRT so PRT water Like this/ water

15

(0.5)

16 T:

Tak puskali ↓vod[u./ so let-out water They poured water

17 (R):

[(Mm)

18 T:

A my: byli vse v kle↓ë:nke./ PRT we were all in plastic-cloth We were all wearing plastic cloth

19 T:

.h I dvizh^enija on takie d- em-pastavil,/= and movements he such choreographed And he choreographed movements that

Focusing on line 13, we can see that the particle -to is used in a turn of talk that functions as a clarification  request, asking the recipient to expand on an earlier part of the narrative. More precisely, the question targets the rather startling announcement  at line 8 (that there was “nastajaschij dosch”/“real rain” on the stage of the theater), which is then further upgraded at line 11 (“prosta lilsja dosch”/“it was pouring”) when no recipient uptake is forthcoming (line 9). Note that Roza’s question is delayed relative to the target utterances and is placed after Tamara has initiated a shift to a new episode (about what the actors were wearing – at line 12). The particle is attached to the noun “dosch”/“rain” – the repeated word from the targeted turn in the earlier part of the narrative  – and serves to mark the current utterance as delayed

Implementing delayed actions

333

relative to where it should have been more appropriately launched, i.e. immediately following its target. After Tamara responds to the question (lines 14–16), she reintroduces the issue (line 18) that was interrupted by the -to-marked question. While in the above case, the -to-marked turn implements an action that was only slightly delayed vis-à-vis its target, the particle -to may operate over larger distances as well. For example, in Extract 2, the -to-marked clarification request is several turn constructional units removed from the trouble source . (2)  Hunting (from Kitaigorodskaia and Rosanova 1994). Alex is telling a long story about his hunting trip to several family members. He has just told the others about how he spent the night in the forest. (5:25) 01 A: I patom pashël,/ vrode ja prasnulsja,/ and then went perhaps I woke-up And then I went/ perhaps I woke up/ 02 rana gluxari zapeli/ eschë gdeta nachala early woodgrouses sang only somewhere early  Woodgrouses started singing early/ only after 03

tret’eva/ uzhe uslyshal slyshu sch:ëlk/ three already heard hear snap two o’clock I already heard hear a snap

04

(1.0)

05 A: °Duma° “Ux ty gospadi/ ne utra zh eschë.”/ think wow you god not morning PRT yet I’m thinking “Wow gosh/ it’s not morning yet” 06

(0.2)

07 A: .h Ja eschë dumal minut tricat’ sorak u menja est’/ I still thought minutes thirty forty with me is I thought I still had thirty forty minutes 08 Slyshu schelcho,k/ i patom tishina./ hear snap and then silence I hear a snap/ and then silence 09

(0.2)

10 A: °Dumaju nu i° (.) prisni,las’/ shto, li/ think PRT and dreamed that PRT I’m thinking maybe I dreamed it up 11

(0.2)

334

Galina B. Bolden

12 A:

Nu skarej kastër zatushi,l/ vrode nedaleko,/ PRT faster fire extinguished perhaps not-far Right away I put the fire out, it seems close by

13

(0.5)

14

tk

A:

15

(0.2)

↑Stal padxadit’/ 16 A: started come I started to come closer 17

(0.5)

18 B: →

A kto schëlknul ta?/ PRT who snapped PRT Who made the snap?

19 A:

[Gluxa:r’/ woodgrouse A woodgrouse/

[Vot tak|oj ] ((two finger snaps)) PRT such Like this ((two finger snaps))

[(A:?)/ Ani zhe [sch[ëlkajut/ 20 V: they PRT snap They do make snaps 21 B: 22

[A:/ da/ oh yes

|Hm-hm/]

(0.5)

Vot takoj zvuk/ ((finger snap)) 23 A: PRT such sound It sounds like this ((finger snap)) 24

(1.0)

25 A:

I vsë/ i tishina/ and all and silence And that’s it/ Silence

26

(1.5)

27 A:

Nu vo:t/ PRT PRT

28

(1.0)

Ja: padxadit’ stal vizhu slyshu vrode on zape:,l/ 29 A: I come began see hear perhaps he sang I started to come closer see hear it seems like it started to sing

Implementing delayed actions

335

Bella’s question in line 18 requests  clarification  of an earlier part of the story, specifically Alex’s mention of a snap he heard in the forest (in lines 3 and 8). The request contains the particle -to indicating its delayed position. The placement of the particle after the repeat of the problematic word “snap” (in its verb rather than nominal form) serves to locate the trouble source in the prior talk. Note that the recipients had multiple opportunities to request clarification  (for example in lines 4, 9, 11, 13, and 15), but apparently waited for the teller to do so himself.7 After the shift to a new part of the telling has been accomplished (at line 16), the ­opportunity for the teller to explain himself passes, and one of the story recipients prompts the clarification  , now from a late position. In a way similar to Extract 1, the -to-marked question suspends the narrative progression to a new part of telling, but after the ­question is answered (lines 19–25), the narrator returns to the point in the story where it was interdicted by the clarification  request (line 29). To summarize, -to-marked questions may be deployed to request a clarification  of an earlier part of the ongoing telling. This prevents, for the time being, the teller’s movement to a new episode. Such -to-marked first-pair part actions thus interrupt the progress of the ongoing line of talk. Resuming a prior course of action Aside from clarification requests , other actions may receive -tomarking when a speaker attempts to resume or reopen a course of action that has been closed or abandoned. In Extract 3, for example, a -to-marked question is used to return to an earlier request (to get the intended call recipient to the phone) after several activity shifts. (3) NG1: call 12. Mura is talking to her teenaged grandson. (0:50) 01 MUR: Kto galodnyj/ Rafi galodnyj/ who hungry NAME hungry Who is hungry/ Rafi ((dog)) is hungry 02

(.)

03 MUR:

[.h

336 04 IL:

Galina B. Bolden [I papa galodnyj/=Vs[e galodnye/ and daddy hungry all hungry Daddy is hungry too/ Everybody is hungry

05 MUR:

[I-

06 MUR:

I vse, galodnye/=A mama gde,/ and all hungry PRT mama where Everybody is hungry?/ Where is mama

07

(0.5)

08 MUR:

.hh[h

09 IL: [Mama tut sidit/Smotrit na (karti-) na nas mama here sits looks at pictu- at us Mama is sitting here/looking at the pictu- at us 10

paka my galodnye/ while we hungry while we are hungry

11 MUR:

.hhh hah-heh-heh

12

.h[hh

13 IL: [Dat’ tebe [ma,mu/ give you mama You want to talk to mama? 14 MUR:

[I ty to?zhe galodnyj/da,/ and you also hungry yes You are hungry too right?

Ne:/ ja uzhe kushaju/ 15 IL: not I already eating No I am eating already ((continue about dog’s food, then some pictures, joking)) (3:20) Ty sl^u?shaesh menja/ ili net/h 16 MUR: you listen me or not Are you listening to me or not 17 IL:

Da da/ [slushaju/ yes yes listen Yes yes I am listening

18 MUR: [Ah-hah ah-gah,/ 19 MUR: .HHH Da/ yes 20

(.)

Implementing delayed actions 21 MUR:

N^u ladna/ PRT okay Okay

22

(1.0)

337

23 MU: → tk! Nu a mama ta vabshe e-m-n- (0.2) .h PRT PRT mama PRT generally So can I ask mama 24

mo – ,zhna- mo – ?zhna priglasit’ k telefonu/ can can invite to phone to the phone

25

(2.5)

Na Rus’/ Ty xochesh le?vaju ili pravuju/ 26 IL: take NAME you want left or right Here you go Rus’/ You want the left or the right one 27

(1.8)

28 IL?:

Heh-heh-heh-heh ((or off the phone))

Ja xachu ma:mu/ 29 MUR: I want mama 30

(1.0)

31 IL:

N(h)u i shto/ xatet’ ne vredna/ PRT and what want not hurtful So what/ You can want all you want

32 MUR:

Huh-huh-huh-huh-huh-huh-huh-

In line 6, Mura issues a pre-request (Schegloff  2006) to get Ilya’s mother on the phone. Ilya indicates his mother’s availability, thus, giving a “go-ahead” for the request that is now due from Mura (line 9). In the second-turn constructional unit (lines 9–10), however, Ilya starts joking about what is happening at home. Instead of launching the projected request, Mura laughs, and Ilya offers  to pass the phone to his mother (line 13). Rather than replying to the offer, Mura continues to joke with Ilya (line 14), significantly expanding the conversation (omitted from the transcript). Several minutes later, Mura resurrects the request (lines 23–24). Her request is now formulated as a -to-marked interrogative (with -to attached to the repeat of “mama”). The deployment of the particle indicates its connection not to the immediately preceding talk but to a prior interactional project that was abandoned prior to completion.

338

Galina B. Bolden

In the above segment, -to-marking is deployed to reintroduce a pending course of action, but it may also be used to return to an interactional project that could have been considered closed. An example of this can be found in Extract 4 where one party (Zhenya) is questioning the addressee’s purchase of a new car, using a -tomarked utterance to reopen her criticism of the purchase. (4) RP call 14. Dina is on her way out of the house when Zhenya calls. 01 ZH: [Nu |ladna/ PRT okay Okay then 02 DI: [.h |Zhen’ka/ [izvini/ NAME excuse Zhenja I am sorry 03 ZH:

[Ty savsem ischezla/ you completely disappeared You’ve totally disappeared

04 DI:

.h da ty znaesh chëta vchera, mashinu kupili/= PRT you know something yesterday car bought You know yesterday we bought a car

05 ZH:

=D[a/ ja zna]ju yes I know

06 DI:   [Celyj den’-] (.) doma ne byla/=A sevodnja entire day home not was PRT today my idëm we go We weren’t home the entire day/ and today we are going 07

.hh eh- abmyvat’ Mishkinu kupch(h)uju(h)/ wash NAME deed-of-purchase to celebrate Misha’s deed of purchase

08 ZH:

Da,/ da/ [da/[( ) yes yes yes

09 DI: [.h [Vot/ubegaem/ ja praspala/ ja usnula/ PRT running I overslept I fell-asleep We are running out/ I overslept/ I fell asleep 10

(.)

11 ZH:

[(Kakuju-) what

Implementing delayed actions

339

12 DI: [Ja utram chëta ochen’ rana I morning something very early For some reason in the morning very early I13 ZH:

Kakuju mashinu vy kupili/ Kamu/ what car you bought whom What car did you buy/ for whom

14 DI:

Mne:/ Ve:n/ me van For me/ A van

15 ZH:

A/=U tebja zh byla [novaja/ vrode by/ oh with you PRT was new seems PRT Oh You had a new car I believe

16 DI:

[t. .h

Nu net/ nu u menja zhe vot eta Santra Nisan/ 17 DI: PRT no PRT with me PRT PRT that Centra Nissan Well no/ I had a Centra Nissan/ 18

Malen’kaja/ Belaya/ small white A small one/ A white one

Nu ana [zh nova[ja/ 19 ZH: PRT it PRT new It was new [.hh [↑Nu my eë atdali↓ Mi:shke/ 20 DI: PRT we it gave NAME We gave it to Misha 21

(.)

22 ZH:

[°A: oh

23 DI:

[A on svaju pradal/ PRT he his sold And he sold his

24

i sdelal daun pejment pod nashu mashinu/ and  made down payment for our car and  made a down payment for our car

a my emu raznicu zaplatili/