Studies in Interactional Linguistics (Studies in Discourse & Grammar)

Studies in Interactional Linguistics Studies in Discourse and Grammar Studies in Discourse and Grammar is a monograph...

Author: Sandra A. Thompson | Margret Selting | Elizabeth Couper-Kuhlen

74 downloads 1646 Views 2MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Studies in Interactional Linguistics

Studies in Discourse and Grammar Studies in Discourse and Grammar is a monograph series providing a forum for research on grammar as it emerges from and is accounted for by discourse contexts. The assumption underlying the series is that corpora reflecting language as it is actually used are necessary, not only for the verification of grammatical analyses, but also for understanding how the regularities we think of as grammar emerge from communicative needs. Research in discourse and grammar draws upon both spoken and written corpora, and it is typically, though not necessarily, quantitative. Monographs in the series propose explanations for grammatical regularities in terms of recurrent discourse patterns, which reflect communicative needs, both informational and socio-cultural.

Editors Sandra A. Thompson University of California at Santa Barbara Department of Linguistics Santa Barbara, CA 93106 USA

Paul J. Hopper Carnegie Mellon University Department of English Pittsburgh, PA 15213 USA

Volume 10 Studies in Interactional Linguistics Edited by Margret Selting and Elizabeth Couper-Kuhlen

Studies in Interactional Linguistics

Edited by Margret Selting University of Potsdam

Elizabeth Couper-Kuhlen University of Konstanz

John Benjamins Publishing Company Amsterdam/Philadelphia

8

TM

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

Library of Congress Cataloging-in-Publication Data Studies in Interactional Linguistics / edited by Margret Selting, Elizabeth Couper-Kuhlen. p. cm. (Studies in Discourse and Grammar, issn 0928–8929 ; v. 10) Includes bibliographical references and index. 1. Grammar, Comparative and general--Syntax. 2. Conversation analysis. I. Selting, Margaret. II. Couper-Kuhlen, Elizabeth. III. Series. P291.S698 2001 415--dc21 isbn 90 272 26202 (Eur.) / 1 58811 0974 (US) (Hb; alk. paper)

2001043028

© 2001 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

Contents

Foreword Sandra A. Thompson Introducing Interactional Linguistics Elizabeth Couper-Kuhlen and Margret Selting

vii

1

Part I. Language structure in interaction

23

Emerging syntax for interaction: Noun phrases and clauses as a syntactic resource for interaction Marja-Liisa Helasvuo

25

At the intersection of turn and sequence: Negation and what comes next Cecilia E. Ford

51

The implementation of possible cognitive shifts in Japanese conversation: Complementizers as pivotal devices Hiroko Tanaka

81

On causal clause combining: The case of ‘weil’ in spoken German Hannes Scheutz

111

Dutch ‘but’ as a sequential conjunction: Its use as a resumption marker Harrie Mazeland and Mike Huiskes

141

On some uses of the discourse particle ‘kyl(lä)’ in Finnish conversation Auli Hakulinen

171

Interactional Linguistics and language development: A conversation analytic perspective on emergent syntax Juliette Corrin, Clare Tarplee and Bill Wells 199

Chapter title

vi

Part II. Interactional order and linguistic practice

227

Fragments of units as deviant cases of unit production in conversational talk Margret Selting 229 Notes on turn-construction methods in Danish and Turkish conversation Jakob Steensig

259

An exploration of prosody and turn projection in English conversation Barbara A. Fox

287

Postposition-initiated utterances in Japanese conversation: An interactional account of a grammatical practice Makoto Hayashi 317 ConWrming intersubjectivity through retroactive elaboration: Organization of phrasal units in other-initiated repair sequences in Korean conversation Kyu-hyun Kim 345 Some arguments for the relevance of syntax to same-sentence self-repair in everyday German conversation Susanne Uhmann 373 Simple answers to polar questions: The case of Finnish Marja-Leena Sorjonen

405

List of contributors

433

Index

435

Foreword Sandra A. Thompson

The title of this volume, Interactional Linguistics, heralds a new direction in the Weld of linguistics. Over the past ten years, there has been an increasing awareness that those aspects of linguistic ‘structure’ that linguists have always been most concerned about, namely phonetics, phonology, morphology, syntax, and semantics, are inextricably intertwined with everyday language use. Growing out of this focus on language in its natural habitat has been a burgeoning body of research attempting to relate issues in linguistics to the Wndings of those scholars who have been devoting themselves to the study of talk-in-interaction, particularly those working in Conversation Analysis. The volume you hold in your hands is a tribute to the success of this exploratory venture: here we see rich discussions of linguistic phenomena explained in terms of the actions that participants are engaged in when they are interacting. These linguistic phenomena include discourse particles, connectives and complementizers, adpositions, the internal structure of clauses and phrases, word order, interrogation, constituency, deixis, negation, language acquisition, and prosody. In each case the researchers show, with careful empirical argumentation, that these are emergent phenomena, dependent on and shaped by not just the context in which they occur, but more speciWcally by the actions which the speakers are undertaking with their talk. Accomplishing these actions means moment-by-moment attention to the developing trajectory of the turn in progress, both so as to predict what action that turn is being used in the service of, and so as to know when it might be relevant for someone else to take a turn to further their projected actions. Recurrent regularities of turn organization and turn sequences are the primary way in which participants’ actions are organized and negotiated. It follows, then, that, to understand our favorite linguistic phenomena, linguists’ attention must also be on the developing trajectory of the turn and the actions which turns-at-talk are accomplishing. In this book, grammar and prosody are shown to be bound up in the organization of turns and sequences. Its papers make a very strong case for seeing the regularities of linguistic

viii Sandra A. Thompson

phenomena as practices which speakers deploy as solutions to recurrent needs in the real-time, situated, everyday human business of interacting and accomplishing the actions that are relevant for them. Its Wndings argue persuasively in favor of a new ‘interactional linguistics’.

Introducing Interactional Linguistics Elizabeth Couper-Kuhlen and Margret Selting

It is only recently that linguists, conversation analysts and anthropologists have become aware of what they stand to gain from joining forces. The fact is that the individual strengths of each discipline — technical description in linguistics, research methodology in conversation analysis and cultural breadth in anthropology — complement each other quite proWtably, as the recent collections Grammar and Interaction (Ochs, SchegloV & Thompson, eds. 1996), Prosody in Conversation (Couper-Kuhlen & Selting, eds. 1996) and Interaction-based Studies of Language (Ford & Wagner, eds. 1996) have shown. Contributors to these volumes were engaged in what we propose to call Interactional Linguistics, a perspective on language structure and use informed by language’s natural habitat in the interaction order. As we see it, it is not just grammar and prosody which lend themselves to study from an interactional linguistic point of view, but all aspects of language structure and use — phonetics, phonology, morphology, syntax, lexis, semantics and pragmatics as well as language variation, language acquisition, loss and disorder. For each of these linguistic subdisciplines and domains, it makes sense to ask how linguistic structures and patterns of use are shaped by, and themselves shape, interaction — be it conversational or institutional, adult or adult-child, with the language-unimpaired or the language-impaired. This then is the interactional linguistic enterprise. In the following we shall outline in more detail what a research program in interactional linguistics might look like. And we shall give a brief overview of current research being carried out in this budding new Weld at the intersection of linguistics, conversation analysis and anthropology.

1.

How has Interactional Linguistics developed?

Before the advent of the late twentieth and early twenty-Wrst century, language was studied primarily in isolated written form, often hypostatized as words and sentences. Students of language fabricated their own material or elicited it from native informants using word lists, questionnaires and translation as

2

Elizabeth Couper-Kuhlen and Margret Selting

tools for cross-linguistic comparison. But as technology has advanced, making speech increasingly accessible and amenable to scientiWc investigation, linguistics has enlarged its perspective to embrace the study of spoken language (Chafe (ed.) 1980, Chafe 1982, Chafe and Danielewicz 1986). With the study of spoken language has come an increased recognition of the role of intonation (Chafe 1994) together with the development of transcription systems sensitive to intonational and prosodic features (Edwards and Lampert 1993). The acceptance of spoken language as an object of investigation in its own right was a Wrst important step on the way to a genuinely interactional linguistics. The second step was made possible by inXuential work in the discourse functional tradition. Here, early on, functionally oriented linguists were looking for — and Wnding — motivated relations between linguistic form and discourse function. Linguistic distinctions, e.g. between word class or aspectual forms, were shown to be motivated by the role they play in narrative discourse (Hopper and Thompson 1984, Hopper 1979), morphological case marking patterns were traced back to underlying discourse strategies such as Preferred Argument Structure (DuBois 1985, 1987), phonological phrasing to a discourse preference for one new idea at a time (Chafe 1979). Functional linguistic research — although it did not focus on conversational interaction — was instrumental in establishing a mindset for the study of language which saw linguistic form as something ‘to do things with’ on situated occasions of use, thereby placing its functional motivation squarely at the center of interest. The third step towards an interactional linguistics came with seminal work in the Weld of conversation analysis. Here as of the 1970’s sociologists were advocating the study of everyday conversation as a locus of social order, and in fact the primordial one (Sacks, SchegloV and JeVerson 1974, Sacks 1992). Other types of discourse — it was argued — were best understood as adaptations of the fundamental conversational order. As a cover term for all kinds of naturally occurring speech exchange in which ‘online’ co-communication is implicated, the term talk-in-interaction was introduced (see SchegloV 1998 for a recent discussion). And a set of methodological tools was developed which entailed radically empirical induction and the reconstruction of categories from collections of genuine, naturally occurring data, enabling practitioners to warrant their analytic categories and interpretations with reference to the orientation of the interactants themselves (see SchegloV 1996 for a demonstration of this methodology). It is the conversation-analytic understanding of speech exchange as social interaction and the conversation-analytic tools of micro-analysis and participant-oriented proof procedures which are instru-

Introducing Interactional Linguistics

mental to the interactional linguistic enterprise. The fourth and Wnal step towards an interactional linguistics came from the anthropological linguistic studies done of speech exchange systems and discourse strategies in cross-cultural comparison (Gumperz 1982, SchieVelin and Ochs 1986, Moerman 1988, Maynard 1989, Hanks 1990). If, as discourse functionalists argue, language is informed by its use in the production of discourse — and vice versa if, as conversation analysts argue, discourse or the conversational order is informed by language use — then it stands to reason that culturally diverse social orders or ways of conducting conversation and organizing discourse will impinge diVerently on language, and that diVerent languages will contribute in distinct ways to the nature of the social order they construct. Accordingly, in addition to the analysis of individual languages as a resource for sequential interaction, a cross-linguistic and cross-cultural perspective is crucial to the interactional linguistic enterprise. Interactional Linguistics thus takes an interdisciplinary and a cross-linguistic perspective on language. Its goal is a better understanding of how languages are shaped by interaction and how interactional practices are molded through speciWc languages. It treats speech as an ongoing or emergent product in a social semiotic event and language as providing one set of resources for the accomplishment of goals or tasks within this event. The resources which language provides are used methodically, it is assumed; with them speakers engage in practices, routine and recognizable ways of carrying out sequentially situated actions and activities (SchegloV 1997). Thus, an interactional linguist asks two sorts of questions which implicate language: (i) what linguistic resources are used to articulate particular conversational structures and fulWl interactional functions? and (ii) what interactional function or conversational structure is furthered by particular linguistic forms and ways of using them? Because language is Wrst and foremost a tool for interaction, an interactional linguist expects it to be shaped in crucial ways by this function, or to use a biological metaphor, to be adapted to the environment which it ‘inhabits’. Conversely, interaction itself is expected to be shaped by language and, on a universal level, to embody subtly diVerent interactional practices depending on language type.

3

4


2.

What is interactional about language?

Some linguists, in particular those inXuenced by the Chomskian paradigm, may wonder why they need an interactional perspective on language. Traditionally, it is competence — the abstract knowledge of linguistic structures and their well-formedness — which has been thought to provide the wherewithal for performance — the realization of language competence in concrete situations of use. On this view, what happens in interaction is merely a contextual, real-time application of structures whose well-formedness is determined out of context by the linguistic system. But precisely this view of language has been called into question by interactional linguistic research. In particular, the seminal work of Goodwin (1981, 1995) has shown that what linguists have traditionally thought of as a sentence — a unit of language presumed to be the product of a single speaker’s competence — is highly sensitive to the securing of recipiency, e.g. via gaze, in face-to-face-interaction. When recipiency is uncertain or when it changes in the process of a speaker’s utterance, the speaker adapts further production of the sentence to the new situation, e.g. by redesigning the sentence for his or her new recipient. So the resulting unit ‘sentence’ is as much the product of the speaker as of the recipient(s) for which it is designed. Ford (1993) makes a related point with respect to subordinate clauses, traditionally thought of as dependent structures which form a complex sentence with some ‘main’ clause. Ford’s observations of adverbial clause use in everyday American conversation show that causal clauses are frequently turns themselves and that rather than being planned as a whole together with the ‘main’ clause, they are instead occasioned by prior talk or its absence — for instance, if a speaker does not receive immediate uptake on a turn hearably doing requesting or inviting or if the recipient initiates repair on that turn. These clauses, then, are again not the product of the speaker’s individual planning, but a product of the situated interaction that speaker and recipient are engaged in. An interactional perspective on language thus entails a radical re-thinking of notions such as competence and performance. Rather than conceptualizing language as an abstract and balanced system of pre-established discrete elements which are combined with one another into ‘sentences’ that are then realized in speech, interactional evidence suggests that language forms and structures must be thought of in a more situated, context-sensitive fashion as actively (re)produced and locally adapted to the exigencies of the interaction at


hand. In this sense they can be conceived of as arising or emerging in use (see also Hopper 1988, 1998). Moreover, linguistic productions — since they take shape in interaction — can no longer be conceptualized as the product of a single speaker. Instead sentence and clause production, indeed speech production in general, must be thought of as an interactional achievement (Goodwin 1981). In this view, syntax, just like prosody and semantics, is a resource that can be relied on as shared knowledge in the speech community and that can be ‘distributed’ across speakers in collaborative productions (Lerner 1991: 141). Thus linguistic structures are both emergent in interaction and heavily context-sensitive, in that their use reXects — and may even contribute to creating — conversational structure. This point has been convincingly demonstrated by Fox’s (1987) work on anaphora in conversation. Although traditionally pronouns have been thought to require an antecedent in the immediately preceding context, Fox’s investigation reveals that the ‘return pop’ (longdistance pronominalization) — where the antecedent of a pronominal form is located not in an immediately prior turn or sequence but in a sequence at some distance from the one which contains it — is a perfectly routine phenomenon. Pronominal forms, Fox argues, signal that a sequence is as yet still open, whereas the use of a full form opens up a new sequence and thus contributes to closing down the sequence of prior talk. Pronominal use thus stands in a reXexive relation to conversational structure: on the one hand, it is ‘determined’ by sequential organization but on the other hand it contributes to creating precisely that organization. Anaphora is not the only type of linguistic form to be at once context-dependent and context-constitutive (see also Silverstein’s (1976) discussion of deixis and other indexical expressions). The compelling conclusion seems to be that context — and in particular, (sequential) interactional context — must be made an integral part of linguistic theory and practice (see also Duranti and Goodwin (eds) 1992).

3.

What is linguistic about interaction?

If linguistics cannot be done properly without interaction, so too the study of interactional conduct is dependent on linguistic generalizations. Take, for instance, the turn-taking mechanism as described by Sacks, SchegloV and JeVerson (1974). The basic unit for the organization of turn-taking in conversation, the turn-constructional unit (TCU), is a linguistic unit: a word, phrase, clause or sentence (Sacks, SchegloV and JeVerson 1974: 702). As recent studies

5

6


have shown, the point at which one of these units is considered to be possibly complete and a change of speakership to be feasible — the transition relevance place (TRP) — depends on a complex interplay of grammatical, prosodic and pragmatic cues (Ford and Thompson 1996; Ford, Fox and Thompson 1996; Selting 1996; SchegloV 1996). Projection of possible turn completion in conversation is thus crucially dependent on cues which are largely linguistic in nature. Likewise, as Fox and Jasperson (1995) and Fox, Hayashi and Jasperson (1996) have shown, the system of repair organization in conversation, in particular same-turn self-repair, is intimately tied up with syntax. Conversationalists depend upon their knowledge of possible syntactic templates in order to recognize repairing segments as departures therefrom. In recycling bits of speech which are repairable, speakers furthermore observe syntactic constituent boundaries — and indeed appear to favor certain ones over others. For instance, Fox and Hayashi’s (1995) examination of same-turn self-repair in English shows that conversationalists make little use of the verb phrase, understood as the verb and its obligatory object complement(s), in self-repairing. For activity sequences, organized around the adjacency pair and its expansions, there is also room, indeed possibly a need, for linguistic generalization. Even if little or no syntax is involved in a conversational move — as is, for instance, the case with free-standing oh (Heritage 1984) — prosodic and phonological generalizations can nevertheless still be made about the way interactants handle the design of such tokens. In the British English materials that Local (1996) examined, for instance, oh as a display of news receipt was typically realized as a diphthong with falling pitch ending low but never terminating with glottal closure, whereas oh in question-elicited informings tended to be realized as a monophthong, occasionally with Wnal glottal closure and rising-falling Wnal pitch. If simple lexical items show this degree of methodic deployment, so a fortiori may more complex syntagms be expected to do so (for other examples, see Ford, Fox and Thompson (in press) and the contributions in Couper-Kuhlen and Selting (eds) 1996; Ochs, SchegloV and Thompson (eds) 1996 and Ford and Wagner (eds) 1996). It is this systematicity which Sorjonen (1996) appeals to when she invokes the notion of an ‘activity grammar’: “semiotic tools through which members of a speech community collaboratively and in a systematic fashion go about constituting activities in and through which they manage their relationships and practical tasks in the everyday life” (1996: 321). Linguistically oriented studies of interaction thus aim to reveal recurrent formal patterns on which the sense-making of conversation depends and on


which participants rely in their conduct of interaction. These patterns are identiWed less on the grounds of statistical frequency than of routine expectation (Levinson 1983). It is the (strategic) use of certain resources — or the (strategic) lack of their use — in sequentially speciWable circumstances which underlies the inferences that participants make in conducting conversation.

4.

Why look at interaction cross-linguistically?

Form-function correlations in conversational interaction are particularly revealing when diVerent languages, and in particular typologically diVerent languages, are compared. In Germanic languages, for instance, syntactic organization in the clause is relatively ‘tight’: not only do turn-initial forms make it possible to project what it will take for the unit to be complete, the probability of this projection being correct also steadily increases as the turn or turn-constructional unit proceeds (Auer 1996: 59). Yet as Tanaka’s research into turn-taking in Japanese conversation shows, this kind of early turn projection is not possible there (1999, 2000). In Japanese, a verb-Wnal language, the parts of the clause do not ‘fall into place’ until the production of the verb at the end of a turn-constructional unit. The clause in Japanese is more loosely organized than in Germanic languages, where obligatory elements are regularly realized in surface structure. Japanese subjects and object complements need not be realized in surface structure if they are contextually given, with the result that turn-constructional units often consist only of optional (adverbial) adjuncts (Fox, Hayashi and Jasperson 1996). This too means that projection must operate diVerently. Japanese TRPs, for instance, are regularly signalled not by possibly complete syntax, but rather by so-called ‘Wnal’ particles (Tanaka 2000). ‘Loose’ syntactic organization in the clause appears to be responsible as well for diVerent repair strategies in Japanese as compared to English or German. Fox, Hayashi and Jasperson (1996), for instance, observe little use being made of the clause as a unit for recycling in same-turn self-repair in Japanese — in contrast to English. On the other hand, they do Wnd conversational selfrepair of bound morphemes in Japanese, a so-called agglutinating language; this practice is unknown in inXectional languages like English or German. The practice of co-construction or co-participant turn completion is also aVected by syntactic organization in the clause or clause complex. Whereas Lerner (1991), for instance, Wnds syntactically deWned two-part formats such as [ if X + then Y ] routinely providing English speakers with opportunities for

7

8


collaborative completion, Hayashi reports that these structures are not used as prominently in Japanese co-construction. Instead, most instances of collaborative completion in Japanese involve ‘terminal item completion’ (Hayashi 1999). Here too the explanation is presumably that early projection is not possible in a verb-Wnal language such as Japanese, but that rather a ‘wait and see’ policy must be adopted (Fox, Hayashi and Jasperson 1996). Clause organization in Korean, also a verb-Wnal language, has been described by Kim (1999) as ‘agglutinating’ in the following syntactic sense: in turn-unit construction speakers tend to chain together phrasal units which themselves are not transition-implicative, often pausing at phrase-unit boundaries to solicit collaborative acknowledgement from recipients. This augmentative practice enables current speakers to mold the turn-under-construction according to the nature of the recipiency which a co-participant signals. At the same time, however, according to Kim, it places an interactional burden on recipients, who are expected to express collaboration before the gist of a turn is perfectly clear. What interactional linguistic research on typologically diVerent languages thus suggests is that the way common problems are dealt with in interaction may be shaped by the distinct linguistic resources which a language provides (see also Ford and Mori 1994, Lerner and Takagi 1999). The prospects which such an avenue of inquiry opens up are vast indeed. One wonders, for instance, how a rhythmic typological diVerence such as stress- vs. syllable-timing might impinge on conversational organization, and in particular on turn-taking (see Auer, Couper-Kuhlen and Müller 1999 for some initial discussion of this question). But further questions suggest themselves: In what ways are prosody and intonation used in diVerent languages for the organization of interaction? Do speakers of tone languages rely less heavily on sentence-Wnal intonation than those of so-called ‘intonation languages’? What impact does rich case morphology have, e.g., on self-repair strategies? How does the syntactic organization of clauses, e.g. in terms of typologically distinct placement of the verb or in terms of more or less Wxed word-order patterns, determine practices of turn-unit and turn organization? In what ways are prosodic, syntactic, lexical, semantic and pragmatic devices combined for accomplishing the projection of turn-holding and turn-completion? What devices are language- or even dialect-speciWc, what others perhaps language-type speciWc or even universal? A cross-linguistic avenue of inquiry suggests that the way interaction itself is conducted may be inXuenced by (typologically) diVerent language practices, a case in point being the distinctive use of recipiency tokens in Japanese, Chinese


and English (Clancy et al 1996). A cross-linguistic examination of interaction thus has clear anthropological implications.

5.

The contributions to this volume

It is precisely such questions that this volume wishes to address. Contributions which start from language structure and investigate its deployment in interaction have been grouped under the heading ‘Language structure in interaction’ in Section I; those which start from the interactional order and identify linguistic practices systematically associated with it have been grouped under the heading ‘Interactional order and linguistic practice’ in Section II. 5.1 Language structure in interaction The papers in this section examine language structures which are primarily syntactic and lexical in nature. They treat these structures as emergent in their context of occurrence and as resources for various types of interactional work.

Clausal and phrasal units Helasvuo examines the clause and the noun phrase in Finnish and English data, aiming to determine how they are deployed as resources for the conduct of interaction. She Wnds little interactional evidence in either language for the clause as traditionally deWned, namely as predicate + core arguments + oblique arguments. Instead, in both Finnish and English, oblique arguments tend to be found in intonation units separate from clause cores and it is often not clear, or important, which surrounding clause they belong to. If clause cores are produced in separate intonation units, it is more likely that subject + verb will form one intonation unit than verb + object, especially if the verb is a mental or communicative one. Interestingly, this Wnding corroborates Fox and Jasperson’s (1995) observations on the lack of relevance of the verb phrase for conversational self-repair but it does so on a diVerent set of interactional grounds. Helasvuo Wnds evidence for the tight internal structure of both clauses and noun phrases in Finnish and English and shows that this structure is used as a means for projecting syntactic completion and ultimately transition relevance. Her data illustrate these units being exploited for conversational tasks such as producing increments, creating lists and co-constructing turns.

9

10


Negation Ford examines the occurrence of negative markers in turn-constructional units in English conversation, identifying a kind of ‘rhetorical combination’ in which units with negative markers are regularly followed up by a next unit which elaborates the work of the negation. Such shorter multi-TCU turns have their own pragmatic projection: they are not treated as complete until the elaborating unit has been produced. Ford’s data shows clearly that where the elaborating unit has been projected but is missing, participants actively pursue resolution, thus attesting to the ‘conditional relevance’ of the second unit. Since some of the negation + elaboration combinations Ford encounters appear to be produced with pre-planning but others clearly emerge through inter-speaker interaction, she sees this rhetorical pattern as being located at the intersection of turn and sequence construction systems. Complementation Tanaka’s contribution is likewise concerned with a linguistic phenomenon situated at the intersection of turn and sequence construction. She investigates the use of the Japanese complementizer to as deployed under real-time conditions in conversational interaction. To appears after the stretch of talk it complementizes and can be used to re-direct or re-formulate a prior bit of talk by retroactively turning it into the direct object of an upcoming verb of thinking, saying or feeling. As Tanaka points out, this may have the cumulative eVect of mitigating, if not un-doing the action of the original turn. Strategic deployment of to in Japanese conversation thus makes it possible to accomplish a type of action-repair without resorting to grammatical repair. Postpositioning of the complementizer in Japanese — as opposed to its prepositioning in Indo-European — brings out clearly how the order in which actions are accomplished via language is dependent on word order: in Japanese the substance of what is being talked about is articulated before the social action bearing upon it becomes clear. ‘Subordinating’ conjunctions Scheutz investigates the use of the German ‘subordinating’ conjunction weil in a large corpus of spoken everyday conversation, settling up with earlier analyses which have postulated a form-function correlation between verb position and type of causality. He Wnds little support in his data for the claim that weil clauses with the verb in Wnal position express propositional causality whereas those with the verb in second position express epistemic causality. Scheutz


argues instead that it is the degree of syntactic integration which correlates with type of causality. Narrow syntactic integration restricts the causal relation to a propositional one; looser syntactic integration allows for a greater variety of semantic and pragmatic functions for the causal. Particularly suggestive from an interactional point of view is Scheutz’s observation that verb-Wnal weil clauses are used primarily as retrospective, sequence-closing devices, whereas verb-second weil can function prospectively to introduce topics and expand turns.

‘Coordinating’ conjunctions Mazeland and Huiskes also address a lexical issue, namely the meaning of the Dutch conjunction maar ‘but’. They isolate one speciWc use of maar, namely as a so-called sequential conjunction, which links turns at talk rather than units within the same turn. Their argument is that maar is used as a device for resuming a line of topic development which has been temporarily abandoned in favor of an extended repair sequence or a competing line of topic development. Maar does the job of framing the turn it introduces as resumptive of the abandoned activity; it reinstalls a position from which more telling can ensue. Mazeland and Husike reXect on what this use of maar means for the age-old linguistic problem of polysemy: their conclusion is that maar is not polysemous but rather that its use contributes to repeatedly occurring, constructionally speciWable types of utterances which can be thought of as practices. They argue that it is the action a practice implements which governs how the linguistic elements of the utterance itself are dealt with when they are made sense of in interaction and not vice versa, thus seriously challenging traditional theories of the lexicon. Particles Hakulinen’s chapter also deals with the use of a lexical item in conversation, the Finnish particle kyllä, a response token typically found in turn- and utterance-initial position. Used with a full-sentence response to a yes-no question, Hakulinen Wnds that kyllä contributes to a display of alignment with the positive alternative. In responses to oVers, it also signals alignment, especially in front position. The alignment becomes less deWnite, however, the further into the turn kyllä is moved. In situations of overt or covert disagreement, kyllä, especially in Wnal position, is used to do away with co-participant doubts or misgivings and also conveys a sense of appealing. Hakulinen’s study suggests that the import of a discourse particle like kyllä depends not only on its

11

12


position in the turn but also — and crucially — on the type of activity sequence this turn is implicated in.

Developmental syntax While the Wrst section of the present volume begins with a paper considering the on-line emergence of syntactic constructions in adult conversation, it concludes with a study of the developmental emergence of grammar in childadult conversation. Corrin, Tarplee and Wells argue that the production of a ‘Wrst sentence’ by the child must be seen as dependent upon the emergence of turn organization, since the latter serves as an occasion for the former. In other words, in order for a succession of semantically structured single utterances to be produced as a whole, the child must secure the Xoor long enough to allow syntax to emerge. The authors identify several strategies which children deploy to accomplish this, including eye gaze and gestural pointing. But one of the most consistent cues to turn projection, they observe, is the use of mid or midlevel Wnal pitch. Even when the syntactic-semantic unit under construction is potentially complete, this pitch signal is shown to function as a turn-continuation device, prompting the mother to withhold a receipt token and allowing the child enough interactional space for a syntactically complex production. Corrin, Tarplee and Wells’ study of the transition from sequence organization to turn organization in the child has compelling implications for the diachronic development of syntax. Moreover, it draws attention to a play-oV between planned syntax and occasioned syntax which synchronic studies of language in interaction have also attested (see e.g. Tanaka and Ford, both in this volume). 5.1 Language structure in interaction The papers in this section investigate a particular dimension of the interactional order and identify linguistic practices which are implicated in and by it.

Turn-unit construction Selting’s concern is the construction of the smallest units of turn organization in interaction, turn-constructional units, and how participants make these recognizable for one another. In order to answer this question, she examines ‘deviant’ TCUs, i.e. stretches of talk that are treated as fragmentary either by being cut oV and followed up by a new unit or by being left incomplete. These fragments of TCUs are the result of various kinds of unfulWlled projection,


syntactic, prosodic, semantic/pragmatic and sequential in nature. In trying to separate out the respective contribution of each type of contextualizing cue to projection, Selting comes to the conclusion that although syntax appears to override prosody when these cues are used discrepantly, neither type is suYcient to cue TCU-completedness or lack of it on its own. Instead, inferences about syntactic and prosodic completion are always made context-sensitively in the light of the semantic/pragmatic and sequential context. What this means is that it is not the units or the fragments themselves to which participants orient, but rather the unit- and turn-constructional practices or acts which they are inferred to be involved in, such as beginning a new unit, projecting the continuation of a unit or completing a unit.

Turn construction Steensig’s paper also addresses the relative contribution of grammatical, prosodic and pragmatic cues to projection but this time with respect to turnsat-talk. He compares and contrasts methods for turn construction in two typologically unrelated languages, Danish (verb-second and inXecting) and Turkish (verb-Wnal and agglutinating). Although speakers of both languages demonstrably operate with turn projection, the respective grammars make quite diVerent contributions: whereas in Danish, clausal grammar permits a relatively reliable projection of what it will take for the unit to be (grammatically) complete early in the turn, in Turkish this information is not available until the very end of the clause, when the verb and an agglutinating marker of Wniteness or non-Wniteness appears. Steensig Wnds, however, that prosody oVers compensatory help in Turkish by creating ‘possible last accents’, which occur before the last syllable and thus permit projection before the actual end of the unit. ‘Possible last accents’ also appear to do projection work in Danish, indeed more so than Wnal vs. non-Wnal intonation. Steensig’s material shows Danish speakers successfully projecting turn completion on grammatical and pragmatic grounds despite the fact that the intonation is non-Wnal. He concludes that ‘possible last accents’ are a more important prosodic cue than Wnal intonational conWguration and, moreover, that for cross-linguistic purposes grammar, prosody and pragmatics should be treated as three independent means of constructing turns and signalling projection. Turn projection Fox also investigates turn projection in her chapter but with special attention to prosody, speciWcally to pitch peaks in two-peaked English conversational

13

14


utterances. She is concerned to explore the claim made in many intonational handbooks that Wnal pitch peaks are more prominent than non-Wnal ones: if they are, she reasons, they could be said to be more projecting of turn completion. The study investigates a data base of 43 utterances, all clauses containing two or more pitch accents with pitch peaks, one of which is close to the end of the turn. Phonetic dimensions such as syllable length, word length, peak alignment, slope, change in pitch, size of pitch step-up or step-down and amplitude are measured systematically in both peaks. Yet the only measure which proves to diVerentiate the peaks signiWcantly is duration: syllables with Wnal pitch peaks are longer in duration than syllables with non-Wnal ones, statistically speaking. Fox concludes that phonetic realizations are likely to be heard not singly but as working in clusters and through the syntactic, semantic and sequential positions in which they occur, a Wnding which can be seen to provide phonetic substantiation for Selting’s (1996) claims.

Turn expansions Hayashi’s chapter addresses the interactional phenomenon of initiating a turn with a recognizable non-beginning (see also SchegloV 1996: 75). The focus of attention is on a distinctly Japanese device for doing so: starting to talk with a postpositional item, which canonical Japanese grammar would normally require to come after some other item in a sentence. Hayashi shows that Japanese speakers use postpositioned items in turn-initial position to build oV of a preceding utterance — whether it be one’s own or that of a co-participant — and to redirect the trajectory of the developing course of action. For instance, a postposition can be used in turn-initial position to latch onto a nominal in one’s own prior turn, thereby making the ‘new’ turn a syntactically integral part of the prior turn. This may be useful for ‘sealing oV’ intervening talk, essentially deleting it from the conversational record. Such a technique may also be used to pursue a response from a recipient by re-invoking a prior syntactic construction and thereby re-implementing an earlier position. Or a postpositional item may be used to latch talk onto that of a co-participant, thus bringing oV a modiWcation or correction of another’s projected course of action without making it the oYcial business of the interaction. Because postposition-initiated utterances are built oV of a preceding utterance and draw on it as a resource for their comprehension, Hayashi’s study shows how situated interactional concerns can shape the grammatical structuring of talk.


Other-initiated repair The focus of Kim’s chapter is other-initiated repair sequences in Korean conversation, in particular the incremental addition of phrasal units to preceding elliptical utterances following other-initiation of repair. As Kim points out, it is the Korean practice of using elliptical forms, and especially referentially non-explicit indexical expressions in topic-initial position, which often makes repair necessary. (This practice may be motivated by a need to give the upshot of an utterance — encapsulated in the clause-Wnal predicate — as early as possible, resulting in informationally ‘light’ reference forms in utterance-initial position.) Faced with other-initiation of repair, Korean speakers will be observed to oVer a phrasal unit as repair, but one which, rather than being a full solution to the problem, is merely a further clue, pointing to a domain of shared knowledge and experience and thus inviting recipients to continue the search for the referent themselves. As with the Japanese use of postpositions to initiate turns, so the Korean practice of adding phrasal units to initially elliptical utterances following repair initiation can be thought of as ‘postpositioning’: the phrasal units are designed to be an incremental continuation of the troublesource turn rather than a response to the repair-initiating turn. They thus have the eVect of not registering, i.e. of sequentially deleting, the intervening repair initiation and allow speakers to appear to be engaging in a retroactive elaboration of their own turns. As in other contributions to this volume (Ford; Tanaka; Corrin, Tarplee and Wells), there is a fascinating — and strategically manipulable — play-oV here between sequence organization and turn organization. Self-repair Linguistic practices are not only implicated in other-initiated self-repair but, as Uhmann’s contribution shows, also in self-initiated self-repair. Uhmann focusses on same-sentence self-repairs in German materials, i.e. on self-repairs in which the sentence under construction contains a repairable and in which the syntactic integrity of the sentence is not destroyed in the repair. She is concerned to work out what rules govern the syntax of self-repair in German. As in earlier work by Fox and Jasperson (1995), Uhmann Wnds that repairing speakers recycle in systematic ways, but she goes one step further in claiming that in German the scope of recycling is constrained by the syntactic feature ‘functional head’. Uhmann argues that speakers recycle from the functional head of the phrase containing the repairable, more speciWcally from the functional head immediately c-commanding the repairable. Additional speciWca-

15

16


tions are given for cases in which the repairable is itself a functional head or is embedded in a cascade of functional heads. Particularly intriguing from a cross-linguistic perspective is her suggestion that the German preference for recycling from the functional head may have something to do with the relatively free word order of German (with varying positions for the Wnite verb) and its relatively rich case morphology (as well as grammatical gender marking). The rich case morphology means that larger parts of the utterance are implicated in cases of repair.

Question responses Sorjonen’s paper deals with a general action type in Finnish, namely answering or responding, as implicated in various kinds of questioning activity sequences. Her study is framed in terms of a typology of answering systems, which classiWes languages with respect to how short answers to polar questions are given. Her argument is that Finnish speakers employ a mixed system: there is a division of labor between repetition and particles when giving positive answers. DiVerent answer types are used to get diVerent types of interactional work accomplished. For instance, in aYrming, or providing new information, the Wnite verb will be repeated in the case of a V-interrogative (=question particle aYxed to the verb). In conWrming, or treating the information as known to the recipient, speakers will rather tend to use the particle joo. With other types of interrogatives, the prototypical response is niin, which conWrms a prior utterance and makes continuation relevant. Interestingly, although each of these forms has its prototypical context of use, joo and verb repeats can also be used non-congruently, in which case a strategic interpretation will be called for but always with reference to the ‘basic’ functions of the respective response types. As in other contributions to this volume (e.g. Mazeland and Huiskes, Hakulinen), Sorjonen adduces interactional evidence for the strategic deployment of linguistic units as resources.

6.

Outlook

Cutting across the contributions collected here are several systematic themes around which current interactional linguistic research appears to be crystallizing. For instance, Ford’s chapter on elaborations following negation as well as Tanaka’s study of the Japanese complementizer to both reXect an interest in language structures with a potential for turn and sequence construction (see also Kim, this volume). The fact that language provides single syntactic construc-


tions which can accomplish these two tasks at once, making it possible to convert one and the same construction from intra-turn into inter-turn talk and vice versa, demonstrates how delicate a tool and how well adapted language is to the situated needs of interactants. The play-oV between planned and occasioned syntax is particularly intriguing from a linguistic point of view because of its far-reaching developmental implications, to which Corrin, Tarplee and Wells’ chapter attests. Language provides not only syntactic structures but also lexical expressions as resources for the conduct of interaction. These items, as the chapters by Scheutz, Mazeland and Huiskes, Hakulinen and Sorjonen testify, are heavily dependent on turn position and on sequential context for their meaning potential. In fact, it may be that lexical expressions are best thought of as parts of larger constructional schemas or patterns which themselves derive their inference-cueing potential from situated use in the service of recurrent interactional tasks. Expressions such as German weil, Dutch maar and Finnish kyllä or joo are thus ways of accomplishing interactional work, i.e. practices for e.g. expanding a turn, recycling an abandoned telling, responding reassuringly to a questioning pre-sequence or treating an answer as known to the recipient, respectively. What cross-linguistic studies of language structure in interaction seem to be saying then is that there is a — perhaps universal — set of tasks, activities and sequence types for interaction but that diVerent languages provide diVerent, albeit commensurable resources for accomplishing them. Viewed from the perspective of the interactional order, the papers collected here reXect a systematic focus of interest on three diVerent conversational tasks. The Wrst might be described as projecting turn or turn-unit completion. Here Selting, Steensig and Fox are in agreement that a complex interplay of cues — primarily linguistic in nature — is at work for the signalling and identifying of upcoming TRPs. Yet the analysis of turn projection in non-Indo-European languages reminds us that for language-internal reasons these cues may be weighted rather diVerently and consequently implicated in diVerent ways in speaker transition. A second focus of interest lies on the interactional practice of starting up turns with non-beginnings. A variety of interactional situations may make this a useful technique including, e.g., the potential derailing of a course of talk into side business by (disruptive) other repair-initiation. As Hayashi’s chapter shows, the Japanese language provides a resource especially well suited for the purpose of pre-emptively deXecting a potentially problematic action, namely postpositionals used so as to tie back to prior talk, eVectively ‘skip-tying’ over an intervening turn or turns. But also in Korean, according to Kim, speakers

17

18


can exploit agglutinative syntax and incrementally add phrasal units which retroactively build on a prior turn, thus deleting intervening talk. Postpositioning is thus a strategy which syntactically ‘agglutinating’, verb-Wnal languages seem to invite. What equivalent devices the typically verb-second languages of Indo-European provide is an open question. (Dutch maar as described by Mazeland and Huiskes, this volume, may accomplish a related task.) Thirdly, current interactional linguistic research focusses on self-repair and in particular on the syntactic regularities evidenced by self-repairing speakers. A preliminary study by Fox, Hayashi and Jasperson (1996) revealed signiWcant diVerences between English and Japanese in the syntax of selfrepair. Uhmann’s chapter complements this study by adding evidence from the more richly inXecting German language. Indeed, as she argues, not only the nature of a language’s syntax but also its morphology will have a signiWcant eVect on speakers’ syntactic practices in self-repair. This underlines a point which all cross-linguistic studies of the interaction order collected here make in one way or another: common interactional tasks may well be serviced by quite diVerent linguistic practices in function of internal — often typologically distinct — language features. The interdisciplinary and cross-linguistic examination of interaction which Interactional Linguistics advocates thus oVers ample opportunity for investigating the trade-oV between language and interaction universals on the one hand and language- and language type-speciWc linguistic practices on the other.

References Auer, P. 1996

“On the prosody and syntax of turn-continuations”. In Prosody in Conversation: Interactional studies, E. Couper-Kuhlen and M. Selting (eds), 57–100. Cambridge: Cambridge University Press. Auer, P., Couper-Kuhlen, E., and Müller, F. 1999 Language in Time: The rhythm and tempo of spoken interaction. New York: Oxford University Press. Chafe, W. 1979 “The Xow of thought and the Xow of language”. In Discourse and Syntax, T. Givón (ed.), 159–181. New York: Academic Press. 1982 “Integration and involvement in speaking, writing, and oral literature”. In Spoken and Written Language: Exploring orality and literacy, D. Tannen (ed.), 35–54. Norwood, NJ: Ablex. 1994 Discourse, Consciousness, and Time. Chicago: University of Chicago Press.


Chafe, W. (ed.) 1980 The Pear Stories. Cognitive, cultural, and linguistic aspects of narrative production. Norwood, NJ: Ablex. Chafe, W. and Danielewicz, J. 1987 “Properties of spoken and written language”. In Comprehending Oral and Written Language, R. Horowitz and S. J. Samuels (eds). New York: Academic Press. Clancy, P., Thompson, S. A., Suzuki, R. and Tao, H. 1996 “The conversational use of reactive tokens in English, Japanese and Mandarin”. Journal of Pragmatics 26: 355–387. Couper-Kuhlen, E. and Selting, M. (eds) 1996 Prosody in Conversation. Interactional studies. Cambridge: Cambridge University Press. Du Bois, J. 1985 “Competing motivations”. In Iconicity in Syntax, J. Haiman (ed.), 343– 365. Amsterdam: Benjamins. 1987 “The discourse base of ergativity”. Language 63: 805–855. Duranti, A. and Goodwin, C. (eds) 1992 Rethinking Context: Language as an interactive phenomenon. Cambridge: Cambridge University Press. Edwards, J. A. and Lampert, M. D. (eds) 1993 Talking Data: Transcription and coding in discourse research. Hillsdale, NJ: Lawrence Erlbaum. Ford, C. E. 1993 Grammar in Interaction. Cambridge: Cambridge University Press. Ford, C. E. and Mori, J. 1994 “Causal markers in Japanese and English conversations: A cross-linguistic study of interactional grammar”. Pragmatics 4: 31–61. Ford, C. E. and Thompson, S. A. 1996 “Interactional units in conversation: syntactic, intonational, and pragmatic resources for the management of turns”. In Interaction and Grammar, E. Ochs, E. A. SchegloV and S. A. Thompson (eds), 134–184. Cambridge: Cambridge University Press. Ford, C. E. and Wagner, J. (eds) 1996 Interaction-based Studies of Language. Special issue of Pragmatics, Vol. 6 (3). Ford, C. E., Fox, B. A. and Thompson, S. A. 1996 “Practices in the construction of turns: the ‘TCU’ revisited”. Pragmatics 6: 427–454. Ford, C. E., Fox, B. A. and Thompson, S. A. In press “Constituency and the grammar of turn increments”. In The Language of Turn and Sequence, C. E. Ford, B. A. Fox and S. A. Thompson (eds). Oxford: Oxford University Press.

19

20


Fox, B. Discourse Structure and Anaphora. Cambridge: Cambridge University Press. Fox, B. A., Hayashi, M. and Jasperson, R. 1996 “Resources and repair: A cross-linguistic study of syntax and repair”. In Interaction and Grammar, E. Ochs, E. A. SchegloV and S. A. Thompson (eds), 185–237. Cambridge: Cambridge University Press. Fox, B. A. and Jasperson, R. 1995 “A syntactic exploration of repair in English conversation”. In Alternative Linguistics. Descriptive and theoretical modes, P. W. Davis (ed.), 77– 134. Amsterdam: Benjamins. Goodwin, C. 1981 Conversational Organization: Interaction between speakers and hearers. New York: Academic Press. 1995 “Sentence construction within”. In Aspects of Oral Communication, U. QuasthoV (ed.), 198–219. Berlin: de Gruyter. Gumperz, J. J. 1982 Discourse Strategies. Cambridge: Cambridge University Press. Hanks, W. F. 1990 Referential Practice: Language and lived space among the Maya. Chicago: University of Chicago Press. Hayashi, M. 1999 “Where grammar and interaction meet: A study of co-participant completion in Japanese conversation”. Human Studies 22: 475–499. Heritage, J. 1984 “A change-of-state token and aspects of its sequential placement”. In Structures of Social Action. Studies in conversation analysis, M. J. Atkinson and J. Heritage (eds), 299–345. Cambridge: Cambridge University Press. Hopper, P. 1979 “Aspect and foregrounding in discourse”. In Discourse and Syntax, T. Givón (ed.), 213–241. New York: Academic Press. 1988 “Emergent grammar and the a priori grammar constraint”. In Linguistics in Context: Connecting observation and understanding, D. Tannen (ed.), 117–133. Norwood, NJ: Ablex. 1998 “Emergent grammar”. In The New Psychology of Language. Cognitive and functional approaches to language structure, M. Tomasello (ed.), 155– 175. Mahwah, NJ: Lawrence Erlbaum. Hopper, P. and Thompson, S. A. 1984 “The discourse basis for lexical categories in universal grammar”. Language 60: 703–751. Kim, K.-H. 1999 “Phrasal unit boundaries and organization of turns and sequences in Korean conversation”. Human Studies 22: 425–446. 1987


Lerner, G. 1991

“On the syntax of sentences-in-progress”. Language in Society 20: 441– 458. Lerner, G. and Takagi, T. 1999 “On the place of linguistic resources in the organization of talk-ininteraction: A co-investigation of English and Japanese grammatical practices”. Journal of Pragmatics 31: 49–75. Levinson, S. 1983 Pragmatics. Cambridge: Cambridge University Press. Local, J. 1996 “Conversational phonetics: Some aspects of news receipts in everyday talk”. In Prosody in Conversation, E. Couper-Kuhlen and M. Selting (eds), 177–230. Cambridge: Cambridge University Press. Maynard, S. K. 1989 Japanese Conversation: Self-contextualization through structure and interactional management. Norwood, NJ: Ablex. Moerman, M. 1988 Talking Culture. Ethnography and conversation analysis. Philadelphia, PA: University of Pennsylvania Press. Mori, J. 1999 Negotiating Agreement and Disagreement in Japanese: Connective expressions and turn construction. Amsterdam: Benjamins. Ochs, E., SchegloV, E. A. and Thompson, S. A. (eds) 1996 Interaction and Grammar. Cambridge: Cambridge University Press. Park, Y.-Y. 1998 “A discourse analysis of contrastive connectives in English, Korean, and Japanese conversation: With special reference to the context of dispreferred responses”. In Discourse Markers: Descriptions and theory, A. Jucker and Y. Ziv (eds), 277–300. Amsterdam: Benjamins. Sacks, H. 1992 Lectures on Conversation. Two volumes. Edited by G. JeVerson, with Introductions by E. A. SchegloV. Oxford: Blackwell. Sacks, H., SchegloV, E., & JeVerson, G. 1974 “A simplest systematics for the organization of turn-taking for conversation”. Language 50: 696–735. SchegloV, E. A. 1996 “Turn organization: One intersection of grammar and interaction”. In Interaction and Grammar, E. Ochs, E. A. SchegloV and S. A. Thompson (eds), 52–133. Cambridge: Cambridge University Press. 1996a “ConWrming allusions: Towards an empirical account of actions”. American Journal of Sociology, 102 (1), 161–216. 1997 “Practices and actions: Boundary cases of other-initiated repair”. Discourse Processes 23: 499–547. 1998 “ReXections on studying prosody in talk-in-interaction”. Language and Speech 41, 235–263.

21

22


SchieVelin, B. and Ochs, E. (eds) 1986 Language Socialization across Cultures. Cambridge: Cambridge University Press. Selting, M. 1996 “On the interplay of syntax and prosody in the constitution of turnconstructional units and turns in conversation”. Pragmatics 6(3): 357– 388. Silverstein, M. 1976 “Shifters, linguistic categories, and cultural description”. In Meaning in Anthropology, K. Basso and H. A. Selby (eds), 11–55. Albuquerque: University of New Mexico Press. Sorjonen, M.-L. 1996 “On repeats and responses in Finnish conversations”. In Interaction and Grammar, E. Ochs, E. A. SchegloV, and S. A. Thompson (eds), 277–327. Cambridge: Cambridge University Press. Tanaka, H. 1999 Turn-taking in Japanese Conversation: A study in grammar and interaction. Amsterdam: Benjamins. 2000 “Turn-projection in Japanese talk-in-interaction”. Research on Language and Social Interaction 33(1): 1–38.

Part I Language structure in interaction

Emerging syntax for interaction Noun phrases and clauses as a syntactic resource for interaction Marja-Liisa Helasvuo University of Turku

This paper explores the use of noun phrases and clauses in various interactional contexts. I will show that conversation participants show orientation to these units when performing certain interactional tasks in conversation, such as adding increments, constructing lists, and making repair. Thus, we could say that noun phrases and clauses are a syntactic resource that can be exploited for speciWc interactional purposes. The paper argues for a view that sees the construction of syntactic structures, such as clauses and NPs, as one kind of conversational activity. Syntactic structures emerge in conversation as a mutual activity by the conversation participants.

1.

Introduction

Discourse studies have shown the robustness of the noun phrase as a syntactic category, both as a clausal constituent and as a free construction, the so-called free NP (or unattached NP; see Helasvuo 1991, in press, Durie 1994, Ono and Thompson 1994, Tao 1996). Furthermore, there is evidence that speakers orient to noun phrases in performing several interactional tasks that are strategically important, such as constructing lists (see Jefferson 1990), adding increments (Ford, Fox and Thompson 2001) and making repairs (Fox and Jasperson 1995, Fox, Hayashi and Jasperson 1996). With the help of examples from both English and Finnish, I will show that noun phrases are a syntactic resource that can be exploited for speciWc interactional purposes. Careful word-by-word analysis of spoken data reveals that the status of the clause as a clear-cut syntactic unit is problematic (for critical discussions on the notion of clause from a discourse point of view, see Cumming 1984, Ono and Thompson 1995, Tao 1996). In the Finnish and English data examined, the

26

Marja-Liisa Helasvuo

clause as traditionally deWned (predicate + core arguments + optional arguments) turns out to be not so clearly discernible as a unit. Instead, it is the clause core (the predicate and its core arguments) that emerges as clearest, both syntactically and prosodically. In both data sets, the clause core was usually produced in one intonation unit. If a core argument was presented in a different intonation unit, however, it was more often the object than the subject. Furthermore, both data sets show that 1st and 2nd person subjects Wgure signiWcantly in the subject role. They may form tight combinations with the predicate, especially with mental and communicative verbs. This Wnding suggests that the bond between the subject and the predicate may be stronger in the two languages than that between the verb and the object. This hypothesis is further supported by evidence on the syntactic organization of repair in conversation (Fox and Jasperson 1995). This paper is organized as follows: Wrst, I will introduce my data. Secondly, I will discuss clauses as syntactic and prosodic units and the ways in which participants orient to clauses as syntactic units in interaction. Thirdly, there will be a discussion of the syntax and prosody of noun phrases and the speciWc interactional tasks they are used to serve.

2.

Data

The data for this study come from Finnish and English everyday conversations, including face-to-face encounters as well as telephone conversations. Everyday conversation is chosen for analysis because I consider it to be the most natural setting of language use. Naturally, the Wndings are representative of conversational language only. The Finnish data come from various sources, but they have all been transcribed by myself (for a more detailed description, see Helasvuo in press). The data come from six naturally occurring conversations between speakers of Finnish. For closer analysis, I have chosen excerpts that I have transcribed in terms of intonation units (henceforth: IUs) following loosely the conventions of Du Bois et al. (1993). The excerpts amount to 40 minutes of transcribed data, consisting of almost 2000 IUs. The English data have been obtained from the Santa Barbara Corpus of Spoken American English, together with the transcripts. The excerpts amount to 40 minutes of transcribed data. The transcriptions are made following the conventions of Du Bois et al. (1993), and they consist of 2000 IUs. Henceforth,

Emerging syntax for interaction

the quantiWcational results are based on these sets of data. In analyzing the data I have tried to be aware of any cross-linguistic implications. Interestingly enough, there do not seem to be any great differences between the two languages concerning the research questions addressed in this study. In the discussion below, however, I will mention questions for future research on potential cross-linguistic variation.

3.

The syntax and prosody of clause construction

The view that IUs correspond to clauses (or sentences) seems to be quite widely accepted. However, it is usually assumed rather than argued for. Only rarely does one Wnd discussions on what is meant by clause. I would like to discuss this question here. I will argue for a view that sees clauses as emergent units. With the help of examples from the data, I will illustrate what I mean by emergence. I will also look at ways in which conversation participants orient to clauses as syntactic units. 3.1 What is a clause? In this section I will discuss the notion of clause and its relation to intonational phrasing. Many transcription theories employ the notion of clause or sentence in describing the transcription conventions of intonation contours. Thus, they speak about sentence-Wnal intonation that is marked with a period and question intonation marked with a question mark (see e.g. Chafe 1994). This may be helpful and illustrative in some ways. Couper-Kuhlen and Selting (1996: 41) see a potential danger in the use of punctuation marks to illustrate intonation contours at unit boundaries. They state: “…the use of punctuation marks (…) for the representation of phrase-Wnal intonation depends on stereotypical associations between sentence-type and intonation which are misleading”. I agree with this, but Wnd it even more confusing that in some transcription systems punctuation marks are used in transcriptions with no explanation as to what they are supposed to mean as transcription symbols. The reader has no way to interpret the punctuation marks other than according to the received typographical conventions. This means imposing conventions and categories of the written language on spoken discourse without analyzing the spoken data in its own right.

27

28


In my view, clauses are emergent syntactic constructions that consist of the predicate and its arguments (on emergent grammar see Hopper 1987, 1988; on “sentences” as emerging constructions, see Goodwin 1979, 1981; Auer 1996 on emergent syntactic gestalts, see also Section 3.3 below). These may be accompanied by optional arguments, but they do not constitute the clause in the same way as the core arguments and the predicate do. By core arguments I mean subjects and objects (note that these grammatical roles may not be relevant cross-linguistically in deWning the clause core). Often enough, there are no syntactic grounds to decide what clause a given oblique argument belongs to, and I would like to suggest that it is an irrelevant question for the emerging syntax-for-interaction. Consider the following example which comes from a conversation between four men discussing the cost of living in suburban Helsinki. (Note that the recording was made in the late 1950’s and the living standards were different from those of today.)1 (1) Alko 1

Olli:

2

Tor:

3

4

Aarre:

5

Aarre:

6

Olli:

7

Tor:

8

Aarre:

tei-llä-hä on kaheksan tu[hatta kuussa.] you-ADE-CLT is eight thousand in.month You pay eight thousand in a month, don’t you? sel]lane [on-ko su-lla is-Q you-ADE a.kind.of Do you have a [^huvi]la siellä tai mite se on. villa there or how it is villa there or what? [joo.] Yeah. kol[me huonetta.] three rooms Three rooms. [<X X X>] matkat lisäks. trips addition-TRA (the cost of commuting) trips in addition? omistaja-n kansa? owner-GEN with With the owner? mm. Mhm.

Example 1 starts out with Olli asking a conWrmation question about the rent Aarre is paying. Partly overlapping is Tor’s question in lines 2 and 3 about Aarre’s lifestyle. These questions are both directed to Aarre. Aarre’s answer to the Wrst question is in line 4, which overlaps Tor’s question. After hearing Tor’s question, Aarre produces the NP kolme huonetta ‘three rooms’ in line 5, which serves as an answer to Tor’s question about Aarre’s lifestyle. Syntactically, the


answer is elliptical and the predicate is recoverable from the question. In line 6, Olli continues the same line of questioning, namely the cost of living, but the part which is in overlap with Aarre’s line 5 is unfortunately not audible. What is audible is the NP + locative construction matkat lisäks ‘trips in addition’. This construction is not elliptical, and it is syntactically free. Of course the inaudible part may contain a predicate, but even without any such predicate the construction is fully comprehensible and syntactically complete. With this construction, Olli continues to query about the various costs of living outside the city. In line 7, Tor produces a postpositional phrase with a rising terminal contour marking it as a question. Syntactically it would be possible as an extension of the clause in lines 2 and 3; however, this does not make sense semantically. In 2 and 3, Tor is asking, “Do you have a villa there or what?”, to which he could add, “With the owner?”, but it would be odd semantically. Rather, Tor’s line 7 builds upon the exchange between Tor (lines 2 and 3) and Aarre (line 5; recall that Aarre’s line 4 is an answer to Olli’s question in line 1 rather than Tor’s). In line 5 Aarre gives an indirect response to Tor’s question: without denying that they have a villa, he states that they have three rooms. If Aarre’s line 5 is relevant as an answer to Tor’s question in lines 2 and 3, we can induce that the three rooms do not exhaust the villa, and that the owner inhabits the rest of the villa. It would be odd if he suggested that they have the villa together with the owner but also that they have the three rooms with the owner. Rather, what he is suggesting is that the owner is living there too. In line 7, Tor bases his conWrmation-seeking question on the whole exchange, not just on either part. My point with the example is that we do not have any clear syntactic, semantic or intonational grounds for assigning the postpositional phrase in line 7 to one clause rather than another in the context. In my view this is not an important question. To sum up, clausal syntactic relations are clearest in the clause core, whereas peripheral arguments often give information about circumstantial relations which are syntactically not tightly related to the clause core. It seems most proWtable to seek correspondences between intonation and clausal constituency by looking at how the constituents of the clause core are distributed across intonation units. Therefore, I have chosen not to include obliques when studying the distribution of clauses across intonation units, but instead, I only looked at clause cores.

29

30


3.2 Relations within the clause core: syntax and intonation In Helasvuo (1997) I looked at the relationship between clausal constituency and intonation using the same sets of data as here. The data showed that there is a clear tendency in spoken discourse for clause cores to be produced in a single IU. I take this to provide evidence for the view that clause cores emerge as units in conversational discourse. I also found that in cases where either of the core arguments or both were in a different IU from the predicate, there was an interesting similarity between the two languages: namely, in both languages, it was usually the object that was in a different IU from the predicate and the subject. Thus, the intonational phrasing suggested a subject–verb grouping on a par with verb–object grouping. In this section, I would like to relate these Wndings to the ways in which syntactic relations in the clause core can be exploited for interactional purposes. Consider example 2, which comes from a conversation between a couple, Joanne and Ken, and their friend Lenore. Before this segment, they have been talking about Ken’s illnesses when travelling in Central America. Lenore, the friend of the couple, asks: “And you both eat the same things?”, meaning “you eat the same things but only he gets sick?”. Joanne answers this by giving a lengthy list of things she eats, and after that Ken gets his turn. He starts by telling what he does not like to eat, and after that he gets into the things that he does eat. (1) Deadly Diseases 1 2 3 4 5 6 7

Ken:

Joanne: Ken:

I eat like, .. you know, .. vegetable=s, and, s- salads and, [(H) that’s the problem though, [m- f- meat from like little corner sta=nds,

In example 2, Ken produces a clause that extends over several IUs. IU 1 contains the subject and the verb. The object is produced as a three-part list, each part of which is produced as an IU of its own. One of the features identiWed with list-formation is parallelism in both syntax and prosody (see e.g. Erickson 1992, Sánchez Ayala forthcoming). Here, all the items of the list are mass nouns and each of them is delivered in a separate intonation unit. Ono and Thompson (1995) have observed in their conversational data that cases where a clause is produced across several intonation units are often attributable to the “one new idea per intonation unit”-principle (observed


inter alia by Chafe 1980, Prince 1981, Terken and Hirschberg 1994). This principle holds for example 2 also. In Section 4.2, we will come back to this example extending the discussion to the larger sequence, and I will point to several interactional factors which may have led to the fact that the NPs are delivered each in their own intonation units. In Helasvuo (1997) I concluded that if we look at grammatical constituency from the perspective of intonational grouping, the data suggest a subject-verb grouping on a par with a verb–object grouping. Thus, the intonational grouping found in the data is in contradiction with Cruttenden’s claim that IU boundaries should occur between “major constituent boundaries”, i.e. between the subject and the predicate (Cruttenden 1986: 130). Nor do the results lend support to the traditional tree structure analysis of constituency which takes as its starting point the assumption that sentences can be divided into two major parts, NP and VP, the latter of which subsumes the possible object-NP. The data suggest a more Xexible view of constituency which allows for Subject– Verb grouping on a par with the more familiar Verb–Object grouping (see Langacker 1997). Further evidence against the reality of verb phrase as an emergent syntactic category can be found in the syntactic organization of repair: Fox and Jasperson (1995) show that when making repair on the object constituent, speakers do not back up to the verb, but rather repair the object NP only. Fox and Jasperson (1995) conclude that speakers do not treat the verb and the object as forming a constituent when making repairs. Finnish has often been characterized as a language with free or “discourseconWgurational” word order (Vilkuna 1989). Support for this claim has been sought in the fact that syntactic permutations rarely yield ungrammatical orderings (Vilkuna 1989: 9f.). However, analysis of word order in conversational discourse has shown that the mutual ordering of subject and verb is very stable with the subject usually preceding the verb (Helasvuo in press). This tendency is very robust, especially in clauses with personal pronouns functioning as subjects. Therefore, it is not surprising to Wnd that subjects are more often presented in the same IU with the predicate than are objects. Thus, Wrst and second person subjects are almost invariably preverbal. Moreover, they most often form a continuous constituent with the verb. In the Finnish data, almost 90% of the 1st and 2nd person subjects were immediately preceding the verb; in the remaining 10% it was mostly small adverbs like nyt ‘now’, sit ‘then’ that were inserted between the subject and the verb, but in some cases, also something bulkier, such as a full object-NP. In English, word

31

32


order is of course more Wxed, and in my data, only adverbs like just or really go between the 1st or 2nd person subject and the verb. In conversational discourse, 1st and 2nd person pronouns Wgure signiWcantly in the subject role; for example in the Finnish data, almost 40% of all subjects were 1st or 2nd person pronouns (see Tao 1996: 123f. for similar Wndings on Mandarin). In Scheibman’s data from conversational English, 47% of all utterances had 1st or 2nd person subjects (Scheibman 2001: Table 3). Discourse studies have indicated that there is a tendency for 1st and 2nd person subjects to co-occur with mental verbs. Inter alia, Kang (1997) found that in her data from spoken French, over half of 1st person subjects occurred with mental predicates (see also Scheibman 2001 and Tao (1996: 152), who shows that pronominal subjects tend to co-occur with mental predicates in Mandarin). In the Finnish data studied here, about 35% of 1st and 2nd person subjects co-occur with mental or communicative verbs. In Scheibman’s data from conversational English, the percentage is around 25% (Scheibman 2001: Table 3). Furthermore, it is interesting to note that in English, 1st and 2nd person subjects form such regular and frequent combinations with certain mental verbs that they crystallize into one Wxed unit, such as I think, you know, I mean etc. These subject + verb combinations have been grammaticized as discourse particles and they are used to serve various discourse functions (for example, in the expression of epistemic stance; see Kärkkäinen 1998, Thompson and Mulac 1991, Östman 1981). Thus, we could conclude that the expression of subjective feelings and attitudes may play a far greater role in the ways language is used, and ultimately, in what grammatical structures are needed for. The data show clearly that 1st and 2nd person subjects have a tendency to appear in formulaic expressions, composed of the subject and the verb (see also Scheibman 2001). We could certainly say that these expressions represent an emergent syntactic construction (or an emergent syntactic gestalt, see Auer 1996), consisting of the 1st or 2nd person subject and the predicate. It is important to note that these constructions may or may not indicate syntactic completion and thus, possibility for turn-transition. For example, the I think construction may appear before or after the utterance it modiWes (for an indepth analysis of this construction, see Kärkkäinen 1998). According to Sacks, Schegloff and Jefferson (1974), speakers recognize certain stretches of talk as forming “turn-constructional units”, which form the basis for smooth turn-transition. Sacks, Schegloff and Jefferson refer to


these units in syntactic terms (they are sentences or smaller syntactic units), but they leave it open as to how syntactic closure should be deWned. Auer (1996) takes the Sacks, Schegloff and Jefferson paper as his starting point and looks at the syntax of turn transition in cases where a syntactically complete turn is continued after the completion point. Auer’s studies are based on an investigation of German conversation and they show that in German there is a “sentence brace” (Satzklammer) that deWnes, inter alia, the positioning of certain syntactic elements (such as inWnitives, verbal aYxes) that are completion-indicative (right brace). The construction is complete at that point. After the right brace there is a slot (Nachfeld) for a possible expansion of the prior clause. Auer views syntax as a “contextualization cue” for turn-taking and states (1996: 63): “The emergence of syntactic gestalts will (…) be monitored by conversationalists in order to locate possible syntactic completions which might qualify as turn-transition places”. My data from English and Finnish suggests that syntactic completion is not so clearly cued as in German, and therefore, speakers have to rely on multiple cueing for turn-transition (see Ford and Thompson 1996). In this section, we have seen that in both sets of data there is a clear tendency for clause cores to be presented in one IU. If an argument is presented in a different IU, it is more likely to be the object than the subject, indicating that the bond between the subject and the predicate may be stronger than that between the object and the predicate. I have also shown that 1st or 2nd person subjects Wgure signiWcantly in the subject role. They may form tight combinations with the predicate verb, especially with mental or communicative verbs. I have suggested that these combinations represent an emergent syntactic construction. 3.3 Clauses as emergent syntactic units in interaction In the discussion so far, I have tried to argue against a view that sees clauses as clear-cut units. Instead, I have proposed that clauses are emergent units where syntactic relations are tightest in the clause core. In this section, I will try to explicate what I mean with emergence of syntax. Looking at the emergence of syntax from a conversation analytic viewpoint, Goodwin (1979: 97f.) has proposed that “[s]entences emerge with conversation. However, in traditional linguistics it has been assumed that the analysis of sentences can be performed upon examples

33

34


isolated from such an interactive process. In opposition to such a view it will be argued here that sentences in natural conversation emerge as the products of a process of interaction between speaker and hearer and that they mutually construct the turn at talk.”2

Since Goodwin’s (1979) paper, there has arisen a new Weld of interactional linguistics (see e.g. papers in Ochs, Schegloff and Thompson 1996, CouperKuhlen and Selting 1996 and this volume) which looks at grammar in interaction, combining the insights of conversation analysis (see e.g. Sacks, Schegloff and Jefferson 1974, Goodwin 1981, Goodwin and Goodwin 1987, Jefferson 1990, Lerner 1991, Schegloff 1987 and 1996, Sorjonen 1997), discourse functional linguistics (see e.g. Chafe 1994, Du Bois 1987, Ford 1993, Fox 1987, Geluykens 1992, Tao 1996) and linguistic anthropology (Ochs 1988, Duranti 1994). I will try to illustrate what I mean by emergent syntactic constructions with the following example, which comes from a conversation between Tintti and her daughter Leena. In the sequence chosen for analysis they are co-constructing a clause complex, with Tintti providing the Wrst part, and Leena adding a but-clause building a contrast to what Tintti said. The conversation has concerned, among other things, ways of spending money. The daughter Leena has been reporting on a conversation she had with a friend about their respective mothers. The friend’s mother had always used money sparingly, but before her death, she had spent money on buying art and expensive clothing, just like Leena’s mother Tintti recently. Tintti starts out by comparing herself with the friend’s mother. (3) Tintti and Leena 1

Tintti:

2

Leena:

@[@ @] [ @

4 Tintti:

[ollu @>,] been [] I-CLT it be-1SG been hell sinful already I sure have been a hell of a sinner.


6

Leena:

mut silti pannu raha-t leveeks. but nevertheless put-PCP money-PL broad but nevertheless spent all that money.

Tintti starts building a contrast with the friend’s mother (line 1): although they have both spent money on buying art, she is different from the friend’s mother. Leena assures her mother that she hasn’t told the friend about this side (lines 2 and 3), and they both laugh at the joke. The mother continues with an upgrade of her assessment about herself (line 5). Leena comes in and co-constructs a compound sentence (line 6), building on the same contrast the mother has introduced (line 1), but, like her mother, upgrading it. The sequence is followed by mutual laughter. The mut (‘but’) clause (line 6) of the compound sentence is elliptical: there is neither a subject nor a Wnite verb, but only the participial form pannu ‘put (participle)’. In a compound sentence, the subject and the Wnite verb can be ellipted in the latter part of the compound, and they are understood to be the same as in the Wrst part. Leena makes excellent use of this syntactic resource: by way of co-constructing a compound sentence, she can avoid mentioning the subject. Had she expressed the subject, she would also have had to choose the subject person (1st or 2nd). The choice of 1st person subject would have resulted in direct reported thought. If the 2nd person subject had been chosen, the utterance would have sounded more like an accusation (but nevertheless, YOU have spent all your money!).3 The interactive resource of co-construction allows her to use the syntactic resources of compound sentences (the so-called “conjunction reduction”). With the help of these resources she manages to continue the talk in the form of mutual joking. In this section, I have shown how “sentences emerge with conversation”, as Goodwin (1979: 78) put it, as a mutual activity by the conversation participants. Each utterance shows an interpretation of prior talk, also in terms of the construction of syntactic structures.

4.

Noun phrases as syntactic and prosodic units

In this section I will Wrst discuss the syntax and prosody of NPs. I will consider NPs functioning as clausal arguments as well as NPs that function as free syntactic constructions, so-called free NPs. Secondly, I will discuss the ways in which conversation participants orient to noun phrases as they are being constructed in interaction. Special attention will be given to the treatment of

35

36


these constructions in the following interactional contexts: turn transition, repair sequences, and co-construction of syntactic units. 4.1 The syntax and prosody of noun phrases In this section, we will Wrst take a brief look at the syntactic characteristics of noun phrases in both Finnish and English. Then we will discuss noun phrases functioning as free syntactic constructions. In Finnish noun phrases, nearly all determiners and modiWers precede their heads and agree with them in case and number (i.e. the modiWers show the same case and number as the head). We could say that this is one of the deWning features of noun phrase formation in Finnish. This means that the determiners and modiWers show the case and number of the whole construction, and in terms of syntactic processing they give an idea of how the clause as a whole will be structured, even though they may not contain enough semantic and discourse referential information. Consider example 4 which comes from a conversation between two brothers. Mikko has been out on a Wshing trip, and he is telling his brother about the trip. (4) Brothers 1

2

Mikko:

mä ^tempasin I caught I caught a,

sielt there

semmose-n, a.kind.of-ACC

^puoltoist kilo-se-n .. (0.9) yli over one-and-ahalf kilo-ADJ-ACC over 1.5 kilo trout from there.

raudu-n. trout-ACC

We will concentrate on the noun phrase semmosen yli puoltoist kilosen raudun ‘a trout over 1.5 kilos’, functioning as the object of the clause. The determiner semmosen appears in the same IU as the predicate verb tempasin ‘caught’. It is a determiner that serves a classifying function (see Erringer 1996). It is followed by a relatively long pause and an adjectival phrase yli puoltoist kilosen ‘over 1.5 kilos (weighing)’ which modiWes the head noun raudun ‘trout-ACC’. In terms of processing, when encountering the determiner semmosen we know that it is going to be part of the object NP of the clause by virtue of its case marking (accusative). At the same time, it projects more, since it cannot form an NP by itself (on the notion of projectability, see Goodwin 1981). Thus we could say that noun phrases are highly projective in Finnish in that the form of the modiWers already reveals the case and the number of the


whole construction (cf. Himmelmann 1997). In contrast, English modiWers do not usually inXect for case or number. Thus, projectability is based on different kinds of facts, such as the order of elements within the noun phrase. Consider example 5 which comes from the same conversation as example 2. Here Ken is telling about the diagnosis his doctor has made and is quoting the doctor. We will focus on lines 9, 12, and 13, especially the object noun phrase some virus, co-constructed by Ken and Joanne. The NP consists of a quantiWer some and a head noun virus. I am arguing with this example that the quantiWer projects “more-to-come”. (5) Deadly Diseases 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Ken: Joanne: Ken:

Joanne: Ken:

Joanne: Ken: Joanne: Ken:

(H) u [=m, [(H) ? and il- -and I -and I think] -and he] doesn’t [2drink2 ]. [2and I think2], I think uh=, you- you [picked up] some, [(H)] … v- [2u=m2], [2Virus2]. virus, like, uh, he named like half a dozen viruses,

⇐

⇐

In the Wrst target line (9), Ken starts a transitive construction you- you picked up some after which there is a slight hesitation (line 11). “Some” is a quantiWer that projects more to come.4 At the point of hesitation Joanne comes in providing the noun virus (line 12), which is a potential head noun for the quantiWer some. Ken repeats it in line 13, and continues with his story, now in the form of an indirect quote. (This extract will be analyzed more closely in section 4.2) When noun phrases function as free constructions, they are most often produced in intonation units of their own. Interestingly enough, the functions of free NPs often relate to the organization of larger sequences rather than speciWc propositions. For example, when a free NP is used to offer an assessment, the “assessed” has usually not been mentioned as a referent in the discourse, but may be a longer sequence, such as a story (for more discussion, see Helasvuo in press). In conclusion, I have shown that the relatively tight internal syntactic structure of noun phrases provides conversation participants with means to

37

38


project syntactic completion, which is important when monitoring potential transition relevance places. We have also seen that they can be used to form units of their own, so-called free NPs, which are independent of clausal syntactic organization. 4.2 Functions of noun phrases in speciWc interactional tasks Noun phrases can be used to perform a variety of interactional tasks. In this section, I will discuss the strategic use of noun phrases in speciWc interactional tasks, namely in constructing lists, adding increments, and making repairs. I will try to show that conversation participants display strong orientation to noun phrases when performing these tasks. Ford, Fox and Thompson (forthcoming) discuss increments in relation to constituency. They distinguish two major types of increments, namely extensions and unattached noun phrases (free NPs in our terminology). As the name suggests, extensions function to extend a prior syntactic construction, and they may consist, for example, of a noun phrase, an adpositional phrase or an adverbial clause. Consider example 6, which comes from the same conversation as example 3, where the daughter Leena is reporting on a conversation with a friend in which they have talked about their mothers. The conversation had concerned ways of spending money, whether it is good to spend money on buying art and expensive clothing. The sequence chosen for this example precedes that in example 3. Leena starts by quoting what the friend said about her mother, who bought a mink coat for herself just before she died. (6) Tintti and Leena 1

2

3

4

Leena:

kuinka sääli oli että, how pity was that, What a pity it was that .. si-tä=, ..(0.5) min.. (0.6) että tota se ei=, that er she NEG+3SG that-PART minminkki-turkki-a mink-coat-PART ehtiny käyttää ku, ..(0.3) pari talvea. have.time use than couple winters .. that er she didn’t have time to use the.. min- mink coat for more than.. a couple of winters. ..(0.8)mut mä sanoin Santulle et, mikä onni että edes but I told Santtu that what luck that even ne. those But I told Santtu that it’s lucky that even those.


5

6

Tintti:

..(0.5) pari talvea. couple winters .. couple of winters. mm. mhm.

On line 4, Leena comes to the point of her story, namely her own view on spending money: according to her, it is okay to use money on expensive clothes, as long as you enjoy them, even if for a short time. At the end of line 4, there is a place for the recipient Tintti to come in and display recipiency. However, there is no response from her, and Leena adds an increment, the time adverbial pari talvea ‘a couple of winters’, thus succeeding in eliciting a response from Tintti (line 6). The increment is grammatically part of the prior clause. Ford, Fox and Thompson (2001) show that noun phrases can also be used to add increments that are grammatically independent and not part of the prior clause. These increments take the form of a free NP (or unattached NP in Ford, Fox and Thompson’s terminology). Jefferson (1990) discusses list-construction as an interactional task. She observes that lists often consist of three parts, and that participants orient to their three-partedness for example in sequential organization (e.g. when monitoring turn completion). She further notes that a list can be constructed by more than one speaker. Consider example 7, which comes from the same conversation as examples 2 and 5, where the participants have been discussing Ken’s illnesses. The example is rather long and contains several examples of three-part lists. (7) Deadly diseases 1 Joanne: 2 3 Lenore: 4 Joanne: 5 Ken: 6 7 Joanne: 8 9 Ken: 10 Joanne: 11 12 13 14 15 16 17

[(H) Twice] in a row, almost deadly diseas [2es2], [2And you2] both eat the same things? .. Yeah [3=3]. [3Yeah3], I just -In fact, [I eat] .. stuff that he doesn’t eat, [<X I don’t know X>]. that, he wouldn’t dare touch. you know, I eat like a ceviche, ⇐ and, and, all kinds of salads, ⇐ and,

39

40


18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64

Lenore: Joanne:

Ken: Joanne: Ken: Joanne: Ken: Lenore: Joanne: Ken: Joanne: Ken: Joanne: Ken:

Joanne: Ken:

Joanne: Ken: Joanne: Ken: Joanne: Ken:

Joanne: Ken:

Gua [ca]. ⇐ [(H)] … I- I eat all kinds of stuff like that. .. you know? [(H)] [(H)] The ce [2viche’s2] full of vegetables, ⇐ [2(Hx)2] ⇐ .. full of cut up tomatoes, and [and]. [I eat most] everything that you do. [2I don’t eat ceviche2], [2Raw fish2]. ⇐ [2Yeah, <X he does X>2]. because [3I don’t3] like the idea of eating [4raw4] food. [3raw3] -[4(H)4] [5but5], [5yeah5], I mean raw m- -like meat, (⇐) and fish, (⇐) and the idea bothers me, (⇐) but, (H) [= I eat every]thing else, [you know], I eat like, .. you know, .. vegetables, ⇐ and, s- salads and, ⇐ [(H) That’s the problem though. ⇐ [m- f- meat from like little corner stands, you’re a lot more, you’re a lot more delicate] -(H) , ⇐ [2in — th -[2(H) And- -%= yeah2]. in -in areas like that2], and like, not be able to s- to have the local foo=d. ⇐ That’s [part of the] whole [2thing. [(TSK) (H)] [2(H) It’s delicious, It’s cheap2], (H) So I eat the local2] food, and get deathly ill.

Example 7 has several examples of lists. First, Joanne lists (starting in line 13) I eat like a ceviche, and, and, all kinds of salads, and. In line 18, Lenore mentions something in Spanish, guaca (for guacamole), possibly offering it as a third item on the list. There seems to be some problem with Lenore’s contribution as


Joanne hesitates (line 19). I argue that Joanne’s utterance in line 19 is structurally not suitable as a third item to the list she has created, since it is not of the right form: given the fact that the Wrst two items are noun phrases, one would expect a third item of the same kind. Joanne’s utterance, however, takes the form of a clause. There is a second list in lines 23, 25 and 29, co-constructed by Joanne and Lenore. In lines 37–38, Ken starts to list items he is bothered about (meat, and Wsh, lines 37 and 38). In line 39, he uses the coordinating conjunction and, which projects an item that is parallel with the Wrst two. However, the third NP (the idea, line 39) is not suitable as a third item to the list that is evolving, because it is structurally and semantically so different from the Wrst two NPs: the Wrst two are uncountable nouns, whereas the third one is a count noun with deWnite reference. Therefore, lines 37–39 do not form a list. The third list is by Ken in lines 43–49. All these lists consist of noun phrases. The fourth list is the only one that is constructed with verbs (lines 52–59, who wants to go and travel and not be able to have the local food). I would like to focus on the list produced by Ken in lines 43–49. One of the properties of lists observed by Jefferson was that speakers orient to their threepartedness so that they know to expect a third part. It can also be noted that list items are usually structurally similar; for example, they may all be inWnitives (go, travel and not be able to have) or noun phrases. This means that in line 47 when Ken produces the latter and, a noun phrase is to be expected. Thus, Joanne’s clause That’s the problem though (line 48) is not constructed as a third item to Ken’s list, because it is not structurally compatible with the rest of the list. Noun phrases provide perhaps the main grammatical resource for constructing lists. But lists may also consist of inWnitival constructions, clauses or larger sequences. For example, Ken and Joanne produce a three-part-list to answer Lenore’s question in line 3 And you both eat the same things?: They Wrst tell what Joanne eats, then what Ken doesn’t like to eat, and Wnally what Ken does eat. Of course lists that consist of larger sequences are not so readily identiWable. Speakers may use some organizational device to display that they have come to a completion. In our example, Ken’s lines 63–64 go back to the topic which this sequence started with, namely deadly diseases, thus closing the answering sequence. During the sequence where the participants talk about the food they eat, they are in a list mode: Wrst Joanne lists what she eats, then Ken tells what he doesn’t like to eat and Wnally, what he does eat. These lists are all produced by giving one new item per intonation unit (on stylized intonation patterns for list-formation in Spanish and English conversation, see Sánchez Ayala forth-

41

42


coming). The intonation and rhythmic patterning used here is congruent with the “one new idea per intonation unit”-principle suggested by Chafe (1980; see also Ono and Thompson 1995). Ono and Thompson (1995) observe several interactional factors which may result in the production of one clause in several intonation units. Interestingly enough, both Joanne’s and Ken’s answers to Lenore’s question take the same structural and intonational form: that of listing the items in separate intonation units. We could say that the participants are orienting to the list-mode and they display this orientation by choosing the same structural and intonational patterns — that of giving one NP per intonation unit — for their contributions. Finally, I would like to discuss the use of noun phrases in co-constructions. In his studies on co-constructions, Lerner (e.g. 1991, 1996) has shown that they typically occur in contexts where there is a complex compound construction; for example, an if-then clausal compound. This is also true of my data. Regarding noun phrases in the data, it is not very often that one Wnds coconstructions where a participant offers a noun phrase to complete a construction some other participant has begun. In my data, there are examples of this in constructing characterizing clauses (see Goodwin and Goodwin 1987). Before the sequence given in example 8, Tintti, the mother, has been telling her daughter about her Christmas preparations. She has told her that she has bought a special kind of liqueur for Christmas, and the daughter has displayed amazement because it is still three weeks to Christmas at the time of the recording. At this point the mother adds that she has bought yet another bottle (example 8, line 1). (8) Tintti and Leena 1

Tintti:

sit mää ostin, m- .. vadelma-mansikka-liköörii. then I bought raspberry-strawberry-liqueur Then I bought str — .. raspberry-strawberry liqueur.

2

Leena:

aha. Uhhuh.

3

..(0.5)

jaa ^jaa . I see, I see.

4

..(0.7)

^no niin. That’s it.

5

Tintti:

ne on, they are They are

⇐

6

Leena:

^suur-hankinna-t. grand-purchase-PL large scale purchases.

⇐


7

Tintti:

ne on ^sievät tuol kirja-hylly-n pääl. they are pretty-PL there book-shelf-GEN on They are pretty up on the bookshelf.

8

Leena:

just. Right.

9

Tintti:

@ @ [@]

10

Leena:

[@]

In line 1, the mother informs the daughter that she has bought yet another bottle, and in lines 2–4 Leena responds to her mother’s informing. There are relatively long pauses that leave the mother room to add new items to her list of kinds of liqueur she has bought, but since nothing is added, Leena closes the sequence with no niin ‘That’s it’ (line 4). Tintti starts off with a characterizing clause ne on ‘they are’ (line 5), which Leena completes with an NP, suurhankinnat ‘large scale purchases’ (line 6). The characterizing NP has an ironic Xavor: Leena seems to be teasing her mother about making such a fuss about her Christmas preparations. Tintti does not accept the contribution Leena has made, and recycles the predicate nominal construction from the beginning, replacing the characterization Leena has given with a new characterizing phrase sievät ‘pretty’ (line 7). The sequence ends with mutual laughter (lines 9–10). Thus, the speakers are co-constructing a characterizing clause, where the Wrst speaker Tintti provides the beginning of the construction (the item to be characterized + the copula) and Leena continues with the characterization. However, the co-construction is repaired by Tintti, who recycles the whole construction (line 7). Predicate nominal constructions thus offer a place for the co-participant to come in and co-construct a characterization. However, co-constructions of this kind are not very frequent in the present data; more frequent are coconstructions of complex constructions combining two complex parts (such as an if-clause and a then-clause). I further assume it is quite rare to Wnd a coconstruction within a noun phrase (see, however, example 2). To summarize, conversation participants show orientation to noun phrases when performing certain interactional tasks in conversation, such as adding increments, constructing lists, and making repair. I hope to have shown that participants display orientation both towards the syntactic structure of the noun phrase and towards the intonation.

43

44


5.

Conclusions

This paper has been concerned with the nature of clauses and NPs as emergent syntactic and prosodic units in conversation. We have seen that the idea of a clause as a clear-cut syntactic unit is problematic. Instead, it is the clause core with the predicate and its core arguments that emerges most clearly in discourse. In both the English and the Finnish data, there was a clear tendency for clause cores to be presented in one intonation unit. I take this to provide evidence for clauses as emergent constructions consisting of the predicate and its core arguments. They may be accompanied by optional arguments, which may or may not appear in the same IU as the core. If a core argument was presented in a different IU, however, it was more often the object than the subject. This Wnding poses problems for the traditional view of constituency but can easily be accommodated in a more Xexible view of constituency which allows for Subject–Verb grouping on a par with the more familiar Verb–Object grouping. I have also shown that 1st and 2nd person subjects Wgure signiWcantly in the subject role. They may form tight combinations with the predicate verb, especially with mental and communicative verbs, and the combinations may even grammaticize into one Wxed unit functioning as a discourse particle (e.g. I think, you know). We have also seen that noun phrases show a relatively tight internal syntactic structure, which provides conversational participants with a means to project syntactic structuring and also, to project syntactic completion, which is important for smooth speaker transition. I have shown that conversational participants show orientation to noun phrases when performing certain interactional tasks in conversation, such as adding increments, constructing lists, and making repair. Thus, we could say that noun phrases are a syntactic resource that can be exploited for speciWc interactional purposes. In this paper I have tried to argue for a view that sees the construction of syntactic structures, such as clauses and NPs, as one kind of conversational activity. Syntactic structures emerge in conversation as a mutual activity by the conversation participants. Conversation analytic work has shown how each utterance shows an interpretation of prior talk, for example, of the sequential organization of the talk. The present study emphasizes that it also shows an interpretation of how the syntactic structures have been construed.


Notes 1. See appendix for transcription symbols and morphological glosses. The third line in the Finnish examples presents a free translation. Note that the free translation should not be read as a “translated transcript”, but rather, as a free translation of the Finnish original. This means that for the exact placement of such interactional and intonational phenomena as overlap, stress, intonational continuity etc. readers are referred to the Finnish transcript (1st line) and its gloss (2nd line). Furthermore, the translations of the response particles are meant to offer only a rough idea of the interactional impact of these responses; by no means are the free translations meant to offer analyses of the particles. For a thorough discussion of response tokens in Finnish conversation, see Sorjonen (1997). 2. It seems that Goodwin is using the term “sentence” in approximately the same meaning as we have been using the term “clause”. 3. In English, the 2nd person can be interpreted as generic, without an indication of accusation. In Finnish, the 2nd person is not used in generic meaning as often as in English. 4. Both some and semmonen could be used alone, in anaphoric relationship with some NP in the preceding talk. But in ex. 4 and ex. 5, there is nothing in the co-text that would create such an anaphoric relationship. Furthermore, when these elements are used as head nouns, they usually carry either primary or secondary stress. This is not the case in either example here.


“On the prosody and syntax of turn-continuations”. In Prosody in Conversation, E. Couper-Kuhlen and M. Selting (eds), 57–100. Cambridge: Cambridge University Press.

Chafe, W. L. 1980 “The deployment of consciousness in the production of a narrative”. In The Pear Stories: Cognitive, cultural and linguistic aspects of narrative production, W. Chafe (ed.), 9–50. Norwood, N. J.: Ablex. 1994 Discourse, Consciousness, and Time: The Xow and displacement of conscious experience in speaking and writing. Chicago: University of Chicago Press. Couper-Kuhlen, E. and Selting, M. 1996 “Towards an interactional perspective on prosody and a prosodic perspective on interaction”. In Prosody in Conversation, E. Couper-Kuhlen and M. Selting (eds), 11–56. Cambridge: Cambridge University Press. Cruttenden, A. 1986 Intonation. Cambridge: Cambridge University Press.

45

46


Cumming, S. 1984 “The sentence in Chinese”. Studies in Language 8 (3): 365–395. Du Bois, J. W. 1987 “The discourse basis of ergativity”. Language 63: 805–855. Du Bois, J. W., Schuetze-Coburn, S., Paolino, D. and Cumming, S. 1993 “Outline of Discourse Transcription”. In Talking Data: Transcription and coding in discourse research, J. A. Edwards and M. D. Lampert (eds), 45–89. Hillsdale, N. J.: Lawrence Erlbaum. Duranti, A. 1994 From Grammar to Politics: Linguistic anthropology in a Western Samoan village. Los Angeles: University of California Press. Durie, M. 1994 “A case study of pragmatic linking”. Text 14 (4): 495–529. Erickson, F. 1992 “They know all the lines: Rhythmic organization and contextualization in a conversational listing routine”. In The Contextualization of Language, P. Auer and A. di Luzio (eds), 365–397. Amsterdam: Benjamins. Erringer, A. 1996 “The functions of demonstrative adjectives semmoinen, tämmöinen and tuommoinen in Finnish conversations”. MA Thesis, Department of linguistics, University of Colorado, Boulder. Ford, C. 1993 Grammar in Interaction: Adverbial clauses in American English conversations. Cambridge: Cambridge University Press. Ford, C., Fox, B. and Thompson, S. A. 2001 “Constituency and the grammar of turn increments”. In The Language of Turn and Sequence, C. Ford, B. Fox and S. A. Thompson (eds). Oxford: Oxford University Press. Ford, C. and Thompson, S. A. 1996 “Interactional units in conversation: Syntactic, intonational, and pragmatic resources for the management of turns”. In Interaction and Grammar, E. Ochs, E. Schegloff, S. A. Thompson (eds), 134–184. Cambridge: Cambridge University Press. Fox, B. 1987 Anaphora and the Structure of Discourse. Cambridge: Cambridge University Press. Fox, B. and Jasperson, R. 1995 “A syntactic exploration of repair in English conversation”. In Alternative Linguistics: Descriptive and theoretical modes, P. W. Davis (ed.), 77– 134. Amsterdam: Benjamins. Fox, B., Hayashi, M. and Jasperson, R. 1996 “Resources and repair: a cross-linguistic study of syntax and repair”. In Interaction and Grammar, E. Ochs, E. Schegloff and S. A. Thompson (eds), 185–237. Cambridge: Cambridge University Press.


Geluykens, R. 1992 From Discourse Process to Grammatical Construction: On left-dislocation in English. Amsterdam: Benjamins. Goodwin, C. 1979 “The interactive construction of a sentence in natural conversation”. In Everyday Language: Studies in ethnomethodology, G. Psathas (ed.), 97– 121. New York: Lawrence Erlbaum. 1981 Conversational Organization: The interaction between speakers and hearers. New York: Academic Press. Goodwin, C. and Goodwin, M. H. 1987 “Concurrent operations on talk: Notes on the interactive organization of assessments”. IPRA Papers in Pragmatics 1: 1–54. Helasvuo, M.-L. 1991 “Syntaktiset rakenteet kertomuksen jäsentiminä” [English summary: Syntactic structures as markers of story structure]. Virittäjä: 33–46. 1997 “What can intonation tell us about syntax?”. Paper given at the symposium on Constituency in Discourse. May, 1997. University of California, Santa Barbara. in press Syntax in the Making: The emergence of syntactic units in Finnish conversation. Amsterdam, Benjamins. Himmelmann, N. 1997 “Projective constituent structures”. Paper given at the symposium on Constituency and Discourse. May, 1997. University of California, Santa Barbara. Hopper, P. J. 1987 “Emergent grammar”. BLS 13: 139–157. Berkeley: Berkeley Linguistic Society. 1988 “Emergent Grammar and the a priori grammar postulate”. In Linguistics in Context, D. Tannen (ed.), 117–134. Norwood, N. J.: Ablex. Jefferson, G. 1990 “List-construction as task and resource”. In Interaction Competence, G. Psathas (ed.), 63–92. Washington, D. C.: International Institute for Ethnomethodology and Conversation Analysis & University of America Press. Kang, A. 1997 “Interpreting on: A discourse-pragmatic perspective on the use of a French ‘indeWnite’ pronoun in personal narratives”. Manuscript. University of California, Santa Barbara. Kärkkäinen, E. 1998 “The marking and interactional functions of epistemic stance in American English conversational discourse”. Ph.D. Dissertation. Department of Linguistics, University of California. Langacker, R. 1997 “Constituency, dependency, and conceptual grouping”. Cognitive Linguistics 8–1: 1–32.

47

48


Lerner, G. 1991 1996

Ochs, E. 1988

“On the syntax of sentences in progress”. Language in Society, 20 (3): 441–458. “On the “semi-permeable” character of grammatical units in conversation: conditional entry into the turn space of another speaker”. In Interaction and Grammar, E. Ochs, E. Schegloff and S. A. Thompson (eds), 238–276. Cambridge: Cambridge University Press.

Culture and Language Development: Language acquisition and language socialization in a Samoan village. Cambridge: Cambridge University Press. Ochs, E., Schegloff, E. and Thompson, S. A. (eds) 1996 Interaction and Grammar. Cambridge: Cambridge University Press. Ono, T. and Thompson, S. A. 1994 “Unattached NPs in English Conversation”. BLS 20: 402–419. Berkeley: Berkeley Linguistic Society. 1995 “What can conversation tell us about syntax?” In Alternative Linguistics: Descriptive and theoretical modes, P. Davis (ed.), 213–271. Amsterdam: Benjamins. Östman, J.-O. 1981 You know: A discourse functional approach. Amsterdam, Benjamins. Prince, E. F. 1981 “Toward a taxonomy of given-new information”. In Radical Pragmatics, P. Cole (ed.), 223–255. New York: Academic Press. Sacks, H., Schegloff, E. and Jefferson, G. 1974 “A simplest systematics for the organization of turn-taking for conversation”. Language 50: 696–735. Sánchez Ayala, I. forthcoming “Constructions as resources for interaction: Lists in English and Spanish conversation”. In Discourse across Cultures, C. L. Moder (ed.). Amsterdam: Benjamins. Schegloff, E. 1987 “Recycled turn beginnings: A precise repair mechanism in conversation’s turn-taking organization”. In Talk and Social Organisation, G. Button and J. R. E. Lee (eds), 70–85. Clevedon: Multilingual Matters. 1996 “Turn organization: one intersection of grammar and interaction”. In Interaction and Grammar, E. Ochs, E. Schegloff and S. A. Thompson (eds), 52–133. Cambridge: Cambridge University Press. Scheibman, J. 2001 “Local patterns of subjectivity in person and verb type in American English conversation”. In Frequency Effects and Emergent Grammar, J. Bybee and P. Hopper (eds), Amsterdam: Benjamins. Selting, M. 1996 “On the interplay of syntax and prosody in the constitution of turnconstructional units and turns in conversation”. Pragmatics 6 (3): 357– 388.


Sorjonen, M.-L. 1997 “Recipient Activities: Particles nii(n) and joo as responses in Finnish conversations”. Ph.D. Dissertation. Departments of Applied Linguistics and Sociology, University of California, Los Angeles. Tao, H. 1996 Units in Mandarin Conversation. Amsterdam: Benjamins. Thompson, S. A. and Mulac, A. 1991 “A quantitative perspective on the grammaticization of epistemic parentheticals in English”. In Approaches to Grammaticalization, E. C. Traugott and B. Heine (eds), 313–329. Amsterdam: Benjamins. Terken, J. and Hirschberg, J. 1994 “Deaccentuation of words representing ‘given’ information: Effects of persistence of grammatical function and surface position”. Language and Speech 37 (2): 125–145. Vilkuna, M. 1989 Free Word Order in Finnish: Its syntax and discourse functions. Finnish Literature Society, Helsinki.

Appendix: Glossing and transcription conventions 1. Glossing The morphemes have been indicated with dashes; in the case of fused morphemes, there is a plus sign in the glossing. Nominal markings: ACC - accusative; ADE - adessive; ALL - allative; GEN genitive; ILL - illative; PART - partitive; TRA - translative; PL - plural; 1SG Wrst person singular pronoun (likewise 2nd and 3rd); 1PL - Wrst person plural pronoun Verbal markings: 1SG - Wrst person singular person marking (likewise 2nd and 3rd); 1PL - Wrst person plural person marking; PST - past tense; CND conditional mood; IMP - imperative mood; INF - inWnitive; PCP - participle; NEG - negation verb Other: Q - question clitic; CLT - other clitics; PX - possessive suYx

2.

Transcription conventions

2.1. Transitional continuity , .

continuing intonation (slightly falling) terminal intonation (falling)

49

50


? --

appeal (slightly rising) truncated intonation

2.2. Accent and lengthening ^ =

primary accent lengthening of the preceding sound

2.3. Quality 〈P P〉 〈HI HI〉 〈LO LO〉 〈% %〉 〈@ @〉

piano: soft higher pitch level lower pitch level creaky voice laughing voice

2.4. Vocal noises @ (THROAT) (TSK) (H)

laughter someone clearing their throat click of the tongue inbreath

2.5. Other [22] 〈X X〉 ..() -

speech overlap (brackets indexed for clarity) uncertain hearing pause (length indicated in parentheses) truncated word

At the intersection of turn and sequence Negation and what comes next* Cecilia E. Ford University of Wisconsin-Madison

This study examines the contextual nature of turn projection by focusing on the intersection of turn construction and the collaborative emergence of interactional sequences. SpeciWcally, observations are made regarding turn elaborations following negative turn components. Such turn formats are related to an attested pattern whereby lack of elaboration after a negation is treated as problematic by interactants, leading to expanded talk across speakers in pursuit of such elaboration. The study supports the notion that turns, as fundamental linguistic structures in conversation, are constructed with reference to the activity sequences in which they occur, and that, in addition to the well-attested practices of grammar and prosody in projection, action, in sequential context, is also at work in the unfolding trajectories of turns.

Overview In this study, I oVer some observations of an order in interaction involving the intersecting systems of turn construction and the expansion of turns into interactional sequences. My observations center on turn elaborations following negations, and the related pattern whereby lack of elaboration after a negation is treated as problematic by interactants, leading to expanded talk across speakers. I will Wrst give a brief overview of the study. In a corpus of American English conversations, I have found that turns initiated with negation — speciWcally, negation that expresses disaYliation or disagreement with prior talk — regularly include continuation beyond the turn-constructional unit containing the negation; that is, negation is followed by elaboration of some sort. For example, in (1) Jake is responding to another participant’s implication that everyone in the room needs a breath freshener. Jake’s Wrst clause contains negation, and in his continuation he oVers an

52

Cecilia E. Ford

explanation: (1) Jake:

Hey. (.) You don’t have to worry about me, I had Listerine this morning.

Talk following negations regularly provides elaboration — explanations or alternatives, and sometimes explicit corrections. In (2) the speaker is correcting another speaker’s talk: (2)

No, You- (.) He said twenty centimeters.

Here are several further cases of negations followed by elaborations: (3) (4)

Nah:, I won’t take- I don’t take second party checks. I-I don’t want to make anything definite. Because I-yi hknow I jis::t thinking tiday ((further talk by same speaker follows)) Not me:, hhuh uh-hhuh .hhh I go in late everyday. No no no no no:, This is water. No. She started with-=no::, she started with (.) an oval.

(5) (6) (7)

In the present corpus, when negation expresses aYliation or agreement, it is regularly oVered without elaboration. However, if a negation is disaYliative but is not followed by relevant elaboration, participants respond by prompting further talk, thus treating unelaborated disaYliative negation as problematic. In (8), Rachel has referred to people in some photographs as relatives of Terry. Terry denies this: (8) Terry :

⇐ unelaborated negation

Terry :

We don’t know these people. (.) ↑>or< these people. You d↑on’t? (.) No.

Abbie:

(.) Do you know the one just below that one though?

Terry : Rachel:

⇐ extension of negation ⇐ continued unelaborated negation

Here then, what could have been done in a single turn as negation plus elaboration is expanded into a sequence in pursuit of elaboration (this instance is discussed in Section 6). In such cases, the relevant elaboration is worked out in an interactional sequence, produced across speakers. My purpose in sharing these cases brieXy here has been to orient the reader to the types of turns and sequences I will be examining. In the body of this paper, I treat in more detail cases involving the work of negation in its sequential contexts and the ways that further talk is treated as relevant. I relate turns

At the intersection of turn and sequence

— as fundamental linguistic structures in conversation — to the activity sequences in which they occur. And I examine the contextual nature of turn projection as it intersects with the expansion of activities into interactional sequences.

Background If one understands turn taking as involving a dynamic, adaptable, and contingent set of resources and practices used by participants in talk-in-interaction, the most convincing accounts of the system are those based on Sacks, SchegloV and JeVerson (1974). What has grown out of that original account are a number of studies reWning our understanding of the multiple inXuences of prosody, bodily movement, gaze, and sequential placement on the articulation of current and next speaker turns (Couper-Kuhlen 1993, Ford and Thompson 1996; Ford, Fox and Thompson 1996; Goodwin 1981; Goodwin and Goodwin 1987, 1992; Local and Kelly 1986; SchegloV 1996a, 1998; Selting 1996, 2000; Wells and MacFarlane 1998, inter alia). By these accounts, interactants treat turns as consisting, at least initially, of one unit of language structure, conventionally referred to as a turn-constructional unit (TCU). Each speaker is allocated one unit — a combination of grammar, prosody, and action — for each turn at talk. Interactants achieve smooth turn transitions, with minimal gap or overlap, through attention to the trajectory of each turn-constructional unit, a trajectory formed through the grammar, prosody, and action associated with that unit. That is, language, as embodied in particular sequences of interaction and through the voices of particular speakers, allows for the anticipation of possible completion before it arrives. Language thus supports a system of smooth turn transition, as well as a system in which gaps and overlaps can be deployed for strategic and meaningful interactional purposes. In the study of conversation, the term projection has been used to describe the manner in which interactants can anticipate the ends of turns before their actual arrival. Projection is also at play when speakers produce longer turns, made up of multiple prosodic and grammatical units, such as stories and explanations. In these cases, a preface or some other projecting device provides a cue as to what the end of the longer unit will be, and the recipients of the longer span of talk participate in a modiWed manner until the primary speaker produces what is recognizable as possible completion, based on the original preface that led into the longer turn.

53

54

Cecilia E. Ford

If speakers orient to the relevance of speaker change at the end of a TCU, then same speaker continuation past such a point represents an interactional achievement. Multi-unit turns have been described throughout the literature on conversation. It is regularly the case, for example, that a same speaker may continue speaking when a potential next speaker does not take the opportunity (Sacks et al. 1974, Sacks 1987, Ford 1993, SchegloV 1996a, Selting 2000); and a speaker may do special work to project a multi-unit turn, as in the case of stories — a common multi-unit accomplishment (Sacks 1974, JeVerson 1978). In the case of shorter multi-unit turns, prosodic devices such as “rush throughs” and accent suppression have been documented (SchegloV 1982, 1987, 1996a; Selting 1996). Lerner (1991, 1996), JeVerson (1990), Atkinson (1984) have looked at projectable shorter structures with internal multi-unit turn organizations. These scholars have shown that speakers regularly produce, and recipients regularly orient to, multi-unit turns involving such rhetorical formats as contrasts and lists. These turns involve internal points of possible grammatical and prosodic completion, but they also have features which allow for the projection of more-to-come based on recurrent patterns of unit combination. In the present study, as outlined in Section 1, I focus on a particular type of short multi-unit turn involving negation followed by elaboration. This pattern is not only positively evident and recurrent in the American English conversations, but it is also, and importantly, oriented to as problematic when its Wrst unit part is produced without the follow-up. My work in the present study is guided by the convergent goals of discourse-functional linguistics and conversation analysis, as both these enterprises aim to account for the basic, recurrent structures and functions of language in use. In a recent review of research in grammar and discourse, Cumming and Ono (1997) encapsulate the discourse-functional approach in the following way: Discourse-functional grammarians view discourse — that is, spoken, signed or written language used by people to communicate in natural settings — as the primary locus for the grammars of the world’s languages, not only as the place where grammar is manifested in use, but also as the source from which grammar is formed or “emerges”. (Cumming and Ono 1997: 112)

While grammar has traditionally been conceived of as located within sentences, clauses, and phrases, in this study I consider the turn as a basic linguistic structure as well. The turn is, after all, a unit that is treated as real by interac-


tants. In a recent essay, SchegloV (1996a) points to the potential for fruitful connections between the functional linguistic and conversation analytic enterprises as regards practices of turn construction: “[T]urns-at-talk,” he reminds us, “are the key proximate organizational niche into which bursts of language are introduced, and to which they may be expected to be adapted”(1996a: 53). In the current study, my focus is on the intersection of turn and sequence structuring, and the emergence of real utterances in real time in real interactions. Let me be speciWc about the simultaneously-operating and interrelated aspects of interactional language use which I assume to be operating: Following Sacks et al. (1974), I assume that there are turns, units of structure, prosody, and action, embodied in utterances contributed by single speakers. Turns have the property of projectability — that is, they have structures the courses of which can be roughly predicted before a unit is completed. Based on projectability, a recipient can predict at what structural, prosodic, and pragmatic (action embodying) place in a turn’s unfolding the turn is likely to end. The projection of a turn’s trajectory can only be determined within particular sequences of actions — turns are oVered at particular slots in particular interactional sequences. Projection allows for the precise placement of recipient next turns at the ends of the current speakers’ turns — bearing in mind that “precise placement” can mean in overlap with, at the very end of, or after some gap following. I emphasize these start up options because they are treated as meaningful in relation to the projected, anticipated end points of particular turns. Interrelated with the structuring of turns, I assume the existence of what have been more traditionally considered as basic structures of language. These are practices, resources, predictable collocations, conventionally described by such terms as word, phrase, clause, and sentence. While these structures are variable, much of such structural material is conventionalized enough that it can provide usable resources for constructing projectable turns at talk.1 Finally, I assume that there are recognizable sequences of actions that interactants orient to and produce together. These patterns are normative, but they are not absolute nor predetermined, that is, they are contingent. In concert with conversation analytic (CA) terminology, I refer to these as sequence structures. Common examples would be greetings following greetings, answers following questions, as well as stories, explanations or other longer spans of single-speaker talk showing interactional structures that allow for and are integrated with opportunities for recipient contributions.

55

56

Cecilia E. Ford

At the intersection of these practices of turn and sequence structuring, I am interested in what sorts of units, which end in points of possible completion, are recurrently followed by elaboration, within what may be interpreted as a single speaker’s turn. The cases of continuation that I am concerned with involve turns in which an initial unit has the normative features associated with turn completion: 1. Grammatical completion. 2. Prosodic unity and possible prosodic completion. Following Selting (1996), a coherent contour without any turn holding device deployed after the Wnal accented syllable is considered to be potentially complete in its prosodic delivery.2 3. Action completion. Within the context of a developing sequence of interaction, the unit could constitute a complete contribution with sequential implicativeness, i.e., straight-forward options for next speaker response. Units with these features have been called turn-constructional units (TCUs, Sacks et al. 1974; see also Selting 2000) or complex turn-constructional units (Ford and Thompson 1996). They can be words, phrases, clauses or clause combinations, and their points of potential completion are projectable before they arrive. My interest is in projectability that goes beyond what have traditionally been understood as units of grammar and prosody; I am exploring a potential grammar of turn formats. In the treatment of projection in CA literature, what has generally been examined is the constitution of single turn constructional units, compound turn constructional units (involving such structures as adverbial clauses and quotative frames), concurrent operations on turns-inprogress, and the work done expressly to project talk past points of possible turn completion. More extended turns have been shown to involve special projection strategies. To project talk past points of possible completion, speakers negotiate a temporary transformation of the turn taking system such that one participant becomes the primary speaker while the others take the roles of recipients of the longer series of turn-constructional units. The possible end of such a stretch of talk is projected not based on language structure and prosody alone, but crucially on what might constitute the end of this longer unit (Houtkoop and Mazeland 1985). Thus, a story, as in example (9), from Sacks (1992b), may be projected to involve a car accident:


(9) A:

B: A:

Say did you see anything in the paper last night or hear anything on the local radio, Ruth Henderson and I drove down to Ventura yesterday, Mm hm And on the way home we saw the:; most gosh awful wreck. (1992: 9)

As Sacks puts it, B, as the story recipient, understands “what she should do from what [Sacks proposes to call] the ‘story preface’” (1992b: 9). Story prefaces allow projection of a longer spate of single speaker talk and they “inform their hearer what the hearer should do when the story is over” (1992b: 9); that is, they let the recipient know what sort of response might be expected at story completion. To determine when the extended telling might be completed, B in (9) might monitor for the “awful wreck.” When that point is reached, a particular kind of uptake will be relevant (such as, “Oh, how awful”). The normal projection of completion at the end of each TCU is, thus, modiWed for the length of the story. Speakers may also expand their speaking slots after they have projected further talk through preliminary bids such as “May I ask you a question?”, as discussed in SchegloV (1980). In such cases, a speaker projects an action such as a question, but then produces something other than the projected question, talk that is preliminary to the actual question. The speaker is thus able to produce an extended span of talk as prefatory to the action that he or she originally prefaced. In the present study I also examine the projection of talk beyond single TCUs; however, I look, in particular, at the possibility that there are shorter multi-TCU turns comprised of something akin to “rhetorical combinations” or “pragmatic units”. These are shorter than typical stories (or other longer single speaker dominated turns) but longer than the standard words, phrases, and clauses most commonly associated with TCUs. The ends of shorter multiunit turns are more local; that is, these turns do not involve long spans of talk in which one speaker takes the role of primary speaker. Instead, with these shorter turns, the current speaker produces an initial action which implicitly projects at least one further unit of elaboration before completion. Shorter multi-unit turns are distinct from longer extended spans of talk in that these shorter turns do not have prefaces and thus do not begin with explicit cues as to what their completion will entail. In this sense, they have more in common with grammatical units — projecting only a small bit ahead. However, as mentioned above, they are more properly associated with rhetorical combinations of

57

58

Cecilia E. Ford

propositions, in that they are not characterized by the tighter internal structure we generally associate with phrases and clauses.

The collected cases As outlined above, the turns which are the focus of this paper involve talk past points of language-structural and prosodic possible completion, without typical longer agendas (i.e., not stories or other prefaced talk). However, more does indeed come, and the more that comes seems familiar and predictable (to me as an analyst). The projection of further talk in these cases is not based on what we usually think of as language structure, nor is it clearly projected through prosody, but rather it is based on recurrent action combinations which follow from the relationship between the action implemented through the negation and the talk which precedes it. This, then, constitutes an intermediate kind of projection, a projection based on recurrent turn formats or recurrently combined action types. I will provisionally call this intermediate kind of projection pragmatic projection (Ford, Fox and Thompson 1996; Steensig 1998). Pragmatic projection will refer to the projection of talk past an initial TCU when that projection is not done through language structure (e.g., an initial adverbial clause), is not done through prosodic turn-holding devices, and is not done through an initial TCU explicitly occupied with projection (e.g., story prefacing). The pragmatically projectable shorter multi-unit turn type which I document in this study has a negatively framed TCU and goes on to include a next unit which elaborates on the work of the negation. The negation in the Wrst TCU may be within a clause, a phrase, or it may be done through a single word (most commonly the particle no or variants such as naw and nope), as long as that unit could be complete in its sequential context. For example, each of the following are negatively framed TCUs in the collection: Clause: Phrase: Word:

We don’t know these people. Absolutely not a drop. No.

In collecting cases, I have allowed for pauses, laughter, and even continuers between the negation and the next TCU. This is because my interest is in the actions that are produced in succession by the same speaker. I would suggest that there is a continuum by which some cases involve more “pre-planning”


and others more inter-speaker negotiation. Nevertheless, they all represent a combination of actions, a second oVered as a relevant expansion of the Wrst. In (10) for example, Abbie has been reporting on a search for a particular birdfeeder at local stores. Rachel asks if Abbie and her shopping partner have tried the pet store “Mounds”. (The points of possible completion to which I want to draw your attention in the examples will be marked by a diamond (“!”)) (10) Rachel: Abbie:

What about- uhm Mounds. (1.2) (We) haven’t tried Mounds, ! (0.4) We can probably call Mounds,

In this case, Abbie does not immediately continue into her next TCU; her continuation follows a pause. There are a number of reasons a speaker may continue in this way, including lack of uptake from the recipient (i.e., next speaker passes the opportunity to speak at the end of a current speaker’s turn (Sacks et al. 1974, Davidson 1984)), or because the recipient uptake has been more minimal than what a speaker is pursuing (Pomerantz 1984b). I include these cases because my interest is in what sorts of actions are produced recurrently and consecutively, whether they be produced smoothly and without a gap (as in examples (1)–(7)) or whether they emerge more contingently from local interactional circumstances (as in (10)). My intention, thus, is to observe a recurrent, sequentially organized combination, regardless of whether the combination appears to be planned or more locally negotiated. In either case, encounters with such combinations would lead users to anticipate particular kinds of next actions to follow particular kinds of initial actions. Thus, the recurrence of this format not only supports the analytic argument I make in this chapter, it also suggests a potential basis on which pragmatic projection is acquired by participants as they form their interactional repertoires within groups. The cases in this collection come from a number of audio and videotaped sources. All involve naturally occurring talk, that is, talk that would have gone on regardless of whether it was being recorded. One is in an institutional setting, a high school physics class, and the rest are in casual circumstances among friends and acquaintances, either face to face or on the telephone. All participants are speakers of American English; some are Spanish-English bilinguals. Initially, I began a general collection of shorter multi-unit turns, but as I noticed a large proportion of these turns involved negation in their initial

59

60

Cecilia E. Ford

units, I began to focus on negatively framed TCUs in particular, collecting shorter multi-unit turns as well as single TCU turns for comparative purposes. As mentioned above, these TCUs involve phrases or clauses with negation, as well as the negative particle, no or its variants.3 In the remainder of this paper, I discuss in some detail the work of negative TCUs and the further talk which tends to follow them. I Wrst turn to the varied sequential environments in which this turn pattern is deployed, looking at the work the negative TCUs are doing in relation to what has come before in the sequences, as well as the work of the elaborations that follow. I then examine cases in which negative TCUs implement aYliative actions; in these instances, no further talk elaborates the negative TCU. Finally, I turn to cases in which participants produce disaYliative negation without elaboration. These instances oVer evidence that when further talk is not produced, recipients treat the talk as problematic and incomplete: participants orient to the need for elaboration after disaYliative negation by initiating an expanded sequence which speciWcally pursues elaboration of the negative turn.

Negation followed by elaboration With respect to the pattern of negation followed by elaboration in the present collection, two observations can be made: 1. The pattern is positively and unproblematically attested in the data; i.e., it is recurrently used by speakers. 2. The pattern is used across a variety of interactional contexts. Apart from the restriction to disaYliative action types, I have not found the format to be limited to particular types of action sequence. 3. Broadly speaking, the work of the negative TCU involves rejection, either rejection of an immediately prior proposition or the rejection and shifting away from a topic or sequence. The work of what follows the negative TCU is to provide a resolution of the rejection: an alternative to what came before, an account for the rejection, a modiWed form of agreement (e.g., to take up a proVered topic while rejecting a proposition), or a shift to another topic or sequence. I oVer several examples to provide a sense of the range of contexts and types of actions that are associated with this turn pattern. Example (11) is from a high school physics class. Kira has been complain-


ing that the classroom is too cold. The teacher counters by teasing her, suggesting that the rest of the class warm her up by breathing on her (Wrst arrow). In Kira’s response (second arrow) she says she will oVer all the others something for their breath, implying that she thinks the others will have bad breath. After the teacher reacts by saying they won’t go near Kira’s nose, Jake, a self-selecting next speaker, also responds to Kira’s implication, asserting that it does not apply to him (third arrow): (11) Kira: ⇒

⇒

⇒

(That fan is) making me colder. ((several lines of talk)) Teacher: Why don’t we all go around Kira and go hhhahhh. ( ): eh huh huh ( ): huh huh huh huh huh ( ): eh huh huh ( ): sss sss sss sss Kira: ‘Cause I’monna offer everybody breath (stine), ( ): huh huh huhTeacher: We won’t go anywhere near [your nose. ( ): [huh huh huh Jake: Huh huh ehuh. Delia: [Huh oh my go:d. ( ): [huh huh huh huh Bruce: She’s got to get the molecules in your body [heated. Jake: [>Hey.< (.) You don’t have to worry about me, ! Listerine this morning ( ): (Yeh I know.[ Taking that ) Teacher: [> You know what I’d like to do, ((teacher continues into teaching topic, i.e., physics problems))

I had

Jake begins with a summons, “Hey”.4 Having turned to face Kira (who is one row over and behind Jake), he points toward her with his pencil and delivers the negative TCU, “You don’t have to worry about me”. This negatively framed TCU implements the action of taking exception, and it comes to a point of possible completion in terms of grammar, prosody, and action. However, the action of taking exception, negating the implication of Kira’s turn, is immediately followed by an explanation for why bad breath will not be a problem in Jake’s case, “I had Listerine this morning”. (Listerine is a mouthwash.) Jake’s turn is topically tied to Kira’s turn, making reference to an implication in Kira’s turn, but his turn is not a second pair part. Jake merely expands the sequence by exempting himself from those who need a breath freshener. Here then, the negation implements the action of taking exception and the elaboration provides an account or explanation for the rejection of Kira’s generalization.

61

62

Cecilia E. Ford

In the next case, which takes place before class begins, Jake is responding to an oVer from Sila, who is selling candy. Jake wants to buy some, but he needs change for a Wfty dollar bill. Sila oVers to give him change in the form of a check that another student has given her. Through his negative TCU (arrow), Jake refuses the oVer. He uses the pattern of negation plus elaboration (talk by Delia is from a separate conversation). (12) Sila : Delia: Sila : ⇒ Jake : Sila :

I have a check. Because if [I don’t make ( ) [Eight fifty. Nah:, ! I won’t take- I don’t take second party checks. eh huh huh huh huh (0.5)

Jake’s turn begins with the negative particle “Nah:,” a TCU which could be complete in itself. He adds an explanation in the form of a rule: he doesn’t “take second party checks”. Distinct from the sequential environment in example (11), where Jake self-selects as next speaker, the negation plus elaboration in (12) is produced as a second pair part to an oVer, for which Jake was the designated recipient. Jake’s negatively framed and disaYliative response implements a refusal (a dispreferred response, Pomerantz 1984a), and it is followed by elaboration in the form of an account for that refusal. In example (13), we Wnd another sequential environment for the work of negation plus further talk. This interchange also occurs before the oYcial start of class. Delia and Bee are Spanish-English bilinguals. Bee has just pointed toward Delia’s face and made reference to it. It is diYcult to make out exactly what Bee says, but she seems to use the word “sudor” (sweat). It is not clear whether Bee has asked a question or made an observation. In either case, Delia is the recipient and is in a position to respond with agreement or disagreement of some sort (Jake’s turn is from a separate interaction involving Sila and the Wfty dollar bill): (13) Bee: Delia: Jake: Delia: Delia:

[(Es sudor.) ((“It is sweat”)) ((points toward D’s face)) No no no no [no:, ! This is water. ! [↑Give me the forty, (.) No es sudor. ! ((“It isn’t sweat”)) (0.4) I put water to keep myself awake.


Delia’s negation plus elaboration here implements rejection, a correction of a proVered understanding, and an account. The Wrst point of possible completion comes after her production of Wve negative particles in rapid succession. Her continuation oVers a correction: she says she has water on her face rather than sweat. Delia’s next TCU uses a further negative, this time in Spanish, “No es sudor”. Here she explicitly states that it is not sweat on her face, and after a 0.4 second pause, she adds the explanation for why she has water on her face. In this case, the elaboration does not follow immediately upon the possible completion of the negative TCU. Instead, Delia comes to add it after a pause clearly long enough to have allowed for uptake. It is perhaps in the environment of lack of uptake, perhaps even a puzzled look (though we can’t tell from the videotape) that this elaboration seems appropriate to add. Again, then, the negation plus elaboration format is attested; again disaYliative negation does the work of rejecting a prior understanding, and elaboration deals with the resolution of the discrepancy, here through both an alternative and an explanation. In (14), from a conversation among friends during a break in the middle of a game of Pictionary, Rachel tries to ascertain whether she or her teammate, Pam, was the last to draw a picture. Terry misinterprets the question to mean who among all the players (a total of Wve) drew last. Rachel’s (arrowed) turn implements repair after a recipient has displayed a (mis)understanding through a next turn. (14) 1 2 3 4 5 6 ⇒ 7 8

Rachel: Rachel: Terry: Terry: Rachel: Abbie : (.)

Uhm (.) who drew the last time. (0.4) [me or her. [I did. (.) Very poorly. No. !=Me or her.= =’t was a long time ago.

Rachel’s turn does what SchegloV (1992) has called third position repair. Terry has responded to Rachel’s Wrst turn in a manner that displays one understanding of that turn — an understanding Rachel then treats as a misunderstanding. Rachel deals with Terry’s misunderstanding through her negation and elaboration. Rachel’s “No” is a possibly complete TCU, after which she adds, “Me or her”, providing an alternative to Terry’s understanding. Rachel’s elaboration is actually a repeat of the me or her she produced as an extension at line 3 after the (0.4) pause. However, Terry’s response (line 4) was delivered in overlap with

63

64

Cecilia E. Ford

that extension. (14), then, contains a negative TCU doing rejection, followed by elaboration which performs a self repair. (15) also involves rejection and correction, but here the speaker is correcting the recipient’s drawing rather than any talk by either speaker. Still discussing the last round in the Pictionary game the friends have been playing, Rachel tries to reproduce a drawing that a now absent participant (Cindy) produced a few minutes earlier. As Rachel draws, Abbie oVers guidance in the form of a negation and elaboration — rejection and correction: (15) Rachel:

⇒

I love that- (.) chimpanzee she drew. [↑How’d she do that. Terry : [Ha heh (.) ((unrelated talk from another room)) Abbie : [No, ! She started with=no::, ! She started with (.) an oval. Rachel: [ºLike that?º

Abbie’s turn is akin to a third position repair, but she is initiating repair based on Rachel’s drawing rather than any previous spoken turn. Thus, whereas third position repair involves a second turn displaying misapprehension of a Wrst (as in (14)), Rachel here displays what Abbie treats as a misapprehension of a drawing technique. It was brought to my attention that there may be further evidence of the unitary nature of the negation plus elaboration turn format here in the fact that the initial “No” is recycled when Abbie repeats her turn (G. JeVerson p.c.). Having delivered the Wrst version in overlap with Rachel’s, “Like that?”, Abbie restarts and includes the initial TCU. However, what is evident in the video is that Rachel has continued to draw during the time she produces her overlapped talk. This means that the repeated version of Abbie’s turn may in fact be responding to a new version of the drawing. In any case, this exchange is yet another example of the negation plus elaboration format, in a diVerent action sequence than previous examples. What the examples so far have in common is that the negation performs a rejection, and the elaboration which follows either provides an alternative to what has been rejected or it oVers an account for the rejection. In another use of negative plus elaboration, (16) contains a case where what follows the negatively framed TCU is elaboration that takes up on the topic of the rejected turn. Thus, the pattern of negation plus elaboration allows a speaker to reject a proposition but follow up the rejection with a move toward


accepting an aspect of the work that the rejected turn was likely doing. In this example, two young women are catching up on each other’s lives. At this point, one topic has come to a close and Bee proVers a new topic, implemented through a question: (16) Bee: ⇒

Ava: Bee :

°(So anyway)° °(I’unno)° .hh Hey do you see v- (0.3) fat ol’ Vivian anymore? No, hardly, an’ if we do:, y’know, I just say hello quick an’ hh y’know, just pass each other in th [e hall.(way) th[e [Is she still hangin’ around with Bo:nny?

Ava’s turn involves a movement toward aYliation after an initially clear rejection. The elaboration, beginning with “hardly”, shows an attempt to aYliate with the topic of “Vivian” after using “No” to reject the proposition that she may see this person anymore. Perhaps at issue is a nuance in the meaning of “see”, which allows Bee to both say no to seeing Vivian as a friend and yes to seeing her in the sense of having visual contact and exchanging brief greetings. My data also contain cases where a negative TCU introduces the return to a prior sequence after an insertion of what is thereby marked as divergent talk. In (17) below, Angela and Bonnie are talking about Wction writing. Bonnie brings up a story they both had been following on the internet, “that story” (Wrst arrow). Apparently, on this website, Wction writers compete for recognition of their work. Over several turns, the talk moves from admiration of contestants to annoyance with a particular writer who throws in famous authors’ names, presumably to impress readers. With the turn in question (second arrow), Angela uses a negative plus elaboration pattern to move back to the subject of the topic-introducing turn by Bonnie (Wrst arrow): (17) Angela: I think we should (.) write an X-Files episode. (.) that entails, a space ship coming to (pick them all up). huh [huh, heh heh heheh Bonnie: [u-huh Bonnie: =Oh, speaking of [writing, I have to um (.) I have to= Angela: [can you just see it? heh heh Bonnie: =go back now when I get back I have to go look at ⇒ that web site, (.) that story, how that’s coming. Angela: It’s coming along really cool. Marcy keeps trying to (.) She’s been entering like everyday= Bonnie: =those- the people that win, those are really, I mean (.) really well written=

65

66

Cecilia E. Ford

⇒

Angela: =except for the Euripedes thing. I’m like,= Bonnie: =O:H (editing that) [I was like Angela: [I was so annoyed by it- Edgar Allan Poe was [good (successful) Bonnie: [That was very (.) James Joyce let me just tell you. Like, I have to throw in this and, famous (.) [you know Angela: [I know. I read that. I’m like, just throw some Euripedes. oh yeah my Uncle James was a (1.0) WHO CARES, (.) [Edgar Allan Poe was like at least (.) Bonnie: [( ) Angela: works his into [( ) his story= Bonnie: [um hmm =um hmm Angela: No it’s been going- I think it’s over this week. Bonnie: O:hh? (1.0) Bonnie: It’s very cool, I’m gonna print it out.

The negative particle plus elaboration in Angela’s turn disaYliates with or “rejects” the continuation of the talk ridiculing the annoying name-dropping author and moves back to the subject of “that story”, the topic originally raised by Bonnie a number of turns earlier (Wrst arrow). In her immediate response, Bonnie aligns with Angela’s proposed shift back; Bonnie provides a change of state token (Heritage 1984) followed by an assessment of “that story”. SchegloV (1996b) also examines cases of negation plus further talk. He documents this turn format in cases where the negative particle marks a move from joking to serious talk. These instances would Wt into my collection in that they involve rejecting a sequence type or a “key” (in Hymes’ sense), and they move on with elaboration in the form of talk in a new mode. Also bearing resemblance to the cases documented in my study are instances discussed by Sacks in relation to sequences involving “correction invitation” devices (Sacks 1992a, 1992c), as well as examples cited by Pomerantz in her investigations of “Wshing devices” and responses to “telling my side” (Pomerantz 1980, 1988). These studies look at the strategies whereby a participant in a sense elicits information from a recipient by providing a turn that invites correction; in such cases, a response may be done through initial negation followed by the correction. Similarities can also be seen in several of the Japanese turn types described by Saft (1998), associated with the particle iya in Japanese interaction. However, all of these studies involve only turns initiated with a negative particle; that is, the studies do not cite cases in which a full clause performs the negation. Furthermore, none of these studies looks at the potential for prag-


matic projection in the sequential contexts in question, though in the case of Saft’s study, the Japanese particle iya always occurs as a turn initial particle, never as a turn in itself — unlike the English particle no, which has this potential (as discussed below). The examples I have reviewed here provide a sample of the positive evidence for the negation plus elaboration turn pattern in these data. These cases illustrate the diverse sequential contexts in which this turn format is deployed. The format is used for taking exception and thus extending a sequence as in (11). After an oVer or a question, the format can do the work of a dispreferred second pair part, as in (12) and (13). It can be the format for a third position repair after a second pair part has displayed some misunderstanding, as in (14). Akin to a third position repair, it can be the format for correcting a non-verbal action by another, as in (10). It can be a format for shaping a turn toward aYliation, as in (16), where the turn Wrst rejects a proposition and then takes up a topic proVered within that proposition. Finally, the format can be used to implement a shift in sequence type or key, as in (17) or in the joke to serious cases described by SchegloV (1996a). Diverse as these cases may be, all involve disaYliation and elaboration toward resolution of some kind, through accounting or providing an alternative to what has been rejected. What follows the rejection may be seen as working toward common ground (Clark 1996), the common ground covering referential work as well as work toward social alignment. In cases of repair, notions of correct reference are at issue, and in cases of refusal or disagreement, issues of interactional alignment are at the fore. Repair mechanisms are fundamental to language interaction (SchegloV et al. 1977), and work on preference structure leads us to expect that dispreferred responses would be followed by accounting talk (Sacks 1987, Pomerantz 1984a). Thus, the elaboration which follows negation, in these environments, can be functionally grounded in previous research on basic structures and tendencies in conversation. What research on repair and on preference structure has not pointed to (with the possible exception of SchegloV (1992)) is the operation of pragmatic projection beyond initial TCUs in turns which do other-repair or which deliver dispreferred actions. Negation can, however, be a vehicle for aligning with previous talk rather than rejecting some aspect of it (JeVerson 1996). Indeed, it is revealing to examine such cases, as they form a set of potential counter examples to the existence of a projectable turn format involving negation plus elaboration.

67

68

Cecilia E. Ford

Single TCU turns and aYliative negation While there is evidence that the format of negation plus elaboration is used by interactants in these data, there are also cases where negation is not followed by any further connected talk by the same speaker. These cases could stand as counter evidence to my claim that after a negatively framed TCU elaborative talk is projected. However, as has been alluded to above, the key in many of the exceptional cases is that the negation is doing agreement or aYliation. When negation is aYliative, that is, when it delivers agreement or alignment with previous talk, then no elaboration seems to be anticipated or delivered. Again this is perfectly in line with what we know about preference structure and the typical shape of agreeing turns. Let me oVer two quite diVerent examples, both of which involve recipient alignment with a previous speaker’s talk. In (18), the high school physics teacher is delivering a longish turn, typical of a teacher-fronted phase of such a class. In this excerpt, he elicits a response from the students and then continues. The students are looking at a problem involving the changing size of an image in a mirror as a man steps away from it. The teacher begins by stating that the whole mirror is not necessary, and then he physically demonstrates this claim by moving away from a wall, asking the students to imagine a mirror on it. In the context, his question is built to receive a negative response; that is, a negative response is aYliative with the claim he has made in the Wrst sentence: (18) Teacher: You certainly don’t need the entire mirror, because as he stands here,(.) Does he need this part of the mirror to see his feet. ⇐ affiliative negation Jake: [No. ! ⇐ affiliative negation Bee: [No. != ⇐ affiliative negation Sila: =No. != Teacher: No. 0.4)>Just< think about it. Now his feet are back here somewhere.(.)Yeh- on the image.

Each of the student turns consists of a single TCU doing negation. However, the context of the response slot is one in which the teacher has already displayed that he anticipates a negative response. He has asserted that one does not need the entire mirror. Thus, when the students produce their negative particles, they are agreeing with what the teacher has just said. Note that the teacher uses a version of the sequence negation plus elaboration in his continuation. However, the teacher’s talk in this instance is not included in my collection since he already had a Wrm and extended hold on the Xoor, and my


interest is in shorter turns and local pragmatic projection. The next example of aYliative negation is in a diVerent sequential environment. In (19), Terry has been telling a story about her partner’s nephew, who stayed with them for Wve weeks during the previous summer. It is now just days before Christmas, and the nephew has only recently sent a thank you card for his summer visit. The highlighted contributions from Rachel and Abbie are negatively framed, but none of them involves more than a single TCU: (19) 1 Terry: 2 3 4 5 6 7 8 9 10

Rachel: Terry: Rachel: Abbie: Terry: ( ): Rachel: Abbie:

11 12

Terry: Abbie:

He was cute,>you know, sending the thank you card before Ch(h)r(h)istm(h)as(h)< hh[uh huh huh huh [Yeah(h) huh huh hah Strategic. We saw right through ‘em. ⇐ negation [He is no:t-a dumb kid.! [Ye[h [eh heh huh huh huh huh huh. .hhh [No sense in sending it in August. ! ⇐ negation [Well at least he didn’t send his list to ⇐ negation Santa. != =That’s right °Yeah.°

Because Terry’s story has been about a boy she has depicted as cute and clever, though somewhat manipulative, aYliation with the story’s upshot would take the form of picking up on some of these aspects of it. Terry’s stance is exempliWed in line 5, “We saw right through ‘em.” Rachel and Abbie provide their uptake and assessments of the story through negatively framed turns.5 Though the turns are negative, they are nevertheless aYliative with Terry’s longer contribution. In a sense, these post-story uptake turns stand as negative assertions for which the story itself has already provided an elaboration. Clearly, however, these turns are not disaYliative, aligning as they do with the stance that Terry has taken in her telling. In cases where negation is done in the service of aYliation, referential or social ground is displayed to be shared, and there is no functional need for elaboration. What remains be considered is what participants do when disaYliative negation is delivered without elaboration. If this occurred unproblematically, it would strongly call into question my claim that elaborative talk is projected by disaYliative negative TCUs.

69

70

Cecilia E. Ford

From turn to sequence: Recipient responses to unelaborated disaViliative negation Thus far we have seen positive evidence in these data that the format of negation plus elaboration is used when the negation implements disaYliation with prior talk, but what other evidence is there that participants treat this pattern as real? Certainly the fact that there is no overlap after the initial negative TCU is one form of evidence for the anticipation that more will follow, but there is even more compelling evidence pointing to the “oriented to” nature of the turn format. This evidence involves the reactions of recipients to turns in which disaYliative negation is presented without elaboration. It is in these cases that we observe the intersection of turn and sequence structuring alluded to in the title of this paper. What could be done through the turn pattern of negation plus elaboration is expanded into a sequence in which the participants work interactionally toward the elaboration of the disaYliative negation. In (20), Rachel and Abbie, guests in the home of Pam and Terry, are noticing pictures on the walls. While waiting for Pam to get oV the phone, the two guests remark on the pictures, guessing about the people and animals depicted. Rachel refers to the people in the pictures as relatives (“Grandma” and “Grandpa”, lines 4–6). In a form of disagreement, Terry claims that she and Pam, the residents of the home, do not know the people in the pictures (line 8). Terry’s action is implemented as disaYliative negation. (20) 1 2 3 4 5 6 7 ⇒ 8 9 ⇒ 10

Terry:

⇒

Rachel:

⇒

11 12 13 14 15 16 17 18

Rachel: Abbie: Rachel: Abbie: Rachel: Terry:

Terry: Abbie: ( ): Abbie:

I see there’s a lo:t of pictures of sheep. Oh yeah. there’s more in the other [r(h)oo:m t(h)oo. [There, there’s sheep there, Right by grandma. ha ha ha Gramma, an’ an’ (.) And and Grandpa, and then the sheep (and [baby sheep) ⇐ negation [We don’t know these people. ! (.) ⇐ extension of _>or< these people. ! negation You d↑on’t? (.) No. ! ⇐ negation (.) Do you know the one [ just below that one though? [just ( ) (0.5) the one in the mid[dle?


⇒ ⇒

19 20 21 22

⇒

23 Terry: 24 25 26 27 28 29

Terry: Terry: Abbie: Rachel:

Rachel: Abbie: Rachel: Terry: Rachel: Abbie:

[No? ! The little cryin’ girl? No. [You just like the >pictures.
serious ‘No’”. Paper delivered at the Annual Meeting of the American Sociological Association, New York, N.Y., 1996. 1998 “ReXections on studying prosody in talk-in-interaction”. Language and Speech 41: 235–263. SchegloV, E. A., JeVerson, G. and Sacks, H. 1977 “The preference for self-correction in the organization of repair in conversation”. Language 53: 361–382. Scheibman, J. 1999 “Local patterns of subjectivity in person and verb type in American English conversation”. Paper presented at the Symposium on Frequency EVects and Emergent Grammar, Carnegie Mellon University, May 28– 30.


Selting, M. 1996

2000 Steensig, J. 1998

“On the interplay of syntax and prosody in the constitution of turnconstructional units and turns in conversation”. Pragmatics 6 (3): 357– 388 “The construction of units in conversational talk”. Language in Society 29: 477–517.

“On “pragmatic” units and “pragmatic” completion in conversation” . Department of Linguistics, Aarhus University, DK. Wells, B. and MacFarlane, S. 1998 “Prosody as an interactional resource: Turn projection and overlap”. Language and Speech 41: 265–294.

79

The implementation of possible cognitive shifts in Japanese conversation Complementizers as pivotal devices Hiroko Tanaka University of Essex

This paper explores the pivotal use of the complementizer to and related elements (tte, to te, to ka, to iu, to itte, etc.) in Japanese conversation for the purpose of modifying an initial trajectory of talk to accommodate possible cognitive shifts in constructing a continuation of one’s own turn. SpeciWcally, it is shown that this item has a singular capacity to accomplish a dramatic rerouting of turn trajectories, which not only transforms the action performed by prior talk, but can accomplish such interactional repair with little or no grammatical repair. Moreover, the continuation regularly incorporates an accounting of the possible cognitive processes that may have triggered the redirection. The remarkable versatility of this particle for implementing the grammatical and interactional redirection of ongoing talk is elucidated by investigating how it is occasioned within the temporal and sequential unfolding of naturally occurring talk, and through an analysis of the place of complementizers within the interpenetration of grammar and interaction in Japanese.

1.

Introduction

While investigating the construction and projection of turns in Japanese (Tanaka 1999, 2000), I came across the following fragment, which seemed to point to the existence of a powerful resource for accomplishing a radical selfrepair of the course of one’s immediately prior utterance, through the adjunction of a bit of additional talk:

82

Hiroko Tanaka

(1) [Tokyo 7, p. 4, mid] multiparty conversation; slightly simpliWed 1

W:

((complains about H))=

H:

= Sore wa soo ne that TOP so FP ‘That’s right, isn’t it’

2

(1.2)

3→ H:

tto iu kara QUOT say because ‘because ((I)) say that ((line 1))’

4

ikenai no wrong FP ‘((is)) what is wrong ((with me))’

5 6

(1.4) ( ):

‘N Yeah’

Prior to the portion shown, W has made a critical remark about her husband H. In response, H Wrst agrees with W’s assessment Sore wa soo ne ‘That’s right, isn’t it?’ (line 1). There is an ensuing silence from coparticipants in line 2, which is followed by a major turnaround in H’s position from the agreement to a self-deprecation: Sore wa soo ne (1.2) tto iu kara ikenai no ‘What’s wrong ((with me)) is because ((I)) say that “That’s right, isn’t it”’ (lines 1–4). Although the cognitive process that H might have undergone is unavailable to us, the modiWcation from an agreement to the self-deprecation appears to have been triggered by the silence in line 2 which may have implicated coparticipants’ disapproval that the husband had so readily gone along with his wife’s criticism. What captured my imagination is that the revision in H’s action is accomplished with remarkable ease and with no grammatical repair: i.e. not through a replacement of a problematic turn or expression (line 1), but simply through a continuation in lines 3–4 of the utterance in line 1. Indeed, through the addition in lines 3–4, H embeds the agreement already produced in line 1 into a new turn construction which takes a stand on the agreement itself. Pursuing this phenomenon further, it increasingly became apparent that such dramatic shifts are regularly — though of course not exclusively — performed through the use of the case particle to and other related elements (tte, tote, to ka, to iu also written to yuu, to itte, etc.) referred to as “complementizers” in the linguistic literature. The particular utility of this class of elements for the manifest implementation of potential cognitive shifts seemed worthy of investigation in its own right. This paper examines features of this particle that vest it with an extraordinary quality to implement a pivotal, wholesale dis-

Complementizers for cognitive shifts

placement of an action which has already been projected or has been completed by the current speaker. Speakers realise such operations by taking advantage of certain basic features of Japanese grammar and interaction — e.g. predicate-Wnal word order, the postpositional structure, and the consequent transformability of turn-shapes (Tanaka 1999, 2000). While drawing on previous linguistic research on complementizers (see Section 2), this paper attempts to make a contribution to the Weld of grammar and interaction in general (Ochs, SchegloV and Thompson 1996, see also Couper-Kuhlen and Selting 1996; Ford 1993; Ford, Fox and Thompson, forthcoming) and speciWcally pertaining to Japanese conversation (e.g. Ford and Mori 1994; Fox, Hayashi and Jasperson 1996; Hayashi 1994, this volume; Hayashi and Mori 1998; Hayashi, Mori and Takagi, forthcoming; Lerner and Takagi 1999; Mori 1994, 1999; Saft 1996; Tanaka 1999, 2000).1 The data for this study are drawn from a number of corpora of multiparty and telephone conversations among native Japanese recorded in both the Tokyo and Kansai areas, of a total duration of approximately Wve hours. The transcription notations are largely derivative of Gail JeVerson’s system, adapted for foreign language presentation. The modiWed conventions and the abbreviations used in the transcripts are provided in the Appendix.

2.

Research on complementizers

Previous analyses on complementizers have focussed primarily on their proscriptive and prescriptive uses, largely based on invented or written texts. Beginning with Kuno’s (1973) characterisation of the complementizer to for marking clauses for which a presupposition of factuality or truth is not made by a speaker, others have variously supported or disputed this criterion (Josephs 1976, Akatsuka-McCawley 1978, Terakura 1984, etc.). Suzuki (1996) applies Kamio’s theory of “territory of information” (e.g. 1979, 1994) to propose that the speaker distances herself/himself from some phrase through the use of to yuu. Matsumoto (1998), on the other hand, draws on the etymological origin of the particle to, and argues how to yuu is most aptly described as a quasiquotation: i.e. that the complement marked by to yuu is treated as if it were a quotation. Maynard (1993: 221–253) furthermore characterises to yuu as a “bridge” for linking the two modes of communication: “saying” and “describing”. Martin (1975) provides a comprehensive cataloguing of the multiple possibilities of the use of to, which serves as an important sourcebook for this

83

84

Hiroko Tanaka

study. In contrast to the other studies mentioned, Hayashi (1997) explores the interactional signiWcance of the particle to in an utterance-Wnal position from a conversation-analytic perspective. The case particle to is known variously as a quotative particle, quotationmarker, an expression-marker (Lerner and Takagi 1999), or complementizer, and is crudely equivalent to the English complementizer ‘that’. There seems to be a consensus among linguists that the particle to is roughly synonymous with the variants (t)te or tote; and (t)te is also used as an abbreviated form of to iu ‘say that’ and to itte ‘saying that’ (e.g. Martin 1975: 228f., 1001). A noun expression or clause marked by to is typically followed by what Martin (1975: 996) calls an “information processing” verb, producing collocations such as to iu (say that …), to kaku (write that …), to omou (think that …), to kanjiru (feel that …), etc. Among these diVerent verbs the construction to iu — alternatively written as to yuu or toyuu — (QUOT + ‘say’) features pervasively in written texts and talk-in-interaction, and has attracted wide interest among linguists. The particle to has variously been noted to have the following features, among others:2 a. To has the characteristic of treating the clause directly preceding it as a quote, and embedding it as a direct object of the verb which follows to (Tsujimura 1996: 172f.). b. The verb following to such as to ((omotte)) ‘thinking/feeling that’ or to ((itte)) ‘saying that’ are often elided or unexpressed (Martin 1975: 997). c. Furthermore, when to is used to quote a preceding phrase, the quotation can be either direct or indirect (Martin 1975: 997; Matsumoto 1996: 247). This seems to diVerentiateto from ‘that’ in English, as the latter normally indicates that a quote is indirect. d. “A quotation can be subdued: …to wa ((iwanakatta)). It can be highlighted: …to mo ((iu)); …to sae ((ieba)), …to sura. It can be restricted by bakari, gurai, made, and dake …, and also by de mo ‘or something’. It can be marked by sika” (Martin 1975: 999). e. By using the particle string to ka (Quotative particle + question particle) forming a question format to ka ‘something to the eVect / or something’, one “can suggest vagueness or doubt: Iki-tai to ka itte imasu ‘They are saying they want to go or something’”. It can also be tacked on to a quoted noun to add a touch of vagueness: Tanaka to ka iu hito ‘a man called Tanaka or something’ or ‘a man called, I believe, Tanaka’. To ka can be used in citing items on a list, with


optional omission after the last item: A to ka B to ka C ((toka)) ‘A and/or B and/or C’ (Martin 1975: 1000, 1021). f. To iu ka sore ((wa)) . . . ‘if you/we ask about . . . it’ is “one of the many roundabout ways of introducing a theme” (Martin 1975: 1015). g. The two compound forms to yuu koto or to yuu no are used as complementizers in place of to when a subject noun clause is formed with to (Kuno 1973: 218). h. A quotation formed through to and its variants such as tte and to iu to can be used for “conditionalization” or “hypothesization” (Martin 1975: 228f.). Although some of the research cited above employs natural conversational data, none with the exception of Hayashi (1997) addresses how complementizers are occasioned within the temporal unfolding of talk-in-interaction. This paper presents a contrast to this body of research by focussing on the real-time deployment of complementizers within the speciWc interactional contexts in which they occur.

3.

Grammar and interaction in Japanese and complementizers

Relevant to this discussion is the postpositional structure and the standard predicate-Wnal turn-structure in Japanese conversation. In contrast to English, which employs prepositions and where the verb normally precedes a direct object, Japanese grammar employs postpositional particles that typically mark the grammatical units that precede them. For instance, case particles are normally attached after a nominal or other unit, specifying the case of the unit (e.g. nominative, accusative, dative), and projecting the production of either a predicate or another nominal, depending on the type of particle. As a concrete example, the case particle ga marks a preceding nominal as a subject of a sentence-in-construction and projects an upcoming predicate as in the following: … tomodachi ga ashita kekkonshiki na no ne= friend NOM tomorrow wedding COP FP FP ‘((my)) friend is having ((her)) wedding tomorrow, you know’

Applying this principle, speakers can Wrst produce a spate of talk and then retroactively specify if it is to be understood as a subject, object, indirect object, etc. Similarly, conjunctive particles may be attached following some expression, to characterize it retroactively as a subordinate clause and to simulta-

85

86

Hiroko Tanaka

neously project a further clausal or sentential unit. In short, certain types of particles (such as case, adverbial and conjunctive) have an integral part to play in the incremental or step-by-step construction and projection of turns, thereby providing considerable Xexibility to direct or revise the course of talk along some path as well as a capacity to incorporate major transformations of the initial trajectory of talk, prior to or even after the current turn has come to a completion (Tanaka 1999, 2000). Indeed, almost any such particle sets the direction of the current speaker’s turn in some way, and can also be used to change the course of talk in interaction with others (see Hayashi, this volume). Among the various types of particles, however, the quotative particle to seems to have a particularly prominent capacity to modify the course of prior talk. As one type of case particle, to and its variants may be employed retroactively to mark a preceding expression as a direct object after the expression has already been produced. Due in part to the verb-Wnal or predicate-Wnal turn-structure, the speciWcation via to of a prior expression as a direct object projects the production of some verb.3 As described in detail below, re-routing is done by exploiting a feature of to to embed a prior expression into a newly emergent sentential or turn structure as the direct object of certain types of verbs (see Section 2). It turns out that participant orientation to word order has an inseparable relationship with the order in which actions are accomplished through that language. A close examination of complementizers within the moment-bymoment evolving of the immediate micro-sequential and grammatical contexts can considerably enrich our understanding of their interactional signiWcance. As suggested in the literature referred to previously, in conversation, complementizers are routinely attached after a spate of talk as a strategy to isolate it as a quote or an expression to be commented on. Depending on the particular context and the speciWc way in which they are enlisted, complementizers can be shown to be integral resources for implementing a wide range of actions. For instance, they are useful for proposing alternative or diVering versions of events, without directly contradicting or disagreeing with coparticipants’ portrayal of the same event (see for instance, fragment (4)). Alternatively, they are sometimes employed as a tool for directly disagreeing with another speaker’s just-completed utterance.4 It has also been pointed out that to yuu ka ‘or I mean/or I would say’ is frequently used to initiate “adjacent” self-repair (Hayashi 1994: 89, n. 7). They can also be grammatically latched on to another speaker’s utterance, to present one’s position as a continuation (see Hayashi, this volume). Complementizers furthermore allow one to dramati-


cally enact some event prior to designating it as a quote. To give a foretaste for how the diVerence in the standard word ordering in English (verb preceding the direct object) and Japanese (direct object preceding the verb) is consequential for the utilisation of complementizers, fragments from conversations in the respective languages are presented below. In each fragment, there is a great deal of quoting of what someone purportedly said or thought. But notice that in the English examples, the speaker projects that upcoming talk will be a quotation (through the formula: subject — ‘said’ — optional ‘that’ — quotation) and moreover regularly speciWes whether a quote is direct or indirect (e.g. through the use of ‘that’ for indirect quotes and the choice of personal pronouns within the quote: ‘I’ for direct and third person pronouns or names for indirect). On the other hand, in the Japanese fragment, the spate of talk is routinely marked as a quote only after the quote has been produced, and complementizers are also employed for direct quotes. An example of direct quotes: (2) [extract of F:TC:1:1:19] from Drew 1998, p. 308; boldface added Shi:

So:- (0.5) one of thee- uh: bouncers came up to me. This new guy. En he said .hhh tell me he said d’you know that girl? .hhh An’ I said su:re. (.) I sedyihknow I know’er, .hh en ‘e said well? (.) said she’s drinking rum’n Coke out’v a water glass.

An example of an indirect quote, employing the complementizer ‘that’: (3) [extract of NB:II:2:19] from Drew 1998, pp. 306–7; boldface added Nan:

A::nd ah,h he seh thet he: w’ss: hed tried t’call’er on Mother’s Day en,h .hhhh …

In the following fragment from a Japanese conversation, three friends A, S and H are having a get-together over some cake, reminiscing about a practical joke that A and S played on H during a previous outing together. A and S tricked H into entering a public toilet, telling her that there was something memorable (one of the milestones from a famous pilgrimage) concealed inside. The three speakers engage in friendly bantering, taking turns to enact/recount their respective recollections of how they experienced the episode by employing direct quotes of what is presented as a “stream of consciousness” at the time. The complementizers (which mark prior talk as a quote or a thought) are highlighted in boldface, and the quoted enactments of their thoughts at the time have been placed within double quotation marks to indicate the location where they occur in the transcript. Note that ‘she’ refers to H.

87

88

Hiroko Tanaka

(4) RKK (p.7: 303-) multiparty conversation; boldface added A:

( ) A:

S: H:

S: H:

A: S: S: A: S:

A:

Imi ga wakaranai no ni [“un” tte yuu] atari meaning NOM not.understand P P yeah QUOT say like ‘Like, that ((she)) would say “okay” even though she didn’t understand what was going on’ [.hhhhhhhhh ] “mimi warui na::” [ t(h)o k(h)a] i(h)-= ears bad FP QUOT QP say‘((made me think)) that “she must be hard of hearing” or something’ [hhhhhhhhhh ] =E:! tte [yuu ka::] what QUOT say QP ‘Wha::t! or should ((I)) say’ [hhh .hhh]hhhh= =”nande minna waratteru n daroo” [ to ka.] why everyone laughing VN COP QUOT QP ‘like “Why on earth is everyone laughing?”’ [ha::. ] [hhhhh ] .hhehh [hhehh] [heh h]eh .hh Iye: nan to yuu ka:: .hhhh “warui Well what QUOT say QP malicious

damasare na [kya ii tricked not good [mmhhehh

ojisan man

ni P

ga kono] hito” CONJ this person hhehh]

H:

to ka [omott(h)e.]= QUOT QP thinking ‘Well how should ((I)) put it, ((I)) was thinking like “((I)) hope this person isn’t going to be tricked by some malicious old man” or something’ [e:::: ehh!]=

S: A:

=heh heh heh [heh heh heh heh heh .hhhh hhhh [hah hah hah hah hah hah

S:

Comparing the Japanese utterance in the Wrst line with the English examples presented above (or compare with the English gloss in the third line), it can be seen that the order of appearance of both the complementizer (e.g. ‘that’ in English and to in Japanese) and/or the verb (such as ‘said’ in English and iu in Japanese) is typically reversed within the utterance-structures in the two languages. In other words, quotation markers (e.g. ‘he said that’ in English, to yuu in Japanese) are normally prepositioned (i.e. before the expression quoted) in English and postpositioned (i.e. after the quoted expression) in Japanese.5 In fragment (4), this feature of to permits a dramatic acting out of some incident


as though it were happening then and there, prior to formulating it as a quotation or as a reporting of a stream of consciousness. This very feature may be exploited in quite diVerent, perhaps even in “unplanned” or opportunistic ways in particular interactional circumstances in which speakers are faced with a perceived need to modify prior talk in Japanese. As is well-documented in the conversation analytic literature (e.g. Heritage 1984), speakers regularly monitor the next turn for an indication of how their turn has been taken by coparticipants. And it may happen that after a speaker completes a turn, the reactions of coparticipants can signal some trouble in responding: e.g. noticeable pause or even a micro-pause. A common way of rectifying such a situation is to repair prior talk to make it more acceptable. (5) [cited in Heritage 1984: 275] (Davidson 1984: ST (detail)) A:

A:

Oh I was gonna sa:y if you wanted to:,= =.hh you could meet me at UCB and I could show you some of the other things on the computer (.) Maybe even teach you how to program Ba:sic or something. .hhh

This fragment documents the trails of a possible cognitive process that A might have undergone in analyzing coparticipant hesitation as “foreshadowing a rejection” and re-issuing the oVer to make it more attractive prior to the coparticipant overtly rejecting the proposal (see Heritage 1984: 275). Notice in this instance that although the second oVer adds to or qualiWes the original oVer (in this case, by making it more concrete), the Wrst oVer still stands unchanged. Now, cases can be found where similar repairing actions are also performed by Japanese speakers as in the example above. In addition to providing alternative and more attractive versions, however, it emerges that the postpositional characteristic of to can alternatively be used as a pivotal device for Japanese speakers to engage in a post-hoc modiWcation of the prior talk itself through an exploitation of the “natural” ordering: a conversational action followed by “quotation-marker” or complementizer. In short, by initiating a self-continuation of prior talk with a complementizer, a speaker can latch grammatically onto the Wrst trouble-source turn to mark it as a quotation or expression to be commented on; and depending on what follows the complementizer, the original turn can thereafter be retroactively re-directed, re-processed, re-formulated, re-constructed, negated, and so forth. The cumu-

89

90

Hiroko Tanaka

lative eVect of such a process can be to mitigate, if not to practically undo, the actions of the original turn. As a variation of this strategy, speakers may incorporate the prior turn into a new turn (i.e. a meta-statement) which directly addresses the speaker’s “cognitive” analysis of coparticipant reactions to his/her Wrst turn.

4.

Tracking “cognition in action”

It has been suggested above that to can be used as a strategy to implement a redirection of talk that can reXect possible cognitive shifts: a pivot that allows speakers to displace the trajectory of talk with little or no grammatical repair, typically by a further addition after these elements. As with fragment (1) introduced at the outset of this paper, this procedure is regularly observed in response to some coparticipant reaction to one’s prior turn. It should be noted, however, that it is quite possible that this same resource may be put into action to modify talk in accordance with a self-reXective reassessment of a prior turn after its completion, quite independently of any coparticipant reaction. This section examines concrete instances in which a speaker initially either completes or projects the course of a Wrst action, but subsequent to some intervening coparticipant reaction such as silence, proceeds by beginning with a complementizer to re-process prior talk. The range of procedures dealt with here (which are by no means exhaustive) include: (4.1) retroactively incorporating prior talk into a meta-statement concerning it; (4.2) re-casting prior talk as a trial formulation to which the speaker is no longer committed, (4.3) retroactively quotationalizing prior talk and attributing it to someone else, (4.4) hypothesizing/ironizing prior talk, e.g. as a tongue-in-cheek comment. Due to limitation of space, only one fragment is presented here for each type. 4.1 Retroactively incorporating prior talk into a meta-statement concerning it To demonstrate the principle by which to may be mobilized to aVect the evolution of the course of a turn and to provide a detailed description of the conversion process involved, the segment of fragment (1) in which a complementizer occurs will be re-analyzed:


Sore wa soo ne (1.2) tto iu kara ikenai no that TOP so FP QUOT say because wrong FP ‘Because ((I)) say that “That’s right, isn’t it” ((is what)) is wrong ((with me))’

The following is a mapping of the step-by-step process which the initial turn (line 1, fragment (1)) is subject to. First, H produces a complete turn: Sore wa soo ne ‘That’s right, isn’t it’ through which H performs the action (labelled ‘X’ below) of agreeing with a prior assessment. The transforming elements for each step are noted in boldface. (a)

Sore wa soo ne that TOP so FP ‘That’s right, isn’t it’

X

Second, by re-starting with tto (which is a phonetic variant of to), the previous action is transformed into an expression to be commented on (to potentially formulate it as a quotation, as something thought about, as something heard, as something felt, etc.). (b)

“X” (1.2) tto X QUOT ‘that “X”’

In this sense, the action performed by a complementizer can be described as preparing the ground for a further operation to be performed on X. Moreover, since an “information processing” verb (see Section 2 above) often follows, the attachment of to can be seen as a preliminary to making a “meta-statement”6 concerning X, i.e. pertaining to some productional or cognitive process that X is being subject to. In the instance under examination, it turns out that the verb iu ‘say’ follows tto, thereby casting utterance X as quoted speech: (c)

“X” (1.2) tto iu X QUOT say ‘say that “X”’

What follows the above has a further eVect in diverting the direction of the turn. In this particular example, the prior talk is processed in two stages: (d)

“X” (1.2) tto iu kara X QUOT say because ‘because ((I)) say that “X”’

(e)

“X” (1.2) tto iu kara ikenai no X QUOT say because wrong FP ‘What’s wrong with ((me)) is because ((I)) say that “X”’

91

92

Hiroko Tanaka

As mentioned above, the further addition of increments in (d) and (e) brings about a radical reformulation of the cumulative action from an agreement to a self-criticism. Grammatically, however, the particle to converts utterance X into a direct object of the verb iu ‘say’, and the meta-statement makes a comment on X. In other words, through the facility of to to embed a prior utterance into a larger utterance, no grammatical repair has taken place to execute this operation. 4.2 Recasting prior talk as a trial formulation to which the speaker is no longer committed Another frequently observed conversion is to produce a turn that takes a stand, and in a continuation of that turn, to recast it as a trial formulation to which the speaker is no longer committed. In the next fragment the participants, who are students, have been discussing what would be involved in analyzing the results of a detailed empirical investigation into processed marine products. In the fragment itself, S and A are debating whether they had previously conducted discussions concerning such investigations in Professor Nakamura’s seminar. In this and the ensuing fragments, the embedded clause is supplied in the English gloss in curly brackets { }. (6) RKK (p. 19: 847–52) multiparty conversation; boldface added 1

S:

[Ma Nakamura sen]see no:: zemi de AP ((name)) professor GEN seminar at ‘Well, at Professor Nakamura’s seminar,’

2

sooyuu no o yatta to yuu no wa: (1.5) that.kind P ACC did QUOT say P TOP ‘as for discussing that kind of ((research)),’

3

m i t a see

4

n a i k a [n a : .] not QP FP ‘((I)) wonder if ((I)) haven’t seen ((it))’

5

A:

6

A:

7

k o t o VN

g a : : : : : (0.7) NOM

[N a i ] k a :. = not QP ‘((Oh, so you)) haven’t seen it’ =Ore mo nai na::. I also not FP ‘I haven’t either’ (0.3)


8→ A:

tte yuu ka ore QUOT say QP I [cumulative gloss ‘Should ((I)) say

wa nai na . TOP not FP of lines 6 and 8:] that {“I haven’t either”}, or I haven’t’

First, S begins to talk about whether the kind of research which is the topic of discussion here was dealt with in Professor Nakamura’s seminar, and toward the end of his turn (line 4), he terminates his turn with the Wnal particles ka na: ‘((I)) wonder’. This combination of Wnal particles is critical for specifying that the action performed by S’s utterance in lines 1–3 is that of searching in his mind whether or not he has seen or heard that kind of discussion. Overlapping its production, however, A enters in line 5 by co-completing S’s turn-ending with a slight time delay, repeating the beginning of S’s utterance in line 4 Nai ka: ‘((Oh, so you)) haven’t seen it’. It is possible that A had not quite heard or registered the ending ka na: which was overlayed by A’s turn beginning in line 5, as is evidenced by A’s continuation Ore mo nai na:: ‘I haven’t either’ (line 6), which is constructed as an assertion. But there is a brief pause in line 7, which could have alerted A that there is some diYculty with the assumption inherent in A’s turn that S had not seen the research. Such a reassessment of the situation may have prompted A to reconsider S’s prior turn to arrive at a re-interpretation that S had perhaps voiced uncertainty about whether he had seen it. At any event, A continues in line 8, this time modifying his prior assertion by replacing the adverbial particle mo ‘also’ (as in ‘also not having seen the research’) with the ‘contrastive’ adverbial particle wa (respectively marked in bold typeface) which has the eVect of presenting his own case in contrast to S, so that the new version no longer contains the assumption that S had likewise not seen the research. Moreover, A highlights the fact that the replacement is with respect to the type of adverbial particle employed by emphasizing wa in line 8 and by a structural repetition of the phrase containing the replaced item: Ore mo nai na:: changed to ore wa nai na. Now, re-analyzing line 8 in relation to the use of complementizers, notice that A employs tte yuu ‘say that’ to reintroduce the theme of the turn he has just completed in line 6 (see Section 2, point f), thereby latching grammatically onto the end of his prior turn. By further adding the question particle ka ‘should ((I)) say’, A Wrst recasts line 6 by embedding the assertion into an utterance which no longer makes an assertion but is more akin to a rhetorical question about the certainty of what was said in line 6, then continues with the alternative version which is now divested of the assumption that S had not seen

93

94

Hiroko Tanaka

the research. This operation can be represented schematically as follows “X” tte iu ka Y X = Ore mo nai na::. Y = ore wa nai na.

where and

‘Should I say that “X” or Y’, ‘I haven’t either’ , ‘As for me, I haven’t’

As in fragment (1), the operation here has not necessitated grammatical repair. This is because the addition of tte iu ka Wrst recasts the original assertion X into just one of many possible formulations he could have produced to which he is no longer committed, prior to proposing the new version Y. Thus, in contrast to the English fragment (5) cited earlier, the speaker uses the complementizer to slightly alter the sense of the original talk prior to reissuing the second version. 4.3 Retroactively quotationalizing prior talk and attributing it to someone else A speaker may Wrst produce a statement which is interpretable as presenting one’s own stance, but subsequently quotationalize the statement, thereby attributing the quoted speech to someone other than the speaker. In the next fragment, prior to the segment shown, Y had been telling her sister-in-law K about her baby son’s illness, and that the consultant at the hospital wanted to see him from time to time to check on his progress and for keeping a record of the illness partly for accumulating a clinical database. The topic seems to have been exhausted for the time being, and after a two-second pause, K comes in again: (7) [Shakujii 1A (#4) p. 9] telephone conversation; boldface added 1

(2)

2

K:

3

Y:

Ano kooishoo tte iu n [jyanaiuhm after.effect QUOT say VN not ‘Uhm, ((it))’ s called an after-effect isn’t it-‘ [So so so so ‘Right right right right’

4→ K:

nai to not QUOT ‘although called an

wa itta kedo = TOP said although what ((they)) said is that {“((it))’s not after-effect“}’

5

Y:

=’N:= ‘Mm’

6

K:

=’Ne mitai n jyanai no = You know want.to.see VN not QP ‘You know, isn’t it that((they)) want to see ((how it turns out))’


Although the same quotative particle to is used to embed prior talk in this instance, the operation which takes place here is subtle but complex. Perhaps as part of what may have been intended as a summing up of the previous exchange, K begins to say something about the possible after-eVects of Y’s son’s illness in line 2. But overlapping the end of this line, Y responds with a series of pulsed acknowledgements (line 3). This somewhat marked reaction may have triggered a reconsideration on the part of K concerning her utterance-so-far, since a reference to after-eVects (which had not been discussed up to that point) can imply that the illness is not without adverse consequences. Now, although there is no deWnitive evidence that this is indeed the cognitive process which K undergoes at this point, it is possible to observe how to is employed as a part of a procedure to bring about a radical re-direction immediately after Y’s reaction can be heard, in the way K recasts line 2 through additions which bring about a step-by-step re-routing of her prior talk involving minimal grammatical repair. In line 4, K emphatically repeats part of the last item which appeared in line 2, nai ‘not’, and then continues with to wa itta kedo. To understand the signiWcance of this minimal repair, Wrst note that the grammatical construction in line 2 can be interpreted as a negative or positive statement depending on the prosodic contours employed. In more detail, the ending jyanai can be taken either as the tag question ‘isn’t it?’ when delivered with a rising intonation, or as ‘it isn’t’ when delivered emphatically without a rising intonation. In other words, whether line 2 is heard as ‘Uhm, ((it))’ s called “an after eVect” isn’t it-’ or ‘Uhm, ((it))’ s not called “an after eVect”’ is largely dependent on the productional features of the turn and especially of the item jyanai. Since the crucial item is overlapped and consequently its prosody is not audible, the outcome of the progress of this turn can be equivocal at the end of line 2. It is thus possible that line 2 can be heard to be projecting the former positive version. Indeed, K could have begun from the outset with a form that is more unequivocally negative such as: Ano kooishoo to wa iwanakatta kedo uhm after.effect QUOT TOP didn’t.say although ‘Uhm, although ((they)) didn’t say that it is an after effect’

In other words, the type of turn that K seems to be projecting toward the beginning of line 2 is at odds with the negating turn that she cumulatively constructs at the end of line 4. This suggests that there may have been some cognitive shift — whether prompted by Y’s response or whether K may have

95

96

Hiroko Tanaka

become aware of the possible implications of her projected turn while constructing it. What is signiWcant here is that this very equivocality becomes the target of the mini-repair that occurs after jyanai. The fact that jyanai gets overlayed by Y’s interjection in line 3 is used as a resource by K to re-issue the critical ending. After the overlap, K vests line 2 with an unequivocally negative sense in line 4 by emphatically re-articulating nai ‘not’ which is the latter part of the utterance-ending jyanai in line 2. The eVect of this is to leave little room for jyanai to be heard as a tag question ‘isn’t it?’. So that at this point, K transforms line 2 which is hearable as the positive version K:

Ano kooishoo tte iu n [jyanaiuhm after.effect QUOT say VN not ‘Uhm, ((it))’s called an after effect isn’t it-’

into the following: K:

Ano kooishoo tte iu n [jyanai- nai uhm after.effect QUOT say VN not not ‘Uhm, ((it))’s not called an after-effect

Then K adds a variant form of the complementizer to wa itta (QUOT + TOP + said) — as described under point d, Section 2 above — to further quotationalize the prior utterance: 4→ K:

nai to wa itta kedo = not QUOT TOP said although [cumulative gloss of lines 2 and 4:] ‘Although what ((they)) said is that {“((it))’s not called an after-effect”}’

Line 2, which can potentially be heard as K’s own assessment of the baby’s condition (with reference to after-eVects), is reframed by the operations detailed above into a quote of the doctors’ diagnosis: that Y’s baby’s condition is not called an after-eVect. In this connection, the very fact that K is now reporting on the doctors’ comments is inconsistent with the fact that all along, it was Y who was the one reporting to K about the doctors’ treatment of her son. The inconsistency in the resultant turn may not be unrelated to the possibility that it was devised as something of an expedient remedy for the original utterance in line 2. At any event, by embedding the original utterance into a larger unit which reports on what the doctor said, K has transferred the source of the potentially problematic comment in line 2 from herself to the doctors as well as to set the Wnal, cumulative tone of the turn.


4.4 Hypothesizing/ironizing prior talk: special case of meta-statement In the Wnal type of conversion to be examined here, the speaker Wrst makes some proposal or demand, but subsequently processes the former action by “hypothesizing” (see Section 2, point h) or ironizing it. In this fragment, three friends A, H and S previously went on an outing together, and are chatting about some photographs that S took of the three of them. After the participants make various comments about how they look in the photographs, A produces the following complaint: (8) RKK (p. 2: 60–62) multiparty conversation; boldface added 1

A:

Ma! (.) dooyuu jijoo de are well whatever situation P is ‘Well whatever the situation may be,’

2

moo ichido shashin o totte morawanai to once.more photos ACC have.take.not unless ‘unless ((you)) take the photographs again’

3

komaru na ore wa. be.bothered FP I TOP ‘I would be bothered’

4

(0.3)

5→ A:

t(h)o k(h)a it[t(h)e: QUOT QP saying [gloss of lines 1–5:] ‘Fancy say(h)ing some(h)thing li(h)ke {“Well whatever the situation may be, I would be bothered unless ((you)) take the photographs again“}’

6 7

H: S:

8

A:

9

[hhh heh heh [.hhh [Nande. ‘Why?’ .hh Iya iya iya Well well well ‘Well what ((I)) mean is’

chotto kami ga nagakatt(h)a kara. a bit hair NOM was.long because ‘because ((my)) hair was a bit too long’

In the foregoing, A complains that the photographs that S took need to be taken over again (lines 1–3). But there is a short pause in line 4, implicating some diYculty in responding to the complaint. This may give A an opportunity to reconsider his prior talk in the light of its possibly sobering implications. A then softens the atmosphere by latching on the quotative particle to to hypothesize/ironize his prior talk, and to lessen its seriousness by transforming it into a light-hearted hypothetical comment.7 That this is indeed what A is

97

98

Hiroko Tanaka

trying to do in line 5 is reinforced by the laughter tokens intermingled with the production of the continuation. (See fragment (9) for another instance of a hypothesizing use of to delivered with laughter.) Coparticipants aYliate with this newly shaped turn by laughing together (line 6) and by displaying interest (line 7). This way of utilizing to can be seen as a special case of (4.1) retroactively incorporating prior talk into a meta-statement concerning it. The meta-statement in this case renders the earlier turn into a tongue-in-cheek statement, and therefore mitigates the implications of performing a possibly dispreferred or face-threatening action such as a complaint, demand, or blame. The dramatic eVect of the adjunction of the complementizer string t(h)o k(h)a itt(h)e: ‘Fancy say(h)ing some(h)thing li(h)ke…’ which relieves the tension and produces laughter is akin to a punch-line of a joke or an unexpected twist at the end of a story.8 This meta-statement achieves an instantaneous reversal, permitting speakers to alter the prior action from a complaint to a statement that quotationalizes it and pokes fun at the fact that they should be making such a complaint.

5.

Complementizers and cognitive shifts

In the section above, I have been referring to possible cognitive shifts, since it is a contentious issue whether or not a mid-course transformation of talk in the fragments examined actually involve such an internal shift — whether it was initially triggered by some external event or not. It is of course also possible that an apparent re-alignment of talk was actually planned, intended, or chosen as part of an interactional strategy for the performance of a particular action. However, it is still possible to analyze the sequential organisation of talk in the immediate vicinity of the deployment of to and to collect various forms of circumstantial evidence to obtain a sense for whether or not it was used to implement a mid-turn change of direction or if it is part of a design which may have been on the agenda from the outset of the turn. For instance, it can be argued that an intervening silence prior to a redirection through to as in fragments (1), (6), and (8) provides fairly strong grounds in support of some change of cognitive state prior to the employment of a complementizer. As is well-known, there are only limited windows of opportunity available for the self-initiation of repair or modiWcation of prior talk: the Wrst being the trouble-source turn, the second chance being the transition


space between the trouble-source turn and the next turn (SchegloV, JeVerson and Sacks 1977). In the fragments mentioned, however, the fact that the current speaker allowed the silence to occur implies that at least the Wrst opportunity (and perhaps even the second) for altering prior talk was not taken — increasing the risk that such a ready opportunity would never present itself again. It is therefore highly unlikely that the speaker intended to redirect the course of talk prior to the gap. By the same token, it can be argued that the redirection is likely to have been triggered by the coparticipant silence itself. On the other hand, an immediately contiguous transformation (i.e. no gap between the Wrst utterance and the transforming turn) may leave open a greater possibility that the complementizer was not necessarily used as a tool for implementing a cognitive shift. Instead, it may (i) be a part of a turn design to execute an action that is projected from the outset of the turn or (ii) be used as a resource to permit the speaker to get one’s say (e.g. coparticipant criticism and accusation, or articulating an imposing demand) but still to be able to mitigate it and show aYliation by complementizing it, or (iii) allow a participant to Wrst aYliate with another participant prior to taking a diVerent position, to name just three possible uses. Two fragments demonstrating these features will be examined here. The following is an instance of the use of a complementizer for hypothesizing/ironizing prior talk, where there is no ostensible cognitive shift. H oVered to send C some documents relating to a group activity they are both engaged in, and asked for C’s work address. When C seems to be fumbling trying to Wnd her name card, H begins to tease her about not knowing her address by heart in the Wrst place. (9) [Shakujii 1B #5, p. 11] telephone conversation; boldface added 1

C:

Ano::gg Uhm watashi no (.) (‘N::) (.) °meishi dasu kara°= I GEN (Mm) name.card take.out CONJ ‘((I))’ ll take out my name card, so’

2

3

H:

=°‘N° ‘Mm’

4

C:

°Are- (.) attcha::° ooops oh.dear ‘Oops, (.) oh dear’

5→ H:

((teasing voice)) Jyuusho gurai oboetoki nasai: address at.least memorise IMP ‘((You)) ought to at least memorize ((your)) address’

99

100 Hiroko Tanaka

[nan’nen hataraiteru no:= how.many.years working QP ‘How many years have ((you)) been working ((there))?’

6→

7

C:

[( )

8

H:

=n(h)a(h)n t(eh)=* ((see below for word-by-word translation)) ‘Fancy saying something like {“((You)) ought to at least memorize ((your)) address. How many years have ((you)) been working ((there))?”}’

9

C:

=Jyuusho wa address TOP ‘As for the address

moo zen[zen oboenai n INT not.at.all memorize.not N I can’t memorize it at all’

10

11

H:

[‘N ‘Mm’

da [yo COP FP [So ↑o ‘Really?’

(watashi wa ) [ dame na n (da) (I TOP) no.good COP N COP ‘I’m useless’ 13 H: [gg’N ‘N ‘Mm Mm’ ––––––––––––––––––––––––––– *Line 8: nan teh is an abbreviation of the following:

12

C:

…nado to ka itte something like QUOT QP saying ‘Fancy saying something like for instance…’

Notice that H employs an emphatic but teasing voice from the beginning of her turn in lines 5–6 to deliver the hortatory admonishment: Jyuusho gurai oboetoki nasai: nan’nen hataraiteru no: ‘((You)) ought to at least memorize ((your)) address. How many years have ((you)) been working ((there))?’ Then, without any pause, H continues in line 8 by adding on the complementizer n(h)a(h)n t(eh) peppered with laughter, which is roughly equivalent to …nado toka itte ‘Fancy saying something like for instance…’. Similarly to fragment (8), this can be regarded as a variant of the hypothesizing/ironizing use of the complementizer (Section 4.4). However, there are at least two points of contrast between the two fragments. First, whereas in fragment (8) there is a gap between the Wrst turn and the issuing of the complementizer, in the present fragment, the speaker does not leave any room for the coparticipant to enter at the end of the complaint/admonishment before it is converted into a hypothesized statement: almost as though the addition is a part of the turn-design. Indeed, in spite of the teasing voice, H may have been reluctant to expose herself to the possibility that

Complementizers for cognitive shifts 101

the admonishment might cause oVense. Second, in (8), the initial turn is only converted into a tongue-in-cheek comment with the adjunction of the complementizer, whereas in the present case, the turn vocally projects from the outset the production of a tease. In short, the use of a complementizer for hypothesization in line 8 above is consistent with, and forms part of the action that H has been engaging in from the beginning of line 5: i.e. delivering a tease (point (i) above). Finally, fragment (9) may perhaps be seen as the use of the transforming feature of complementizers as a resource to permit a speaker to even articulate potentially face-threatening actions such as coparticipant admonishment with the assurance of being able to overturn it at the end (point (ii) above). The next fragment illustrates how the convertibility of a prior by means of the complementizer tte can be incorporated in a turn design which allows the speaker to aYliate with a coparticipant’s suggestion while not giving in completely (point (iii) above). Recall Wrst that an operation similar to those described so far may be accomplished by tte alone, while leaving unexpressed the ensuing verb iu and even the question particle ka (see Section 2, point b). Here, Y is debating whether or not to go to the beach the next day. The person referred to as niichan ‘big brother’ is Y’s husband. (10) [Shakujii 1A (#4) p. 19] telephone conversation; boldface added 1

Y:

=‘N:: de - niichan ga umi ni Yeah and big.brother NOM beach LOC

iku go

to ka itteta n dake[do dooka ne:= QUOT QP said VN although how FP ‘Yeah, and although big brother said that he was going to the beach or something, but ((I)) wonder’

2

3

K:

[‘N ‘Mm’

4

Y:

=chotto: kazeppoi kara little windy since ‘since ((it))’s a bit windy, [ ne ((sniff)) FP you know’

5

6

K:

7

Y:

hoo ga [ii yo ( ) [‘N yameta Yeah call.off way NOM good FP ‘Yeah, it’s better to call it off’ [‘N: yameyoo kana:= Yeah call.off maybe ‘Yeah, maybe ((I))’ll call it off’

102 Hiroko Tanaka

8→ Y:

=t[te mada wakan nai kedo= QUOT not.yet don’t.know though ‘should I say that {“Yeah, maybe ((I))’ ll call it off}, though ((I)) still don’t know yet’

9

[‘N ‘Mm’

K:

In lines 1–2 and 4–5, Y introduces the topic of going to the beach, and mentions her indecision about whether or not to go since it is windy outside. In response, K comes in with a suggestion that it would be better if she called oV the trip (line 6). To this, Y constructs a very skilfully designed turn that simultaneously displays strong aYliation with K’s suggestion while not committing herself entirely to the suggestion. Y begins to produce an early, preferred response even before K completes her turn: ‘N: yameyoo ‘Yeah, ((I))’ ll call it oV’ (the Wrst two words in line 7), through which she aYliates with K’s turn. Y continues, however, with the two Wnal particles ka na: ‘maybe’, which puts a little twist at the end of line 7 dampening the agreement which has been emerging as very forthright until that point. In other words, even at the end of line 7, Y has both agreed while rendering that agreement somewhat tentative. Then, without any break, she latches on the complementizer tte — which can be heard as an abridged version of tte iu ka (QUOT + say + QP) — to embed line 7 into a new unit which again partially discounts the agreement: ‘Should ((I)) say that “Yeah, maybe ((I))’ ll call it oV” though ((I)) still don’t know yet’ (line 8). Importantly, however, the deployment of the complementizer in the turn design has allowed Y — albeit only momentarily — to Wrst aYliate with K (in the immediately preceding segment of talk) prior to marking it as a trial formulation (see Section 4.2). Thereafter, Y is able to return to her original position that she is still debating about whether or not to go to the beach (in line 8). As with fragment (9), the complementizer in this fragment is attached contiguously (without any gap) to the expression it marks. If Y had waited for K to reinforce her point that Y should call oV the trip to the beach, it could of course have made it more diYcult to retroactively characterize her agreement as a tentative one, for reasons already discussed. Note also that the choice of the abridged complementizer tte in fragment(10) which leaves unexpressed the question particle ka ‘rather’ as in tte iu ka ‘rather than saying…’ further mitigates the somewhat contrary aspect of the complementizer itself by vesting it with a touch of laxness. Thus, it can be argued that the complementizer contributes to and implements the overall agenda of performing an aYliative action while still having her say (points (i) and (iii)).


The diVerences between the fragments dealt with in this section and those in Section 4 pertaining to the deployment of the very same grammatical resource illustrate how one might go about the detective work of determining whether or not a cognitive shift may have taken place. To reiterate, however, the “evidence” examined here is essentially circumstantial, and the full extent of the mental processes involved in these cases remains inaccessible.

6.

Discussion and conclusions

In each of the fragments sampled in Section 4, a current speaker’s initial turn is followed by some coparticipant reaction, which is then followed by a change in the trajectory of the talk through a continuation of the prior turn initiated by a complementizer. The direction which the original utterance was progressing in (or had attained) is retroactively re-routed by transforming it into another type of grammatical unit (direct object), and the original utterance then ends up being taken along a diVerent course. In sum, to complementize prior talk with to (and its variants) is to prepare to take it along another path — much like a trapeze artist establishing hand clasps with another performer prior to bringing the latter along in one’s own swing. To take the analogy further, the complementizer to acts as a pivot just as the performer can do a quick rotation to enable her to latch onto another swing which is moving along a diVerent trajectory. Now, once the pivot has been cleared, the complementizer permits the adjunction of an entire range of verbs — having to do with processing the prior turn — and this allows the speaker to take it along even more varied paths. The freedom with which a complementizer can be latched on to almost any prior utterance as well as the Xexibility of the operations which can subsequently be performed on a complementized utterance is suggested by the following: A sentence can be quoted by adding the particle to ‘(says/thinks) that’….The sentence quoted can be negative, desiderative, etc.; it can be perfect, tentative, etc.; and the quotational verb that follows can undergo all conversions independently of the quoted sentence. (Martin 1975: 996)

This paper has examined a number of complementizing operations, including: incorporating the prior into a meta-statement which may variously comment on it, hypothesize it or ironize it; converting an assertion into a trial formulation; quotationalizing and attributing prior talk to someone other than the

104 Hiroko Tanaka

speaker. In sum, the complementizer permits the speaker to reproject an utterance in progress or even a completed turn. This Wnding is consistent with earlier work on turn-taking in Japanese (Tanaka 1999, 2000) which has discussed the relationship between aspects of conversational grammar (e.g. use of postpositional particles and the predicate-Wnal turn structure) on the one hand and features of turn-taking (e.g. incremental and delayed projectability, turnWnal projectability, and the constant transformability of turn-shapes). Moreover, I speculate that the retroactive operations enabled by Japanese complementizers may be more diYcult to accomplish in English. Previous research indicates that English syntax facilitates an early projection (relative to Japanese) of the type of turn being produced, since the social action performed by a turn is typically made available early in the progress of a turn. In other words, the substance of what is being talked about is commonly produced after the turn-shape has already been projected. Roughly the reverse can be said to hold in Japanese. Partly as a result of the predicate-Wnal orientation and postpositional grammar, turns in Japanese are massively structured so that the substance of what is being talked about is articulated before the social action bearing upon that substance is made known. (Tanaka 1999: 141)

English has a standard Subject – Verb – Object turn structure and employs prepositions. These features are likely to result in an early commitment to a turn-shape, and provide relatively less Xexibility for participants to change the turn-shape once it has been projected without engaging in overt repair. Such potential cross-linguistic comparisons are left for future investigation. It was observed that among other uses, complementizers are useful for changing the course of a turn or redirecting a turn to reXect possible cognitive shifts. Now, the type of verb which can follow a complementizer has largely been reported to be verbs which pertain to saying, thinking, feeling, etc. (see Section 2 above), in other words, the range of verbs include those that bear upon cognitive processing. As a result, the type of action or operation prevailing upon prior talk (through complementization) invariably addresses what the speaker or someone else might have said, thought, felt, etc. about the prior turn, i.e. for making a meta-statement. It should come as no surprise, then, that complementizers are a potent resource for referring to, displaying, or implementing a potential cognitive process or displacement. What then are the sorts of interactional activities for which such a resource might be valued? I believe that one answer can be found through comparison of the types of redirection or reprojection of action which take place between the initial turn and the redirected turn, for instance:


agreement with self-criticism → self-deprecation [fragment (1)] mistaken assumption → shedding the assumption [fragment (6)] suggesting a sensitive issue → divesting responsibility for it [fragment (7)] coparticipant complaint → mitigating the complaint [fragment (8)] From this small sampling of shifts in action trajectories, it is possible to draw a tentative conclusion that complementizers are an eVective device for the posthoc self-reconstruction of actions which the speaker retroactively realizes to be potentially misinformed, face-threatening, disaYliative, dispreferred, or more generally disruptive of a sense of solidarity with coparticipants. This particular use of complementizers, therefore, enables speakers to perform the curious operation of carrying out a type of action-repair without resorting to grammatical or other prima facie repair.9 But as suggested, they can alternatively be mobilized as part of a larger turn-design, not necessarily accompanied by an ostensible shift in action or cognitive position. Moreover, they are enlisted in both argumentative (fragment in footnote 4) and aYliative turns (fragment (4)). We are once again reminded that complementizers, like many other grammatical devices, have no single overarching role, but are powerful and versatile resources for meeting a wide range of interactional contingencies.

Notes 1. I am extremely grateful to Makoto Hayashi for his insightful comments on an earlier version of this paper, for allowing me to use a number of corpora of conversational data, and for sharing with me a pre-publication copy of his work on postposition-initiated utterances (Hayashi, this volume), which provided a major source of inspiration for my paper. 2. For citations from Martin (1975), the parentheses and notched brackets in the original text (which both denote optionally sayable items) have been replaced in this paper by double parentheses. 3. As noted above, linguists have been debating at length what kind of verb can follow to. 4. An example is provided in the fragment below. The wife W is insinuating that her husband H likes having her as a wife because she is convenient for him (lines 1 and 2). H begins in line 4 to confront W by using the complementizer to iu (line 5) as part of a turn designed to directly disagree with W. The complementizer to iu is followed by yori ‘more than/rather than’ and the contrastive wa, which sets what H is going to say in contrast to W’s prior statement.

106 Hiroko Tanaka

[Tokyo 7, p. 5] multiparty conversation 1

W:

ba[kashi yatte i- .hh [kitta kara exclusively doing because ‘((I))’ ve continued to serve as ((your)) gate, so ((you))’ ve got to admit, ((I))’ nience’

2

3 4

(De) onsukedachi onmigawari and assistant surrogate

( ): H:

benri jyanai convenient not assistant or surrove been a conve-

[Uhn: hm: uhn: [Iya iya (.) sore:: (.) no no that

6

W:

sore to iu [yori wa ne (.) that QUOT say rather.than TOP FP ‘No, no, rather than saying that, you know’ [Hhhhhhhhhhhhhh ((laughter))

7

K:

‘N:

8

H:

Sore wa .hhh soiu fuu ni iware tara sukoshi that TOP that.way P say if.then a.bit ‘As for that, if ((you)) put it that way, then ((it))’ s a bit’

5→ H:

mo: .hhh ijiwa [ru ( ) INT unkin [d ‘nothing but unkind’

9

10

W:

[Okori nasai yo angry IMP FP ‘Get angry then!’

5. It is conceivable that the distinction between direct and indirect quote might in some instances be an analytical category which may not necessarily be oriented to by Japanese participants. At any event, since to in Japanese can be employed to mark direct or indirect quotes and because personal pronouns are frequently left unexpressed, it may not always be possible to make the distinction. 6. Similar terms, ‘meta-uses’ or ‘meta-forms’ are frequently used in the literature to refer to the functions of complementizers, for instance, to connote that ‘the speaker feels distance from the entities’ (Suzuki 1996: 321). See also Takubo 1989. The term ‘meta-statement’ used here, however, does not make any assumptions about the ‘feelings’ of the speaker. 7. This use of to is consistent with some previous research, which suggests that to iu marks the non-factuality or lack of supposition of truth in a proposition (e.g. Kuno 1973: 213; Matsumoto 1998: 244). 8. Although I do not have any recorded fragments at hand, I believe that a similar action can be performed in English by a speaker making a statement and then following it with an expression such as ‘s/he says’ with respect to one’s own prior statement. 9. In other words, with one exception (7), the fragments in Section 4 do not involve the more unequivocal types of self-repair such as word replacement, cancellation of something already produced, cutting oV and restarting, etc.


References Akatsuka-McCawley, N. 1978 “Another look at no, koto, and to: Epistemology and complementizer choice in Japanese”. In Problems in Japanese Syntax and Semantics, J. Hinds and I. Howard (eds), 172–212. Tokyo: Kaitakusha. Couper-Kuhlen, E. and Selting, M. (eds) 1996 Prosody in Conversation: Interactional Studies. Cambridge: Cambridge University Press. Drew, P. 1998 “Complaints about transgressions and misconduct”. Research on Language and Social Interaction 31 (3/4): 295–325. Ford, C. E. 1993 Grammar in Interaction: Adverbial Clauses in American English Conversations. Cambridge: Cambridge University Press. Ford, C. E., Fox, B. A. and Thompson, S. A. (eds) forthcoming The Language of Turn and Sequence. Oxford: Oxford University Press. Ford, C. E. and Mori, J. 1994 “Causal markers in Japanese and English conversations: A cross-linguistic study of interactional grammar”. Pragmatics 4 (1): 31–61. Fox, B. A., Hayashi, M. and Jasperson, R. 1996 “Resources and repair: A cross-linguistic study of syntax and repair”. In Interaction and Grammar, E. Ochs, E. A. SchegloV and S. A. Thompson (eds), 185–237. Cambridge, UK: Cambridge University Press. Hayashi, M. 1994 “A comparative study of self-repair in English and Japanese conversation”. In Japanese/Korean Linguistics, Vol. 4, N. Akatsuka (ed.), 77–93. Stanford: Center for the Study of Language and Information. 1997 “An exploration of sentence-Wnal uses of the quotative particle in Japanese spoken discourse”. In Japanese/Korean Linguistics Vol. 6, H.-M. Sohn and J. Haig (eds), 565–581. Stanford: Center for the Study of Language and Information. Hayashi, M. and Mori, J. 1998 “Co-construction in Japanese revisited: We do ‘Wnish each other’s sentences’”. In Japanese/Korean Linguistics Vol. 7, S. Iwasaki (ed.), 77–93. Stanford: Center for the Study of Language and Information. Hayashi, M., Mori, J. and Takagi, T. forthcoming “Contingent achievement of co-tellership in a Japanese conversation: An analysis of talk, gaze, and gesture”. In The Language of Turn and Sequence. C. E. Ford, B. A. Fox and S. A. Thompson (eds). Oxford: Oxford University Press Heritage, J. 1984 GarWnkel and Ethnomethodology. Cambridge, UK: Polity.

108 Hiroko Tanaka

Josephs, L. S. 1976 “Complementation”. In Syntax and Semantics 5: Japanese Generative Grammar, M. Shibatani, (ed.), 307–69. New York: Academic. Kamio, A. 1979 “On the notion speaker’s territory of information: A functional analysis of certain sentence-Wnal forms in Japanese”. In Explorations in Linguistics: Papers in Honor of Kazuko Inoue, G. Bedell, E. Kobayashi and M. Muraki (eds), 213–31. Tokyo: Kenkyusha. 1994 “The theory of territory of information: The case of Japanese”. Journal of Pragmatics 21: 67–100. Kuno, S. 1973 The Structure of the Japanese Language. Cambridge, Mass.: MIT. Lerner, G. H. and Takagi, T. 1999 “On the place of linguistic resources in the organization of talk-ininteraction: A co-investigation of English and Japanese grammatical practices”. Journal of Pragmatics 31: 49–75. Martin, S. 1975 (reprinted in 1988) A Reference Grammar of Japanese. Tokyo: Tuttle. Matsumoto, Y. 1998 “The complementizer toyuu in Japanese”. In Japanese-Korean Linguistics, Vol. 7, N. Akatsuka, H. Hoji, S. Iwasaki, S.-O. Sohn and S. Strauss (eds), 243–55. Stanford: Centre for the Study of Language and Information. Maynard, S. K. 1993 Discourse Modality: Subjectivity, Emotion and Voice in the Japanese Language. Amsterdam: Benjamins. Mori, J. 1994 “Functions of the connective datte in Japanese conversation”. In Japanese/Korean Linguistics, Vol. 4, N. Akatsuka (ed.), 147–63. Stanford: Center for the Study of Language and Information. 1999 Negotiating Agreement and Disagreement in Japanese: Connective Expressions and Turn Construction. Amsterdam: Benjamins. Ochs, E., SchegloV, E. A. and Thompson, S. A. (eds) 1996 Interaction and Grammar. Cambridge, UK: Cambridge University Press. Saft, S. L. 1996 “Reassessing cross-cultural comparisons of back-channel behavior in Japanese and English: Arguments for an expanded notion of context”. Linguistics and Language Teaching: Proceedings of the Sixth Joint LSHHATESL Conference, 169–187. SchegloV, E. A., JeVerson, G. and Sacks, H. 1977 “The preference for self-correction in the organization of repair in conversation”. Language 53: 361–82. Suzuki, S. 1996 “An analysis of tte- and to yuu no wa- marked phrases in Japanese discourse”. Word 47 (3): 321–337.


Takubo, Y. 1989

Tanaka, H. 1999 2000

“Meishiku no modaritii” [The modality of noun phrases]. In Nihongo no Modaritii [Modality in Japanese], Y. Nitta and T. Masuoka (eds), 211– 33. Tokyo: Kuroshio. Turn-Taking in Japanese Conversation: A Study in Grammar and Interaction. Amsterdam: John Benjamins. “Turn-projection in Japanese talk-in-interaction”. Research on Language and Social Interaction 33 (1): 1–38.

Terakura, H. 1984 “Noun modiWcation and the use of toyuu”. Journal of the Association of Teachers of Japanese 18 (1): 23–55. Tsujimura, N. 1996 An Introduction to Japanese Linguistics. Oxford: Blackwell.

Appendix: Transcription conventions The original Japanese talk appears in line 1 of the transcripts, word-by-word translation/grammatical description in line 2 and an English gloss in line 3. Double parentheses are used not only to encase transcriptionists’ comments but also for supplying unexpressed elements in the English gloss in line 3. Bold typeface is added to highlight parts of fragments and is not meant to indicate aspects of delivery. A full stop between words in the word-by-word translation in line 2 indicates that the English translation of a Japanese word consists of two or more words (connected by the full-stop).

Abbreviations in transcripts acc ap conj cop fp gen imp int loc

accusative particle appositional conjunctive particle copula Wnal particle genitive particle imperative intensiWer locative particle

N NOM

p qp quot sfx top vn

nominaliser nominative particle particle question particle quotative particle Wnal suYx topic particle verb nominaliser

On causal clause combining The case of weil in spoken German* Hannes Scheutz University of Salzburg

In written German, causal adverbial clauses introduced by the connector weil must be expressed in canonical verb-Wnal word order. In spoken German, however, main clause verb-second word order is used as well. Recently, there has been intense discussion about the syntactic and semantic properties of these diVerent weil-clause structures. It has been postulated that weil-verbWnal clauses represent propositional, weil-verb-second clauses epistemic causality. This hypothesis is the starting point of our analysis, which deals with the semantic and pragmatic properties of weil-clauses based on a large empirical corpus of about 400 weil-clauses taken from spontaneous everyday conversation. The results do not support the hypothesis of a correspondence between verb position and the semantic type of weil-clause. There seems to be another decisive parameter on which a form-function-mirroring is based — the degree of syntactic integration, ranging from a fully integrated intraclausal prepositional phrase at one extreme to a syntactically and prosodically non-integrated autonomous sentence at the opposite end of the scale. The narrower the syntactic integration of a causal is, the stricter is the semantic restriction to a propositional causal relationship; increased loosening of syntactic integration leads to a greater variety of semantic and pragmatic functions for the causal. On the interactive level, one Wnds a corresponding situation. While weil-verb-Wnal clauses are clearly the device of choice for retrospection and sequence-termination, weil-verb-second clauses are much less limited in their functions: they can be used as themeintroducing and turn-expanding structures as well.

1.

Introduction

In this paper we will be dealing with word order in causal clauses introduced by the conjunction weil. In spoken German we Wnd two diVerent structures: one exhibiting typical subordinate clause word order with Wnal position of the

112

Hannes Scheutz

Wnite verb (cf. (1)), the other exhibiting main clause word order with verbsecond position (cf. (2), representing a headline in an Austrian newspaper imitating spoken statements).1 (1) A:

B:

(2)

stell dir vor (-) und dann (-) habm wir ihn umdreht, (-) imagine and then we turned it around und dann habm wir ihn wieder anrennen lassn, and then we tried to get it going again dann is er angsprungen zum glück. (2sec) [so then it started fortunately [ja(-)zerst wird er net angsprungen sein, yeah at first it didn’t start weil er noch zu warm war. because it was still too warm Ist das überhaupt Kunst? Weil — ordinär ist das schon! Is that art? Because — it is vulgar!

The clause-Wnal position of the verb is the canonical word order as prescribed in Standard German normative grammars; the verb-second position has been considered a special phenomenon restricted to Southern German, or even to Bavarian-Austrian German. More recently, a number of investigations have conclusively demonstrated that verb-second in weil-clauses is a more general phenomenon of spoken German, and not dialect-speciWc. Traditional grammars oriented toward the written language have either completely ignored weil-verb-second or stigmatized it as colloquial. More liberal descriptive accounts have viewed weil-verb-second and weil-verb-Wnal as optional variants of one and the same semantic relationship without adding detailed functional speciWcations. However, more recent discussions of this word order problem have developed novel approaches (cf. Küper 1991; Schlobinski 1992; Keller 1993, 1995; Wegener 1993, 1999; Günthner 1993, 1996; Gohl and Günthner 1999; Uhmann 1996, 1998; Pasch 1997; Scheutz 1998; Selting 1999).2 A common denominator of some of these positions is the postulation of a more or less strict correlation between formal and semantic properties. In principle it has been claimed (with special emphasis by Keller 1993 and Wegener 1993) that the weil-verb-second pattern is used to express epistemic causality (i.e. oVering an account for the utterance itself), whereas the weil-verb-Wnal pattern is employed for propositional causality (i.e. accounting for the state of aVairs referred to in the main clause) (see 2.2.).

On causal clause combining

2.

Weil-clauses: Syntactic, semantic and pragmatic properties

In this section I present a survey of the most salient properties connected with weil-clauses, as a summary of the current state of the research. 2.1 Syntactic properties The contrastive investigation of weil-verb-Wnal vs. weil-verb-second highlights a series of formal diVerences, which make clear that the two types of weilclauses are two distinct syntactic entities: i. only postposed weil-clauses can be expressed with verb-second: (3) (3')

Ich zahle es nicht, weil ichs nicht einsehe. Ich zahle es nicht, weil – ich sehe es nicht ein. I won’t pay for it, because I can’t see any reason for it.

(4) (4')

Weil ichs nicht einsehe, zahle ich es nicht. *Weil – ich sehe es nicht ein, zahle ich es nicht. Because I can’t see any reason for it, I won’t pay for it.

This fact is already an important indication that in the verb-second case, we are not dealing with a dependent clause, as all subordinated adverbial clauses can be moved to the preverbal position of the main clause (Vorfeld). A further indication in this direction is the fact that one can position diVerent constituents after weil within the weil-clause. Only verb-second clauses (cf. (6) vs. (5)) allow this word-order variation: (5) (5') (5'')

…, …, …, …,

weil ich bis morgen das Geld nicht auftreiben kann. ?? weil bis morgen ich das Geld nicht auftreiben kann. ?? weil das Geld ich bis morgen nicht auftreiben kann. because I can’t scrape the money together by tomorrow.

(6) (6') (6'')

…, …, …, …,

weil – ich kann das Geld bis morgen nicht weil – das Geld kann ich bis morgen nicht weil – bis morgen kann ich das Geld nicht because I can’t scrape the money together

auftreiben. auftreiben. auftreiben. by tomorrow.

ii. It has been assumed in recent work and has been widely accepted that the integration of a weil-clause into the main clause by means of a prepositional phrase is possible only with verb-Wnal structures: (7)

Die Heizungsröhren sind geplatzt, weil es Frost gegeben hat. (Pasch 1983: 232) The heating pipes burst because there was frost. → Die Heizungsröhren sind wegen des Frostes geplatzt. The heating pipes burst because of the frost.

113

114

Hannes Scheutz

(8)

Die Heizungsröhren sind wohl geplatzt, weil — der Keller ist überflutet. The heating pipes have evidently burst, as the basement is flooded. → *Die Heizungsröhren sind wohl wegen der Kellerüberflutung geplatzt. *The heating pipes have evidently burst because of the flooding of the basement

The same holds for the possibility of a pro-form deshalb for the weil-clause in the main clause: (7')

Die Heizungsröhren sind deshalb geplatzt, weil es Frost gegeben hat. The heating pipes burst for the reason that there was frost.

(8')

*Die Heizungsröhren sind wohl deshalb geplatzt, weil — der Keller ist überflutet. *The heating pipes have evidently burst for the reason that the basement is flooded.

As I will show later, this is just an epiphenomenon. iii. Another widely accepted thesis concerns the prosodic properties of weilclauses: it has been assumed that diVering intonation structures mirror diVerent states of syntactic integration of the matrix and the weil-clause. According to this view, the syntactic and semantic integration of weil-verb-Wnal clauses correlates to a cohesive intonation contour over both matrix and weil-clause (continuous pitch progression, no pause). On the other hand, the lack of integration of verb-second clauses is reXected in a non-cohesive intonation contour (pitch break, a clear pause after weil). Our empirical analyses, however, will demonstrate that this account is an oversimpliWcation. 2.2 Semantic properties The formal diVerences between weil-verb-Wnal and weil-verb-second have triggered a series of speciWc semantic interpretation hypotheses; the most widely adopted position views the formal diVerences as parallel to a semantic opposition: propositional vs. epistemic causality. Compare (7) and (8) again. In (7) the proposition of the main clause is accounted for by means of the state of aVairs expressed in the weil-clause — the existence of frost caused the bursting of the pipes — whereas in (8) the content of the weil-clause by no means accounts for the proposition of the main clause. The fact that the basement is Xooded is not the cause of the bursting of the pipes but rather the


consequence of it. In the weil-clause this consequence is asserted. By means of a reductive conclusion (consequence → cause) the speaker derives the reason for his assumption that the heating pipes have burst. The assertion of the weilclause in (8) supplies us with an account of or justiWcation for the statement uttered in the main clause. 2.3 Pragmatic properties The syntactic independence of weil-verb-second is mirrored on the pragmatic level by the fact that it has its own illocutionary force, whereas syntactically dependent weil-verb-Wnal clauses have no illocutionary force of their own. (9) (9')

(10)

Kommst du zu mir, weil du Hilfe brauchst? [oder gibt es einen anderen Grund?] *Kommst du zu mir, weil du brauchst Hilfe? [oder gibt es einen anderen Grund?] Are you coming to me because you need help? [or is there another reason?] Komm zu mir! Weil — du brauchst dringend Hilfe. Come to me! Because you need help immediately.

A common problem area in the pragmatic analysis of weil-clauses seems to be the issue of information structure. Gaumann (1983) assumes that weil-verbWnal clauses are given and weil-verb-second clauses are new. Keller (1993: 229) supports this thesis, claiming that “in weil-verb-Wnal the proposition of the weil-clause is presupposed, in weil-verb-second the weil-clause is asserted. What is presupposed is assumed to be familiar to the dialogue participants and in a very broad sense it can be considered as given.” Günthner (1993), however, claims that weil-verb-Wnal clauses have to be focused, whereas the concomitant main clause consists of given material. Küper (1991: 140) (cf. also Pasch (1983, 1997)) distinguishes between falling and rising intonation contours in the main clause preceding weil-verbWnal clauses: (11) (11')

Peter könnte gestürzt sein↓, weil seine Verletzung noch nicht ausgeheilt ist. Peter könnte gestürzt sein↑, weil seine Verletzung noch nicht ausgeheilt ist. Peter may have fallen down, because his injury has not healed yet.

In (11), the main clause or at least the predicate of the main clause is said to be rhematic (the speaker expresses — for the Wrst time — the supposition that

115

116 Hannes Scheutz

Peter might have fallen down), whereas in (11') Peter’s falling down is assumed to be ‘old information’ and the main clause is therefore thematic.

3.

Representation of weil in our corpus

Our collection encompasses about 400 weil-clauses taken from a corpus of everyday conversations, each with two to four participants from diVering social backgrounds. The language employed is essentially colloquial Austrian German, in some cases a more dialectal variety of Middle Bavarian in Upper Austria. The Wrst signiWcant result is quantitative: two thirds of the collected data are cases of verb-second, only one third is realized as verb-Wnal. This distribution diVers signiWcantly from the distribution found in investigations conducted in other German-speaking regions. In Gaumann’s (1983) analysis, based on data from the Freiburg Corpus, one can Wnd the following distribution: out of a total of 126 weil-clauses, only 12 instantiate verb-second; out of a total of 15 speakers, three speakers do not produce a single verb-second weil-clause, whereas the rest of the speakers produce just one each. Schlobinski (1992), who analyzed North German data, noted 75 verb-Wnal and only 21 verb-second weil-clauses. A contrastive analysis by Norbert Dittmar (personal communication), which involves West vs. East Berlin data, shows a similar preference for verb-Wnal clauses for West Berlin speakers. In East Berlin one can hardly Wnd any verbsecond weil-clauses, presumably because of the strong inXuence of norms and linguistic conservatism. The discrepancy between these findings and our results is striking. Whereas in northern regions weil-verb-Wnal clauses dominate, in our corpus, weil-verbsecond clauses prevail over weil-verb-Wnal clauses in the ratio of 2:1. The comprehensiveness of our data excludes the possibility of statistical coincidence or text-type (genre) speciWc restrictions. These results constitute serious counterevidence to the recently alleged regional unspeciWcity of the use of weilverb-second. At the same time, they refute the traditional view that verb-second weil-clauses are limited to Southern German or Bavarian. The following presentation of our analysis treats the distinction between propositional and epistemic weil as discussed in 2.2.; we wish to test the hypothesis that the topological diVerence of weil-verb-Wnal vs. weil-verb-second should be seen as a formal reXection of the diVerence between propositional vs. epistemic causality.


First, we should note that there are clear cases of propositional and epistemic causality like the aforementioned example (7), which expresses propositional causality, and (12), which contains an epistemically grounded justiWcation: (12)

der hat sicher wieder gsoffen. (0.3) he’s definitely been drinking again weil — sie läuft total deprimiert durch die Gegend. (Günthner 1993: 42) because — she is going around totally depressed

But there are also numerous examples in our corpus which cannot be interpreted unambiguously. This holds true even after a very careful analysis of contextual aspects, cf. (13): (13)

einmal ist mir einer vor der haustür hint auffi gfahrn. (—) ein schifahrer. (—) once in front of the house I was rammed from behind by a skier hat mirs schon recht gstiert. (--) it really irritated me hab ich wirklich nichts dafür können, I really couldn’t do anything about it weil ich hab lang gnug vorher blinkt. because I had my blinker on long enough beforehand

Both the propositional interpretation (I couldn’t do anything about it because I had my blinker on long enough beforehand: ‘the fact of my blinking is the reason for my innocence’) and the epistemic interpretation ([I am sure that] I couldn’t do anything about it because I had my blinker on long enough beforehand: ‘the fact of my blinking is the reason for my conviction of being innocent’) are equally plausible. 3.1 Weil-verb-Wnal clauses 3.1.1 Propositional weil i. As expected, weil-verb-Wnal clauses mainly occur in propositional contexts. All of the examples display the following prototypical features: – In addition to the topological criterion of verb position, we Wnd the prosodic feature of a cohesive intonation contour encompassing both the matrix and adverbial clause. There is a level or low rising intonation at the end of the main clause and the contour continues without interruption in the weil-clause. – The primary accent of the whole sentence is located in the weil-clause. – The Wnal feature is semantic, namely, that the state of aVairs expressed in the weil-clause constitutes a suYcient condition for accepting the state of aVairs

117

118

Hannes Scheutz

expressed in the matrix-clause as true. (14)

und nachher habms net einmal mehr recht umadumfahrn können mitm auto, and afterwards they couldn’t even drive around in the car weils kein geld nimmer ghabt habm. because they no longer had any money

(15)

ich möcht auch nicht einifahrn. I wouldn’t want to drive there either na ja sie (-) gezwungenermassen mit ihrem freund halt mit, well she involuntarily with her friend weil der furchtbar gern fischt und (-) because he really likes to fish and

ii. In contrast to the preceding examples with prototypical intonation contours, in (16) and we Wnd main clauses with falling intonation: (16)

momentan bin i auch ein weng erschrockn gwesn. at first I was also a little bit frightened weil i mir denkt hab, because I thought na jetzt hat er sich weh tan. now he’s hurt himself

(17)

jetzt kimmt sie ja nimmer so oft. now she doesn’t come around so often weil sie aa nimmer so beweglich is. because she doesn’t get around so easily

This intonation pattern has no eVect on the semantic interpretation; (16), (17) are as propositional as the aforementioned prototypical examples. We can also Wnd epistemic weil-verb-Wnal clauses displaying the same intonation contour: (18)

A: da war i aber erst ein jahr alt. I was just a year old then B: naja (-) mhm (--) ganz klein bist [halt dort gewesen. yeah mhm you were pretty small then A: [jaja (--) ja weil wir (-) yeah because we weil wir ein jahr beim binder ungefähr waren. because we were at Binder’s about a year

The discourse topic is the local situation in the early Wfties and the correct chronology of events. The content of the weil-clause is intended to corroborate the accuracy of Speaker A’s memory. This weil-clause is a clear instance of epistemic causality, where the weil-clause does not account for the proposition of the preceding main clause but instead, states the evidence which serves as the speaker’s reason for the assertion of the main clause.

On causal clause combining 119

iii. Our examples demonstrate that the intonation structure of the whole complex sentence interacts directly with the information structure: depending on the position of the focus accent and the extension of the focus domain, we get distinct intonation patterns in the main clause — as proposed by Küper and Pasch — but without any impact on the semantic interpretation of the causal relationship. This can be illustrated by means of the (hypothetical) example Karl hat sich am Knie verletzt, weil er hingefallen ist (‘Karl hurt his knee, because he fell down’); diVerent focus domains can be elicited by special questionanswer contexts (focus domains are underlined): (19)

i

A: B:

ii

A: B:

iii

A: B: B:

iv

A: B:

v

A:

B:

Was ist los? (What happened?) Karl hat sich am Knie verletzt→, weil er HINgefallen ist. Warum hat sich Karl am Knie verletzt? (Why/How did Karl hurt his knee?) Karl hat sich am Knie verletzt→, weil er HINgefallen ist. Was ist Karl zugestoßen? (What happened to Karl?) Karl hat sich am Knie verletzt→, weil er HINgefallen ist. Karl hat sich am KNIE verletzt↓. weil er HINgefallen ist. Was ist mit Karl passiert, weil er hingefallen ist? (What happened to Karl because he fell down?) Karl hat sich am KNIE verletzt↓. (weil er hingefallen ist.) Wer hat sich am Knie verletzt, weil er hingefallen ist? (Who hurt his knee because he fell down?) KARL↓ (hat sich am Knie verletzt weil er hingefallen ist.)

As we can see, the main clause ends with a level or low rising intonation pattern if the primary accent (as marker of the focus exponent) lies within the weilclause. If at least one focus exponent lies in the main clause, then this clause ends with a falling intonation pattern. In our data we Wnd that if the main clause displays a level or low rising intonation pattern and the primary accent lies in the weil-clause, then we have an example of a propositional causal relation. The falling intonation pattern can occur with epistemic causality as well, but there has to be an additional marker of the increased non-integration between matrix and causal clause, i.e. a pause inbetween. This clear indication of non-integration results in a diVerent syntactic structure, which could be classiWed as a kind of “afterthought”:

120 Hannes Scheutz

(20)

Karl hat sich am Knie verletzt. (-) [und zwar vermute/behaupte ich das deshalb] weil er hingefallen ist. Karl hurt his knee. [and the reason for my assumption/assertion is] because he fell down.

By displaying a lower degree of integration, this structure can enter a wider spectrum of possible semantic relations with preceding clauses. iv. This discussion of diVerent variants of information structure opens up a new perspective on the issue thematicity vs. rhematicity in weil-clauses (cf. 2.3). Main clauses ending with level or low rising intonation need not be associated with ‘given’ material. Out of all examples cited so far, only the main clause in (1) contains given information. This contradicts Küper’s (1991) thesis that a main clause ending with a rising contour is obligatorily ‘given’. But Gaumann’s (1983) hypothesis that weil-verb-Wnal clauses are ‘given’ gets even less empirical support: in examples (19i-iii) the weil-clauses are clearly rhematic. Examples (19iv, v), where the weil-clauses are indeed thematic, can only be construed as repair-sequences if we are to have an acceptable interpretation (speaker A has not understood a preceding passage and then poses a question) and are thus rather marginal phenomena. v. As far as conversation organization is concerned, weil-verb-Wnal clauses are often used as cohesive devices for turns. Because of the syntactic dependency of the utterance (such as a subordinate clause), a close relationship to the preceding utterance is created. This results in textual cohesion even in cases where the linear ordering of the matrix clause and weil-clause is interrupted, that is, when the string is not produced as a planned complex sentence (main clause + weil clause) by a single speaker. In our conversational data this is well documented by interactively produced turn-pairs like (21); in this way one can retrospectively establish a close textual connection with utterances that do not immediately precede them. (21)

A:

B:

richtig entrisch rennts einem auffi, you get a really weird feeling wenns so mit der em pi a so (-) angschlagen dastehen und (-) drei gleich. when they stand there with their m g ready and three at once des des (-) geht dir gar net [ein. you just don’t get it [ja weil mans gar net gwohnt is. yeah because one isn’t used to it


(22)

A:

B:

A:

ja um des gehts eben. (3sec) yeah that’s what it’s all about sonst hätt i mitn berg schon lang ghandelt. otherwise I would have negotiated with Berg weil (2sec) [weil sonst because because otherwise [vielleicht setzt morgen diese verhandlungen [fort. maybe these negotiations will be continued tomorrow [ja (-) ja per gaudi (-) yeah just for fun weils ja recht gschickt is, net, (-) because it is quite favourable right weils neben der straß is und alles haut ja hin. because it is next to the street and everything is ideal

In (22) the conversation has to do with a piece of land which speaker A wants to obtain in order to keep ducks. The cited excerpt is immediately preceded by the discussion of legal diYculties that have arisen, which argue against the purchase. Earlier in the conversation speaker A had expressed his desire to buy this piece of land. After concluding the legal diYculties at issue and closing the parenthesis (which was opened by speaker B’s proposal and includes A’s response), speaker A resumes the topic of the advantages, listing them in the weil-clauses cited: weils ja recht gschickt is / weils ja neben der straß is und alles haut hin. These weil-clauses do not have any meaningful connection to the immediately preceding utterance; they represent a continuation of something not mentioned immediately before. This seems to be especially noteworthy: it clearly shows that the syntactic dependency or incompleteness of subordinate verb-Wnal-clauses operates as a searching instruction for the conversational participant. He/she is given the signal to look in preceding utterances for adequate antecedents or clauses on which the subordinate clauses can depend. 3.1.2 Non-propositional weil Although the propositional use of weil-verb-Wnal clauses clearly dominates, we also Wnd examples deviating from this pattern in our corpus. We can Wnd genuine epistemic uses as in the aforementioned example (18); however, they are extremely rare. A special kind of epistemic use can be seen in (23): (23)

A: B:

is sie beim öa em te tse? [auch net. is she an auto-club member not even that auch net. (-) [nichts. not even that nothing at all

121

122 Hannes Scheutz

A:

weil die unter umständen vielleicht auch einen rechtsanwalt stellen stellen. because under the circumstances they could perhaps provide a lawyer

Here the weil-clause gives a cause or a justiWcation for the utterance of the speech act represented by the preceding main clause (‘I ask if she is an auto-club member because I want to inform you that they could provide a lawyer’). Therefore, this use is often classiWed as “speech act conjunction” (see e.g. Sweetser (1990), Küper (1991), Günthner (1993)). At Wrst glance, the categorical diVerentiation between propositional, epistemic and speech act causality seems to be plausible. However, there appears to be no substantial diVerence between the epistemic and the speech act use. In the ‘usual’ epistemic case, the speech act type of the main clause is an assertion; to assert (that p) counts as an expression of belief (that p); the weil-clause provides the justiWcation for this belief — or, in speech act theoretical terms, shows that at least one of the preparatory conditions for assertions (“S has evidence for the truth of p”) is fulWlled. Exactly the same is true in the case of the speech act speciWc use, the only diVerence being the fact that these weil-clauses are connected to nonasserting main clause speech acts (e.g. questions, assumptions, advice-givings, etc.). As a consequence, diVerent preparatory conditions are expressed by the respective weil-clauses, all of them including causes for the utterance of the main clause. In the majority of non-propositional cases, the relationship expressed by weil-verb-Wnal clauses is merely pseudo-causal: (24)

na aber des des (-) well that des spielt heut schon sehr viel mit, that plays an important role today weil (-) die frau viel zu selbständig ist ist. because women are too independent

In cases such as this one cannot Wnd any causal connection established by weilverb-Wnal. Weil in (24) simply serves the purpose of marking syntactic dependency on the main clause and can easily be replaced by the semantically unspeciWed subordinator dass (‘that’) with no resulting semantic change. Another form of non-propositional use is demonstrated by weil-verb-Wnal clauses with no discernible semantic relation between the weil-clause and the main clause. In these cases we are dealing with a kind of metacommunicative comment.


(25)

A:

B: A:

so (-) jetzt geht unser auto wieder, (-) now our car is working again weil i s grad (-) hör, (-) because I just heard it dreitausend schilling (-) sind wir leichter worden. (-) we are out three thousand shillings hats leicht was ghabt? was something wrong with it mei des hat s uns zeigt. (-) in jugoslawien. ( — ) it left us in the lurch in yugoslavia weisst aber d schuld haben ja die (-) depperten mechaniker. you know but the stupid mechanics are to blame weil i schon so sagen muss. (-) because I have to tell it like it is wennst es beim mechaniker ghabt hast, if you have had it at the mechanic’s, dann hast die gewähr, dass es dann was hat. then you can be sure that something will be wrong with it

In such cases, the speaker explains the reason for his/her contribution, i.e. its relationship to the overall topic of discourse. The weil-clause has no preceding main clause to which it could attach, and therefore we could speak of an elliptical causal clause here. One could try to Wll the gap by assuming a propositional relationship to a hypothetical main clause ([I propose that] because …) 3.2

Weil-verb-second clauses

3.2.1 Propositional weil The hypothesis that weil-verb-second clauses are used exclusively for epistemic interpretations is equally untenable empirically. The distribution actually found deviates signiWcantly from the predicted results in an even more profound way than in the case of the weil-verb-Wnal clauses discussed above. The empirical facts oVer no support for this hypothesis, even if we try to weaken it considerably. The weaker hypothesis that weil-verb-second clauses are limited IN MOST CASES to epistemic use does not hold either. Weil-verb-second clauses, like weil-verb-Wnal clauses, are used predominantly for expressing propositional causal relationships. (26)

sonst (-) bei die reifen bin i net so kleinlich. but otherwise as far as the tires are concerned I am not so fussy weil (-) mir kommt vor, because I think für einen pkw is halt ein komma sechs (-) eh noch a recht a gutes profil. for a car one point six is a rather good tread

124 Hannes Scheutz

(27)

na aber weißt, i fürcht den bauernschnaps insofern, (-) but you know I am afraid of the peasant’s schnaps to the extent weil (-) i mein halt allweil, because I always think sie sind ein wenig (-) großzügig mitn vorlauf. they are too generous with the first distillation

(28)

er is zwar gleich kommen, (-) he came immediately hat gsagt, i muss mit neue reifen kommen. (-) and said I have to come with new tires aber i kann bis morgen net, but by tomorrow it is impossible weil i muss mirs erst (-) besorgen. (-) gell übermorgen. because I have to buy them first the day after tomorrow

All these examples are undoubtedly cases of propositional weil. The speaker explains the reason for the proposition in the main clause (cf. (26) ‘The reason I am not so fussy as far as the tires are concerned is because…’; (27) ‘The reason I am afraid of the peasant’s schnapps is because…’, etc.). From a formal point of view we should point out that the intonational features ascribed to weil-verb-second clauses can only be considered at most the statistically predominant ones. Neither the Wnal falling intonation pattern of the main clause nor the pause after weil prove to be obligatory. An utterance like that in (28) diVers in no signiWcant way from a prototypical weil-verb-Wnal clause as far as intonation is concerned. Far more important than the formal marking of the diVerence between propositional and epistemic explanation is, in my opinion, the fact that weilverb-second instantiates a syntactically and pragmatically independent utterance. This independence reXects a preference for the reduction of syntactic complexity in spoken discourse, as well as opening up all the possibilities for syntactic organization and diVerentiation of information structure which autonomous structures have. Of particular importance in this context is the preposing of an (often focused) argument in the clause to a position preceding the front Weld (Vorfeld), traditionally known as ‘left dislocation’. This left dislocation structure is restricted to verb-second clauses and functions as a topic-introducing, -expanding or -resuming resource mainly in spoken discourse (for details, see e.g. Scheutz 1997). The following examples give an idea of some of these conversational functions (preposing of contrastive focused elements in (29), resuming of referents previously mentioned in discourse (30), establishing new discourse topics in (31)). In addition, there is a strong, syntactically motivated preference for the use of weil-verb-second (and pre-


posing structures) in complex weil-clauses involving an embedded complement clause as in (32).3 (29)

weil a fahrlehrer allein der darf ja gar nicht unterrichten. (-) net? because as a simple driving instructor he is not allowed to give lessons is he

(30)

i bin net dabeigwesn, I was not around wie sie telefoniert hat nach der reihe. (3 sec) when she kept calling nana des glaub ich sicher. no I’m sure about it weil die wilma di lasst sich das bestimmt nicht entgehn. because Wilma, she doesn’t pass up such a chance

(31)

A:

B:

A:

(32)

A:

B:

A: B:

i glaub halt jetzt eh wieder mehr auf die sachen. now I believe in such things more weil [ja (-) der hundertjährige dem gehts heuer nach. (-) gell, because the farmer’s (century) almanac it will become reality this year [a na geh really ja (-) dem gehts heuer nach. yeah it will become reality this year und drum glaub i wieder was. gell, (-) and for this reason I believe in it again derweil hab i glacht übern hundertjährigen. in the meantime I used to laugh at the almanac und alles andere kommt (-) wenns schneien anfangt (-) in d (-) kühltruhen. (--) and all the other stuff comes in the freezer as soon as it starts snowing aso. (-) da derfst ihnen aber dann keinen namen geben. (-) oh I see. but you must not give them a name weil sobald amal a viech an namen hat, (-) because as soon as an animal has a name iss persönlich [gell? it becomes personal right [ja bringt mans nimmer um. yeah one can no longer kill it

With regard to the aforementioned interactive (topic-introducing) functions of preposing, (31) is especially instructive. The claim which is formulated by means of the weil-clause is addressed in the following turns; the “hundertjährige 〈Kalender〉” (farmer’s/century almanac) remains the topic of discussion for a sequence of 12 turns. In addition to this clearly conversational function of preposing, another characteristic property of weil-verb-second

126 Hannes Scheutz

clauses also becomes clear. In contrast to weil-verb-Wnal clauses, which primarily function backwards, serving to continue or close preceding statements, the independent statements expressed by means of a weil-verb-second clause function in a potentially turn-expanding fashion. They look ahead by constituting — in many cases — the starting point of further contributions to discourse. 3.2.2 Epistemic weil Apart from the great number of propositionally used weil-verb-second clauses, we also Wnd a group of weil-clauses which — along the lines of the hypotheses stated in 2.2. — establish epistemic causal relationships, as the following examples illustrate: (33)

A:

B: A:

(34)

A:

B: A:

[des pflegepersonal da muss schon (-) an idealismus habm.(-) [sondergleichn. the hospital staff must have an incomparably idealistic attitude [na (-) was glaubst denn. [unbedingt. really definitely weil des kann net a jeds. because not everybody is able to do this weil die müssn schon a lieb mitbringen für sowas. because they must love doing such things der war a achtervierziger. oder a neunervierziger jahrgang sowas war er. (-) he was born in the year forty-eight or forty-nine something like that ah so really ja. weil i (-) bin fuffzig. yeah because I was born in fifty

Here, the preceding claim is corroborated or supported by supplying presuppositions or assumptions as evidence. For instance, in (33) the claim that the hospital personnel must be idealistic is supported by the conviction that not every one can do that, and that one must have a special love for this kind of task. In (34) the assumption that someone was born in the year 1948 or 1949 is corroborated by the statement that the speaker was born in the year 1950, etc. In all these cases the syntactic and illocutionary independence of the weilclause is evident; apart from the semantics speciWc to these constructions, there are no noticeable diVerences from the properties of the propositional weilverb-second clauses discussed above.


3.2.3 Indirect causal connections In a great number of weil-verb-second examples the relationship between the explaining and the explained clause appears implausible. In these cases we are very often dealing with forms of indirect discourse which would result in false or meaningless causal relations if the hearer did not reinterpret them properly. (35)

A:

B: A:

no und des hab i eh (-) so ein fasserl mit dreiundzwanzig liter, (-) and I have like a twenty-three-liter barrel dreiundzwanzig und an halbm liter, des hab i angfüllt. (-) twenty-three and a half liters this I filled up no bin neugierig, (-) i hoff er wird was. gell, (-) I am curious I hope it will be good kriegst einen richtigen apfelwein. You will get a good apple wine weil (-) so frühäpfel des is ja net des ideale. because early apples they are not the best

By uttering I am curious, I hope it will be good, the speaker expresses a wish and at the same time his hopes for good results. The fact that he also leaves open the possibility of a negative outcome, that he is actually concerned about it, becomes clear from the following weil-clause, which refers speciWcally to this negative aspect. The weil-clause underlines the skepticism (already potentially available in the foregoing utterance) by referring to a possible reason for failure. The whole utterance could be paraphrased as I hope my apple wine will be good, because early apples are not the best for it, which is however meaningless in a literal reading. What we need is a reasoning process which reinterprets the literal meaning according to the Gricean concept of conversational implicature to determine the actually intended meaning. With this reinterpretation of the main clause I hope it will be good → I fear it could be bad, the causal justiWcation can be directly applied. 3.3 Weil as a discourse particle Finally, there are cases in which an antecedent main clause is no longer reconstructable, however hard one tries; weil evidently has merely the function of signalling general syntactic continuation or of establishing cohesion. This frequent junctor function, virtually empty from a semantic point of view, is conversationally very important as a Xoor-holding device. By this means, the speaker signals that he/she intends to continue the turn. The causal conjunction weil is here reduced to its inherent textual cohesive function and is inserted as a kind of discourse particle (cf. Gohl and Günthner 1999). This

128 Hannes Scheutz

point becomes clear in (36), where the reactions of the hearer support the Xoor-holding intention of the speaker expressed by weil and do not take over a turn for themselves; in (37) the continuation of the turn is emphatically claimed by the use of weil. (36)

A:

B:

A:

(37)

A:

B:

4.

des kann schon sein auch. (--) that may be dass es da auch a weng hapert. (2 sec.) that there are some difficulties aber es muss nicht unbedingt [sein. (-) weil [ahh (2 sec.) but it need not necessarily be so because [na eh net. [na a net.(-) a wieder net. really not really not really not oft einmal wirklich leut von (-) there are often people of sagn mir angenommen wieder wie bei meinem schwager say lets assume in the case of my brother-in-law ja jetzt geh i heim, weil [(2 sec.) weil i eben heim gehn muß. now I am going home because because I simply have to go home weil was. [weil because what

Conversational functions

Though our analysis is essentially based on conversational data, the question if diVerent weil-clauses are used to fulWll particular interactional functions still remains open. Such an investigation must occur on two levels: one must investigate, Wrst, the basic conversational function of a clause introduced by weil, and second, the possible functional diVerences between weil-verb-Wnal and weil-verb-second clauses. 4.1 General functions of weil-clauses We found that all of the causal weil-clauses occur exclusively AFTER the matrix clause,4 irrespective of the position of the verb. This is the canonical form for the weil-clauses with verb-second position. However, it would also have been possible for the weil-verb-Wnal clauses to occur in the front-Weld (Vorfeld), before the Wnite verb of the main clause (cf. 2.1. i)).


A cursory examination of weil-clauses in written language also shows a strong tendency to position causal adverbials after the main clause but one can Wnd a certain number of weil-clauses in the front-Weld.5 The interesting phenomenon that the front-Weld position was not present in my data at all must be a consequence of the speciWc requirements of spontaneous everyday communication. Auer’s (2000) thesis that “causes iconically precede their eVects” seems to be reversed, at least in spoken language: the speakers Wrst make a statement and only afterwards give the reason for the assertion or assumption. It is especially noticeable that this order reXects successive cognitive processes precisely in those cases where the weil-clause occurs after a hesitation following a main clause ending with falling intonation. These turn-continuations after a (possible) completion point could be considered “afterthoughts”, explanations, justiWcations etc. It is at the end of the main clause that the interlocutor’s reactions (conWrmation or contradiction of the speaker’s statement, explicit inquiries, etc.) are found; these evoke the turn-continuation that then follows in the causal clause. A continuation by the same speaker is preferred. One can Wnd a typical example in (34), where the interlocutor’s question evokes a turncontinuation by A. However, it is also possible for the interlocutor to supply a causal continuation of the previous turn: (38)

A:

B:

aber jetz hört man eigentlich nichts mehr [so. (-) is es (-) ruhig worden. but now one actually no longer hears anything it has become quiet [nja weils uninteressant wird.(-) mit der zeit. because it becomes less interesting as time passes

In this example, the connection with weil is strengthened by the syntactic criterion of dependence (verb-Wnal), initiating the search for a suitable antecedent. Finally, the turn-initial position is the usual location for demonstrating the relevance of the current turn to prior utterances. This means that this type of weil-clause has a decidedly backwards orientation; it brings a thematic unit to a close. Examples like (22) are prototypical for this type of clause, sentences in which the antecedent clause has preceded much earlier, i.e. the causal clause completes the previously opened thematic frame only after an intervening passage. The connective function of weil becomes especially obvious when the sentences introduced by weil do not demonstrate any type of causal connection

130 Hannes Scheutz

to previous utterances at all. The connective function inherent in the speciWc causal semantics is exploited: the meaning of weil is reduced to an unspeciWed connective function, in this case weil functions as a general relevance marker in conversation. Metacommunicative or other comments which merely suggest “causality” in the form of weil-clauses provide numerous examples (cf. (24)–(25)). This function of weil as a relevance marker also provides the basis for its use as a ‘Xoor-holding device’ with which the speaker assures his/her right to continue the turn or claims the right to speak after an intervening turn. In passages in which the conversation lags, this function is often used to avoid an anticipated ‘lapse’ after the ending of a turn. If no other speaker takes over the Xoor, the current speaker continues his/her turn. An afterthought introduced by weil serves as an ideal ‘subtle’ expansion of an already completed turn, since it can also be used to suggest that the speaker has used the pause after the completion point for reXection and had no intention of really ending his/her turn. The numerous hesitations, reformulations and syntactic breaks occurring after an introductory weil demonstrate that it is indeed used in this way. It is clear that the speaker did not intend to continue the conversation by giving a reason for a previous statement but that weil is used here merely to maintain the Xow of conversation. Often a speaker can Wnd a relevant contribution to the ongoing exchange within the turn-continuation introduced by weil only after several attempts and false starts. The weil-clause is interrupted and replaced by another syntactic construction signalling the beginning of a new topic (cf. (36)). In addition, the large number of incomplete weil-clauses (approximately 15% of the entire corpus) — often immediately broken oV after weil — demonstrates that the search for causality frequently takes place only after the utterance of a statement and is therefore often unsuccessful. 4.2 Functional diVerences between verb-Wnal and verb-second clauses The conversational implications discussed up to now are generally valid for all causal clauses introduced by weil. The next step is to investigate possible functional correspondences reXecting the syntactic diVerence between the verb-Wnal and verb-second positions. i. The verb-Wnal position, indicating a syntactically and pragmatically dependent clause, expresses a stronger cohesion and connection to the previous utterance than the independent weil-verb-second clause does, since the latter


has an illocutionary force of its own. Although our data do not exclude other possibilities, there seems to be a deWnite tendency for verb-Wnal clauses to be sequence-terminating. They often express subordinate themes (commentary, background information etc.) tangential to the primary discourse topic. (17')

und sagn wir mit dem dirndl, (--) and as far as the girl is concerned jetzt kimmt sie ja nimmer so oft. now she doesn’t come around so often weil sie aa nimmer so beweglich is. (-) because she doesn’t get around so easily aber früher is sie jedes wochnend da gwesn, (-) but she used to be here every weekend hast ganz vernünftig redn können. you could really talk to her

The topic of A’s turn (continued in subsequent turns) is a handicapped child with whom one can talk intelligently in spite of her handicap. Parenthetically inserted into the framing sentence is background information indicating that the girl no longer shows up as regularly as she did earlier. The verb-Wnal clause providing the reason for the diminishing frequency of the visits is attached to the main clause within the parenthesis. This clause clearly deals with information not integral to the main topic — it is not previously mentioned nor is it later referred to. There appears to be a correspondence between syntactic/pragmatic dependency and downgrading of the thematic relevance of the weil-clause. On the syntactic level, the weil-clause closes the structure initiated by the main clause; comparably, on the conversational level, it signals the completion of the conversational activity connected with the main clause. This is thus the interactive counterpart of the syntactic “Gestalt”-closing function of weil-verb-Wnal. In the vast majority of cases, the reason expressed in the weil-clause serves to support a statement and thereby contributes to the development of the primary discussion topic without becoming a discussion topic itself. ii. At Wrst glance, causal clauses in verb-second format do not diVer unambiguously from the verb-Wnal clauses discussed above. They can also be used to express reasons which function only as adjuncts to the primary discourse topic, background information, parenthetical inserted commentaries etc.; cf. (39), where the discourse topic is a story about troubles at the Austrian-Czech border. In the excerpt cited — the exposition of the narration — the speaker tells how they came to the border via several towns. The inserted weil-clauses

131

132 Hannes Scheutz

constitute a secondary level, which discusses the party’s motivation for going to these towns. (39)

s nächst sind wir (-) d evi und ich und ihr freund recently we evi me and her boyfriend sind wir in kollerschlag draust gwesn we were at kollerschlag weils (-) aah (-) da bei dem hofman von kollerschlag hat hats den küchnbelag kriegt because from hofman at kollerschlag she got the kitchen flooring sind wir draust gwesn das aussuchn we were there selecting one und dann sind wir auf wegscheid weitergfahrn and then we continued driving to wegscheid weil ja (-) aah (--) da kaufn kaufns ja gern einen tabak because they like to buy tobacco there weisst zum wuzln so einen you know the kind you can roll a cigarette with und den kriegens nur drausst den sie mögn and the kind they like they can only get it there sind wir da über d grenz gfahrn so we crossed the border

In contrast to verb-Wnal clauses, however, verb-second clauses, as syntactically and pragmatically independent utterances, often precipitate the introduction of a new thematic aspect into the conversation: (40) 〈The discourse topic is schnapps, its production and storage, etc.〉 A: du aber das mit der farb, (-) say, about the color weil das sagt allweil der n.n., because n.n. says da tun’s was dazu zum schnaps. they add something to the schnapps weil der jugoslawische slibowitz, der is doch gelb. (-) gell, because the yugoslavian slibowitz it is yellow you know right B: ja aber wahrscheinlich durch (--) yes but probably by richtig gelb wird er durchs ablagern. it really becomes yellow in storage A: das glaub ich aber nicht. i don’t believe that [20 more turns follow about manipulation and the yellow color of the schnapps]

Even an inserted statement formulated as causation may subsequently become the theme of conversation, as in (40). In doing so, the verb-second format, as an autonomous form of expression, shows its potential for expanding turns and/or sequences.


5.

Conclusions

In recent years, various form-function correspondence analyses have been proposed for the data discussed here. As a conclusion of our investigation of weil-clauses, we propose that none of the strong versions of the usually propagated accounts can be veriWed stricto sensu by our empirical facts. This is particularly true of the theses by Keller (1993) and Wegener (1993), who claim that the diVerence between propositional and epistemic causality is in direct correspondence to the syntactic opposition of weil-verb-Wnal vs. weil-verbsecond clauses. The diVerent types of weil-clause behave diVerently to the degree that the spectrum of functions they realize varies. The weil-verb-Wnal pattern is most often used in propositional causal relationships and, in general, proves to be the functionally more restricted form as compared to weil-verbsecond clauses. But it is simply not the case that the use of weil-verb-Wnal is Wxed and specialized to propositional causality, and weil-verb-second to epistemic causality. Rather, it seems to be the case that a unidirectional implicational relationship exists between the tightness of syntactic integration and the semantic interpretation of a causal adverbial. The tighter the syntactic integration of a causal adverbial is, the stricter is its specialization to the semantic type of propositional causality. But, on the other hand, the establishment of propositional relationships is not bound to the actualization of syntactically integrated causal adverbials. This can be easily shown by means of some prototypical examples. We have maximal integration in the realization of a causal adverbial as a prepositional phrase in a simple sentence. This form is exclusively reserved for the implementation of propositional reason relationships, which refer to the causal circumstances of the action or state of aVairs expressed by means of the verbal phrase. (41)

Wegen des Frostes sind die Heizungsröhren geplatzt. Because of frost the heating pipes burst.

If we try to express epistemic reason relationships this way, we get unacceptable results. Without exception, prepositional phrases force propositional interpretations (cf. the ungrammaticality of (42), where an epistemic relationship is expressed by a prepositional phrase). (42)

*Wegen der geplatzten Heizungsröhren hat es Frost gegeben *Because of the burst heating pipes there has been a frost.

134 Hannes Scheutz

On other occasions when an epistemic weil-clause is reformulated as a prepositional phrase, we Wnd semantic recategorization (cf. the epistemic weil-verbsecond clause of (43) vs. the reformulations in (44, 45) with clearly propositional readings). (43)

Er ist wieder betrunken, weil — sie läuft deprimiert durch die Gegend. He is drunk again, because she is going around in a depressed mood.

(44)

Er ist wegen ihrer Depression betrunken. He is drunk because of her depressed mood.

(45)

Sie ist wegen seines Betrunkenseins depressiv. She is in a depressed mood because of his drunkenness.

The next step is to express the causal adverbial by means of a syntactically subordinate clause. When the focus exponent lies in the causal clause and a coherent intonation phrase encompasses the matrix and constituent clause, then the reference of the causal adverbial to the VP of the main clause is obligatory. (44')

Er ist betrunken, weil sie depressiv ist. He is drunk because she is in a depressed mood.

(45')

Sie ist depressiv, weil er betrunken ist. She is in a depressed mood because he is drunk.

This obligatory propositional reading can be annulled only in special cases, for instance when the main clause is epistemic (46) or in restricted contexts as in (47): (46)

(47)

Er scheint heute wieder betrunken zu sein, weil sie so deprimiert durch die Gegend läuft. He seems to be drunk again today, because she is going around in a depressed mood. A: B: A:

Er ist betrunken. He is drunk. Das kannst du doch gar nicht wissen. You can’t know that. Er ist betrunken, weil er so dahintorkelt. He is drunk because he is staggering.

Increasing syntactic non-integration in a kind of extraposition construction, which is signalled by means of the existence of autonomous intonational phrases over the matrix and the constituent clause, leads to the abandoning of a strictly propositional, intrasentential causal relationship; an utterance such as


(48)

Er ist betrunken. (-) weil sie so deprimiert ist.

allows for both readings. Both the propositional reading (‘he is drunk because of her depression’) and the epistemic reading (‘her depression causes the assumption/assertion that he must be drunk’) seem possible. Verb-second position in the weil-clause is a further step towards a syntactically non-integrated structure. It leads in the end to an autonomous statement, which is connected with the preceding statement to form a complex construction. With this construction, an explicit reason-relationship is established, but the domain of reference of the weil-clause and the speciWc interpretation of the reason relationship are not dictated by the formal features. We can see this clearly if we change our prototypical example of an epistemic reason relationship (43) to the formally equivalent statement (49): (43)

Er ist wieder betrunken, weil — sie läuft so deprimiert durch die Gegend. He is drunk again, because she is going around in a depressed mood.

(49)

Er ist wieder betrunken, weil — er hat die schlechte Nachricht einfach nicht verkraftet. He is drunk again, because he simply couldn’t take the bad news.

In (49) the verb-second weil-clause can hardly be interpreted as an epistemic reason relationship. What seems to be crucially important for the obligatory propositional interpretation in this connection is the semantics of the two participating clauses. These insights enable us to reinterpret the aforementioned ungrammaticality of (8') vs. (7'). (7')

Die Heizungsröhren sind deshalb geplatzt, weil es Frost gegeben hat. The heating pipes burst for the reason that there was a frost.

(8')

*Die Heizungsröhren sind deshalb geplatzt, weil — der Keller ist überflutet. *The heating pipes burst for the reason that the basement is flooded

The possibility of explicitly referring to the causal clause by means of the proform deshalb (for this reason/for the reason that), which, strictly speaking, transforms the weil-clause into an attributive clause, is a decisive indication of a very close syntactic relationship, since only propositional causality can be expressed this way. So the ungrammaticality of (8') is not primarily due to the

135

136 Hannes Scheutz

syntactic autonomy of the causal clause — as is assumed by Wegener (1993) — but to the close syntactic integration created by the pronoun — a constituent sentence structure which is not consistent with non-propositional causality. The crucial role of the semantics of the two participating clauses for the establishment and adequate interpretation of a speciWc causal relationship becomes even more obvious if we delete the explicit operator weil and then compare the clauses introduced by the conjunction with the same clauses not introduced by any conjunction at all. The propositional interpretation of (50'') doesn’t change, in contrast to (50, 50'). This is also true in the case of the epistemic interpretation of (51''), in contrast to (51, 51'): (50)

(50') (50'') (51)

(51') (51'')

Die Heizungsröhren sind gegeben hat. The heating pipes burst Die Heizungsröhren sind Frost gegeben. Die Heizungsröhren sind gegeben.

geplatzt, weil es heute nacht Frost

because there was frost last night. geplatzt, weil es hat heute nacht geplatzt, es hat heute nacht Frost

Er ist noch nicht nach Hause gegangen, weil in seinem Büro noch Licht brennt. He hasn’t gone home yet, because in his office the light is still on Er ist noch nicht nach Hause gegangen, weil in seinem Büro brennt noch Licht. Er ist noch nicht nach Hause gegangen, in seinem Büro brennt noch Licht.

It is evident that the interpretation of the overall sentence is essentially based on the partial meanings of the two clauses immediately aVected, which are related to one another according to the Gricean maxims of conversation. Provided that the speaker is adhering to the conversational maxims, or at least the cooperative principle, the uttering of part B (in his oYce the light is still on) after assuming part A (he hasn’t gone home yet) is acceptable if and only if B is in a content relationship to A (cf. * He hasn’t gone home yet, in his oYce the light is still on, but the one fact has nothing to do with the other). The hearer has to work out an adequate interpretation relying on the context, linguistic or otherwise, of the utterance, and on other items of background knowledge. This interpretation procedure is essentially based on pragmatic principles and does not depend on the use of explicitly expressed causal conjunctions. On the conversational level, introducing a clause by weil generally indicates a high degree of connectedness and relevance; the exclusive use of this very function is demonstrated by the numerous examples showing weil as a


semantically empty connector, merely marking the relevance of the present turn to prior conversation without any causal relationship. The syntactic diVerence between weil-verb-Wnal and weil-verb-second clauses has no deWnite and exclusive counterpart on the interactive level; nevertheless, there is a clear preference for using verb-Wnal as a primarily retrospective and sequence-closing device, whereas verb-second can function as a theme-introducing and turn-expanding structure as well. Summarizing our results, there actually seems to be a general rule of correspondence between the formal and functional properties of weil-clauses. The tighter the structural connectedness and dependence of a construction is, the more restricted are its functional possibilities. This applies on both the semantic and the interactive levels: verb-Wnal clauses (as opposed to verbsecond clauses) are the more “marked” construction and therefore have a more limited spectrum of functions.

Notes * I am very grateful to Ioannis Fykias and Kandace Einbeck for correcting my English. 1. Examples from my corpus of spoken German are given in a rather broad transcription; only those features are noted exactly which are relevant to a detailed description of the phenomenon discussed; many dialectal (Middle Bavarian) features which are of no interest for our purposes have been deleted. Some transcription conventions: – simultaneously spoken passages are listed one above the other in diVerent lines, their respective beginnings are indicated by [ – pauses of up to one second are noted with (-) or (--); for longer intervals, the exact duration in seconds is indicated – hesitation signals (‘Wlled pauses’) are noted with ahh – intonation contours at the end of syntactic or intonation phrases are indicated as follows: . falling , level or low rising ? rising 2. This paper is a moderately revised version of a paper presented at the Sixth International Pragmatics Conference in Reims in 1998; papers published afterwards are not taken into consideration. 3. The left dislocated elements are underlined. 4. This result agrees completely with that of Ford (1993) for causal clauses in spontaneous English conversation.

138 Hannes Scheutz

5. In Kafka’s novel “Das Schloß” the proportion of fronted to post-positioned weil-clauses is 1:7.


“Pre- and post-positioning of wenn-clauses in spoken and written German”. In Cause, Condition, Concession, Contrast: Cognitive and discourse perspectives, Elizabeth Couper-Kuhlen and Bernd Kortmann (eds), 173– 204. Berlin: Mouton de Gruyter.

Ford, C. E. 1993 Grammar in Interaction. Cambridge: Cambridge University Press. Gaumann, U. 1983 “Weil die machen jetzt bald zu” Angabe- und Junktivsatz in der deutschen Gegenwartssprache. Göppingen: Kümmerle Verlag. Günthner, S. 1993 “ ‘...weil — man kann es ja wissenschaftlich untersuchen’ — Diskurspragmatische Aspekte der Wortstellung in WEIL-Sätzen”. In Linguistische Berichte 143: 37–59. 1996 “From subordination to coordination? Verb-second position in German causal and concessive constructions”. In Pragmatics 6 (3): 323–356. Gohl, Ch. and Günthner, S. 1999 “Grammatikalisierung von weil als Diskursmarker in der gesprochenen Sprache”. In Zeitschrift für Sprachwissenschaft 18 (1): 39–75. Keller, R. 1993 “Das epistemische weil - Bedeutungswandel einer Konjunktion”. In Sprachgeschichte und Sprachkritik. Festschrift Peter v. Polenz, H.-J. Heringer and G.Stötzel (eds), 219–247. Berlin: de Gruyter. 1995 “The epistemic weil”. In Subjectivity and Subjectivization, D. Stein and S. Wright (eds), 16–30. Cambridge: Cambridge University Press. König, E. and Auwera, J.v.d. 1988 “Clause integration in German and Dutch conditionals, concessive conditionals, and concessives”. In Clause Combining in Grammar and Discourse, J. Haiman and S. A. Thompson (eds), 101–133. Amsterdam: Benjamins. Küper, Ch. 1991 “Geht die Nebensatzstellung im Deutschen verloren? — Zur pragmatischen Funktion der Wortstellung in Haupt- und Nebensätzen”. In Deutsche Sprache 19: 133–158. Pasch, R. 1983 “Untersuchungen zu den Gebrauchsbedingungen der deutschen Kausalkonjunktionen da, denn und weil”. In Linguistische Studien, Reihe A 104: 41–243.


1997 Scheutz, H. 1997

1998

“Weil mit Hauptsatz – Kuckucksei im denn-Nest?” In Deutsche Sprache 25: 252–271. “Satzinitiale Voranstellungen im gesprochenen Deutsch als Mittel der Themensteuerung und Referenzkonstitution”. In Studien zur Syntax des gesprochenen Deutsch, P. Schlobinski (ed.), 27–54. Opladen: Westdeutscher Verlag. “weil im gesprochenen Deutsch”. In Beiträge zur Dialektologie des ostoberdeutschen Raumes. Referate der 6. Arbeitstagung für bayerisch-österreichische Dialektologie 1995 in Graz, C.-J. Hutterer and G. Pauritsch (eds), 85–112. Göppingen: Kümmerle Verlag.

Schlobinski, P. 1992 Funktionale Grammatik und Sprachbeschreibung. Eine Untersuchung zum gesprochenen Deutsch sowie zum Chinesischen. Opladen: Westdeutscher Verlag. Selting, M. 1999 “Kontinuität und Wandel der Verbstellung von ahd. wanta bis gwd. weil. Zur historischen und vergleichenden Syntax der weil-Konstruktionen”. In Zeitschrift für germanistische Linguistik 27 (2): 167–204. Sweetser, E. 1990 From Etymology to Pragmatics. Metaphorical and cultural aspects of semantic structure. Cambridge: Cambridge University Press. Uhmann, S. 1996 “Nur ein Sturm im Lexikonglas. Zur aktuellen Verbstellungsvariation in weil-Sätzen”. In Wuppertaler Arbeitspapiere zur Sprachwissenschaft 13: 1– 26. 1998 “Verbstellungsvariation in weil-Sätzen: Lexikalische DiVerenzierung mit grammatischen Folgen”. In Zeitschrift für Sprachwissenschaft 17 (1): 92– 139. Wegener, H. 1993 “weil - das hat schon seinen Grund. Zur Verbstellung in Kausalsätzen mit weil im gegenwärtigen Deutsch”. In Deutsche Sprache 21: 289–305. 1999 “Syntaxwandel und Degrammatikalisierung im heutigen Deutsch? Noch einmal zu weil-Verbzweit.” In Deutsche Sprache 27: 3–26.

Dutch ‘but’ as a sequential conjunction Its use as a resumption marker Harrie Mazeland and Mike Huiskes* University of Groningen/University of Utrecht

In Dutch, the connective maar (‘but’) is recurrently used as a sequential conjunction linking turns. In the type of use we will focus on, it alludes to a contrast at the level of discourse organization. The speaker resumes a line of development that was abandoned in favor of a competing line of talk, and proposes that the abandoned line becomes the focus of talk again. Speakers can return in diVerent ways to an abandoned line of talk. Continuations present the return as an unproblematic next step, whereas resumptions deal with the return as a task that is not an unproblematic thing to do (Section 2). We discuss two prototypical environments in which resumptions occur: after repair-sequence expansion, and after a competing line of topic development. They have in common that the preceding line of talk is oriented to as an alternative to the line that is resumed (Section 3). The turn-constructional format is examined in Section 4. Apart from the level of linguistic form — the use of the connective as the turn’s Wrst term -, two other levels of format analysis are relevant. First the level of establishing a relation with the context (the way the tying with an earlier line of talk is accomplished). Second, the level of unit positioning within current turn (the resumption occurs in turn-initial position and it is preliminary to a next turn-constructional unit in which the resumed line of talk is continued). We compare the construction format with a similar construction type that is used as a vehicle for proposing topic closing in Section 5. In the concluding section (Section 6), we summarize our analysis and discuss its consequences for the semantics of conjunctions. Instead of a monosemic or polysemic analysis from a perspective that takes its starting point in the language system, we advocate an approach in which utterance types are characterized in terms of sequential features.

142 Harrie Mazeland and Mike Huiskes

1.

Introduction

A conjunction such as but, and or because can be used as a device for linking units within the same turn, but can also be used for linking turns. We use the term sequential conjunction to refer to the use of connectives — conjunctions, conjunctive adverbs and other types of lexicalized expressions — for specifying relations between turns.1 A sequential conjunction gives an indication of current turn’s relationship to the preceding talk. For the time being, we want to diVerentiate this type of use from the one in which it is employed to link clauses or phrases within the same turn (e.g., when a speaker is relating the current turn-constructional unit to the preceding one). Connecting turnconstructional units within the same turn might be diVerent from the work a speaker does when indicating a relationship between the current turn and its interactional context.2 Several types of use of the Dutch connective maar (‘but’) as a sequential conjunction can be discerned. A prominent type of use we will not deal with in this paper is its use in objections, as in line 66 of the following fragment3 (the Dutch original is followed by an English translation): (1)

Girls’ argument. Ans and Bea are teenager girl friends in high school. Ans has accused Bea’s boyfriend of the theft of her motor bike.

58 Bea: 59 60 Ans?: 61 Ans: 62 Bea?: 63 64 Bea: 65 66 Ans:→ 67 68 Bea: 69 Ans: --------58 Bea: 59 60 Ans?: 61 Ans: 62 Bea?: 63 64 Bea: 65 66 Ans:→ 67 68 Bea: 69 Ans:

hij he↑ef’t ↓niet expres gedaan 0.4 mhhHEHhh= =>WAAROM zegt ie dat dan OMDAT IE GEWOON DOOR EEN MEI:SJE IS GE↓BEL:D 0.3 >maar dat kan ie toch ook verzonnen °heb↑beh 0.3 waarom zou die dat verz[inneh [dat weet ik ni↑et

he didn’t do it on purpose 0.4 mhhhuhhh= =then why did he say this about this girl who called him= =Hh 0.3 because he was called by a girl of course. 0.3 but he could have made that up, right? 0.3 why would he make up a thing like [that [how should I know?

Dutch ‘but’ as a sequential conjunction 143

In line 66, Ans objects to the claim her friend made in the preceding turn. The objection-making relationship with prior turn is marked from the start by prefacing the turn with maar (‘but’). The type of use of maar we will examine in this paper does not have this objection-making quality, at least not at the level of substantive stance-taking. An example of the type of use we will concentrate on is given in the following fragment. It is from a conversation of the same group of high school pupils as fragment (1). In the arrowed turn (lines 332–33), the story-teller Jan resumes the telling of a story that was suspended after a repair initiation of his recipient (lines 317–331). The turn in which the return is made begins with maar: (2)

The-boy-from-my-former-school story. From the same group of high school pupils as fragment (1). Two boys and two girls are doing homework, sitting around the same table. During work, they talk about other things.

313 Jan: 314 315 316 317 Ans: 318 319 Jan: 320 Ans: 321 322 Jan: 323 324 Ans: 325 Jan: 326 327 Ans: 328 329 Ans: 330 331 Ans: 332 Jan:→ 333 334 335 336 337 338 339 Bea: --------313 Jan: 314 315 316 317 Ans:

weet jeh, (.) >op me ouwe school ↑hè (0.4) daar zat zo’n go:zer en die(jie-) die mocht ik niet zo↓:h (0.3) en da was heel grappig: [(want e:h) [(°hoe heet(t)↓ieh)° 0.5 ↑wat= =en hoe heette die (.) alexander, 1.3 >op welke school heb je geze:t:n dan.= =op sint maart↓‘ns:. (.) o↓:h. 0.3 (lage ge↑:) 0.2 goe[d. [(°>ga verder)] [maar di[e e : : : : ]:h (.) die e:h (0.6) die (°jongeh°) (0.3) die mocht ik niet zo.= =>da was ook ‘n heel vreemde vogel:,< ·h en laa↓:tst >toen: zag ik zij↓n:- (.) zijn adres en z’n naam bij CHIN↓a express ↑hè (0.4) omdat ie zich had opge- opgegeveh. (.) bij mijn wer:k, ·h[h (°om)dat ie d’r o↑ok wou werke [hm°↓m

listen, (.) at my former school, you know (0.4) there was this guy and he (uh-) I didn’t like him very much. (0.3)and this was really hilarious [(because uh) [(°what was his name.°)


318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339

Jan: Ans: Jan: Ans: Jan: Ans: Ans: Ans: Jan:→

Bea:

0.5 what?= =°and what was his name (.) Alexander, 1.3 which school was it. =Saint Martin’s. ((name of the school)) (.) o:h. 0.3 (Lower Quay,) ((name of the school’s area)) 0.2 [(go ahead) ] rig [ht. [but thi [s u: : : : ]:h (.) this u:h (0.6) this (°boy°) (0.3) I didn’t like him very much. it was really a very strange guy, ·h and not so long ago I saw his- (.) his address and his name at China Express you know, (0.4) because he had si- signed himself up (.) at my work, ·h[h (because) he wanted to work there too, [hm°↓m (…) ((story continues))

In the turn that begins in line 332, maar is used as a resumption marker. It prefaces an initial turn-constructional unit in which the speaker returns to his story. Inbetween, the story’s delivery is stopped because of the repair initiated by its recipient (lines 317–331). Contrary to the type of use documented in fragment (1), maar is not prefacing a disagreeing utterance here. Note that it does not tolerate (pre-dis-) agreement tokens like yes or no in front of it. Substantial disagreement is not what is at stake here. Rather, the connective signals some kind of contrast at the level of the organization of the discourse. The current turn is not seamlessly continuing the activity the participants were involved in in last turn, but is resuming an activity that was abandoned in the talk that led up to it. It is this resuming use of maar we will analyse.

2.

Continue versus resume

If a next speaker moves on by returning to material further back than prior turn, he can either present his turn as a continuation of this earlier talk or do the return as a resumption (cf. JeVerson 1972: 319 V.). The decision to present the next turn as one or the other is not made automatically. Depending on how a speaker packages a next turn, the preceding talk is either dealt with as some-


thing that can be integrated unproblematically into the ongoing activity (continuation) or as something which is moved away from when the abandoned line is picked up again (resumption). Compare, for example, the diVerent ways in which the story-teller in the fragment below goes on after an inserted repair sequence. In lines 29–30, the speaker continues the telling of his story with en (‘and’). In lines 53–54, however, the telling is resumed by recycling a story component in a turn beginning with maar. We will gloss the moves that are of interest in the right margin of the corresponding lines in the translation: (3)

Hiking story. From a phone call between friends

22 Bob: 23 24 25 Eric:→ 26 Bob: 27 28 29 Bob:→ 30 31 32 33 34 35 36 37 38 Bob: 39 40 41 42 43 44 Eric:→ 45 Bob: 46 47 Eric: 48 49 Bob: 50 51 52 Eric: 53 Bob:→ 54 55 56 Eric: ----------

nou↓eh:: (.) we zijn daar aa::ngekomen (0.7) en eh: (.) we zijn naar ‘t (.) Lac du la:Gileppe gerejen (.) repair initiation ‘t wa:t= =Lac du la Gileppe (.) dat is een heel groot stu:wmeer 1.5 continuation ·hh en eh::: (.) nou: da (.) daar was ‘t oeh:: (.) hEE:l erg regenweer (.) daar was ammel snee::uw en eh ha::gel en zo (.) toen hebben we op een of a:ndere:: (.) mt (.) afdakje hebben we (.) onder gezeten (.) hebben we erwtensoe:p gemaakt ·hhh en hebwe:: (.) ‘n heleboel gegeten we hadden voor twee: dagen eten bij (.)·hh en eh:: 2.0 nou: (.) toen- toen werd ‘t droo:g (.) wonder boven wonder (.) hh toen zijn we aa::ngelopen (.) toen was ‘t al na:cht (.) of toen was ‘t al avond toen was ‘t al donker (.) toen zijn we aangelopen (.) en eh:: (.) he’we ongeveer vie:r of vij:f kilometer gelopen of zo:= repair initiation =in ‘t donker?= =ja (.) oh (.) wij veronderstelden (.) dat (.) dat jullie uitgewerkt waren (.) zo gauw ‘t donker waren (.) was= =nou (.) we zijn toe:n gewoon doorgelopen want we hadden nog geen slaapplek (.) >jaja= =·hh maar toen eh (.) nou toen zijn we gewoon resumption de b↑o:ssen ingelopen (.) we hadden geen lamp bij en da was wel heel lastig jaha


22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56

Bob:

well uh:: (.) we got there (0.7) and uh: (.) we then drove to the (.) Lac du la Gileppe (.) repair initiation Eric:→ the what= Bob: =Lac du la Gileppe (.) that’s a very big artificial lake 1.5 continuation Bob:→ ·hh and uh::: (.) well there (.) there toothe weather uh:: (.) very rainy weather (.) it was all snow and uh hail and stuff (.) then we sat at some (.) mt (.) canopy we sat (.) under it (.) made pea soup ·hhh and ate (.) lots of it. we had two days of food with us (.) ·hh and uh:: 2.0 Bob: well: (.) then- then the rain stopped (.) wonder of all wonders (.) hh then we started to walk (.) it was already night by then (.) or already evening time it was already dark then (.) then we started to walk (.) and uh:: (.) we did about four or five kilometers or something:= repair initiation Eric:→ =in the dark?= Bob: =yes (.) Eric: oh (.) we supposed (.) that (.) that you guys were finished (.) as soon as it were dark (.) was4= Bob: =well (.) we just walked on then because we didn’t have a place to sleep yet (.) Eric: >yes yes= resumption Bob:→ =·hh but then uh (.) well then we just walked into the woods (.) we didn’t have a flashlight with us and it was really very difficult Eric: yhes.

The talk after the repair sequence in lines 25–28 is marked in several ways as a continuation. Bob’s weather report in lines 29–30 no longer deals with the trouble source that was remedied in the repair sequence, but is designed so as to be heard as a next component of the telling that was suspended by the repair initiation. The unit that delivers the weather report begins with en (‘and’, line 29). Used as a sequential conjunction here, the connective links the upcoming talk to the preceding talk and adds to it. The weather report does not add to the directly preceding repair sequence, however, but to the telling of the hiking adventures. After the inserted sequence in lines 44–52, on the other hand, the teller designs his next turn as a resumption of earlier talk. The teller is not returning to the telling by simply delivering the next component of the story. Instead, he


recycles the story component that occasioned the intervening talk. The event that was already talked about in line 41 (then we started to walk) is Wrst reinstated into the talk and then used as a set-up for going on with the telling (lines 53–56: but then uh (.) well then we just walked into the woods (.) we didn’t have a Xashlight with us …). Continuations provide their own “interactional uninterestingness” (JeVerson 1972: 319). Going on is dealt with as an unproblematic thing to do next. Resumptions, on the other hand, indicate that there is “a problem in accomplishing a ‘return’” (1972: 319). In the case of the resumption, the picking up of the telling line is not only framed as a kind of contrast with the foregoing (but …), but is observably done as a re-instalment of the position from which the abandoned telling can be ‘continued’. Fragment (3) illustrates each of the diVerent ways to go on with talk that is suspended. The inserted sequence in lines 25–27 is treated as an incidental supplement that is necessary to understand the ongoing story. As a subsidiary sequence, it is integrated into the ongoing talk as something that no longer causes a problem. It is something that can be forgotten because it is “deleted” (JeVerson 1972: 319) in and through continuing the telling of the story. The talk occasioned by the recipient’s understanding check in lines 44–52 is not dealt with as something that can be integrated smoothly and seamlessly into the ongoing talk. Rather, it is dealt with as a digression from which it is necessary to return in order to be able to go on with the telling. It is oriented to as talk in its own right. It is terminated orderly and collaboratively (see, for example, the co-operative manner in which the story-teller responds to the accounting in lines 49–50). It is treated as a diVerent line of development that requires active re-instalment of the abandoned telling. Resumptions accomplish non-obvious returns to a line of interaction that was abandoned because something else came inbetween. By using maar as the turn’s Wrst term, the speaker signals the non-incidental, not-unproblematic character of this type of return.

3.

Environments of use

The instances of resuming maar we came across in our corpus had in common that they all occurred after a kind of digression from the line that was returned to in the resumption. Without claiming to be exhaustive, we shall discuss two types of such environments: (i) after expansion of a repair sequence that is


inserted in the course of an ongoing telling, and (ii) after a competing line of topic development. 3.1 Environment of use (i): after an expanded repair sequence One type of context of resuming maar-turns resembles the Wrst two parts of the triplet structure JeVerson uses to characterise side sequences (1972: 316). An on-going sequence (O) is abandoned for a side sequence (S), after which a return (R) to the on-going sequence is made. Schematically: (4)

JeVerson’s side sequence: O - S - R

In fragment (2), for example, the Ongoing Sequence part is the telling in lines 313–16 (the relevant part of the fragment is repeated below). A participant projects a ‘hilarious’ story about this boy from his former school. The story’s delivery is halted when the recipient inquires after the name of the boy (line 317). This inquiry opens the Side sequence part, which runs until line 331. The teller returns to the story in line 332–33 (Return): (2') Repeat of the English translation of fragment 2: The-boy-from-my-formerschool 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334

Jan:→

listen, (.) >at my former school, you know (0.4) there was this guy and he (uh-) I didn’t like him very much. (0.3) and this was really hilarious [(because uh) [(°what was his name.°) Ans:→ 0.5 Jan: what?= Ans: =°and what was his name (.) Jan: Alexander, 1.3 Ans: * which school did you go to then. Jan: =to Saint Martin’s. ((name of the school)) (.) Ans: o↓:h. 0.3 Ans: (Lower Quay,) ((name of the area)) 0.2 [(°go ahead) ] Ans: rig [ht. Jan:→ [but thi [s u: : : : ]:h (.) this u:h (0.6) this (°boy°) (0.3) I didn’t like him very much. it really was a very strange guy,

Ongoing sequence Side sequence

Return

The environments in which maar-resumptions are done have more in com-


mon than merely being side sequences, however. The side sequences are launched by a repair initiation of the recipient (SchegloV et al. 1977). In fragment 2, for example, this is the question in line 317. Second, the repair initiation begins a repair sequence that is not closed as soon as repair is provided. Instead, the sequence is extended by non-minimal postexpansion(s), as in lines 323–31 of (2). Because the notion post-expansion (cf. SchegloV 1990 and 1997c) is central to the argument, we will brieXy elaborate on it. The extended repair sequence in lines 317–331 of fragment (2) has developed from a base sequence with an adjacency pair structure through diVerent types of sequence expansion. The Wrst pair part of the base sequence is the question that initiates the repair (what was his name?, line 317), and the answer that is given to it is its second pair part (Alexander, line 322). Sequences with adjacency-pair structuring may be expanded before the Wrst pair part (pre-expansion), between the Wrst and the second pair part (insert-expansion; e.g., the inserted repair sequence in lines 319–20), and after the second pair part (post-expansion, as in lines 323–31). The post-expansion in the repair-sequence in fragment (2) is non-minimal. Instead of minimally registering the repair’s successfulness, the repair-initiator reacts with yet another question (line 324). This leads to an expansion of the sequence with at least two more positions (the question turn in line 324, and the answer to it in 325). Problems in achieving a preferred type of second pair part are a major source for sequence expansion. But even if no direct problem is manifestly observable, sequences may be extended beyond minimal postexpansion. In fragment (3), for example, the recipient of the repair does more than just register its informativeness with the news mark oh. The recipient continues and dwells upon the issue that was re-focused in his understanding check (in the dark? line 44). He gives a reason for initiating repair (lines 47–48), which promotes further explicatory work on the matter from the storyteller (lines 49– 50): (3') Repeat of the English translation of a detail from fragment 3: Hiking story 41 42 43 44 45 46 47 48

Bob:

it was already dark then (.) then we started to walk (.) and uh:: (.) we did about four or five kilometers or something:= repair initiation Eric:→ =in the dark?= Bob: =yes repair outcome (.) expansion Eric:→ oh (.) we supposed (.) that (.) that you guys were finished (.) as soon as it were dark (.) was=


49 50 51 52 53 54

Bob:

Eric: Bob:→

=well (.) we just walked on then because we didn’t have a place to sleep yet (.) >yes yes= =·hh but then uh (.) well then we just walked into the woods

resumption

So both in (2) and in (3), we see an inserted repair sequence that has nonminimal postexpansion. Schematically: (5)

TELLER:

TELLER:

Ongoing telling Side sequence:

RECIPIENT: Repair-initiation TELLER: Repair + non-minimal postexpansion Return to the telling through Resumption

Non-minimal postexpansion of an inserted repair sequence is not obligatorily oriented to as a context for returning through resumption. See, for example, the repair sequence in the beginning of fragment (3). It has a repair initiation as its Wrst pair part (the what, line 25), and a possible repair as its second (Lac du la Gileppe, line 26). Although the second pair part position itself is expanded with an explication — perhaps due to a lack of recipient uptake —, the return to the ongoing sequence is still done as a continuation, not as a resumption: (3'') Repeat of the English translation of a detail from fragment 3: Hiking story 23 24 25 26 27 28 29 30

Bob:

and uh: (.) we then drove to (.) Eric:→ the what= Bob: =Lac du la Gileppe (.) → that’s a very big artificial 1.5 Bob: ·hh and uh::: (.) well there well the weather w- uh:: (.)

the (.) Lac du la Gileppe

lake (.) there as very rainy weather

repair initiation repair outcome expansion continuation

So not every type of repair-sequence postexpansion provides an environment for doing a return through resumption. There are speciWable features of repairinitiation types, however, that seem to promote the type of postexpansion after which returns are done as resumptions. Note Wrst that questions like what was his name? (fragment 2) or in the dark? (fragment 3) actively pursue topical interests of the recipient. They bring about a focus shift. In this respect, they diVer from the halting, time-marking type of repair initiation such as the ‘the what?’ question in line 25 of fragment (3). This latter class of repair initiation does not promote a recipient-controlled line of topic development. The recipi-

Dutch ‘but’ as a sequential conjunction

ent primarily acts as the listener who tries to keep track of what is being said in prior turn. The repair initiations that promote side-sequence expansion, on the other hand, propose a recipient-governed shift of focus. If there had been no inquiry after this particular aspect, it would not have been topicalised at this very moment in the interaction. The slightly diVerent line started by this kind of topically-steering repair initiation becomes even more pronounced when it is followed by yet another sequence elaborating upon it. Instead of minimising discontinuity, the intervening talk is extended with a trajectory of at least two more turns. The side sequence begins to lose its primarily subsidiary character and its topic may become consolidated as talk in its own right. If the talk in a side sequence is clearly subsidiary, the participants work towards a point of return to the main line. As soon as the problem that occasioned the insertion is solved, returning is the appropriate thing to do. A potentially competing line of talk, on the contrary, may be elaborated upon until it is exhausted for other reasons. It supplies its own measurement system for closure readiness. Participants show their orientation to the side sequence’s greater potential for independence when they do work that is specialised for terminating it. In fragment (2), for example, extensive closing work is done before there is a return to the suspended telling (lines 327–31). The side sequence is not dealt with as a sequence from which an unproblematic return can be made by simply continuing the suspended line of talk. Resumptions attend to this digression potential. They retroactively formulate the elaboration of the repair sequence as a possible departure from the line of talk that the resumption is picking up again. To summarise: one type of environment in which maar-prefaced resumptions occur is (i) after a side sequence that suspends an ongoing sequence; (ii) the side sequence is launched by a repair initiation of the recipient; (iii) the repair initiation is of a type that has topic-shifting potential; (iv) after the delivery of the repair, there is non-minimal postexpansion of the repair sequence, (v) which is oriented to as a possible departure by doing the return as a resumption. 3.2 Environment of use (ii): after a competing line of development Maar-prefaced resumptions are also used as a device to return from a recipient-proposed focus shift that is alternative to the line pursued by teller. The pattern of this type of use is characterised in (6):

151


(6)

TELLER: RECIPIENT: TELLER:

Teller’s line Competing line Return to teller’s line

Apart from describing sequential progress primarily in terms of topical development, the main diVerence from the pattern for returning from an expanded repair sequence is the second step in the diagram. In the side-sequence instances, the talk intervening between the abandoned telling and its resumption is subsidiary to the delivery of that telling. It is a side line that takes oV with a problem that has to be solved in order to be able to pursue the main line. In the type of case outlined in (6), however, there is not a return from a sequence that was subsidiary to an ongoing sequence. Rather, a competitive line of topic development is returned from. See the fragment below. The steps that were outlined in (6) will be marked again in the right margin of the transcription. The fragment is taken from a phone call between a skipper-family mother — calling from the family’s barge — and her 16-year-old son, who is in a boarding school for skipper children. The Teller’s line starts in line 107, where the mother tells her son that an acquaintance who is a skipper — Uncle Pascal — is coming to the harbor where the family’s barge is anchored. At Wrst, the son does not really promote this line of development, but then accommodates to it when he asks whether another skipper is coming too (line 115). The part with the Competing line begins here. The son’s question causes the mother to tell the latest bits of news about the movements of this other skipper (lines 119–131). As soon as she has told all the relevant details she can about the other uncle, she returns to telling more news about the uncle she herself had volunteered to talk about (line 133). This is the Return-to-teller’s line part: (7)

What-about-Uncle-Karel? From a phone call between mother and son. The mother is calling from the family’s barge. During weekdays, the son is in a boarding school for children from bargee people. He left the family’s home the evening before, on Sunday night.

103 104 105 106 107 108 109 110

Ma: Son: Ma:→ Son:

d’r zijn hier >hel:maal geen sche:pen joh! (.) ne↑eh 0.2 ne↓e. oom Pascal is ook onderweg hier naar toe↑: 0.2 o:↓h. (.)

Teller’s line


111 Ma: 112 Son: 113 114 Ma: 115 Son:→ 116 117 Ma: 118 Son: 119 Ma: 120 121 Ma: 122 Son: 123 124 Son: 125 Ma: 126 Son: 127 128 Ma: 129 130 131 132 Son: 133 Ma:→ 134 135 136 137 Son: 138 Ma: 139 140 141 142 Son: --------103 Ma: 104 105 Son: 106 107 Ma:→ 108 109 Son: 110 111 Ma: 112 Son: 113 114 Ma: 115 Son:→ 116 117 Ma: 118 Son: 119 Ma: 120 121 Ma: 122 Son:

jah. want d’r e: [h [daar lag niks in de vluchthaven (tenminste) toen ik ‘r met de trein langs ging.= =ne:↑e[↓:h [oom Karel dan: Competing line 0.7 oom Karel (Stijn [↑:) [jah die achter (Gebien:)= =jah. die moet hier ↓loss:en. (.) [en in Utrecht. [a↓hah. 0.6 jaA [h [die was vanmorgen dan ook nog even op de beurs:= =°↓jah↓ 0.2 enne:h ·h jah die kwam geloof ik vandaag nog niet leegh hier en dan eh: (j: jah) die hoopt d’r ook zo gauw moglijk (leeg te zijn.) vooral omdat ‘r nou zo weinig schepeh zijn.= =u↓huh.= Return =·hhh maar oom Pascal komt hier o↑ok naar toe want as ie hier vanmiddag geweest was dan had ie al aan de kolen gekund naar Baar↓n 0.2 o↓::h [: [want dat is ‘n eh schip(er eh) >in IJsselstein èrh ((clears)) in IJsselstein heeft dat aange°nomen. dus eh: hh die zal vannacht hier o↓ok wel zijn↓: 0.3 uhuh.

there are no ships at all over here, you know! (.) noh, 0.2 no. Uncle Pascal5 is on his way here, too: Teller’s 0.2 line o:h. (.) yes. because there u: [h [there was nothing in the refuge harbor (at least) not when I passed there on the train.= =no: [:h [so what about Uncle Karel Competing line 0.7 Uncle Karel (Stijn [:6) [yes the one behind (Gebien:7) =yes. he has to unload here. (.) [and in Utrecht. [ahah.

153


123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142

Son: Ma: Son: Ma:

Son: Ma:→

Son: Ma:

Son:

0.6 ye [s [he was at the exchange this morning, too= =yes 0.2 and u:h ·h well I don’t think he’ll be empty today and then uh: (w:- well) he too hopes to be empty as soon as possible. especially since there are so few ships here at the moment.= =uhuh.= =·hhh but Uncle Pascal is coming this way too because if he’d been here this afternoon then he could have done a coal transport to Baarn already 0.2 o::h [: [because there is a skipp(er uh) in IJsselstein urh ((clears throat)) in IJsselstein he has accepted this. so uh: hh he will arrive here too tonight. 0.3 uhuh.

Return

What is happening here is a kind of technical topic competition (cf. Sacks 1992b: 348 V.). The scrimmages already start at the moment the mother tells the news about ‘Uncle Pascal’. The son does not really embrace this topic (see lines 112–13). Yet — on second thought, so to speak — he goes along with the route his mother took, but not without heading for a slightly diVerent course, however. He inquires after another member from the class of persons his mother was talking about (so what about Uncle Karel?). Making a shift to a coclass member — from ‘uncle Pascal’ to ‘uncle Karel’ — is a very elegant way of proposing an alternative focus while still doing topical talk (cf. Sacks 1992a: 756 V.). The mother at Wrst fully complies with the line her son is pursuing. At great length she tells the latest news about Uncle Karel (lines 119–131). Yet when her report reaches a possible saturation point — coming back to something that was already talked about before (since there are no ships at all here / there are so few ships here, line 103 and line 131 respectively) —, she again picks up the topic of Uncle Pascal. She recycles the announcement of his coming and then continues with an explication of the ins and outs of why this is so. Returning to Uncle Pascal by way of a maar-prefaced resumption retroactively formulates the shift to Uncle Karel as a competing line of topic development.8 The talk about Uncle Karel was not just a topical alternative, but a line of development that pushed aside the line the mother was pursuing when she launched an elaboration of the Uncle Pascal news.


Thus, maar-prefaced resumptions are also used to mark the return from a competitive line of topic development. Although a competing topic is not subsidiary to the talk it is following, there are similarities with the pattern described for returns from expanded repair sequences. First, when the mother launches into an elaboration of the Uncle Pascal news (see her because there u:h, line 111), something which is about to become ‘ongoing’ comes to a halt. Although this onset is interrupted without treating its abandonment as suspending an ongoing sequence, it is later on repeated as a re-instatement of a line of talk that could not get through the Wrst time. After the mother has recycled the news of Uncle Pascal’s coming, she continues with a next turnconstructional unit that has the same beginning as the one that was abandoned earlier on (see lines 133–34: but Uncle Pascal is coming this way too. because …). This might be a way of showing ‘this-was-what-I-was-about-to-say-the-Wrsttime’. Second, as in the side-sequence cases, the ‘owner’ of the abandoned line fully complies with pursuing the line of topic development that is proposed by the recipient. Again, the focus shift is proVered in a way that makes it hard not to comply with. It is formatted as a question. Note, however, that the mother is still the one who does the telling, albeit with respect to a referent that is supplied by her son. Third, the line pursued by the son is oriented to as talk in its own right that is to be terminated before the abandoned line can be returned to. Only after the recipient has acknowledged the relevancy of the news about Uncle Karel (line 132) does the mother return to her own line of telling. So both types of environment discussed in this section correspond with respect to the following features: an ongoing activity is abandoned for another one (subsidiary and/or alternative), which is oriented to as an activity in its own right. This line Wrst has to be terminated. The ensuing return to the abandoned line is done as a resumption, not as a continuation.

4.

Tying modes: skip-connecting and resuming

The default device to tie one turn to another one is simply placing it after the one it is referring to. JeVerson (1978) calls this next-positioning. Unless there are indications that point to the opposite, a recipient will assume that current turn is to be interpreted against the background of prior turn. A speaker has to use special devices for tying a current turn to some other-than-last turn.

155


Some of these devices are discussed in Harvey Sacks’ Lectures on Conversation (Sacks 1992a/b). Particularly the tying mode he calls skip-connecting has several features in common with how resumptions tie to earlier talk. In this section we will compare skip-connecting with resuming as a way to discover the relevant similarities and diVerences. Sacks characterises skip-connecting as follows: “… a speaker produces an utterance that is indeed related to some prior utterance, but it’s not related to the directly prior utterance, but some utterance prior to the directly prior utterance” (Sacks 1992b: 349). Although this description allows for the inclusion of a whole range of ways of tying a current turn to a pre-prior one, Sacks seems to reserve the notion for a particular type of tying. He discusses a fragment from a multiparty discussion in which there is competition between two participants with respect to the line of development that gets taken: (8) Example from Sacks 1992b: 348 Roger: ((About the new Pike)) Oh this place is disgusting. [ Any day of the week. Jim: [ I think that Pop is [depressing, it’s justRoger: [But you go- you go- takeJim: Those guys are losing money. hehh Roger: But you go down- dow- down to th’New Pike there’s a buncha [ people, oh:: an’ they’re old, an they’re pretending ((Jim)) [((cough)) Roger: they’re having fun, but they’re really not.

Roger’s remarks are about the amusement park called the Pike, and Jim’s are about PaciWc Ocean Park, another amusement park that gets called ‘Pop’. In each next turn, current speaker relates his utterance to his last one, not to the prior one of the other speaker. Sacks (1992b: 348) diagrams this as: Roger: Jim: Roger: Jim: Roger:

Pike Pop Pike Pop Pike

Sacks notices two sorts of facts about skip-connecting utterances: speakers tend to skip-connect to one of their own utterances; and they do not skipconnect unless their own utterance was last-but-one (1992b: 349). We diagram this one more time in (9):


(9)

(mutual) skip-connecting: speaker B skip-connects from turnl to turnj speaker A: turni / speaker B: turnj / speaker A: turnk / speaker B: turnl speaker A skip-connects from turnk to turni

Sacks’ Wrst observation — speakers tend to skip-connect to their own utterances — also holds for resumptions: speakers tend to resume their own tellings. In fragment (2) it is the teller of the story about the boy who resumes the telling; and in (7), it is the mother who returns to telling about Uncle Pascal. However, resumption of an abandoned line of talk may be done by its recipient, at least under the kind of conditions we will describe for the fragment below. It is taken from an interaction in a tattoo shop. The fragment begins just after the tattooist has made a remark about the name of a Dutch rock musician on his customer’s T-shirt (Herman Brood, in some Dutch music scenes a real legend). The customer then starts talking about a performance he once attended (lines 43–53): (10) Tattoo-shop. The tattooist is preparing his customer’s skin for a tattoo. From time to time, he and his customer also engage in talk about things other than the tattoo. Further persons present are Rose (the customer’s girlfriend), and John (a friend of the tattooist). 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57

Customer:

>d’r was op gegem:n moment e::h was die erg’s e:hm (1.6) w↑aht waz ↓‘t (0.4) [E m m e h ] Rose: [(°>’k was ‘r ni]et bij. oh god!) 0.3 Customer:→ <Emmeh Compascum= Tattooist: =>jhUh! (.) o↓h Customer: 1.5 Customer: jaah, da’r was e:h- (.) [‘n dik fee:↑st Rose: [>( ) niet te hoog? >↓oh sorry. mag ik me m’rmee bemoeie? 0.4 Tattooist: °ja hoor. (…) ((1.5 minutes))

134 135 136 137 138

Customer: Rose: John: John:

(moet) ↑niet gaan trainen. nou niks meeh doen.= =·h heh ↑H [Uh [geen eh swartseneger wordeh, 0.9 °da mag nie.°

157


139 1.2 140 Tattooist: nouh! (0.3) [(plof ma-) ] plof maar effe ne:↓er hh: 142 Customer: [Oké↑: ] 143 3.6 144 John: ↑woe:oeh ((imitates sound of tattoo needle)) 145 2.4 146 Tattooist: turn it off agaihain, ((singing)) 147 2.6 148 Tattooist:→ MAar- (.) °was. (m-) °>hij was in Emmer Compascu↑um 149 0.2 150 Customer: ja:h joh. (0.5) daar was toen ‘n de een of andere feest eh wee:k. met eh ·hh jah norma↑al die trad 152 153 op. en eh de sie:n. herman brood. ------------43 Customer: there was at some point u::h he was 44 somewhere u:hm (1.6) where was it (0.4) [E m m e h ] 45 46 Rose: [(I wasn’t the]re. God!) 47 0.3 48 Customer: Emmer Compascum= ((name of a little town)) 49 Tattooist: =jhuh! 50 (.) 51 Customer: oh 52 1.5 53 Customer: yeah, there was uh- (.) [a huge party [( ) not too high? oh 54 Rose: ((about the position of the tattoo)) 55 sorry. you mind if I meddle? 56 0.4 57 Tattooist: no not at all.

(…) ((1.5 minutes about other things)) 134 Customer: (should)n’t start working out now. leave it alone.= 135 Rose: =·h heh h [uh [not uh become a Schwartzenegger, 136 John: 137 0.9 °I wouldn’t want him to.° 138 Rose: 139 1.2 140 Tattooist: right! (0.3) [(just si-) ] just sit down hh: [okay ] 142 Customer: 143 3.6 144 John: woo:ooh ((imitates sound of tattoo needle)) 145 2.4 146 Tattooist: turn it off agaihain, ((singing, an English song line)) 147 2.6 148 Tattooist:→ but- (.) was. (m-) >he was in Emmer Compascum 149 0.2 150 Customer: yes man. (0.5) there was some 151 festival uh week. with uh ·hh well Normaal* played. 152 and uh the Scene*. Herman Brood*. ((* Dutch pop groups))


In line 148, the tattooist recycles the Wrst part of the telling his customer had begun about 1.5 minutes before: but … he was in Emmer Compascum repeats the customer’s at some point he was somewhere … Emmer Compascum in lines 43–48. The delivery of this telling was interrupted by a remark from the customer’s girlfriend about the position of the tattoo (line 54). After this, the tattooist goes on with his work, but now and then, they also engage in talk about other things, as in lines 134–38, where the customer talks with the participants in the ‘audience’ about the consequences of a fresh tattoo and working out. Note that the tattooist recycles a telling part that is immediately recognised and directly taken up by the participant who told it the Wrst time. The mentioning of the fact that he9 — the singer — was in Emmer Compascum suYces to prompt the former teller to continue to tell about the musician’s performance. In his response, he recycles the sole other orientational description he has already given (there was u:h … a huge party, line 53), but in such a way that new information is also supplied (there was some festival uh week, lines 150–51). The tattooist’s role as prompter of the return may be a way of ‘doingbeing-a-party’ within the setting of the tattoo-shop. He and his customer have their own participation framework. It can be intruded on from outside (as in line 54), it can dissolve into a plenary participation conWguration (as in lines 134–38), but it can also be re-installed by resuming talk that was done primarily as something between the two of them. Sacks’ second observation about skip-connecting utterances — speakers tend to skip-connect to an utterance of theirs that was last-but-one — does not hold for resumptions. Resumptions usually go back to the sequence that came before another — subsidiary, alternative, and/or competitive — sequence came inbetween. Unlike the ‘turn-over-turn’ type of skip-connecting Sacks discusses, resumptions do tie over long distance, albeit still under some kind of a ‘last-before-intervening-line-of-talk’ constraint. Schematically: (11) Resumption: participant A: linei / B and A: linej / speaker A: return to linei return to the abandoned line of talk

Resumptions do long distance tying. But they do not skip-connect. They reinstall a move from the abandoned activity. This explains why the practices by which they are done are so diVerent from the techniques that are used to skip-


connect. The skip-connecting in the fragment from Sacks is done by way of such techniques as full noun repeat (Roger’s New Pike), or adding particulars that can only apply to the referent the current speaker was already talking about. The prevalent way for doing a resumption is a same-speaker recycle of the last telling unit before the speaker’s telling was abandoned in favour of a diVerent line of talk. In fragment (7), for instance, this was the news about Uncle Pascal (lines 107 and 133). As a rule, the last telling unit before the telling came to a halt is recycled. Not always, however. In fragment (3), for example, the speaker recycles the utterance about starting to walk again (line 41), and not the very last utterance of the telling before it was suspended (about having walked about four or Wve kilometres, line 43). Although the telling part that is recycled should be as late as possible in the telling so far, it is only eligible to be recycled if it can serve as a basis for continuation of the telling. Resumptions typically are done as the Wrst turn-constructional unit in a multi-unit turn. After this initial unit, the speaker continues the abandoned line of talk. When the unit with the recycled telling line is possibly complete, the speaker does not wait for a response that attends to it independently, nor does the recipient treat the recycled unit as an action that should be receipted separately. The resumption is oriented to as ‘preliminary’ to a speciWc type of next (cf. SchegloV 1980). It is targeting a next action that uses the recycled telling part as the base to which a next telling part can be attached. Most of the resumptions we have discussed inherit this property of ‘activity-incompleteness’ from when they were done the Wrst time (cf. Houtkoop and Mazeland 1985). The utterances that are recycled in (2), (3) and (10) are recognisably not done as independent, stand-alone announcements, but as a part of a larger telling that is not yet complete. Even if the recycled telling does not inherit the property of ‘ongoingness’ from the source telling, it is oriented to as prefatory, as shown by the telling about Uncle Pascal in fragment (7). The repeating of the ‘news’ about his coming after a competing line of talk (Uncle Karel’s coming) and its framing as being in contrast with this line by prefacing it with maar makes it analysable as preliminary to the delivery of more tellables about the refocused event. In sum: As a rule, resumptions are done by the speaker of the abandoned line of talk. They do long-distance tying by recycling the last telling component that is suited to serve as a basis for the kind of continuation the speaker is working towards. They are done after possible termination of the intervening line of talk. They not only inherit a property of ongoingness, but as a resump-


tion, they are oriented to as prefatory to some kind of elaboration. Most recycles are done by repeating material from the source turn. The recycles in (2), (7) and (10) have almost the same wording as when they were said the Wrst time. A repeat is a tying technique that is relatively independent from a turn’s actual position. It locates some prior turn as its source by the matching of lexical and constructional materials. What action the repeat is doing, however, is dependent on how it is framed and formatted, where it is used and by whom. Repeats can implement diVerent sorts of actions. When participants repeat a part of the turn of a prior speaker in their next turn, for example, these repeats do such diVerent actions as initiating repair on a trouble source in prior turn, registering the receipt of what prior speaker has said, or targeting ‘a next action’ — such as rejection or correction — in the next turnconstructional unit (cf. SchegloV 1997a: 527 V.). The kind of repeat we are dealing with is done by the same speaker and over a long distance. It not only locates an earlier turn-constructional unit by same speaker as its product item (cf. JeVerson 1972). It recycles the action for which it was the vehicle and reinstalls it in the context of the activity for which it was abandoned. The abandoned line of talk is retroactively revealed as alternative to and competing with the activity that took its place. In sum, the position in which the recycle is done (long distance after the possible termination of an intervening sequence), the relation of its speaker to the telling that is re-installed (same speaker), the type of features of the original action that are inherited by the recycle (ongoingness, retroactive formulation as prematurely abandoned), the format of the turn-constructional unit (recycling an integral telling part and framing it as an action that is in contrast with its context), and the continuation without delay with a next turn-constructional unit in which a next telling part is delivered (constituting the recycled telling as preliminary to continuation of the activity it is re-installing), are all features that jointly contribute to the interpretability of the utterance as a resumption.

5.

Furthering versus closing the topic

The construction format [turn-initial maar + recycled telling component] thus appears to be used as a practice that implements a speciWc type of action in speciWable types of environments. Quite similar formats can be used to do very diVerent actions in other types of environments. In the fragment below, a set of

161


similar features is to be found in an utterance that proposes topic closure. A comparison with the use of this latter type of turn will help to understand what kind of practice it is that speakers use as a vehicle for doing resumption. The fragment is from the same mother-and-son call as fragment (7). It opens with a topic proVer from the mother (were you in time for the train yesterday?, line 212). At the end of the fragment, the mother initiates closure of the talk about this topic by re-addressing the very same issue she started with (but anyway you were on time, weren’t you, line 238): (12) In-time-for-the-train? From the same phone call as fragment 7, but more towards the end of the call. 212 Ma:→ 213 214 Son: 215 Ma: 216 217 Son: 218 219 220 Ma: 221 Son: 222 223 Ma: 224 225 Son: 226 227 228 Ma: 229 230 231 Son: 232 233 Son: 234 235 Ma: 236 237 Son: 238 Ma:→ 239 Son: 240 241 Ma: 242 243 Son: 244 Ma: --------212 Ma:→ 213 214 Son: 215 Ma:

was je op tijd gistere voor de tre↑in: 0.6 e:h: ja↓::h. >°nou jah.= =jah. was ‘t nie zo ve↑r: 0.4 neu↓:. ‘k geloof dat het dertig kilometer ↓was. °zag ‘k op de borden staan. Leiderdorp. zesentwintig kilometer. °dus eh: 0.3 <j:aA↓jah. (.) nou jah. >jaAh, ·h >maar ja goed. >dat is misschien ‘t centrum. hè:? >dan ben je natuurlijk ook weer ‘n eindje van het centrum af:.= =NEE. TOEN ZATEN WE:eh- jaA:h. (is ook weer iets.) 0.2 °ja↓:hh (.) ja↑: 0.3 (jah)°↓ja maar je was in ieder geval wel °↓op tijd. j:a:h (.) okÉ. 0.2 jah? nouh, verder heb ik geen bijzonder(s).

were you in time for the train yesterday 0.6 u:h: yes. well yes.= =yes. it wasn’t that far, was it?


216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244

Son:

Ma: Son: Ma: Son:

Ma:

Son: Son: Ma: Son: Ma:→ Son: Ma: Son: Ma:

0.4 no. I think it was about thirty kilometers. still that much. (.) still. yes 0.4 o:h (.) or twenty six or so. saw it on the signs. Leiderdorp. twenty six kilometers. so uh: 0.3 well well. (.) yes okay. yes, ·h okay, fine. but that’s probably the centre, isn’t it? so you’re some distance away from the centre of course.= =no. then we were uh: yes. (that could be.) 0.2 yes (.) yes 0.3 (yes)yes but anyway you were on time, weren’t you. yes (.) okay. 0.2 yes? well, I’ve got nothing special left to tell.

The mother proposes topic closure by returning to the question that launched the topic. Returns to the start of the sequence or the topic are among the most common turn types that serve to initiate sequence-closing (see SchegloV 1997c: 189 V.). The return is packaged as a request for conWrmation (… you were on time, weren’t you?, line 238). By conWrming it, the son aligns with the closure proposal it is implementing (yes, line 239). The ensuing talk — in particular the mother’s announcement that she does not have anything left to tell (line 244) — shows the participants’ orientation to the current state as next to a possibly closed topic. The construction format of the turn in which the mother proposes topic closure is very similar to the one used for doing resumptions. The utterance occurs as the Wrst turn-constructional unit, it is prefaced by maar and it is followed by a repeat of earlier materials. The wording of the mother’s original question is copied almost literally. However, this time not as a question, but as a request for conWrmation (of something the addressee has already committed himself to, by the way). Although there is a return to the issue the initial action was about, the action itself is not recycled, nor is the activity it was part of re-


installed. The properties of an utterance do not have to be inherited when only its content is repeated. In sum: although at Wrst sight very similar to the construction type of resumptions, both the construction format of maar-prefaced closings and the environment in which they are used do diVer with respect to essential sequential features. These account for the fact that this type of turn implements a very diVerent type of action.

6.

Discussion

We have analysed the properties of a speciWc type of use of the Dutch connective maar (‘but’). When it prefaces a turn it marks a relation between turns, not between same-turn units. In order to characterise its use in utterances that resume an abandoned line of talk, we Wrst distinguished diVerent procedures deployed to return to a suspended line of talk (continue versus resume). We then described two types of environments in which resuming maar occurs: after repair sequence expansion, and after a competing line of topic development. In Section 4, we showed that, as a rule, resumptions are done by the speaker of the abandoned activity and over a long distance. The resumption is accomplished by recycling the last unit of the abandoned activity that is suited as a base for its continuation in the next turn-constructional unit. In text-linguistic literature on the use of maar, it is pointed out that maar usually triggers the conversational implicature that the maar-part of an [assertion1, maar-assertion2]-utterance is to be considered as the carrier of the utterance’s relevant information (Foolen 1993: 117–118 and Spooren 1989). By prefacing a turn with maar, a speaker proposes a similar kind of ranking, but now at the level of sequential and topical organisation (see also SchiVrin 1987: 173–75). Resumptions bring about a relevancy structuring of participant agendas. By maar-prefacing a re-instalment of his own line of talk, the speaker formulates the intervening talk — that is to say, the talk between the resumed and the resuming utterance — as alternative to, competitive with and/or a departure from the line of talk that is picked up again. In an attempt to maintain a uniWed description of the meaning of maar that would also account for its discourse-organising uses, Redeker (1994) proposes the following characterisation: ‘Maar as a segmentation signal constrains the expectation that the ongoing discourse segment will be continued’10 (1994: 219). We think that this account is too narrow. It might even turn out to


be inappropriate. In most of the fragments we have discussed, the participants have already been working towards closure of the intervening sequence (see the discussion of fragment 2 and 3 in Section 3). It seems rather unlikely that the participants would foster the expectation that the current discourse-segment will be continued in an environment in which they are already engaged in terminating it. If there is anything that is contradicted in resuming-maar turns, it is rather the expectation that the re-installed activity has been terminated. Although it was abandoned, it is still on the agenda of the current speaker. Resuming-maar re-structures local discourse-organising relevancies. It lends priority to the resumed activity and formulates the intervening sequence as a competitive side line. Our criticism of an attempt to provide a uniWed description of some kind of ‘basic meaning’ of the conjunction does not imply, however, that we are in favour of a polysemic approach, that is, a modelling of the meaning of the conjunction according to which diVerent types of use are based on diVerent meanings that are interrelated in more or less systematic ways. The polysemy model puts too much weight on the meaning of isolated words as part of the language system. Instead, some types of uses of the conjunction may contribute to repeatedly occurring, constructionally speciWable types of utterances. They can be thought of as practices that accomplish speciWc types of actions in sequentially speciWable types of environments (cf. SchegloV 1997a: 504 V.). The construction format [maar + recycling an abandoned telling] used in an environment of the possible closure of a potentially alternative line of development is one such practice. It may be used as a device for resuming an abandoned line of development that came before the alternative one. The construction format [maar + return to the start of the topic] is another type of example. It may be used to propose topic closure in an environment of topic talk that has reached the stage of topic exhaustion (see JeVerson 1981). What a word is doing in an utterance must eventually be determined by sequential reasoning: why is this said by this participant in this position in this manner? The action an utterance implements governs how the linguistic elements of the utterance itself are dealt with when they are made sense of. It is not the other way around. The action quality of an utterance is not to be deduced from linguistic form. Linguistic form contributes to the utterance’s construction type, but the features that make it a construction type are at least partially and unavoidably derived from its situated use in talk in interaction.


We want to conclude with two methodological remarks. First, an utterance property such as the use of a technique to recycle an earlier utterance appears to be central in determining the features of resuming-maar utterances. An utterance can only be recognised as a recycle or as a return, however, when the utterance’s environment is taken into account. Analysing an utterance as recycling a previous one is a function of the pair. It is not an inherent property of the utterance per se. In order to be able to describe at least some kinds of uses of connectives in turns at talk in interaction, it appears to be necessary to include sequential properties into the description of the utterance-type that is examined. Second, we looked at just one type of use of maar as a sequential conjunction. It was essential to describe the sequential properties of the utterance-type in order to be able to determine how it is used. We think this might prove to be a productive approach for research into the use of connectives. Instead of claiming a basic meaning for the connective per se, its workings in sequentially speciWable types of utterances have to be examined. The meaning of words may then really turn out to be a function of their use in utterances in turns at talk in speciWable sequential environments.

Notes

* The paper is a thoroughly revised version of Mazeland and Huiskes 1997. It is partially based on Mike Huiskes’ master’s thesis (Huiskes 1996). We are very grateful to Gail JeVerson for commenting on some of the analyses we had doubt about. We presented the paper at the CCCC workshop at the University of Konstanz in October 1998 (Cause, Condition, Concession and Contrast: Cognitive and discourse perspectives). We thank the participants of this workshop, particularly Peter Auer, who acted as the discussant for this paper, for their comments and a stimulating discussion. Finally, we thank Claire van den Donk and Marjolijn Verspoor for taking a look at the way we use(d) the English language. And, of course, we thank the editors for their careful readings. A part of Harrie Mazeland’s research on the subject of this paper was made possible by a three-month stay in autumn 1998 as a guest professor at The International Graduate School in Language and Communication, Odense University, Denmark. 1. The Wrst time we encountered the term sequential conjunction was in JeVerson (1981). 2. Heritage and Sorjonen (1994) describe how and-prefaced questions do not just link a question to the preceding turn, but to the preceding question/answer-pair(s) as a next step in an agenda-based course of action in interactions in institutional settings. The use of the conjunction is analysed by taking the interactional context into account, “including, most


importantly, the sequential and activity context” (1994: 2) together with the speech act type of the utterance that is prefaced by the connective. 3. The transcription fragments (1) and (2) come from a corpus of high school pupil talk in a homework class, recorded and transcribed by Henrike Padmos. Fragment (3) comes from a small corpus of telephone conversations between friends transcribed by Heidi van Mierlo and Chris Driessen. Fragment (10) is from recordings made in a tattoo-shop, transcribed by Rozemarijn de Kruijf. The fragments (7) and (12) come from a corpus of phone conversations transcribed by Harrie Mazeland, who also did a second transcription round for the other transcriptions. The base corpus that was used for this analysis consists of two small corpora with recordings of phone calls. In the course of our research, the collection was extended with a corpus of high school pupil conversations and a couple of transcriptions that were made for student assignments. 4. By saying was after his turn has already arrived at a possible completion point, Eric does a transition-space self-repair substituting the plural ‘were’ with the grammatically correct singular ‘was’ (cf. SchegloV 1997b). 5. We have the impression that in this family — or perhaps in this skipper subculture —, some of the acquainted skippers are referred to as ‘Uncle X’ in interactions between parents and their children (where X usually is a surname, that may be followed by the family name if further identiWcational material is needed, as in line 117). 6. ‘Uncle Karel (Stijn)’: `Stijn’ is probably a family name. 7. ‘Gebien’: perhaps a geographical name. 8. Gail JeVerson pointed out that the mother also packages the return to the telling about uncle Pascal as a next when she says that Uncle Pascal is coming this way too (line 133), — as if she is adhering to her son’s topic (Uncle Karel), instead of the other way around. 9. The tattooist’s resumption in fragment (10) exhibits an interesting instance of the use of a personal pronoun as a tying device. In ‘but- … he was in Emmer Compascum’ (line 148), the locally subsequent reference form ‘he’ is used on an occasion that is not an obvious locally subsequent reference position (cf. SchegloV 1996: 450 V.). This ‘mismatch’ between reference form and reference position, however, contributes to the turn’s interpretability as a resumption. One and a half minutes after the talk about this person is abandoned, the tattooist refers again to the rock musician with a personal pronoun (line 148). Inbetween, several other, possibly competing referents were talked about (see, for example, the mention of Schwarzenegger in line 136). Still, the shop-owner’s utterance supplies suYcient cues to enable the recipient to identify the rock musician as the referent of he. By saying that he was in Emmer Compascuum, lexical material is repeated that is unmistakably associated with the customer’s former telling about the rock musician. However, referring via a next-time reference form to a person not talked about for some time shows that the talk about this subject was not closed yet (see Fox 1987: 16–32). The incongruent, and thereby marked use of a locally subsequent reference form in a locally initial reference position contributes to the utterance’s recognisability as a resumption. The use of a marked reference form thus may have discourse-organising eVects. (See also Mazeland 1996.)


10. Our translation. The original quote goes as follows: “Maar als segmentatiesignaal legt een restrictie op de verwachting dat het lopende discourse-segment gecontinueerd zal worden” (Redeker 1994: 219).

References

Foolen, A. 1993

De betekenis van partikels. Een dokumentatie van de stand van onderzoek. Met bijzondere aandacht voor maar. Dissertation University of Nijmegen.

Fox, B. Discourse Structure and Anaphora. Written and conversational English. Cambridge: Cambridge University Press. Heritage, J. and Sorjonen, M. 1994 “Constituting and maintaining activities across sequences: and-prefacing as a feature of question design”. Language in Society 23: 1–29. Houtkoop, H. and Mazeland, H. 1985 “Turns and discourse units in everyday conversation”. Journal of Pragmatics 9: 595–619. Huiskes, M. 1996 Maar als segmentatiesignaal. MA Thesis, Dept. for Dutch Linguistics and Literature, University of Groningen JeVerson, G. 1972 “Side sequences”. In Studies in Social Interaction, D. Sudnow (ed.), 294– 338. New York: Free Press. 1978 “Sequential aspects of story telling in conversation”. In Studies in the Organization of Conversational Interaction, J. Schenkein (ed.), 219–248. New York: Academic Press. 1981 “Caveat Speaker”: A preliminary exploration of shift implicative recipiency in the articulation of topic. Final report to the (British) SSRC. London: Social Science Research Council. Mazeland, H. 1996 “Pragmatic aspects of the use of pronouns in wh-questions”. In Coherence & Anaphora, W. de Mulder and L. Tasmowski (eds.), 185–206. Belgian Journal of Linguistics 10. Amsterdam and Philadelphia: Benjamins. Mazeland, H. and Huiskes, M. 1997 “Maar als sequentiële conjunctie. Het hernemend gebruik”. Taalbeheersing 1997/3: 265–277. Redeker, G. 1994 “Maar nu even iets heel anders — maar als segmentatiesignaal”. In 1987


Nauwe Betrekkingen, R. Boogaart and J. Noordegraaf (eds.), 213–221. Amsterdam: Stichting Neerlandistiek VU. Sacks, H. 1992a/b Lectures on Conversation. Volume I and Volume II. G. JeVerson (ed.), Oxford: Blackwell. SchegloV, E. 1980 “Preliminaries to preliminaries: ‘Can I ask you a question?’” Sociological Inquiry 50/2: 104–152. 1990 “On the organization of sequences as a source of ‘coherence’ in talk-ininteraction”. In Conversational Organization and its Development, B. Dorval (ed.), 51–77. Norwood, NJ: Ablex. 1996 “Some practices for referring to persons in talk-in-interaction”. In Studies in Anaphora, B. Fox (ed.), 437–85. Amsterdam and Philadelphia: Benjamins. 1997a “Practices and actions. Boundary cases of other-initiated repair”. Discourse Processes 23/3: 499–547. 1997b “Third turn repair”. In Towards a Social Science of Language, Vol. 2, G. R. Guy, C. Feagin, D. SchiVrin and J. Baugh (eds.), 31–41. Amsterdam and Philadelphia: Benjamins. 1997c Sequence Organization. Department of Sociology, UCLA (mimeo). SchegloV, E., JeVerson, G. and Sacks, H. 1977 “The preference for self-correction in the organization of repair in conversation”. Language 53/1: 361–83. SchiVrin, D. 1987 Discourse Markers. Cambridge: Cambridge University Press. Spooren, W. 1989 Some aspects of the form and interpretation of global contrastive coherence relations. Dissertation University of Nijmegen.

On some uses of the discourse particle kyl(lä) in Finnish conversation* Auli Hakulinen University of Helsinki

In this paper I analyse the ways in which a very frequent discourse particle, namely kyl(lä), is used in everyday Finnish conversation. In aYrmative answers to yes/no questions, a typical place for the particle kyllä is initial position. Here, it makes a full-sentence answer less like a repair than a ‘bare’ SV-sentence would. I argue that there is a common denominator in the usage of the particle in diVerent kinds of turns; for example, the use of kyllä in proverb-like statements which can be viewed as reactions to non-expressed doubts bears a resemblance to kyllä as used in answers to yes/no questions. Another variable discussed in the paper is the position of the particle within a turn. In other kinds of turns than in answers, kyllä may also occur in the middle of an utterance or at the end of one. It is suggested here that the position iconically reXects the stance of the speaker: the further back the particle is placed in the utterance, the clearer the resistance of the speaker to the issue at hand. This is not the whole truth, however. It may well be that the Wnal position (‘post-possible completion’) carries other senses besides that of the resistance of the speaker.

Introduction From the perspective of sentence grammar, discourse particles1 are regarded as optional, and their explication is normally left to dictionaries. However, from the point of view of interactional linguistics, and for the understanding of turn design (the grammar of turn construction), these elements cannot be called optional. As elements that mostly have the whole turn in their scope, discourse particles are not only obligatory but often play an essential role in the speciWcation of the action performed by the turn. It is therefore one of the essential tasks for interactional linguistics to describe and explicate the behaviour of discourse particles as resources for speakers to convey their stances to the talk and their positions as participants in the conversation they are engaged in. In

172 Auli Hakulinen

studying particles, the immediate grammatical (sentential) context is clearly not enough; the work particles do is inextricably connected with the actions that they occur as part of. In my paper, I am going to look at the import in conversation of the Finnish discourse particle kyllä (with a colloquial variant kyl), which up to now has not been so much as mentioned in Finnish grammars but is only dealt with as a lexical item (with the category label “adverb”) in dictionaries. The particle kyl(lä), carrying roughly the (dictionary) meanings ‘sure’, ‘really’, ‘indeed’, is among the most frequently used particles in Finnish conversation. Sequentially, its use is restricted: it cannot occur in questions or directives, but only in (responsive) statements. Within a turn, kyl(lä) can occur in several alternative places, and it does not normally form an intonation unit of its own. Typically, it occurs (i) turn- and utterance-initially, but it is also found (ii) in the middle of an utterance, and even (iii) turn- and utteranceWnally.2 In other words, kyl(lä) appears in positions that are typical sites for dialogue particles, as well as in positions where modal particles tend to occur. Moreover, it also occurs at the end of turns, where what SchegloV (1996: 92) has called ‘post-completion stance markers’ show up. By analysing the behaviour of a particle with a wide array of choices for its placement within the utterance, it is possible to investigate whether the alternative positions of the particle contribute to the interpretation of the turn, and if so, in what way. The turn-initial kyllä Wrst caught my attention when I was charting out the conditions for the occurrence of sentence-formatted answers to yes-no questions. The two prototypical or conforming ways to build answers to this type of question in Finnish are by repeating the verb of the question, or by using the response particles joo or niin (Sorjonen 1997: 71; also this volume). In addition, as in any language, full-sentence answers may be used in certain circumstances, and with certain consequences for the trajectory of the sequence. Unlike verbs and particles, sentences have, in principle, no linguistic element in them that would allow us to recognize them as answers (cf. Sacks 1992: 25). However, sentences that begin with kyllä can be recognized as answers, or at least as responsive turns, by virtue of the prefaced particle. I want to begin by brieXy presenting one type of full-sentence answer. With that as a background, I shall then investigate the import of the particle kyllä in full-sentence answers. Subsequently, I shall look at some examples where kyllä is used in utterances that perform functions other than answers to Wnd out if there is a common denominator in turns that include the particle.

The discourse particle kyl(lä) in Finnish conversation 173

My data come from the corpus of Finnish conversations collected at the department of Finnish in the University of Helsinki. The majority of examples are from telephone calls. The current database of non-particle answers consists of 175 positive and 78 negative answers. Of the former, there are 55 instances with a SV-order. The database of sequences with the particle kyllä consists of appoximately 120 excerpts, taken both from telephone conversations and from doctor-patient consultations.

Full-sentence answers Plain SV-sentences Prototypical aYrmative answers comply with their questions, accepting the propositions conveyed by them. But, as is well known, questions are put to multiple uses in conversation. Besides (merely) requesting information, they may act as topic openers oVering a space for the co-participant to tell a story or to go on with one. Invitations, oVers and numerous other Wrst pair-parts may be dressed in an interrogative form. As a consequence, also answers are used to do more than ‘just’ oVer a complying minimal answer to the preceding question (cf. Raevaara 1993, Lindström 1999, Raymond 2001). Here I will look at two examples in which the question seems to be a genuine request for information. However, in both, the answer consists of a full sentence instead of a mere repeat of the Wnite verb in the question. In the Wrst example, the answerer makes substantial changes in the sentence provided by the question; in example (2), the changes are lexical ones. The Wrst extract comes from a phone call between two expert metal workers. Speaker Sami is giving a detailed account of his polishing-up work on a wedge, with further advice to the recipient about how to give Wnishing touches to the object. (Key to glossings and transcription symbols will be found in the Appendix.) (1) 1

[Enemy, 2] Sami:

2 3

Joo et tota mä kattelin ei (siin) jos siin alkaa Yeah so um I was looking no (there) if you begin to viilaa niin siin meneep (.) ik- ikä pe[rkele. file it so it will take up all your life dam[ned it

Vesa:

[Siin on niin [There are so

174 Auli Hakulinen

4 5

monta eri kulmaa many different angles Sami

6 7 8

kyl[lä.h KYL[LÄ. h

[Siin on ne kul]mat ja sit se: [There are the]angles and then the se ellän kulmaki on terävä nin sit (.) si(i)t ainaki the L-shaped angle is sharp so (.) from that one at least

siit yläosast hyvä ottaa vähä pyöreämmäks ja, from the top part it is good to make it a bit rounder and, Vesa:=> Sai-t sä taivu-tettuu kunnolla ne, properly them, get-PST-SG2 you bend-CAUS ‘Did you get them bent properly,’

9

Sami:=> Ne taipu iha:n siististi they bent qui:te neatly ‘They bent just neatly yes

10

Vesa:

11

Sami:

[et siin ei oo] (.) [ so there not is (.) [so there is no ]

p- rautasahalla et se oli:#:# .hh (.) ei siihen b- iron saw-ADE so it wa:#:s# .hh (.) not it -ILL w- with an iron saw so it was .hh (.) it took no more menny tota ni (0.6) ei went PRT PRT (0.6) not than an hour to

13

ne teki ja taivutteli. them ø did and ø made-bend. do them and bend them.’ Vesa:

joo yes

[Joo joo:. [ Yes ye:s.

12

14

]

siihen it-ILL

menny tun:tii-kaa ku went hour-CLIT when

°↑Just joo:.° °↑Quite yeah:.°

To Vesa’s question, ‘did you get them bent properly’ (line 8), Sami gives an aYrmative but a full sentence answer, with some systematic grammatical changes as compared to the question. The agentive subject sä ‘you’ is replaced by an inanimate patient subject ne ‘they’ (line 9), and the transitive (causative) verb taivu-tta-a (bend-CAUS-INF ‘to get bent’) is replaced by the corresponding intransitive one, taipu-a (bend-INF) ‘to bend’. In addition, the verb construction which in the question includes a modal element (sai-t ‘could-SG2’), conveying the sense of overcoming a diYculty, is replaced by the single, nonmodal verb taipu ‘bent’, which emphasizes the non-problematic nature of the job in question. Even the adverb kunnolla ‘properly’ in the question is changed in the answer to siististi ‘neatly’. The implications of the job being a diYcult one are thereby wiped out with the nonchalant description of an eVortless event (‘they bent just neatly’). What the example illustrates is the fact that, even when the answer given is an aYrmative one, the recipient can distance him/ herself from some of the senses or implications included in the question. This

The discourse particle kyl(lä) in Finnish conversation

is achieved by presenting a sentence of one’s own instead of a verb repeat which uses the elements available in the question. The utterance ends with an aYrmative response particle joo, which marks it grammatically as an answer. It should be pointed out here that the answerer subsequently pursues the topic, and gives detailed evidence of how and why the job was so easy, thereby treating his answer as one that departed from the implications of the question. This Wnding agrees with the claim made by Raymond (2001) that ‘non-conforming’ answers tend to require an account or reason for departing from the minimal answer. That an account is not always necessary, however, will become evident in the next example. Example (2) illustrates another context typical for a full-sentence answer. Earlier in this call, the main speaker Mari has been telling her recipient Eeva about the sad occasion of her dear workmate leaving and moving to another part of the country. Lines 1–4 are part of Mari telling about another friend who has been helping her. With the exclamation at the end of line 4 she can be heard to return to the issue of the departing workmate. Subsequently, the recipient directs a matter-of-fact question to her about the time of her friend’s (Riitta’s) leaving. (2) [Fox, 5] 1

M:

2

ruottiks [kommentoida] ja [tämmöst ] to comment in Swedish and [ that sort of ]

3

E:

4

M:

5

E:

6

E:

[>Nii nii:.Yes yes:.< ] näinnik.kääst .että .hh kind of a .way .so .hh

[Joo:. ] [ye:a.. ] [.hh ] @Voi °sentää.hh°@ [.hh ] @Oh my °goodness° .hh@ [Joo::. ] [Yea:: ]

No tuota< (.) vielä siitä Riitasta ni lähtöö-kö Well PRT< (.) still that-of 1stName-ELA so leave-SG3-Q ‘Well um< (.) still about Riitta is she going to se ↑nyt sitte heti:. is ↑now then immediately. leave right away then.’

7

8

.hhh Joo::. Ja sit ku se vielä ossaa vähä .hhh Ye::s. And so when she even knows a bit how

M:=>

[Tavarat lähtee e Pieksä[mäelle.] Se muuttaa huomenna. [things leave-SG3 uh PlaceNAME-ALL. it move-SG3 tomorrow ‘She is moving tomorrow’ [Her baggage is setting off for Pieksä[mäki.’

175

176 Auli Hakulinen

9

E:

[.tj

[Voi luo]:ja. ‘[Oh my Go:]d.’

Instead of giving a simple aYrmative answer by repeating the verb of the question (lähtöö), Mari answers with a full sentence that has the same syntactic frame as the question in lines 6–7. She uses an anaphoric pronoun se ‘(s)he’ as the subject phrase in her answer (line 8), tying it to the person (Riitta) mentioned in the question. Yet she replaces both the verb and the adverbial with ones more speciWc than those used by her interlocutor. In general, using a full sentence as an answer allows the speaker to change the verb. This is done in example (2): from lähtee ‘leaves’ to the verb muuttaa ‘moves’, which requires a human subject and implicates ‘not returning’. In the next utterance, Mari adds information about the ‘baggage leaving’, using here the verb oVered by the questioner. The full-sentence answer, then, seems to achieve two goals at once: while oVering an aYrmative answer, it also more or less discreetly produces a repair of something in the question by changing several of the linguistic details in the description of the state of aVairs presented in it.

Full-sentence answers with an initial kyllä As I mentioned above, a sentence-formatted answer is frequently prefaced by the particle kyllä. The authoritative Dictionary of Contemporary Finnish Nykysuomem sanakirja (1961, s.v. kyllä I, 1.) states that it is used ”especially in turns of talk that express concession, or agree with the thought of another person; it makes the claim more emphatic; and corroborates aYrmative answers”. Let us next look at some examples from the database in order to see to what extent Wndings in the naturalistic data attest to the claims made in the dictionary. In the next two examples, like in examples (1) and (2), the respondents make changes in the verb oVered in the preceding question. Yet, the answer more clearly aligns with the positive alternative of the yes/no question than in the Wrst two examples. In example (3), the speakers are negotiating about putting up a job announcement on the departmental notice board. (3) [Notice, 1] 1 2

I:

.hh- (.) ↑voisik sä panna sen. .hh- (.) ↑could you put it. ((up on a notice board)) (.)


3

M:

4

I:

5

6

hjoo kun sanot vaan että mitä siihen panna [an. hyeah when you just tell me what to put in [it. [o- on-ks sul [i- is-Q you-ADE [‘Are you’ nyt niin kiire että sä- e:hdi-k-sä ottaa. now so busy that you- have-time-Q-you take. so busy now that you- have you got the time to take it.’

M:=>

↑kyllä minä kerkiä-n ottaa. joo ↑Kyllä I have-time-SG1 take. yeah ‘Kyllä I have the time to take it. yes’

The caller Ilona asks if the recipient could put up an announcement (line 1). Having received a positive response (line 3), she goes on to ask if Maija has the time to ‘take’ (i.e. write down the information needed for the announcement) it right away (nyt ‘now’, line 5). In her answer, Maija conWrms having the time but does this with a verb choice of her own, using a synonym taken from her own vernacular. The context of this example diVers from the ones discussed so far. We already noted that Ilona does not proceed right away to the question that we are focussing on but she has already checked the circumstances with her Wrst question at line 1 by asking Maija to do the favour for her. With her answer to Ilona’s question, at line 3, Maija agrees to do the favour but sets conditions on doing so. Ilona’s turn at lines 4–5 can be analysed in relation to Maija’s conditional agreement. Dictating the announcement is potentially time-consuming, and there is the possibility that Maija might not have the time. (In fact, there is a question about time already in the opening routine of the call; data not shown.) The turn consists of two parts: Ilona begins her question by asking whether Maija is occupied (‘are you so busy now ‘). She then interrupts herself after the conjunction että ‘that’ and the subject pronoun sä- ‘you’ thus aborting the projected negative subordinate clause (‘that you cannot take [down the facts]’). As a self-repair, Ilona subsequently produces a more straightforward question (‘have you got the time’). In this instance, Maija’s full-sentence answer with the initial modal particle kyllä seems to function as a reassurance, producing a more explicitly conWrming answer than a mere verb repeat would, thereby doing away with the doubts created by the preceding exchange. The rise in tone when she utters the particle kyllä adds to the sense of reassurance that the action carries. The kyllä-prefaced full-sentence answer is thus more explicitly forthcoming, aligning more clearly with the positive alternative oVered in the question, than a ‘plain’ SV-formatted utterance would be.

178 Auli Hakulinen

The reassurance must be seen not only with respect to the question at the end of the previous turn but with respect to the entire previous turn: the turnWnal question is preceded by another one expressing doubts about the possibility of getting Maija to do the job right away. Thus, Maija’s answer is orienting to both parts of the turn at once. With her choice of the synonymous modal verb (kerkiää ‘have the time’, line 6; cf. ehtii, line 5)3 Maija avoids a verbatim repetition of the question. It is hard to pin down the diVerence in Xavour that the change of verbs brings about. In view of the explanation given for example (1) above, one could claim that the forming of a sentence of her own may reXect the independence of Maija’s assessment of the time she can devote for the task at hand. In the next example, the caller Kantola presents a question that, like in the previous case, opens up a presequence. The question projects an invitation for a lunch date (line 4). In her answer, the recipient, Arja, responds with a full sentence which makes use of and modiWes the elements in the question. She uses the modal verb voida ‘can’, in a subjunctive mood, not used by the questioner. The modal verb, the stress on the verb, and the subjunctive mood implicate her conditional willingness to negotiate the proposal. (4) [Lunch date, 1] 1

AR:

Arja Rautiainen 1ST NAME SURNAME ,

2

K:

Kantola täällä hei? here hi, SURNAME

3

AR:

↑no ↓hei. ↑Well ↓hi

4

K

Olek-ko sä tänäpänä syömässä täällä be-SG2-Q you this-day eating-INE here..’ ‘Are you eating here today.’

5

AR:

Kyllä mä voi-si-n olla.=joo. KYLLÄ I could-SG1 be.=yeah. ‘KYLLÄ I could be. =yes.’

6

K:

Erinomasta.=mutta mihin aikaan. Excellent. =but at what time.

7

AR:

Mikäs sulle sopis.= What would suit you.=

8

K:

=> hetiRight away

sen lämmitysjärjestelmää ja .hhh oisin kysyny sitä että it-GEN heating-system and .hhh be-CON-SG1 ask it that its heating system and I wanted to ask if

7

mh ö vanhoihi laittaa-ko (.) Kuhmon Sähkö täm:mösiä install-Q (.) Kuhmo-GEN El thi:s-kin:d mh er old-PL-ILL the Kuhmo Electricity installs in old stoves

8

niitä sähkövastus sähkövastus,h (0.2) .hhh uuniin .hhh stoves-ILL those electrical resistor, h (0.2) those electrical resistors

9

#ö# °lämmitys juttuja.° lämmitys (.) juttuja. #um# °heating (.) things.° for heating.’


10

P:→

Ky:llä me niihi öh niitäkin tehdä-än °että.°= KY:LLÄ we they-ILL er those-PRT make-PAS °so°.= ‘KYLLÄ we make them for those as well so.’

11

T:

=Joo. .hhh No (.) mites se o saako sen yösähkööhh. =Yes. .hhh well (.) how is it can you get it on night electricity

12

P:

.hh Ky:llä kyllä KYLLÄ .hh KYLLÄ

13 14

saa °et ei siinä,h° you can get (it) °there is nothing, h°

(.) P:=>

15

Se että vaa: se saattaa k- öö sehän p- mittauskeskuksen:: it that only: it may h- um it-CLIT f- electric meter:: ‘It is just that you might you know the m- electric meter öh siinä saattaa joutua uusimaan °sillon että.°= uh there may have to renew °then that.° = might have to be replaced in that case so.’

16

T:

=Joo. .hhh Mutta että tällanen on hh (.) mahdollista. =Yes. .hhh But so this kind of thing is hh (.) possible.

17

P:

Ky:llä on mahdollista. KY:LLÄ is possible. ‘KYLLÄ it is possible.’

To Tuula’s inquiry about installing electric heating to traditional stoves (lines 7–9), the electrician replies with a kyllä-prefaced turn in which he has changed the verb in the question laittaa-ko ‘does install’ (line 7) for its synonym (me) -- tehdään ‘(we) make’ (line 10). The situation diVers from the ones in the two previous examples in that there is no doubt in the question nor in the overall context. Rather, it may be the case that the state of aVairs presented in the question is not an ideal one from the point of view of the respondent.5 What is potentially relevant here is his pronunciation of the particle ky:llä. There is a slight prolongation in the Wrst syllable (indicated by the colon). The prolonged particle gets repeated in his next answer about the use of night electricity (line 12), as well as for the third time, in reassuring the caller about the possibility of the enterprise (line 17). Thus, while he is aligning with the question, and by using the initial particle, going against some doubts, Pitkänen marks his hesitancy with the pronunciation of the particle. The sequence gets a diVerent continuation than the previous ones. As a matter of fact the hesitation seems to be projecting a problem or an obstacle of some sort: later on (see lines 14–15) Pitkänen indeed brings up a possible problem of having to do more than just the installation, namely the need to replace the electric meter as well.6

181

182 Auli Hakulinen

Sequential environments for the two answer-types Questions that receive full sentences as answers in the database can be roughly grouped into three types. The most frequent type is genuine requests for information. The majority of kyllä-prefaced answers are given as responses to these information-seeking questions. The second largest group is formed by questions that story recipients make in order to get a clariWcation of something in an ongoing narrative, thereby letting the main speaker continue or elaborate on the narration. To this type of question, there were no answers beginning with kyllä; this was a favourite position for the plain SV-type answer. Thirdly, to questions that open up a presequence, projecting a request, invitation or interrogation, kyllä-type answers were favoured. Answers formulated with a SV-type utterance tend thus to turn up at places where the answerer has been the main speaker and is being enticed to go on by the co-participant. Continuing one’s own topic is an activity where a full sentence is to be expected instead of a mere verb repeat that is parasitically attached to a question by the coparticipant. On the other hand, answers beginning with kyllä tend to occur in presequences, or as responses to questions that are in some way backgrounded, and it is the questioner who continues the talk and not the participant who produced the answer.

Kyllä in other environments Next, I shall move on to look into some other uses of the particle kyllä, Wrst in turns that act as second pair-parts, and then in some other positions within a sequence or in a turn. As was pointed out in the introduction, the utterance-initial position is not the only possible one for the particle kyl(lä). Let us next see under what conditions the other positions are available for the particle.

Responses to oVers The next two examples illustrate the import, for the overall meaning of a turn, of the positioning of kyllä in second pair-parts other than answers. In both examples, the Wrst pair-part to which the kyllä-prefaced turn responds is an oVer. In each case, the response is constructed with an evaluative utterance but the particle is not placed in the same position within the turn in the two cases.


In example (7), the recipient responds with a turn initiated with exclamations plus an attention-getter (kuule ‘listen’) followed by an utterance prefaced with kyllä. In example (8), the answer begins with a prolonged response particle (j:oo:.) plus a smack (.mt). These initial elements are followed by an utterance where kyllä occurs after the initial subject se ‘it’. In (7), the speaker Tellu is acting as a mediator in arranging for her friends to get cloudberries from Lapland at a reasonable price — an activity engaged in among friends in Finland in the autumn. (7) Cloudberry, 1 1 2

Tellu: Eija ja Outi tulivat (.) eilen kuule Lapista ja ne NAME and NAME came (.) yesterday PRT from Lapland and they oli saaneet ylimääräistä ja< had got some extra ((berries)) and
Ai jai ja kuule kyllä vaan oikeen mielellään. .hh PRT real pleased. .hh Oh wow and listen KYLLÄ ‘Oh wow and listen KYLLÄ indeed with great pleasure. .hh Tosi ö tosi herkku A real um real yummy

juttu oikeen. thing indeed.’

The oVer (at lines 4–5) is formulated in a vague manner (‘they’d be selling some if it would interest (you)’), indicating the possible delicacy of the action in question. The boldfaced utterance contains a zero object (Hakulinen and Karttunen 1973; Laitinen 1995): in the ‘if’-clause, there is no constituent expressing who is the one who would be interested. However, in the context, the utterance is hearably an oVer made to the current recipient. Ulla’s acceptance shows eagerness; it contains evaluative elements (‘with great pleasure’), making use of various kinds of aVective material, and with both the subject and the verb missing. The particle kyllä can be heard as responding to the delicate aspect of the oVer.

184 Auli Hakulinen

Example (8) comes from a slightly more business-like call, where the buyer, Ossi, has been making inquiries about where to get the best discount for kitchen equipment. He gets an oVer of a Wfteen percent discount from M, the saleswoman he has obviously been talking to on an earlier occasion. (8) [Kitchen, 12] 1

M:

2

Siin on tota sitten toi vaku- vakuutushan tulee sitte There is PRT then that insu- insurance will come then noista,= from those,=

3

O:

4

M:

=Tulee lisää pää[lle joo. =(It’ll) come in [addition on top yes. [noista, noista plus yks elikkä tota: hmhh [from those, those plus one that is PRT hmhh viidentoist prosentin, ‘a fifteen percent,

5 6

O:

J:oo. Yes

7

M:

prosentin alennuksen sitten myöntää>. percent discount then . percent discount one can grant from it then.’

8

O:=>

hyvältä vältä, J:oo:. mt Se kyl kuulostais Ye:ah:.mt It KYL sound-CON-SG3 good-ELA, ‘Yeah. mt It KYL would sound good,’

9

S:

Mm::. Mm::.

The oVer is presented at lines 5–7; just as in the previous example, the oVer here too is formulated as a zero-person construction (‘(one) can grant a Wfteen percent discount’).7 Compared with the enthusiastic response of the recipient in (7), the response here by Ossi is heard as quite reserved. It is also formulated as an evaluative utterance, but several of its features — the prolonged form of the turn-initial response particle j:oo:, the following smack (mt), plus the subjunctive mood of the verb — mark the response as a rather reserved receipt of the oVer. There is no explicit acceptance like ‘OK' or ‘that's Wne with us'. What is of interest for our present purposes is that the aligning particle kyl here is not placed immediately in utterance-initial position but after the subject. Through its delayed placement and its truncated form kyl seems to be adding to the resistant character of the turn. Thus the overall design of the turn amounts to an interpretation that speaker Ossi is taking a stand as a customer with the right to evaluate the oVer critically. The alternative formulation with


an initial kyllä (kyllä se kuulostais hyvältä ‘KYLLÄ it would sound good'), on the other hand, if it were possible at a place like this, would more explicitly convey an emphatic appreciation of the oVer. (For a parallel case, see example 4, line 9 above.) On the basis of the two previous examples illustrating the use of the particle kyllä in receipts of oVers, we could set up a tentative hypothesis: the placement of the particle within an utterance iconically reXects the degree to which the speaker is going along with the suggestion or other action presented by the co-participant. The further to the front the particle is placed, the more deWnite the alignment.

Going against opposing opinions In the examples discussed in the previous section, the responsive utterance with kyllä was performing an evaluation of the oVer made by the co-participant. In this section, we shall slightly change the point of view by taking up some examples of assessments in which the kyllä-prefaced utterance is not turn-initial but occurs in the middle of the main speaker’s turn. These are turns that do not seem to form a second pair-part within a sequence. In example (9), there are two utterances by the same speaker Pete, who is telling Kuikka, the recipient, about a big change in his life: he is going to leave town for life on a country estate. In lines 5–6, he complains about the tiresome job of clearing out his previous house; in lines 12–13, he produces a more optimistic idiom-like statement with which he ends the complaining sequence. (Cf. Drew and Holt 1988.) (9) [Kuikka, 6] 1

P:

No< hän: aina vielä kerkii kaiken kiireen keskellä.= Well she always finds the time in the midst of everything.=

2

K:

=Nii:::= =Yea:::=

3

P:

=Jo [o it is. =And don’t you get

[ .hhh ] ↑No: um er: and (we) get quite a bit even at times but

11

.hh sit ku näkee uuden nin ei kyl sit katoo taas .hh then when you see the new one so it disappears again

12

=>

ja .h kyl se niin on et ei tä-s elämä-s mittään saa and .h KYL it so is that not this-INE life-INE nothing øgets and .h KYL it is so that you don’t get anything in this life jos nyt yhteen tonttiin ↓juuttuu. if PRT one-ILL plot-ILL ↓get-stuck if you just cling on to one plot.’

13

14

K:

Aijjaha. oh I see.

15

P:

=nii::. =ye::s.

16 17

(.) K:

juu mutta uskollisuus on myöskin yks: #e::# arvo. one #um::# value.’ yes but loyalty is also

Both of Pete’s utterances convey an evaluation or an assessment: both the utterance at lines 5–6 (following a particle chain that accomplishes a change of topic) and at 12–13, are initiated with kyl. The particle occurs here in a truncated form like in the previous example, where it was suggested that it adds to the interpretation of the turn as resistant. Notice, however, that in both instances here it is stressed and placed initially. Now, it is quite clear that neither of these two kyl-utterances is used as a response to what was said previously. In line 5, speaker P accomplishes a change of topic by means of the particle chain ei mut ‘no but’ (Sorjonen 1989), and the utterance in lines 12–13 occurs within the ongoing turn of the current speaker, linked with the connective ja ‘and’ to what he himself has said before. In fact, in both these instances, the kyl-prefaced utterance seems to act as a Wrst pair-part, creating a space for the recipient to come up with a second assessment. Thus, there must be some motivation for these utterances to be kyllä-prefaced other than aligning or going along with


something that is in the immediately preceding turn of the co-participant. What these two utterances by speaker P seem to have in common is that both of them formulate a stance to a state of aVairs as belonging to common experience or general knowledge. The Wrst one of them gets an immediate second assessment from the recipient: ‘to be sure it is’ (line 7).8 The second one, constructed as a verb-Wnal statement the way idioms and proverbs are,9 is not followed by an aligning turn but receives a change-of-state token (Aijjaha) from Kuikka (line 14). Subsequently, Kuikka comes up with a turn that conveys another general truth, ‘loyalty is also a value', thereby making the disagreement explicit. On the basis of examples of this kind, it can be argued that kyllä-prefaced stance takings that are formulated as proverbial truths can be used to anticipate or refute alternative opinions that are not explicitly expressed but that are somehow lurking in the air. What I'd like to suggest here is that there is a link to the kinds of assessments that were oVered as responses (examples 7–8): in the former instances the turn was responding to some problematic aspect of the turn by the previous speaker; in (9), the speaker is working against a possible disagreement, which may or may not be put in words, by expressing a general truth.

Doing away with doubts Through the examples described above, we were able to capture some of the functions of kyllä in three diVerent sequential environments. The picture is, naturally, simpliWed, as space does not allow more examples to illustrate some additional aspects of its use. But I still wish to discuss the claim made in the Dictionary of Contemporary Finnish, mentioned above, according to which kyllä makes “the claim more emphatic; corroborating aYrmative answers”. The kinds of aYrmative answers we encountered so far were found not so much corroborating as reassuring and doing away with doubts. The utterances that occurred in positions other than that of second pair-parts could be emphatic but did not have to be. In this section, I will discuss two additional cases that will shed light on the issue of ‘emphasis’, whatever that in actual practice might mean. The Wrst one is exempliWed by (10a). Here, the caller R is making inquiries about equipment that would enable him to send rubbish from a renovated apartment down to the street level. He is thinking of renting a tube that could be Wxed from a window down to the ground level.

188 Auli Hakulinen

(10a)

[Renovation, 3]

1

R:

Ja tota niin m- mites- kuinka pieniks osiks tää menee and PRT so wh- what- how small-TRA piece-TRA this goes ‘And um so h- in wh- to how small pieces can this be taken

2

=>

et kulkee-ks se tavallisella henkilöautolla. so goes-Q it ordinary-ADE person-car-ADE. that is does it fit in an ordinary passenger car.’

3

L:

Ei It

4

R:

[Ei [kulje [It [does not

5

L:=>

[Kyl se o- kyl se vaatis: vähintään [KYL it i- kyl it deman:d-CON at least [‘KYL it would require at least

6

kuljehh[h does not h[hh

peräkärryn et s’on, (.) s’on ku s’on supuss:a. cart so it’s, (.) it’s when it’s closed-INE. a cart so it’s, (.) it’s when it’s pulled in.’

To the question of the possibility of carrying the tube on top of or inside a small car, the leaser L Wrst gives a negative answer (line 3), which gets repeated by the caller. As a negative answer is a dispreferred one, it is followed by a explanation through a turn prefaced with kyllä (lines 5–6). Here, the function of the particle seems to be one of doing away with the misgivings of the recipient; it is corroborating the speaker’s previous negative answer and going against the client’s assumptions. After some further negotiating about the details and functions of the tube (29 lines omitted), the same speaker L again makes use of a kyllä-initiated utterance (example 10b, lines 40–41), this time when making an estimate of the weight of the tube. (10b) 36

R:

37 38

sen suunnill(h)een it will approximately L:

41

kuinka how

paljo sä arvioit much do you estimate

painava. weigh.’

mt .hh No siinose painaa jotain: kymmenkunta mt .hh ‘Well there’s- it weighs around: ten-ish kiloo (.) ja, ja niit on viistoista kilos (.) a and, and there are fifteen

39 40

Milläs sitte-, t- ku‘How thent- h-

=>

kappaletta siinä että Kyllä se semmosen KYLLÄ it kind-of piece it-in so pieces in it so that KYLLÄ it will mfff sataviiskytä kiloo painaa suurin piirtein. mfff hundred-fifty kilo weighs rough lines weigh around hundred and fifty kilos more or less.’


As a response to the seemingly neutral, information-seeking question (lines 36–37), the kyllä-prefaced utterance is quite clearly doing more than just oVering an answer. Through the particle, the utterance is also orienting to the doubts of the caller, who does not seem easily to want to give up the idea of taking the tube home in his car. Example (10) thus illustrates the use of kyllä in talking about issues where the participants have diVering opinions or even express disagreement. Here, the function of the initial particle seems to be to weigh the speaker’s own argument in one direction, and thereby to do away with the doubts or misgivings of the co-participant. The doubts may, however, be raised by the speaker himself/herself as well. This state of aVairs can be observed in the following example, which comes from a call between two friends who are planning a tent trip. Pekka reports that he has not started packing yet, but has only tried to Wx the roof rack on his car. (11)

[Packing, 3]

1

S:

No huomennako meinaat pakkailla.= Well were you thinking of packing tomorrow.

2

P:

=.hh Joo en mä oo viel mitään tehny ↑Mä ö< sain =.hh Yes I haven’t done anything yet ↑I er< got

3

sen e< (.) tossa< äkkiä laitoin sen toisen< (0.8) it f- (.) there< quickly I put the other< (0.8)

4

>kattotelineen< paikalleen ja,h (.) >roof rack< on its place and, h (.)

5

=>

6 7

kyl se nyt sopii siihen joten kuten .hhh KYL it PRT fits in there somehow .hhh ‘KYL it will fit there somehow .hhh’ (1.3)

S:

Nii, Yeah,

Here the speaker is in a way talking to himself. He is late in his packing; he has performed the Wxing of the roof rack in a hurry: äkkiä ‘quickly’ (line 3) — something that could have resulted in a bad Wt. The kyl-prefaced utterance at line 5 may be seen as doing away with his own doubts — possibly also those of his mate — about how well the rack is Wxed to the roof. One of the elements that contribute to the interpretation of ‘talking to himself’ rather than trying to convince the recipient is the use of the particle nyt. In earlier research (Hakulinen 1998) the particle was found to occur frequently in statements that convey personal opinions. So far, we have seen that the deWnition in the Dictionary of Contemporary

190 Auli Hakulinen

Finnish, quoted earlier on, does not capture all of the aspects of the use of the particle kyllä. Rather than “expressing concession” or “corroborating aYrmative answers”, the common denominator that has been found among its uses in diVerent sequential environments is the function of doing away with doubts, misgivings or disagreements which are either explicitly or implicitly present in the context. It was also suggested that the placement of the particle within the turn and within an utterance plays a role in its interpretation. The utteranceinitial particle favours certain types of turn within a sequence; in particular, responses to questions. In a mid position, it was suggested, kyllä reXects, iconically, resistance to agreeing or aligning with the co-participant. Let us next move on to look at examples where kyllä is placed in utterance-Wnal position.

Kyllä in the Wnal position The last two examples come from the set of data with an utterance-Wnal kyllä. Example (12) is from a phone call between two brothers, one of whom (Saku) is planning to do some levelling work on his yard with a caterpillar. His brother Veke, on the other hand, thinks the ground is too soft for such a heavy vehicle. (12)

[Brothers, 12]

1

Veke:

2

Saku:

Vahinko vain et nyt on noin pe:h [meetä (jo). ] ‘A pity only that (it) is now so so:f[t (already).’ [mthh [mthh

] Joo ] Yes

3

mä just kattelin se se lupas ei:len (.) tai huomise-lle I just watched it it promised yesterday (.) or tomorrow-ALL ‘I was just watching they forecast yesterday (.) I mean for

4

päivä-lle ni .mhh [plus y:ks plus ka]ks. tomorrow PRT .mhh [plus o:ne plus tw]o.

5

Veke:=>

[Se:hä menee< ] Se:hä menee [i:t-CLIT goSG3< ] ‘See it will go nur:mikko ihan paskaks.h law:n quite shit-TRA all mucky the lawn.’

6

7

men:nä kyllä [joo Saku:=> °Voi joo°. ] Voi se vähän joo KYLLÄ [ yea°. °can it a little go: ‘It may do so a little bit KYLLÄ yes.’

8

Veke:

[Ky:l se] menee Nyt se on [KY:L it will go Now it is


9

niin (.) se on niin liejua toi pin:ta et, so (.) it is so sludgy the su:rface that,

10

(.)

11

Saku:

12

mt (Mut jos mä) kokeilen kyl KYL mt (But if I’ll) try

se (m-) (.) jonkun verran it w(.) to some extent

jäässä on °(viel:,)E icy is (still,)

The two brothers Wrst seem to be in agreement about the conditions: the weather forecast predicts a temperature of +1–+2° centigrade, with the consequence that there will be hardly any ground frost left. On the basis of this fact, Veke expresses a strong opinion about the outcome of the job: the caterpillar will make the lawn go all mucky (lines 5–6). He even uses the enclitic particle hän (se:hä line 5), which indicates shared knowledge. The relevant response to the utterance is agreement, and this is what Saku produces, as indicated by the turn-Wnal response particle joo (cf. the similar case in example 4 line 5). The turn has several features that reXect an unwillingness to agree, however. It is pronounced sotto voce; it is modiWed by the initially placed modal verb voi ‘may’ indicating possibility and by the adverb ‘a little’.10 In contrast with example (4), where the response begins with kyllä, the particle here is placed at the end of the utterance. This might be an instance of “expressing concession” in the sense mentioned by the Dictionary of Contemporary Finnish. As a conspicuous contrast to this turn, note the use of the particle kyl in the following two turns as the arguing continues. At line 8, Veke holds on to his opinion with a kyl-prefaced utterance, conWrming his own previous turn. Instead of showing any more agreement, Saku for his part hangs on to his view that the ground is still a little bit frosty (line 11). Extract (13) is from a call between two acquaintances, Sara and Vera. Sara has called Vera, who has the right to buy provisions at a cheap price in a wholesale store, to ask if they could do some shopping there together. (Data not shown here.) Having agreed to this, Vera in turn oVers her friend the possibility of buying a bucket of honey at a lucrative price. The oVer leads to a lengthy negotiation about the details of getting hold of the honey.11 (13) 1

[Honey, 4]

Vera:

Nytpä on muu]ten tuo:ta toi hunaja'haku et mä en Now there's by the way PRT the honey fetch so I don't

2

tiiä ootteks te (.) ollu kiinnostuneita yleensä know if you've (.) been interested in general

3

[(ottamaan). [(in taking (any).)

] ]

191

192 Auli Hakulinen

4

Sara:

5

[Ai:< ↑Joo mä oon] joskus ottanu ja [‘O:h< ↑yes I have] sometimes taken and [kyllä näköjää:n ] vois ajatella. =e Millos se on. [KYLLÄ apparently:] one could consider (it). =e When is it.

6

Vera:

[°(Joo::)° [°(yea::)

7

Vera:

No< se ois itse asiassa nyt aika ↓kii:rekki jo.hh Jos, Well< there’d be in fact now quite ↓hurry already.hh If,

8

Sara:

Nii et[tä ois pitäny jo ilmottaa [vai, ] So that (one) should have already notified or,

9

Vera:

10

] ]

[( ) [( ) v- periaatteessa a- in principle

[↑Nii:h] >kyllä [↑yea:h] >KYLLÄ niinku like

juhannukseen mennessä mut by midsummer but

-------((50 lines omitted, re timing, location, amount and price of the honey, etc.)) -------61

Vera:

62 Sara:

64

Vera:=>

[Joo. ] Joo [joo.] [Yes. ] yes [yes.] ir:veen [(Se)] hi (Se)] on ollu h [ (it) has been terr:ibly hyvää good

65

67

se on ollu it has been

hiukan halvemp[aa.hh] a bit cheap[er .hh ]

63

66

[ Nii]:. Et[tä päivvaston [ yes:. so on the contrary

Sara:

hunajaa kyllä että= KYLLÄ so= honey

=Joo:. .hh ↑Joo no kiva kun (.) kuulin ku tosiaan =Yes:. .hh ↑Yes well nice to (.) have heard as really (.) täss on .hh ollu näit muita hössötyksiä nin, so,’ (.) there have .hh been these other hassles

The piling up of the details increasingly points to the unwillingness of Sara to commit herself to buying any honey. As a last eVort, Vera produces a recommendation in order to convince her resistant interlocutor (lines 64–65). The recommendation is formulated as an evaluative statement. It diVers, however, from the ones discussed earlier (cf. examples 7–8) in several ways. First of all, it is not oVered as a response that shows evaluation or appreciation of what was previously said by the co-participant. On the contrary, due to its sequential position, the turn is presented as an appeal to the resisting recipient. Also grammatically the utterance is built diVerently. The verb of the utterance is in the present perfect tense. This tense is typically used in turns that justify assessments and opinions by referring to personal experience (Seppänen 1998).


The utterance may thus carry the implication that, as it has been found good previously, the honey will be good this year as well. What this kyllä-utterance has in common with examples (10–11) is that it is used to do away with doubts. The Wnal position12 of the particle, however, seems additionally to convey the sense of appealing. This Wts well with the observation made by SchegloV (1996: 91) that at the place of post-possible completion, various stance markers tend to get placed. (Cf. also Hakulinen and Seppänen (1992) for the placement of the appealing attention-marker kato at this position.) Earlier, I suggested that the placement of the particle — whether in the initial or mid position — iconically reXects the stance of the speaker. The last two examples seem to support this suggestion only in part. What the examples have in common is the fact that there is overt or covert disagreement between the interlocutors. But the interpretation of the import of the particle seems, in the last resort, to depend on the sequential position of the turn. In (12) the Wnal position was used in a turn conveying reluctant agreement with a preceding claim by the co-participant. In (13), we found Vera placing the particle in the Wnal position in a turn that appealed to her unwilling co-participant. The interplay of the sequential position of the turn that contains kyllä and the position of kyllä in the turn is something that obviously requires further investigation.

Conclusion In the analyses presented in this paper, a common denominator was detected between the uses of kyllä in various kinds of sequential environments. I started oV by explicating why kyllä appears initially in sentence-formatted answers to yes/no questions. It was possible to proWle its use here in contrast to answers that are formed with a mere repeat of the verb in the question and to those that consist of a ‘plain’ statement-formatted answer. What kyllä seems to be doing in answers is that it more clearly aligns with the positive alternative of the yes/ no question than a full sentence, and gives the answer the Xavour of an independent opinion of its speaker as compared with answers that are formed as mere verb repeats. Subsequently, instances that have kyllä in other types of turns, and in other than initial positions were discussed. On the basis of the analyses, it turns out that the sense of reassurance or aYrming that kyllä seems to convey is often a reXection of and reaction to something that is in the environment: it is a way of dealing with delicate actions, doubts, or diVering

194 Auli Hakulinen

opinions. In addition, as one of the outcomes of the analysis, it was found that the position of the particle kyllä is an indisputable element in its interpretation. The closer to the beginning of the utterance the particle is placed, the more clearly its speaker shows alignment and agreement with the co-participant. The initial position was found to be favoured by answers as well as by proverbial expressions of general truths, whereas when kyllä occurs in Wnal position the utterance conveys some trouble in the exchange: either the speaker’s unwillingness to comply with the ongoing activity, or an eVort to appeal to a resisting co-participant.

Notes * I am grateful to the two editors of this volume and especially to Marja-Leena Sorjonen for many constructive points of criticism on earlier drafts of this paper. I also beneWtted much from the comments made by Marja-Liisa Helasvuo, in particular with respect to the examples. The remaining faults are my responsibility only. 1. The diversity of terms used for discourse particles is due to the wide range of approaches used for describing them (cf. Jucker and Ziv 1998). In my view, the principal line of division should be drawn between dialogue particles (yes, mm, etc) that can form an utterance and a turn on their own, and discourse particles that typically occur within a turn. The latter may, for speciWc purposes, be further divided into subgroups like modal particles vs. connectives, etc. In the present context, it suYces to say that kyl(lä) is a particle that has characteristics of modal particles as well as of those that in German linguistics have been called Abtönungspartikeln (cf. Weydt 1969), roughly translatable as ‘tone particles’ or ‘shade particles’. 2. In institutional dialogue in particular, you may Wnd kyllä as an answer, forming a turn of its own, in the way dialogue particles like niin or joo ‘yes’ are used. In everyday talk, kyllä only seldom forms a turn of its own — and there, interestingly, it is often a topic killer, in the repeated form kyllä kyllä. 3. The stem of verb ‘to have time’ is ehti, and an inXected form of the stem, in certain positions, is ehdi. A peculiarity of the Finnish system of modal verbs is the large number of lexical alternatives expressing dynamic possibility, i.e. ability. In addition to the basic modal verb voi ‘be able; may’, there are close synonyms conveying senses like ‘be able because of time, energy, courage; be warm enough’, etc. These verbs are especially used in requests and in accounts for not doing something (cf. Flint 1980). There is a danger of reading too much into the change of verbs here. However, it is not without interest to note that kerkiää, the verb used in the answer, has a more general sense of availability than ehtii, which only means ‘have the time’. I was kindly provided with this information from the archives of the dialect dictionary by Eeva-Leena Seppänen. 4. Space does not allow me to discuss another possible variant, a verb-initial full-sentence answer. It would be, in both (3) and (4), a grammatically possible alternative, but perhaps


lacking the sense of reassurance that the variant with the initial particle has. This issue requires more careful research. 5. I am grateful to Marja-Leena Sorjonen for this observation. 6. The prolongation of the particle is a symptomatic but not a necessary ingredient in a reserved answer, however. The format kyllä -- mutta ‘yes but’ is frequent enough as a partly agreeing answer to have been recognized by lexicographers. The Dictionary of Contemporary Finnish [Nykysuomen sanakirja] puts it in this way: ”Especially in adversative connections, meaning ‘indeed’: He is kyllä a good hunter but a bad physician.” Note, however, that in the formula, kyllä does not stand in initial position. (Cf. the discussion of examples 8 and 12.) 7. The notions ‘missing person’ or ‘missing subject’ construction refer to sentences with no overt subject (or object or other core argument of the verb) but with an understanding that the ‘missing’ referent is human. (Hakulinen and Karttunen 1973; Laitinen 1995.) Sentences that contain the missing person can mostly be translated into English by using the generic pronouns one or you. 8. Compare this with a further example of kyl occurring in the Wrst assessment of an adjacency pair (from Tainio 1996, 103): A: B:

[Party] Kyl se on: syksy on mahdottoman kaunis.= kyl it is: the autumn is awfully beautiful.= =On se,

9. An example from outside the present data will corroborate the Wndings made here. One conventionalized type of proverb begins with the particle kyllä, e.g. Kyllä elävä palavansa tuntee. ‘KYLLÄ the living one feels if (s)he is burning’, literally: ‘KYLLÄ living burning feels’. Proverbs represent knowledge and attitudes taken to be shared within a community in a fossilized and compactly formatted manner, and they are typically used as reactions to something that is salient in the context — not necessarily something that was said. The proverb quoted here, as well as so many others beginning with kyllä, would be ungrammatical without the initial particle that conveys the sharedness of the experience required by the genre of proverbs. 10. Depending on the context, a verb-initial utterance, according to Vilkuna (1989: 114), may be interpreted either as aYrming or as correcting. 11. We may note in passing that in Sara’s initial response to the oVer, there is a kylläprefaced utterance at line 5: ‘KYLLÄ one could seemingly consider (it)’. The design of the turn is somewhat contradictory, however. While the initial kyllä can be seen as a sign of alignment, the odd employment of the adverb näköjään ‘seemingly’ conveys a diVerent message, that of disinterestedness. As will be shown in the analysis of the example, Sara seems, in the end, unwilling to go along and buy the honey for her family. 12. There is, to be sure, another particle että ‘so’; ‘so that’ placed at the very end. This particle typically occurs as a marker of paraphrase in turn-initial positions. When placed at the end, it leaves room for the co-participant to draw the consequences from the current turn. However, the turn could have ended after the predicate complement ‘terribly good honey’. This is a possible point of completion, after which the particle kyllä can be seen as a post-completion stance marker.

196 Auli Hakulinen

References Drew, P. and Holt, E. 1988 “Complainable matters: the use of idiomatic expressions in making complaints”. Social Problems 35(4): 398–417. Flint, A. 1980 Semantic Structure in the Finnish Lexicon: Verbs of Possibility and SuYciency. Helsinki: Finnish Literature Society. Hakulinen, A. 1998 “The use of Finnish nyt as a Discourse Particle”. In Discourse Markers: Descriptions and Theory, A. Jucker and Y. Ziv (eds.), 83–96. Amsterdam and Philadelphia: John Benjamins. Hakulinen, A. and Karttunen, L. 1973 “Missing persons: on generic sentences in Finnish”. In Papers from the Ninth Regional Meeting of CLS, 157–171. Chicago: Chicago Linguistic Society. Hakulinen, A. and Seppänen, E.-L. 1992 “Finnish kato: from verb to particle.” In Journal of Pragmatics 18, 527– 549. Laitinen L. 1995 “Nollapersoona”. [`The zero person’]. Virittäjä 99: 337–357. Lindström, A. 1999 Language as social action: Grammar, prosody and interaction in Swedish conversation. [Skrifter utgivna av institutionen för nordiska språk vid Uppsala universitet 46]. Uppsala: Institutionen för nordiska språk vid Uppsala universitet. Nykysuomen sanakirja 1951–1961 [Dictionary of Contemporary Finnish]. Porvoo; WSOY. Raevaara, L. 1993 Kysyminen toimintana. Kysymys — vastaus — vieruspareista arkikeskustelussa. [‘Questioning as an activity. On question-answer adjacency pairs in everyday conversation.’] Unpublished licentiate thesis. Department of Finnish Language, University of Helsinki. Raymond, G. 2001 The structure of responding: type-conforming and nonconforming responses to yes/no type interrogatives. Ph.D dissertation, Department of Sociology, UCLA. Sacks, H. 1992 Lectures on Conversation: Volume II, G. JeVerson (ed.). Cambridge: Basil Blackwell. SchegloV, E. 1996 “Turn organization: one intersection of grammar and interaction”. In Interaction and Grammar. [Studies in Interactional Sociolinguistics 13], E. Ochs, E. SchegloV and S. A. Thompson (eds.), 52–133. Cambridge: Cambridge University Press.


Sorjonen, M.-L. 1989 “Vuoronalkuiset konnektorit: mutta” [`Turn initial connectors: but’]. In Suomalaisen keskustelun keinoja I [`Characteristics of Finnish conversations I’]. [Kieli 4], A. Hakulinen (ed.), 162–176. Department of Finnish, University of Helsinki. 1997 Recipient Activities: Particles nii(n) and joo as responses in Finnish conversations. Ph. D. dissertation, Department of Applied Linguistics, UCLA. Tainio, L. 1996 “Kannanotoista arkikeskustelussa”. [Òn assessments in everyday conversations.’] In Suomalaisen keskustelun keinoja II [`Characteristics of Finnish conversation II’], [Kieli 10], A. Hakulinen (ed.), 81–108. Department of Finnish, University of Helsinki. Vilkuna, M. 1989 Free Word Order. Its syntax and discourse functions. Helsinki: Finnish Literature Society. Weydt, H. 1969 Abtönungspartikel. Die deutschen Modalpartikeln und ihre französischen Entsprechungen. Bad Homburg: Gehlen.

Appendix Key to the glossings and transcription symbols For the target lines, written in bold letters, both glossings and free translations are given. For the rest of the examples, there are only free translations. When relevant, suYx morphemes have been separated from the stems with a dash (). Unmarked forms: nominative case, singular, active voice, and present tense are not labelled. Abbreviations used in the glossings are as follows: Verb forms: 1st person singular SG2 2nd person singular SG3 3rd person singular ø zero person ‘one, you’ ÌNF inWnitive SG1

Case endings: ADE adessive ‘at, on’ ALL allative ‘to, for’

PL2 PAS CON CAUS

ILL INE

2nd person plural passive, PST past tense conditional (= subjunctive mood) causative suYx

illative ‘into’ inessive ‘in’

198 Auli Hakulinen

ELA GEN

elative ‘from, out of’ genitive

TRA

translative ‘to; becoming’

Other: CLIT PRT Q

clitic particle free standing particle interrogative particle

Transcription symbols: . falling Wnal intonation , level intonation ? rising intonation ↑ rise in pitch ↓ fall in pitch can emphasis indicated by underlining : lengthening of sound < word ending in arrowhead Wnished abruptly [ beginning of overlapping talk ] end of overlapping talk (.) micropause: 0.2 seconds or less (0.5) silences timed in tenths of seconds = two adjacent utterances ‘latched’ to each other without silence >< talk inside arrowheads spoken faster than the surrounding talk talk inside arrowheads spoken slower than the rest .hh inbreath (number of h’s reXects the length of the inbreath); .ettd inhaled sounds or words hh outbreath .mt .tj smack ## creaky voice °° soft or whispery voice @@ spoken with a changed tone of voice

Interactional linguistics and language development A conversation analytic perspective on emergent syntax Juliette Corrin, Clare Tarplee†, Bill Wells University College London

The transition to constructed multi-word speech oVers a developmental window onto emergent syntax. Child language research has typically investigated the linguistic aspects of transitional sequences of child talk. Particular attention has been paid to ‘successive utterance sequences’ which closely approximate Wrst ‘sentences’. This chapter introduces possibly the Wrst interactional account of such sequences in natural child-mother data. The central analytical concern is with turn organization: in eVect, how the child holds a turn to produce a successive sequence. Conversation analytic methods explicate the collaborative shaping of successive sequences by both child and mother. A range of non-vocal resources are found to facilitate the occasioning of the child’s sequence as a single turn, including the coordination of eye gaze and gesture, and play action. However, prosodic resources prove central to the analysis. SpeciWcally, it is pitch height that provides a consistent account of the data: pitch is held at, or narrowly contoured around, mid height at the end of non-Wnal utterances that are internal to the successive sequence. It is interactionally evident that this pitch device accomplishes turn continuation for the child. By comparison, Wnal utterances within a successive sequence show a prominent pitch excursion, most typically a rise-fall movement, which brings the sequence to a close. The mother’s subsequent receipt treats the child’s sequence as a complex proposition within a single turn. Parallels with Wndings in adult data are discussed, highlighting the insight brought by developmental studies to our understanding of turn-constructional units, and the interplay between syntax and prosody in shaping turns.

200 Juliette Corrin, Clare Tarplee, Bill Wells

Introduction Following the acquisition of a Wrst vocabulary, children enter a transition period during which constructed multi-word speech gradually emerges. Dore, Franklin, Miller and Ramer (1976: 15) memorably conceptualized it as a period during which there is “… on the one hand something more than one-word speech and on the other something less than syntax”. A traditional concern of linguists has been to characterize the expressive language behaviours which typify this transition period. These include, for example, children’s use of rotelearnt multi-word utterances (Peters 1983; Pine and Lieven 1993), their tendency to lengthen utterances by adding non-lexical material to conventional words (Bloom 1973; Dore et al. 1976) and patterns of reduplicating single words under one intonation contour. But perhaps the most notable for its syntactic-like quality is the child’s expression of a semantically structured proposition in a succession of hearably separate single utterances (Scollon 1979; Veneziano, Sinclair and Berthoud 1990). What one witnesses is the incremental emergence of something approximating a ‘Wrst sentence’: the child is, as it were, lapping at the edges of syntax. In fact, some believe that it is only articulatory Xuency that bars the way to full sentential status (Branigan 1979). The following extract from the seminal work of Bloom (1973) illustrates the phenomenon.

Traditionally, in the studies of developmental linguists, sequences such as these are foregrounded from their talk context and scrutinized for emerging semantic-syntactic merits. Thus, “Allison / baby doll / truck” might be accredited for

A CA perspective on emergent syntax 201

its thematic relations of ‘agent – theme – location’. The phenomenon is thus viewed as a new child language skill on display. The focus is a mentalistic one, in which the central concern is the child as a language learner. However, what has received little attention in prior research is the interactional basis of the child’s successive utterance sequences; in eVect, the most fundamental question is how they arise. It is this that we take as our central concern. In fact the very term ‘successive utterance sequence’ confesses to the one-sided perspective of past research, that is to say, the focus on what the child said. If such sequences are re-contextualized, it is conceivable that a very diVerent picture will emerge, and that the orderliness that has hitherto been identiWed by researchers as “successive” is in fact an interactional order in which both child and mother are implicated. One way of bringing into focus the interactional perspective on successive sequences is to take a developmental step backward to look at a type of sequence typical of the early transition period, namely the non-successive utterance sequence. In this case, the expression of semantically related thoughts is scaVolded by intervening talk from the mother. Consider the nonsuccessive utterance sequence below. The phonetic symbols used in this and subsequent extracts follow the IPA format.1

In this extract, the child talks about the unfolding action in a series of single turns: he makes a star shape in the play dough: “stars”, covers it over: “oh”, and sees the result: “gone”. Further, it is structurally obvious that the sequence is non-successive; the utterances are separated by intervening talk from the


mother. However, it is the placement of that intervening talk which is analytically incisive. Two points emerge which are necessarily basic to the story that follows. The Wrst is that each utterance is receipted by the mother. The second is that each is a local receipt of only the child’s immediate prior, not a summation of the child’s talk-so-far, thus each utterance is treated as a satisfactorily complete proposition. The non-successive nature of the sequence is therefore a collaborative outcome. What the participants have designed here is three organizationally independent exchange pairs: [“stars” → “stars” receipt] — [“oh” → “oh” receipt] — [“gone” → “gone” receipt]. Each of the child’s one word turns constitutes a single turn-constructional unit that ends in an operative transition relevance place which is accordingly taken up by the mother (Selting 2000). In addition to this structural observation, further evidence that the child designs such utterances as interactionally complete turns comes from instances where a lack of incoming receipt by the mother is treated by the child as accountable. Take extract (3) below, for example:

After a silence of 2.6 seconds, the child repeats his utterance more loudly, shifts focus from “monkey” to “gone” and adds [I] at the start (presumably in the role of determiner). This repair in response to the mother’s silence meets with prompt incoming by her. The child’s Wrst utterance [‘m˜wadig4n] is clearly being treated as transition implicative. In summary, then, our data suggests that children’s utterances within these non-successive sequences function as turn-transition points. Evidence comes from cases typical of our data in which the mother does receipt each utterance on completion, and cases in which the child holds the mother accountable if she does not. It is against this developmental background of non-successive sequences that we re-introduce the contrastive and later emerging pattern of successive utterance sequences which are the topic of this chapter. Re-consider extract (1) from Bloom’s data, in which Allison produced the sequence “Allison / baby


doll / truck” in three hearably separate utterances. Successive sequences of this sort constitute a logical problem as an emergent ‘object’ of interaction. Given the mother’s prior developmental expectation of her child as a participant who talks about play actions in single utterances (usually single words), and the associated interactional pattern of non-successive sequences in which each of those single utterances are receipted by the mother, how are interactional opportunities occasioned during which the child can reveal the newly emergent skill of expressing a complex proposition in a succession of single utterances, since it follows that the successive organization of such sequences is as much dependent upon the mother’s intervening silences as upon the child’s talking? As a further example, take extract (4) from our own data. Here, the child expresses a vaguely intelligible proposition about putting a farmer in a tractor in a succession of four hearably separate utterances, glossed as “farmer / on / X / here”.2

In contrast to the non-successive sequences as illustrated in extract (2), here the mother’s participation necessarily involves non-receipt during emergence of the four part sequence. How does the child hold the turn in such a way as to secure this continuous interactional space? And (as a necessary corollary of interaction) where does the mother place her eventual incoming receipt? This chapter presents initial Wndings from a doctoral study, using conversation analytic techniques in an attempt to characterize the sequential nature of children’s successive utterance sequences, their projected completion, and their subsequent treatment by the mother.


The data The data derive from video and audio-recorded play interaction between 3 child-mother pairs at home in the South East of England, collected longitudinally as the children underwent a normal transition towards constructed multiword speech. The context of play is of particular note in the extracts under analysis. The talk is largely about play; it is not the case, for instance, that the partners are explicitly working on talk, as might be the case in certain kinds of picture book interactions (see Tarplee 1993), and this makes the subtle interactional accomplishment of opportunities for linguistic advancement all the more intriguing. Further, within this particular type of social activity, the context of play is one which assumes intra-interactional relevance (SchegloV 1992). That is to say, play is a primary factor shaping the interaction between child and mother; it provides a context in which talk frequently occurs, but need not. A further preliminary comment on the speciWc form of talk under consideration is also necessary. The general practice of early talk-in-play subsumes a range of interactional forms in which talk is diVerentially implicated. For example, at times talk is conversationally non-implicative: it might be a resource used by the child to self-narrate actions, or an instructional device for the mother whilst the child executes play actions. The point is that at these ends of the spectrum, there is no evident sequential implication for receipting turns at talk from the co-participant. However, at other times the child and mother are clearly engaged in conversation, in the technical sense of the term (SchegloV 1998). And it is this particular context of talk which we will examine here, more speciWcally that form of conversation which addresses an unfolding play action.

The phenomenon of successive utterance sequences As described in the introduction, the point that ignites the interest of child language researchers is that successive sequences are potential home to the child’s ‘Wrst sentences’, bar the Xuent articulatory delivery under a single intonation contour. That is, the focus of these researchers is on the linguistic status of the emerging sentence. However, what is of potential interest to readers of this volume on interactional linguistics, and arguably more germane, is that such sequences are at the developmental cutting edge of grammar and turn organization: they are a developmental site at which emergent grammar and emergent turn organization are shaped in real time. In his discussion of the reXexive relationship between grammar and turns at talk, SchegloV


(1996: 55, our emphasis) refers to “the basic natural environment for sentences [being] in turns at talk, but the grammatical units which occur in turns are not all sentences”. This is never more so the case than at this stage of development. As we shall see, what is critical to our story about successive utterance sequences is not their degree of sentential mastery, but rather that, regardless of its sentential status, the sequence may become interactionally constituted as a single turn. Here, we choose the term ‘turn’ to refer neutrally to a stretch of child talk ending in an operative transition relevance place (TRP); the question of whether these stretches of talk that we recognize as ‘successive utterance sequences’ are to be interpreted as single or multiple turn-constructional units awaits the outcome of the analysis which is to follow. Take, for example, the contrasting extracts presented below. In extract (5), the semantic/syntactic cohesion of the two separate utterances “duck / it go(es) there” might be projected by the contextual implications of the mother’s prior question. Having asked explicitly “where does the duck go” earlier in the sequence, she repeats the question in the immediately prior turn as the child approaches the puzzle space: “goes over there”, with rising pitch on “there”. The Wrst utterance in the child’s subsequent sequence, “duck”, would not be a contextually complete response; this comes only with the second utterance “it goes there”.

However, in extract (6) the semantic/syntactic structure of the child’s successive sequence cannot be projected from the mother’s prior talk. The two separate utterances “I say hello / smoke/” are at odds with prior talk, being an amalgam of two diVerent topics, the conversation regarding telephones and


the play regarding funnels. The mother’s question would presumably predict the answer “I say goodbye” or minimally “goodbye”. Yet in both extracts (5) and (6) the successive sequences appear to come oV as single turns.

SchegloV (1996) points out that grammar, although elemental, is but one form of organization for the turn-constructional unit. If not grammar, then, what other forms of organization could function as resources for the participants in constituting successive sequences as single turns? Most fundamentally, how does the child secure the continuous interactional space prerequisite to the incremental expression of cohesive thought? In the exposition that follows, we formulate a possible answer to this question by considering the inXuence of both non-vocal and phonetic resources on the design of successive sequences.

Non-vocal resources We saw in the non-successive utterance sequence illustrated in extract (2) how hearable silences are routinely treated as turn transition relevance places, each being followed by an incoming receipt from the mother. From the child’s perspective then, such hearable silences would be points of vulnerability in the potential construction of a successive sequence. Given that the construction of talk itself is vulnerable within successive sequences, what other forms of com-


municative signal might the child use as a resource in structuring an incremental turn? One candidate for investigation is the co-ordination of gestural pointing and eye gaze (Goodwin and Goodwin 1992; Streeck 1992, 1993; Kendon 1980). In our data, these non-vocal signals are frequently, but variably co-ordinated in children’s play activities. As the subject of analysis here, we have chosen the most promising scenario in which eye gaze and gestural pointing are synchronized throughout the emergence of the sequence, thus synergistically oVering turnholding potential. Take extract (7) as an example. Mother and child are completing a puzzle. The topic of talk concerns the placement of the ‘spider’ piece. In prior talk, the space has been identiWed and the piece named by the mother. Here, she explicitly shows the child the spider piece. Whilst pointing to the spider, he produces a two-utterance sequence, the gloss of which is uncertain: it could be “it’s a / there”, or that [7˜ça‘˜] is a naming attempt at spider, thus “spider there” (this is how the mother treated it). Whatever the case, the focal point of analysis here is the accompanying deployment of eye gaze and gestural pointing. The transcript format is based on work by Streeck (1993), and highlights the temporal relationship between the non-vocal and vocal aspects of the successive sequence.3 For this particular analytical purpose, the relevant transcript lines (marked with arrows) present a horizontal, rather than a vertical, display of child and mother talk. The relevance of the trajectory of the child’s eye gaze and hand gestures are thus shown not only in temporal relation to the child’s own talk, but also in relation to the following silence and receipt by the mother.

-


The physical trajectory of the point gesture is synchronized with the emergent successive sequence so that the child’s hand reaches its target point position as the Wnal utterance in sequence is spoken “there”. After delivery of the sequence, the child’s point is held over the course of the following silence and receipt from the mother, subsequently decaying. It is also noticeable that the child’s eye gaze is constantly focused on the location of the point. The continuity of this co-ordinated gesture and eye gaze pattern across the intervening silence of 0.7 seconds invites the view that these non-vocal signals act as turnholding devices for the child, providing a link between the hearably separate utterances within the sequence: the unfolding gesture projects more talk to follow. However, our data cautions against this inviting hypothesis as a general account of turn-holding in successive utterance sequences. We have also to account for extracts such as (8) below, which tell a contrasting story. The topic of conversation concerns the space within a puzzle where the clown piece should Wt. In an earlier ‘clown’ sequence, the mother explicitly models this new word for the child, but without success. In this sequence, the child notices the empty clown space and whilst looking and pointing at it, he produces two utterances. The Wrst is “there”. The second is probably “a ball” — a reference to the ‘pom-pom ball’ on the clown’s hat, that is the part of the clown he points to. As in extract (7) the child’s eye gaze remains Wxed on the relevant puzzle


piece; however, in this case, the trajectory of the gestural point is rapid and the child’s hand reaches its target point position by the end of the Wrst utterance in sequence, and certainly before the intervening silence. Streeck (1993) suggests that metacommunicative gestures, such as pointing, visualize the illocutionary force of the utterance with which they co-occur. In this conceptual context, it is clear that the child’s co-ordinated eye gaze and gestural point have delivered their illocutionary force within the Wrst utterance of the sequence; no more talk is necessarily projected by their continued state of being ‘held’ during the intervening silence and delivery of the Wnal utterance in sequence. In fact, their continued state during the intervening silence might conversely be construed as transition implicative, awaiting receipt from the mother, rather than turn-holding. What our analysis does suggest, however, is that such patterns may accomplish retrospective turn cohesion: once the child’s successive sequence is complete, the hearably separate utterances within it may be retrieved by the mother as one cohesive proposition on the basis that they were ‘packaged’ together with one gesture. In summary, our data suggests that co-ordinated patterns of eye gaze and gestural pointing are variably synchronized with the temporal unfolding of successive utterance sequences and do not consistently function as turn-holding devices. Broadening out the enquiry, what other non-vocal resources could be at work? Some clues are provided by a more detailed analysis of extract (4), which concerned a proposition about a farmer placed in a tractor.


Apart from the talk, the strikingly continuous property of the sequence is the trajectory of play. One resource available to the child, then, is conceivably the alignment of play and talk action in such a manner as to have embryonic ‘sentences’ scaVolded by the continuity of a progressively unfolding play action. In extract (4) above, for example, the intention of putting the farmer in the tractor is clear from the start, but the execution of the play action takes time. Primarily, then, the child appears to be constructing a turn at play, which houses emergent talk. The concept is a double-sided one: whilst being the business of interaction, play is also witnessed to be a device within that interaction. And this orientation by the child to synchronize the trajectory of play and talk is our analytical device: play action becomes a demonstrable form of turn organization that can provide co-occurring evidence of the continuity of successive sequences of talk.4 This interpretation is supported by the mother’s subsequent treatment of the sequence. It is notable that she does not respond immediately; an approximate 2.8 second silence follows prior to her incoming with “is he going to be the farmer today”. The timing of her receipt is thus oriented to a transition point in the play action. Whilst the trajectory of a singular play action can often be inferred from its apparent practical intent, e.g. Wtting two pieces of a toy together, there are other cases in which the trajectory arises from the inherent sequential organization of the play activity. As illustration, take extract (9) below:

Each cycle of play action in completing the puzzle involves a two-part sequence: selecting a piece and then placing it. It is plausible, therefore, to suggest that the pending trajectory of the second part of the sequence (noted by the arrow symbol) could function as a turn continuation device — essentially a play turn continuation device that simultaneously provides the opportunity to deliver not only the second part at play, i.e. placing the puzzle piece, but the second part of talk, the utterance “there”, thus bringing oV the successive sequence: “dog a go / there” as a single turn.

A CA perspective on emergent syntax

Although a case for the sequential continuity provided by play actions may be built around extracts such as (8 ) and (9) above, it cannot provide a general account of how children hold a turn during the incremental delivery of successive sequences; clearly, not all play actions have a singular intent that is progressively executed, or have inherent sequential organization. In extract (10) below, for example, the child’s action of picking up a toy is accompanied by a sequence of two utterances. The Wrst utterance is either “more” or “four” — both are supported by contextual gloss. The second utterance is “a/the soldier”. The accompanying action itself is immediate: there is no trajectory — no part of the action still forthcoming. The topic of soldiers is well established and it could be argued that the action of picking up another soldier and saying either “four” or “more” is a complete proposition open to receipt by the mother. Yet the child successfully holds his turn to deliver a second utterance “a/the soldier”. The mother’s incoming after the second utterance treats the sequence as complete, acknowledging both of the child’s prior utterances. Her candidate hearing is “more soldiers” and she requests conWrmation of this by deploying a rising pitch pattern. For neither of the co-participants, then, was the end of the Wrst utterance an operative TRP (Selting 2000).

If, as illustrated by this extract, play action does not consistently function as a turn-holding device, one is led to question what other resources might conceivably be critical to the continuity of successive sequences? We started this section by looking at the talk itself — the ‘grammar’ of sequences — and made the point that this form of organization is obviously free to vary since it is the developmental outcome for the child, and therefore of our enquiry. But if not the ‘grammar’ per se, could there be another aspect of the talk which has a determining inXuence over sequence continuity?

211


Phonetic resources A variety of studies of ordinary adult-to-adult conversation have shown that prosody can mark speaker continuation. For instance, Local (1992) demonstrates how in British English a speaker may splice together the beginning of a new turn with the end of a foregoing interrupted or discontinued turn by aligning pitch level, loudness or tempo. Taking the evidence into the crosslinguistic domain, Selting (1995, reported in Auer 1996; cf. also Selting 1996) reveals how, in German, mid-level pitch height is deployed as a way of signalling non-completion across turn-constructional unit boundaries. In a speciWc study of the way in which speakers mark syntactically tagged-on material as integral with the utterance-thus-far, Auer (1996) shows pitch, tempo, loudness, rhythm and pause to be implicated as contour-delimiting devices in German. Thus the literature robustly shows prosodic features to be a set of devices which can accomplish the interactional task of marking continuation between a speaker’s turn-constructional units, even across speaker turns. It is a not unreasonable hypothesis, then, that children’s successive utterance sequences are marked as non-complete during their emergence by some prosodic device. In fact, a speciWc lead in this direction was provided by Branigan (1979), although no follow-up is evident in the literature. Branigan’s study compared the prosodic contours of three utterance types: single word, successive single word and multi-word. Data comprised recordings of three children evidencing a transition to syntax. Formal measures were made of pitch and duration features. In both respects, successive single-word utterance sequences were found to share similar markings to conventional multi-word utterances, whilst both were distinct from single word utterances. Of particular interest were the pitch Wndings: whilst the fundamental frequency contour of single words typically showed a terminal fall, this was not the case for non-Wnal single words within a successive single word sequence. In fact, non-Wnal single words within a successive sequence were strikingly similar to non-Wnal words within multi-word utterances in that neither showed a pattern of terminal fall, this only being apparent on the Wnal utterance of the sequence. The combined adult and child literature thus point to a wide range of phonetic resources deployed in the service of turn continuation. Within the scope of this chapter, we focus primarily on the pitch features of our successive utterance sequence data. Pitch is presented in an impressionistic notation. This reXects a consensus derived from independent perceptual transcriptions prepared by each author, combined with evidence from F0 traces obtained using PRAAT speech analysis software.


As a starting point, we re-present extract (2) — a non-successive sequence — with added prosodic transcription and invite a comparison of pitch features with the following extract (11) — a successive utterance sequence.


Extract (2) shows a clear pattern of rise-fall pitch movement on each of the child’s utterances. As previously noted, each utterance is evidently designed as complete by the child and treated as such by the mother, who receipts it accordingly. In eVect, then, each turn by the child comprises a single turnconstruction unit that ends in an operative TRP (Selting 1998). This contrasts with extract (11). The Wrst utterance glossed as “it’s a” diVers from those typical of non-successive sequence utterances in two related respects. Firstly, it is hearable as being spoken at mid height in the child’s pitch range, and secondly it is held level. A silence of 1.2 seconds follows prior to the next and Wnal utterance, glossed as “smoke”, which has a Wnal falling pitch movement that extends towards the child’s baseline. Given similar Wndings in both the adult and child literature cited earlier, there are grounds for claiming that mid, and level, pitch on “it’s a” functioned as a turn-continuation device for the child. Whilst the prosodic features exempliWed in extract (11) are salient, it could be counter-argued that turn continuity was in fact projected by the semantic/ syntactic design of the sequence. The mother asked the question “what comes out of the funnel” and the child replied “it’s a / smoke/”. The Wrst utterance “it’s a” would not have been taken as complete by the mother on linguistic grounds alone; clearly, there was more to follow and “smoke” was the awaited semantic focus of the reply. However, this kind of successive utterance sequence is scarce in our data, compared to cases where each of the two utterances is semantically potentially complete. This is hardly surprising: semantic/ syntactic aspects of turn construction are emergent at this age. They are in a


state of developmental transition, and so cannot provide a reliable basis for the production of complete turns at talk by the child and their recognition by interlocutors. As illustration, take extract (12).

In this case, the mother takes the semantic focus of the child’s response to be the Wrst utterance in the sequence — his rendition of ‘fence’ — so conceivably this could have been a point of recognitional incoming for her (JeVerson 1983; Wells and Macfarlane 1998), particularly as the child picks up the piece of lego fence at this point. There is no semantic/syntactic projection of further talk to follow. Why, then, does she withhold receipt through the following 1.2 second silence and delivery of the second and Wnal utterance? Here again, pitch marking provides evidence of functioning as a turn-continuation device, but with an added degree of subtlety. As with extract (11), we note the feature of mid pitch height on the Wrst utterance glossed as ” (det) fence”. Unlike extract (11), pitch is not held on a level — there is a rise-fall movement, but nevertheless the movement is delivered within a notably narrow range of pitch excursion that returns to mid pitch height. In summary, semantic-syntactic features of utterance construction do not oVer the mother a reliable basis for recognizing that the child’s turn is complete; these are the very features that are being formed within, rather than forming, the turn. By comparison, prosodic turn projection by pitch marking appears a consistent feature. As illustrated in extracts (11) and (12), the pitch movement on the non-Wnal utterance is either held at, or narrowly contoured around, mid pitch height. In both cases, a pitch fall at the end of the Wnal utterance is treated as sequence-closing by the mother.


Another perspective concerns the conversational ‘structure’ of such sequences. In the case of extracts (11) and (12), both sequences function as ‘question – answer – receipt’ exchanges. The design of the sequence is therefore structurally implicative of an answer from the child, and of a following receipt from the mother: the co-participants are accountably engaged in ‘conversation’. A more rigorous test of pitch as a continuity device concerns sequences initiated by the child, where talk is not anticipated, or not anticipated as a conversational contribution requiring receipt. Earlier, we noted how talk-ininteraction at this stage of child development subsumes a range of interactional forms in which talk is diVerentially implicated. In particular, talk is often a child’s resource for self-narration of play actions with no evident sequential implication for receipting turns. How, then, does the child frame a self-initiated narrative of play action in such a way as to have it received as a conversational contribution? It follows that, in such sequences, it is not only the intersubjective accomplishment of receipt placement at the designed point of turn completion that is impressive, but the very fact of receipt, i.e. that the mother orientates to the sequence as a conversational contribution at all. As illustration, take extract ( 13) below:

In this extract, the mother is helping her child to complete a puzzle. She searches for the ‘dog’ piece and gives it to the child. All that is then required is for the child to Wll out a constituent of play — to place the piece in the puzzle. The child’s two-utterance sequence “Elsa / go there /” is therefore not anticipated nor required by prior context. As it happens, the child refers to the ‘dog’ piece as “Elsa” after the name of their own dog, and thus this Wrst utterance


could well have been treated by the mother as a straight picture label — naming the dog as he placed that piece in the puzzle. Given this vulnerability of contextual mis-interpretation and of the following 0.7 second silence, how did the child hold his turn to produce the second and Wnal utterance “go there“? Once again, pitch marking provides a plausible answer. Whilst the utterance glossed as “Elsa” shows a rise on the Wrst syllable, it levels at mid pitch height on the second syllable, which has the added quality of being prematurely terminated by glottal closure. The ‘midness’ of pitch height gains further relative value in contrast to the particularly low pitch incoming of the Wrst two syllables of the Wnal utterance, glossed as “go a”. The Wnal word of the Wnal utterance, “there”, shows a nuclear pitch movement that falls toward the child’s baseline, bringing the successive sequence to a close. It is our view that mid pitch contours have a dual role in sequences of this sort: Wrstly, against the background of ‘free-Xoating’ play action they project continuation of talk not simply as play narrative, but contextualize it as a relevant conversational contribution; secondly, in combination with the Wnal fall of the Wnal utterance, they mark the sequence as a complete and cohesive conversational contribution. In other words, in addition to the independent role as continuity marker on non-Wnal utterances that we have already noted, mid pitch height has a dependent role in being internally related to the terminal pitch movement of the Wnal utterance. In this extract, it is the utteranceWnal pitch sequence of mid height marking followed by a fall towards the baseline that signals the turn to be complete and internally cohesive (see Selting 1996 for a parallel in adult data). The mother’s subsequent incoming certainly provides evidence that she treats it as such — as both complete and cohesive. Her receipt “OK put it there then” incorporates an acknowledgement of both utterances in the child’s successive sequence: her pronoun “it” refers to the puzzle piece named “Elsa” by the child, and her phrase “put it in there then” embellishes the “go there“; she thus acknowledges the semanticsyntactic cohesion behind the incremental emergence of the sequence. The extracts examined thus far have illustrated successive sequences in which the co-occurring cues which delimit turn construction are aligned: the prosodic pitch markings are synchronized with the emergent semantic-syntactic unit under construction. In a more rigorous test of the primacy of pitch marking, we now investigate a scenario in which pitch marking and ‘sentence’ marking are non-aligned, that is to say, where the semantic-syntactic construction is potentially complete, but the prosody is not. This phenomenon is widely acknowledged in studies of adult-to-adult interaction (Selting 1998; Auer 1996). It is taken as basic to conversation analytic enquiry that, whilst complete


sentences may occupy turns-at-talk, they by no means deWne them (SchegloV 1996; Ford and Thompson 1996). For example, Wells and Macfarlane (1998) found syntactic completion to be a poor guide for recipients in locating the Wnal, major pitch accent which projects a TRP; rather, recipients oriented to the phonetic characteristics of the pitch accent itself. It was the deployment of prosodic resources that set the boundary for turns at talk by projecting an upcoming TRP, not the utterance syntax. Further still, even an interactionally deWned concept of syntactic completion has been found to be inadequate in projecting turns at talk. Ford and Thompson (1996: 143) set their deWnition of syntactic completion relative to discourse context as “a potential terminal boundary for a recoverable ‘clause-so-far’”. However, even with this degree of deWnitional adaptation, syntactic completion was not a reliable boundary indicator: whilst intonation and pragmatic completion points were nearly always found to be synchronous with syntactic completion, the reverse was not true; i.e., intonation and pragmatics had primacy in determining the turn boundary. Given these Wndings in the adult data, we turn now to investigate this relationship between prosodic and semantic-syntactic turn boundary indicators in successive child utterance sequences. How does the child accomplish the incremental expression of a potentially complete ‘sentence’ and then add yet further syntactic material without being overlapped or interrupted by the mother? As illustration, take extract (14) below: :


In this three-part successive sequence there are two interactionally vulnerable sites at which our hypothesis regarding the work being accomplished by mid pitch cues can be tested, each marked with an arrow. The Wrst two utterances are linked to the child pointing to a picture of a plane: the semantic force is something akin to “this a / plane/”. Whatever the case, the pitch movement at the end of the Wrst utterance [6˜ j6 j6‘˜] is clearly at mid height and held on the level, linking it to the second utterance [pe˜Iñ]. But at this point the sequence presents us with an analytical twist. On the one hand, there are numerous features which suggest this to be a possible point of turn completion from the child: the play action of pointing is certainly complete, the picture has been named, the primitive sentence akin to “this a / plane/” would be hearable as contextually complete, and the prosodic pattern shows a complex pitch movement with extra duration and loudness, suggestive of a nuclear tone — as if “plane” was the complete semantic focus.5 However, on closer inspection, it is apparent that the second and non-Wnal utterance [pe ˜Iñ] terminates around mid height, being extended on the level by the prolonged duration of the vowel. Thus, there is projected continuation despite a nuclear pitch movement. The turn is thus left open for the third and Wnal utterance which adds further syntactic material. It expresses the expanded proposition of planes being located in the sky with an accompanying point gesture. Auer (1996) explains that syntactic expansions are vulnerable to overlap and this makes the intersubjective accomplishment of continuity noteworthy. Mid pitch would thus appear to be a powerful enough continuity device to override semantic-syntactic boundaries: whilst a successive utterance sequence may be linguistically complete in its play context (albeit in a primitive form), delicate deployment of pitch resources can project and secure a local opening for further talk from the child, thus functioning as the prime TRP marker. The concurrence between Wndings in our child data and those in the adult data is noteworthy in two developmental respects: Wrstly, the primacy of prosodic TRP-marking appears to allow the child the developmental scope to negotiate the transition to syntax within independently assured units of interaction; secondly, it invites a theoretical view of grammar as being a system shaped by interactional forces from the point of developmental inception with the ‘sentence’ being viewed as a linguistic adaptation to its habitat of talk-ininteraction (SchegloV 1996).


Conclusion Successive child utterance sequences are held to be a signiWcant characteristic of expressive language during the transition to syntax. Whilst frequently identiWed as a developmental phenomenon in the traditional child language research literature, their interactional basis has been side-lined. The study presented in this chapter has redressed this imbalance by taking the view that such sequences are at the cutting edge not only of emerging grammar, but of emerging turn organization. It has attempted to characterize the real-time interactional contingencies that shape the mutual dependency between these two forms of organization by analyzing how successive child utterance sequences are occasioned. Analysis has focused on the child’s accomplishment of turn-continuity and cohesion, and the mother’s subsequent treatment of the successive sequence as a single turn. A variety of expressive resources have been found to be implicated. The temporal alignment of non-vocal features such as co-ordinated hand gesture and eye gaze, and play actions, may assist the ‘syntax’ enterprise in two ways: Wrstly, they may project turn continuation during the intervening silence(s) of a successive sequence; secondly, they may retrospectively contextualize the utterances within the sequence as a single, cohesive proposition. However, these non-vocal features cannot provide a general account of our data. Since they are not consistently deployed, they may facilitate the process, but cannot be critical to it. We have found the most consistent design characteristic of successive utterance sequences to be the deployment of particular prosodic features. Our central Wnding concerns pitch height. Non-Wnal utterances that are internal to successive sequences consistently end at mid pitch height. Further, the pitch marking is frequently held on a level, or if directional, then of a notably narrow range of excursion. Our analysis has shown that this marking accomplishes turn continuation for the child, securing an interactional space into which syntax can emerge. By contrast, Wnal utterances within a successive sequence show a prominent pitch excursion, typiWed in the particular extracts presented here by a rise and subsequent fall towards the child’s baseline. In fact, a variety of pitch sequences are found in our data: whilst mid pitch consistently marks non-Wnal utterances, the pitch movement of the Wnal utterance is free to vary in achieving a variety of pragmatic ends, at times being a rise-fall, a rise-fall-rise and so forth; what is critical is that it is heard as not being a mid pitch contour, relative to the prior non-Wnal utterance. This


sequence-Wnal prominent pitch excursion is reliably treated by the mother as a completion marker, signalling a transition relevance place for her receipting turn. The combined interactional force of the sequence of pitch movements, that is a mid pitch contour on the non-Wnal utterance followed by a prominent pitch excursion on the Wnal utterance, accomplishes sequence cohesion. This is evidenced by the mother’s treatment of the sequence as a complex proposition within a single turn. Whilst pitch marking is primary within our data, we acknowledge that it is by no means exclusive; other phonetic devices are also at work. Glottal closure has been shown to accomplish turn continuation in adult-to-adult talk (Local and Kelly 1986) and similar Wndings emerge in our data. Other phonetic features such as duration and rhythm are no doubt implicated and would repay future investigation. Our Wndings concur with the child data of Branigan (1979), whose measurements identiWed mid-pitch height at the ends of non-Wnal utterances in successive sequences. We have built on this Wnding, showing how pitch is deployed as an interactional device by the child and how it is oriented to by the mother. Taking this interactional angle shifts the analytical perspective away from what the child said, to examine how a collaborative occasion arose in which the child could say it: the semantic-syntactic diVerentiation of successive child utterance sequences that aVords their sentence-like quality is thus reconceived as an endpoint, a Wnal step along an already travelled interactional pathway. Analysis suggests that prosodic devices (of which pitch is primary) consistently cluster together with a variable array of non-vocal devices to contextualize successive child utterance sequences as single turns, and that these turns provide a pre-existing interactional environment within which nascent syntax can advance. Our Wndings regarding pitch as a turn-continuation device also resonate well with previous studies of adult-to-adult talk (Local and Kelly 1992; Auer 1996; Müller 1996; Selting 1996). In that research context, the role of prosody in marking turn construction boundaries has challenged the importance conventionally assigned to syntax as determiner of turn completion. The perspective oVered by this child study adds further weight to that challenge by showing prosody to have developmental primacy in shaping interactional opportunities for successive child utterance sequences. We have shown how prosodic features delimit such sequences as single turns, i.e. with an operative TRP arising only at the end of the sequence. This leaves the semantic-syntactic construction of the turn free to emerge in real time, and free to evolve over development


time. In our data, the majority of sequences appear to be constituted by multiple turn-constructional units (TCUs). In these cases, each utterance in the sequence functions as a semantically retrievable and potentially complete proposition in context, i.e. as a single turn-constructional unit, but prosodic features override the possible TRP at the end of each utterance, thus contextualizing it as a non-Wnal TCU and projecting more talk to follow (Selting 1996, 2000). The Wnal utterance in the sequence then functions as a Wnal TCU with an operative TRP, bringing the turn to a close. Developmentally, the transition to syntax can be seen as one in which these multi-TCU successive sequences gradually evolve into single TCU sequences: over time, the semantic-syntactic construction of the sequence itself becomes suYciently elaborated to provide turn projection cues; prosody need now no longer be primary, but is seen to function in increasingly subtle partnership with the semantic-syntactic aspects of utterance design.

Notes † Dr Clare Tarplee died as the result of a cerebral haemorrhage on 16 November 1999, while this chapter was in its Wnal stages of preparation. She was 36. Clare gained a Wrst class B. A. degree in English Language at the University of Newcastleupon-Tyne, then moved to the University of York to undertake graduate study. In her doctoral research she used Conversation Analysis to investigate how young children’s linguistic skills are shaped collaboratively through interaction with adults. After three years as lecturer in sociolinguistics at Newcastle, in 1992 she moved to the National Hospital’s College of Speech Sciences (now the Department of Human Communication Science, University College London), where she lectured on children’s language development, phonetics and conversation analysis. Her research took a clinical direction, analyzing interactions involving children with various types of communication diYculty. She co-organized the Wrst international conference on Conversation Analysis and Communication Disorders, held at UCL in 1997, and from 1998 until her death she coordinated an ESRC-funded seminar programme on the same topic. Through her own research and through her inspiring teaching Clare fostered the application of Conversation Analysis in speech and language therapy practice. She was a much loved colleague and friend, and will be greatly missed. 1. The International Phonetic Alphabet (IPA 1999) notation is applied to give an impressionistic transcription, with particular attention to the micro-sequences under analysis. Extended notation used includes: { } span of speech to which the notation applies; < > swell; f loud; p soft. A gloss of the pertinent child utterance(s) is given on the righthand side. Where this is not recoverable with reasonable certainty the word appears in brackets ( ). The convention (X) indicates unintelligible speech.


2. Throughout the chapter extracts, the child’s successive utterances are transcribed on separate lines, with silences noted in between. This departs from conversation analytic convention, in which the utterances of one speaker that constitute their turn are written in consecutive text with incorporated silences. The motivation for the transcript format shown here reXects our central concern with the construction of the child’s turn. In the sense that transcription can embody analytical preconceptions (Hutchby and WoYtt 1998: 76), the transcription of the child’s utterances on separate lines is a theoretical default position that avoids the danger of assuming, a priori, that the child has in any sense ‘constructed’ a single multi-word utterance; this would be the developmental case to be proven. 3. The successive sequence of child utterances in this extract (7), and the following (8), are transcribed on one line in order to highlight the display of non-vocal notation and make the analysis more accessible to the reader. 4. In Wootton’s (1989: 247) explanation, the term ‘co-occurring evidence within the turn’ refers to other words within the turn that accompany the item under analysis. Here, by extension, we take the non-vocal actions of play to be an equivalent type of co-occurring evidence. 5. Nuclear tone refers to a system of choices for the types of pitch pattern that form the intonational nucleus (for further detail, see Couper-Kuhlen 1986).


Bloom, L. 1973

“On the prosody and syntax of turn-continuations”. In Prosody in Conversation, E. Couper-Kuhlen and M. Selting (eds.), 57–100. Cambridge: Cambridge University Press. One word at a time: The use of single word utterances before syntax. The Hague: Mouton.

Branigan, G. 1979 “Some reasons why successive single word utterances are not”. Journal of Child Language 6: 411–421. Couper-Kuhlen, E. 1986 An Introduction to English Prosody. London: Edward Arnold. Couper-Kuhlen, E. and Selting, M. 1996 “Towards an interactional perspective on prosody and a prosodic perspective on interaction”. In Prosody in Conversation, E. Couper-Kuhlen and M. Selting (eds.), 11–56. Cambridge: Cambridge University Press. Dore, J., Franklin, M., Miller, R. and Ramer, A. 1976 “Transitional phenomena in early language acquisition”. Journal of Child Language 3: 13–28.


Ford, C. and Thompson, S. A. 1996 “Interactional units in conversation: syntactic, intonational, and pragmatic resources for the management of turns”. In Interaction and Grammar, E. Ochs, E. A. SchegloV and S. A. Thompson (eds.), 134–184. Cambridge: Cambridge University Press. Goodwin, C. and Goodwin, M. H. 1992 “Context, activity and participation”. In The Contextualisation of Language, P. Auer and A. di Luzio (eds.), 77–100. Amsterdam and Philadelphia: Benjamins. Hutchby, I. and WooYtt, R. 1998 Conversation Analysis. Cambridge: Polity. JeVerson, G. 1983 “Two explorations of the organisation of overlapping talk in conversation, 1: Notes on some orderliness of overlap onset”. Tilburg Papers in Language and Literature 28. Tilburg University, Tilburg, Netherlands. Kendon, A. 1980 “Gesticulation and speech: Two aspects of the process of utterance. In The Relationship of Verbal and Nonverbal Communication, M. R. Kay (ed.), 207–228. The Hague: Mouton. Local, J. and Kelly, J. 1986 “Projection and ‘silences’: Notes on phonetic and conversational structure”. Human Studies 9: 185–204. Local, J. 1992 “Continuing and restarting”. In The Contextualisation of Language, P. Auer and A. di Luzio (eds.), 272–296. Amsterdam and Philadelphia: Benjamins. Müller, F. E. 1996 “AYliating and disaYliating with continuers: Prosodic aspects of recipiency”. In Prosody in Conversation, E. Couper-Kuhlen and M. Selting (eds.), 131–176. Cambridge: Cambridge University Press. Peters, A. 1983 Units of Language Acquisition. Cambridge: Cambridge University Press. Pine, J. and Lieven, E. 1993 “Re-analysing rote-learned phrases: Individual diVerences in the transition to multi-word speech”. Journal of Child Language 20: 551–571. SchegloV, E. A. 1992 “In another context”. In Rethinking Context: Language as an interactive phenomenon, A. Duranti and C. Goodwin (eds.), 191–228. Cambridge: Cambridge University Press. 1996 “Turn organisation: one intersection of grammar and interaction”. In Interaction and Grammar, E. Ochs, E. A. SchegloV and S. A. Thompson (eds.), 52–133. Cambridge: Cambridge University Press. 1998 “ReXections on studying prosody in talk-in-interaction”. Language and Speech 41: 235–263.


Scollon, R. 1979

Selting, M. 1995 1996

2000 Streeck, J. 1992

1993 Tarplee, C. 1993

“A real unzippered condensation of a dissertation on child language”. In Developmental Pragmatics, E. Ochs and B. SchieVelin (eds.), 215–227. London: Academic Press. Prosodie im Gespräch. Aspekte einer interaktionalen Phonologie der Konversation. Tübingen: Niemeyer. “On the interplay of syntax and prosody in the constitution of turnconstructional units and turns in conversation”. Pragmatics 6 (3): 357– 388. “The construction of units in conversational talk”. Language in Society 29, 477–517. “Previews: Gestures at the transition place”. In The Contextualisation of Language, P. Auer and A. di Luzio (eds.), 135–157. Amsterdam and Philadelphia: Benjamins. “Gesture as communication 1: its coordination with gaze and speech”. Communication Monographs 60: 275–299.

Working on talk: The collaborative shaping of linguistic skills within childadult interaction. D. Phil thesis, Department of Language and Linguistic Science, University of York. Veneziano, E., Sinclair, H. and Berthoud, I. 1990 “From one word to two words: Repetition patterns of the way to structured speech”. Journal of Child Language 17: 633–650. Wells, B. and Macfarlane, S. 1998 “Prosody as an interactional resource: Turn-projection and overlap”. Language and Speech 41: 265–294. Wootton, A. J. 1989 “Remarks of the methodology of conversation analysis”. In Conversation: An interdisciplinary perspective, D. Roger and P. Bull (eds.), 238– 258. Clevedon: Multilingual Matters.

Part II Interactional order and linguistic practice

Fragments of units as deviant cases of unit production in conversational talk* Margret Selting University of Potsdam

This analysis aims to further investigate the ‘turn-constructional unit’ (TCU). I use the conversation analytic technique of deviant case analysis for the investigation of ‘fragments of units’ as a way to reconstruct our interpretation of ‘units’ in talk in conversational interaction. I investigate more closely the diVerent roles and functions of prosody and syntax as well as the role of semantically and/or pragmatically built up projections in the construction of fragments in their semantic, pragmatic and sequential context. The (interpretation of the) fragmentariness of stretches of talk is reconstructed as the result of non-fulWlled context-sensitive projections. If this is true for fragments of units, it also corroborates prior analyses that units cannot be deWned with respect to either syntax only or prosody only but are the result of the interaction of both in their sequential context. The interpretation of units, then, is the result of fulWlled context-sensitive projections. Both fragments of units as well as units are context-sensitive inferences. They are, however, not oriented to as such by participants: they are only the epiphenomena of practices which organize talk in turns.

1.

Introduction

In previous papers I have analysed the interplay of syntax and prosody in the production and interpretation of turn-constructional units (TCUs) in their lexico-syntactic, semantic, pragmatic and sequential context (Selting 1996a, 1998, 2000). I described the role of syntax as a projection device with scope over the current TCU till the end of a possible sentence or other possible syntactic construction in the given context, and the role of prosody as a projection device with the ability to locally project continuation beyond the current TCU, with further syntactic, lexico-semantic, pragmatic and activitytype speciWc projections projecting larger turns. This analysis was used to clarify the notions of the turn-constructional unit (TCU) and of transition

230 Margret Selting

relevance places (TRPs) at the possible completion points of possible turns. TCUs were deWned, largely in agreement with Sacks, SchegloV and JeVerson (1974), as the smallest linguistically possible complete units in their sequential context, with TRPs being blocked or suspended at the ends of non-Wnal TCUs till the projected (Wrst) possible completion point(s) of possible Wnal TCUs in the turn. If, however, we want to further ask what the resources are in detail that participants use in order to construct their units in talk, we encounter diYculties. As Sacks, SchegloV and JeVerson (1974) noted, in principle any item can be used in, say, a phrasal or a single-word unit, signalled via intonation. For this reason, it will be diYcult to warrant our analysis of the ingredients in a possible TCU positively. Therefore, in this paper I will use deviant cases in the production of units, i.e. fragments of units, in order to (a) further investigate the kinds of devices and knowledge that we use to construct and make both fragments of units as well as units in talk interpretable, and to (b) further support and validate my prior analyses of ‘units’ in talk. Fragments of units — stretches of talk that do not constitute complete units but are left unWnished — occur quite frequently in talk. How are such unWnished fragments recognizable as unWnished? How can participants distinguish units from fragments of units? And what, in consequence, makes Wnished, complete units recognizable as such?

2. The analysis of ‘fragments of units’ as deviant cases as a way to reconstruct our interpretation of ‘units’ Fragments of TCUs are those stretches of talk that are treated as unWnished by participants. As any unit-in-progress can be continued, even after a pause of considerable length, by producing a morpho-syntactic and prosodic continuation, fragments of units only become interpretable as such retrospectively, i.e. when the speaker cuts them oV and continues with a new beginning or when an unWnished construction is abandoned if, e.g., the Xoor is relinquished. In general, the interpretation of stretches of talk as ‘unWnished fragments’ seems to be the result of opened up and yet unfulWlled projections of various kinds. The interpretation of a stretch of speech as a ‘fragment of a unit’ entails that it has not reached a point of possible TCU completion, regardless of whether this TCU is a turn-internal or a possible turn-Wnal one (cf. Selting 1998, 2000). I want to ask what kinds of projections and knowledge are involved in the

Fragments of units as deviant cases 231

interpretation of a stretch of talk as ‘unWnished’. Following received analytical practice in CA and applying it to the question at hand, the analysis of deviant cases can be used in order to investigate participants’ underlying expectations in the production of complete TCUs (for deviant case analysis, see Sacks, SchegloV and JeVerson 1974, Levinson 1983, Wootton 1989). The analysis of deviant cases in TCU production can lead us to infer the kinds of knowledge that are involved in the interpretation of a stretch of talk as a TCU. The result of this investigation is the deconstruction and reconstruction of the signalling systems and practices that are routinely used in order to make fragments interpretable as ‘fragments’ and that make units interpretable as ‘units’. There are at least three distinct classes of fragments of units in which the interpretation of the stretch of talk as fragmentary rests on diVerent criteria: a. cut-oV of a projected unit plus new start and beginning of a new unit, b. non-completion of a syntactically and prosodically projected unit, c. non-completion of a semantically and/or pragmatically projected unit. These classes have been ordered according to the place in which the fragmentary character of the unit becomes obvious for the recipient: cut-oVs do something to the projected talk-in-progress, new starts and beginnings of new units follow immediately. Both the non-completion of syntactically and prosodically as well as lexico-semantically, pragmatically and/or activity-type speciWcally projected units only become recognizable in retrospect, i.e. after a new unit has been begun. I will give examples of each of these classes in the following sections. For the purposes of the present paper, it is in most cases suYcient to diVerentiate between syntactic, semantic and pragmatic projection within a given sequential context (see Selting 1998, 2000, for more detail). I will Wrst look at examples of the most frequent cases in which syntax and prosody in co-occurrence are involved in the production of fragments. Following that, I will show, however, that fragments of units cannot be distinguished from complete units with reference to either syntax or prosody alone, nor without reference to their semantic, pragmatic and sequential relation to their contexts. I will Wrst deal with ‘clear cases’, then present examples of less clear cases. The data are taken from several informal conversations between three participants each. The auditive description was veriWed by instrumental phonetic analysis (for more detail, see Selting 2000).

232 Margret Selting

The analysis of fragments of units yields insights into speech production and interpretation processes and resources. My analysis of fragments of TCUs will (1) corroborate my earlier analyses of TCUs in conversation and (2) make it clear that neither fragments of TCUs nor TCUs are the categories that participants orient to as such. The production of TCUs is only an epiphenomenon of, and contingent on, the constitution of activities in conversation; fragments of units are only deviant cases of unit production. As participants in conversation orient to interactionally relevant conversational activities, not to single TCUs and not to single linguistic construction devices, we cannot expect to be able to always make use of recipient responses as warrants of single TCUs or even their single production devices. This methodological problem must be overcome by the systematic reconstruction of the signalling resources that are used by participants to manage the production of fragments and units in talk. As JeVerson (1974) has shown, the production of fragments can be used by participants as a resource for activity constitution, e.g. by recognizably projecting an activity without actually completing it, thus hinting at it and then leaving the actual production of the hinted-at activity open to interactional negotiation. My analysis looks below the interpretation of such possible ‘hints’ by reconstructing the resources and conditions that make fragments interpretable as fragments in the Wrst place.

3.

The roles of syntax and prosody in fragment and unit production

In the following, I will Wrst look at the most frequent cases in which syntactically and prosodically incomplete structures co-occur in fragments of units; after that I will try to separate the individual roles of syntax and prosody for the interpretation of fragments. 3.1 The co-occurrence of syntactic and prosodic incompletion in their sequential context In most cases, both syntactic and prosodic devices are involved in the production of fragments of units. The prosodic devices, however, can be quite diVerent, as the sets of fragments in a. and b. show:


a. Cut-oV of a projected unit plus new start and beginning of a new unit Instances in which a projected unit in progress is cut oV and the speaker starts a new unit are exempliWed in the following extracts (English translations are given separately after the Wgures; the transcription conventions are explained in the appendix): (1) K4: 530–531 ((Sie reden über Vorbereitungszeit für Lehrveranstaltungen))

((Talk about the time the participants need for the preparation of seminars.)) Eli: Lea:

Eli: Lea: Eli:

although it’s stopped being like that now well for me it was sometimes alr’= =well for me it’s worse this semester than it was the last one= =yeah yeah no I’m improving

In (1), the unit in progress, aso ICH hab das manchma scho’ (‘well for me it was sometimes alr’), is cut oV with a glottal closure in the middle of the word schon, which itself is not a possible completion of this unit, and speaker Lea immedi-

234 Margret Selting

ately starts a new unit. Besides the cut-oV and the latching of the new unit, she produces a downstep in pitch for the beginning of the new unit and produces the Wrst word of the new unit, aso, with fast tempo. Syntactically, too, a new construction is recognizably begun; the beginning of this new construction cannot be heard as a continuation of the old one. So, here, we Wnd syntax and prosody constituting the break in talk that makes the fragment interpretable as a fragment and the start of a new unit as the start of a new unit. Retrospectively, it becomes evident that Lea is recycling the beginning of her Wrst unit; she produces the items aso ICH hab again and then chooses a new formulation. (2) K1: 5–6


Nat:

Ida: Nat: Ida: Nat: Ida:

no I was also so the er ‘m I don’t know the first three four meetings I was there (.) [ because it got on my nerves afterwards [ were you there when I presented my paper no then I was [ t ‘I ] think that was [ hm ] somehow the next [ meet ] ing (..) [ hm ] did you stay there longer

In (2), we Wnd several fragments of TCUs. The speaker starts with ich WAR auch (‘I was also’), but then adds aso: with sound stretch and level pitch, thus holding the turn and projecting ‘more to come’. Aso (=also, ‘so’) with this prosody cannot be heard as the continuation of the prior syntactic construction. The prior syntactic construction itself, ich WAR auch, could well be a possibly complete syntactic construction in other sequential environments, e.g. after a possible prior utterance such as ‘und dann erzählte sie du wärst da gewesen’ (‘and then she told us that you had been there’ — ‘I was’). In the sequential context given here, however, it is not a possibly complete syntactic construction. After this, Nat creates a break in talk by producing the particle aso:, which is very often used as a signal to project reformulation (Gülich/ Kotschi 1987). But this, too, is left as a fragment, and Lea delivers another beginning of a new unit die er with a downstep, but cuts this oV again with a glottal stop and a nasal m. Then, she produces the parenthetical phrase weiß

236 Margret Selting

nich (‘I don’t know’) with very low pitch and low voice. Finally, the beginning of the unit that is ultimately Wnished, die ersten drei vier SITzungen war ich da: (‘the Wrst three four meetings I was there’), is set oV with an upstep in pitch and a return to normal loudness, thus making this beginning of a new unit recognizable as such. Another cut-oV is produced in line 06 further down in the transcript, when Nat starts with NEE: da war ich, then produces the sound t and immediately after that a clearly audible glottal stop — more prominent than the normal glottal stop expectable here for phonological reasons — which produces something like a prosodic break: a ‘hard’ start of the ich as the signalling of the start of a new unit. Furthermore, the item ich realizes a downstep in pitch and is syntactically incompatible with the unit that had been begun before, i.e. it must also be interpreted on syntactic grounds as the beginning of a new unit by the recipients. (3) K3: 27–28

Nat:

no there is also well I had also at first no relationship to my father=and then it was built up bit by bit


Another fragmentary unit can be seen in (3). After a Wrst unit NEE (‘no’), speaker Nat begins a second one with da IS au, which is produced with fast tempo. But after this, she produces a downstep in pitch with the reformulation signal also and thereby creates a prosodic break and begins a new unit. Here, thus, a downstep in pitch, the change in tempo from faster to normal, and the start of a new syntactic construction with the reformulation signal also constitute the break and signal the beginning of a new unit, while the prior one is left unWnished. (4) K4: 733–736 ((über Studenten))

238 Margret Selting

((Talk about students)) Lea: Eli: Lea:

?:

because I always wondered why they’re so= =they always seem to me to be so depressed you know the students yes because I don’t believe you know they always all say they are lazy= =they won’t talk and [ so=and they don’t do anything [hhh hhh

In (4), we Wrst see an example of a fragmentary unit that is cut oV by quickly starting a new unit, i.e. by latching the beginning of a new unit onto the old one at just the point where it is abandoned: weil ich IMmer darüber NACHgedacht hab warum die so=die KOMmen mir immer so melanCHOlisch vor ne (‘because I always wondered why they’re so=they always seem to me to be so depressed you know’). Later, in line 05, there is another fragment: Lea’s stretch of talk weil ich GLAUbe nich (‘because I don’t believe’). In this context this construction needs to be complemented with a subordinate clause specifying what Lea does not believe, i.e. it semantically projects more-to-come. It also prosodically ends with a level pitch accent, i.e. it also prosodically projects continuation. But here, too, Lea begins a new unit immediately after ich GLAUbe nich by beginning a recognizably new syntactic construction WEIß die sagen immer alle die sind FAUL: (‘you know they always all say they are lazy’) and by changing to faster tempo for the new unit. In these examples (1) through (4), the following syntactic and prosodic devices are used in co-occurrence to constitute and contextualize the beginning of a new unit and thus to retrospectively make inferrable that the previously begun unit has been abandoned and left unWnished: syntactic cues: – unWnished unit-in-progress prior to the break and – recognizable start of a new syntactic construction, prosodic cues: – glottal closure or alveolar stop as a means to cut-oV a unit-in-progress and/ or – change of tempo and/or – upstep or downstep in pitch. In contrast to these examples with fairly clear and saliently displayed breaks to signal the beginning of a new unit when a prior one is abandoned, the next extract demonstrates a less clear case.


(5) K3: 103–104 ((Nat. über ihren Vater))

((Nat about her father)) Nat:

only when it happened that I needed his help ((inbreath)) no matter how then it worked somehow it didn’t work then then there was always some reason or other on his part why he couldn’t help me ((etc.))

In (5), line 04, Nat produces the possible beginning of a sentential unit dann GING das (‘then it worked’). Although in other contexts this could very well be a complete possible sentence and unit, in this case, where the preceding sequential context has projected a negative statement, this construction is heard as unWnished, i.e. still lacking a negation (cf. below). And even though the next word, I:Rgndwie (‘somehow’), might be heard as continuing the prior syntactic construction, it is produced with an upstep in pitch to constitute an — albeit not very salient — prosodic break. However, it is the continuation with the formulation of yet another verb form in GINGs dann nich (‘didn’t work’) that shows retrospectively that Nat has reorganized her possible sentence from the initially projected dann GING das (I:Rgndwie) nich (‘then it somehow didn’t

240 Margret Selting

work’) to the Wnally produced I:Rgndwie GINGs dann nicht (‘somehow it didn’t work then’). As, however, the prosodic break was not contextualized clearly and saliently — it could as well just have been a prominent pitch accent — this re-organization only becomes inferrable in retrospect and what results is a syntactic construction that is called ‘apokoinu’ (cf. also Scheutz 1992, Selting 1995b), a special kind of merged syntactic construction used to construct special kinds of TCUs.1 This example shows that the use of only a single prosodic contextualization cue such as an upstep in pitch in a construction that still can be interpreted as syntactically cohesive is not suYcient to signal a clear break and make the beginning of a new unit interpretable. Nevertheless, this kind of less salient contextualization seems to be a design feature of the apokoinu construction examined here: the upstep subtly contextualizes the point at which reorganization of the syntactic construction begins. The reconstruction is, however, only recognizable in retrospect. b. Non-completion of a syntactically and prosodically projected unit Some fragmentary units are the result of the non-completion of a projected syntactic and prosodic construction. The speaker stops her or his unit without beginning a new one immediately. This is illustrated in the following extracts: (6) K1: 106 ((Laufnr. 046)) ((Ron. hatte gesagt, Überrepräsentation von Frauen in Seminaren sei normal))


((Ron just said that an overrepresentation of women in seminars was normal)) Nat: Ron:

what is your subject then (.) I mean it always [ depends on [ I am studying German and music

In (6), line 01, Nat addresses a question to Ron. When Ron does not take the turn in the short gap after her question, Nat begins another unit in order to specify her question. Midway through her unit, however, Ron now takes the Xoor to give his answer. As a result of this, after a few more words Nat drops out and gives the Xoor to Ron. Both syntactically and prosodically, Nat’s unit is recognizable as an unWnished one. Syntactically, the syntagm ich mein es KOMMT ja auch immer: (‘I mean it always depends on’) is, in this context, not a possible syntactically complete construction but only the beginning of one (in another context, however, this formulation is well imaginable as a complete TCU). Prosodically, the last word ends with a sound stretch and leaves a projected globally falling intonation contour hanging at about mid pitch; i.e., the speaker does not produce something like a terminal falling or rising pitch as a contextualization cue of a possible unit or turn ending, but ends with almost level pitch as a prosodic unit- and turn-holding cue. Both these prosodic devices contextualize that this is not a possible ending of a complete unit. As, however, this projection of more-to-come is not fulWlled, retrospectively, Nat must be heard as relinquishing her unit and turn after interruption by another speaker. (7) K3: 208–209 ((Ida über die Einladungen ihres Vaters))

242 Margret Selting

((Ida about her father’s invitations)) Ida:

the first times I couldn’t go because in most cases I had to work (.) and then I discovered (.) one time when he called I answered the telephone myself ((etc.))

Extract (7) shows a less clear case. After the construction un DANN is mir ma aufgegangen (‘and then I discovered’), syntactically, a subordinate clause is necessary for the construction to be complete; this could be added and prosodically integrated into the same continued intonation contour. At the end of un DANN is mir ma aufgegangen such a continuation is projected syntactically. Prosodically, the signalling seems to be less clear than in the previous cases: the pitch in the entire stretch is globally falling, and in the last syllables of this unit it is still falling, if only very Xatly and slightly. The ending of the fragment is not accompanied by other holding devices. As continuation has been projected here syntactically, the speaker seems to be free to use equivocal prosody, signalling neither unit completion nor unit holding clearly. In this way she gains maximal freedom as to how to continue, i.e. whether to prolong this unit or begin a new one. Yet, after a brief pause Ida chooses a new unit by beginning a recognizably new syntactic construction EINmal hatt ich ihn SELBST am t am TElefon (‘one time when he called I answered the telephone myself’), starting with an upstep in pitch in relation to the prior talk and constituting a new intonation contour. Retrospectively, the prior stretch of talk is recognizable as a fragmentary unit that nevertheless fulWlls its function of holding the turn and projecting more-to-come and of announcing the story that Ida then tells about what she discovered about her father’s behaviour towards her.2 In (6) and (7), then, a begun and syntactically projected TCU is not cut oV, but simply abandoned before a possible syntactic completion point of a possible syntactic construction has been reached. And the prosody in most cases


contextualizes just that: the intonation contour is left hanging at mid pitch without a falling, rising or level possible last pitch accent contextualizing possible unit or turn completion (see Selting 1996a); in some cases additional devices such as sound lengthenings project turn holding, which, however, is not fulWlled after all. In one of the cases, however, syntactic projection of moreto-come is accompanied by prosodically equivocal cues, syntactic projection thus clearly overriding prosodic equivocality here. In short, the cues contextualizing a fragment here can be summarized as follows: syntactic cue: – abandoning a projected syntactic construction before a possible completion point in the present sequential context, prosodic cues: either – pitch left hanging without possible ending of a contour and – optionally other turn-holding devices such as sound lengthening or – equivocal cues between unit and turn holding or completion. The extracts presented in this section showed examples of the most frequent kinds of fragments of units, i.e. those in which both syntactic and prosodic construction schemata are cut-oV or are abandoned and remain unWnished. In the next sections I will try to clarify the relation between syntax and prosody further by trying to separate the individual roles of syntax and prosody in unit production. 3.2 Trying to separate In order to investigate the individual roles of syntax and prosody, I looked for fragments in which the two signalling devices are used discrepantly: one of the two signalling devices is used as if to signal a complete unit, but the other signals incompletion at the end of the fragment. This search, however, yielded results only in one of the possible cases. 3.2.1 The role of syntax In the previous sections we saw cases in which syntactically incomplete fragments of units also ended with holding pitch or prosodically and/or intona-

244 Margret Selting

tionally incomplete contours. In order to show that syntax can be the only decisive cue, and thus that units cannot be distinguished from unWnished fragments of units with reference to syntax only, I looked for sequences in which it is only the prosody that projects more-to-come at the end of an otherwise possibly complete phrasing unit or TCU. This search, however, yielded no results. In fact, cases in which a syntactically complete TCU is signalled prosodically as incomplete, are interpreted diVerently: in these cases, holding prosody is not interpreted as signalling the incompleteness of the current and just complete TCU, but it is interpreted as projecting more-to-come. This more-to-come can then be added both as a continuation of the just completed unit or as a new unit, depending on the kind of syntactic and prosodic continuation. This shows that in cases of possibly complete syntax, prosody cannot be used in order to contextualize this unit as incomplete. Prosody cannot be used against syntactic structure. In these cases, then, (a) syntax seems to be stronger than prosody, and (b) prosody is not interpreted as signalling the state of the production process with respect to the signalling of completion or noncompletion, but as signalling turn holding for a continuation, regardless of how this continuation will be added. Turn-holding, however, does not deal with the formation of units as such but is a completely diVerent kind of activity that deals with turn organisation beyond the unit-in-progress (see below). This in turn corroborates my earlier results showing that while syntax is used in order to project possible completion points of single TCUs and reaches as far as the possible completion points of syntactic constructions such as possible sentences, clauses, phrases etc., prosody is used in order to more locally contextualize the state of the speaker’s production process within the current unit and beyond that into the next such unit. Syntax and prosody thus seem to be used with diVerent scope and with diVerent functions in the organization of conversational talk (Selting 1996a). In cases of discrepancy, however, syntax seems to be able to override prosody but prosody does not seem to be usable against syntax. 3.2.2 The role of prosody As we have seen, not every fragment of a unit is contextualized as such by a cutoV. Some fragments in the examples given so far ended with a kind of hanging pitch that, although perhaps still falling slightly, did not reach a projected possible ending of the intonation contour that was begun before. Some other


fragments in the examples given so far ended with level pitch, with or without sound lengthenings, thus projecting turn-holding. Level pitch for turn-holding, however, is of course not a pitch contextualizing the fragmentariness of an unWnished fragment as such, but a pitch contextualizing the current state of the formulation process as unWnished in order to project intended continuation. As such, level pitch is used both at the ends of units in order to project another unit to come, and at the ends of syntactically and/or semantically and/ or pragmatically unWnished units in order to project the continuation of the unit-in-progress. Only at the end of syntactically and/or semantically and/or pragmatically unWnished units can level (or slightly rising) pitch be used as a cue to signal intended turn-holding for a continuation of the unit and turn. Whether this continuation then turns out to be a continuation of the unit under way or a continuation of the turn by beginning a new unit, can only be analysed retrospectively. It is only the start of a new unit that retrospectively makes inferrable that a prior unit is not being continued and Wnished, but abandoned and left as a fragment. In order to investigate the separate role of prosody, and in order to show that units cannot be distinguished from unWnished fragments of units with reference to only prosody, I looked for sequences in which syntactically incomplete fragments were nevertheless prosodically displayed as if the unit were complete. This search did yield quite a lot of cases in which fragments end with a pitch conWguration that could on its own very well be used as the possible end of a possible contour. In the extracts presented so far, the following examples occur. (8) K1: 5–6

246 Margret Selting

Nat:

Ida:

no I was also so the er ‘m I don’t know the first three four meetings [ because it got on my nerves afterwards I was there (.) [ were you there when I presented my paper

After prior talk about whether Nat and Ida know each other from a seminar they both participated in, the syntactic construction ich WAR auch (‘I was also’) in (8), line 01, cannot be interpreted as a complete unit here. Nevertheless, the fragment ich WAR auch shows a clear example of falling pitch in the accented word WAR and the following unaccented word auch, which could on its own very well be the ending of a possible intonation contour that could be used as a possible unit- or turn-yielding contour. This pitch looks very much like that used for possible unit- and turn-endings, for instance the Wnal pitch accent in Nat’s later unit weil mir das dann geSTUNken hat (‘because it got on my nerves afterwards’). Nevertheless, despite its falling pitch, this construction could also have been continued by adding further material with continuing prosody. But again, it is only the following constructions that make it clear retrospectively that the construction ich WAR auch has been left unWnished and a new construction has been begun with die ersten drei vier SITzungen war ich da:. (‘the Wrst three four meetings I was there’). The Wrst noun phrase of this construction, die ersten drei vier SITzungen, could still be interpreted as the continuation of the prior construction, after inserting some other material (aso: die er ‘m weiß nich ‘so the er ‘m I don’t know’). Just as in those cases in which possible sentences that are produced with non-cohesive prosody are nevertheless interpreted as one unit (Selting 1998, 2000), if here the fragment had been continued to Wnally produce a possible sentence, syntax would then,


too, override the non-cohesive prosody and the result would have been the interpretation of a single unit with an internal self-initiated self-repair in which die er ‘m is self-interrupted and followed by Wrst the parenthetical phrase weiß nich and then the repaired die ersten, etc. Here, however, retrospectively, the upstep for die ersten drei vier SITzungen war ich da: must be reinterpreted as contextualizing the beginning of a new unit. Notwithstanding all this, the pitch conWguration at the end of the fragment ich WAR auch is very similar to that at the end of weil mir das dann geSTUNken hat and, leaving the rather long tail out of consideration, also similar to that ending the unit after the fragment, i.e. die ersten drei vier SITzungen war ich da:. The slope and the depth of the falling pitch accents and their tails are quite similar. (9) K1: 175–178

((Talk about problems in feminist groups; Ron is speaking in Nat’s voice, proVering his own interpretation of Nat’s prior descriptions)) Ron: Nat: Ron: Nat:

we are women and we must have the same opin [ion [yes or uh (.) [we all have the s [a [we have it [we must

In (9), Ron proVers his understanding of the view which, according to Nat, many women in feminist groups hold, namely wir SIND frauen und MÜSsen alle dieselbe meinung habm (‘we are women and we must have the same opinion’). After this, he begins a new unit in which he sets out to formulate the alternative possibility, namely that these women might indeed have the same opinion voluntarily, but after the words oder äh: (.) wir HAben alle diesel cuts this oV in the middle of the word dieselbe. Here, however, no glottal stop is used

248 Margret Selting

and the intonation is falling from the pitch accent in the word HAben to a fairly low pitch that could in other contexts well be a designed unit- and turnyielding pitch. After Nat tries to take the Xoor, Ron simply seems to trail oV until he Wnally stops with a pitch that in other circumstances could be used for unit- or turn-completion. These examples show that fragments need not necessarily end with holding devices or some other kind of prosody that makes the yet unWnished unit recognizable as unWnished. They can end with pitch contours that could also be used as possible completions of possible intonation contours and are similar to those used for designed unit- and turn-completions. Thus, in principle, fragments cannot be distinguished from complete units with reference to prosody only. Nevertheless, the kind of prosody that is chosen does diVerent kinds of interactional work, i.e. diVerentiating between holding or non-holding the yet unWnished unit and turn. In cases in which the speaker produces a prosodic break in order to contextualize the beginning of a new unit and then immediately begins that new unit, no holding devices need be used but the very beginning of the new unit cancels the prior projection and now projects the new unit-in-progress. Cut-oVs, in some cases with saliently held glottal stops, sound lengthenings, and level or slightly rising intonation are used in order to signal unit- and turn-holding, no matter whether the construction-so-far is a syntactically possibly complete unit or not. In this case the speaker prosodically projects more-to-come and thus secures the turn for him/herself to continue. Locally falling or rising intonation, either as in designed contour endings or as in trail-oV pitch, is used in order to signal non-holding of the unit and turn. At the end of a possible unit, and if there has not been a more global semantic or pragmatic projection built up that still needs to be fulWlled (cf. below), such non-holding locally falling or rising pitch is the unmarked pitch that contextualizes designed turn yielding (cf. Selting 1995a, 1996a). If used as trail-oV pitch, this pitch contextualizes that the speaker is not attempting to hold the turn but is yielding it to the recipient. If this recipient has already come in competitively before (French/Local 1983), the speaker now simply relinquishes the turn and leaves it to the competitor. If no competitor has been competing for the Xoor, the recipient is invited to take over and continue. Just as in the case of turnholding, this last option of signalling non-holding of the turn can be used regardless of whether the construction-so-far is a syntactically possibly complete unit or not. That means that, although it can be used in order to display a stretch of talk that ultimately remains an unWnished fragment, the prosodic


contextualization of holding or non-holding of the turn is used as an independent resource in order to contextualize the speaker’s management of the unit and turn. As a result, there is no evidence of a particular prosody signalling and ‘doing fragments’. But there is, Wrstly, a prosody signalling and ‘doing turnholding’ and ‘doing possible turn-yielding’ at the end of yet unWnished units, as well as, secondly, a prosody signalling and ‘doing continuing’ and ‘doing beginning of a new unit’ after a prosodic or other break within or after a yet unWnished unit. Ultimately, this conWrms that it is not the units themselves that are relevant for participants, but the practices related to the construction and delimitation of units-in-turns, i.e. the practices of beginning new units-inturns, of projecting and continuing begun units with or without internal gaps or breaks, and of possibly completing units and yielding the turn. 3.3 Fragments as the result of semantically and/or pragmatically unfulWlled projections in their sequential context In some cases, a syntactic (and prosodic) construction that could in other contexts very well be a syntactically (and prosodically) complete construction cannot be interpreted as complete because in the given sequential context it contradicts the expectations built up in the preceding context. Examples of this occur in the previously presented extracts (2) and (5). In (2), the fragment da war ich t (‘then I was t’), in line 06 of the extract, is a case in point. This fragment, which, like many beginnings of a unit before its Wrst accented syllable, is produced with level pitch, is cut oV with the sound t and the new unit is started with a downstep in pitch and the use of a more salient-than-usual glottal stop as the signal of a prosodic break at the beginning of the new unit ich glaub das war daNACH irgnwie die sitzung (‘I think that was somehow the next meeting’). In this case, however, it is not just the syntax and/ or the prosody that make this stretch da war ich t recognizable as a fragment of a unit. After Ida’s question WARS du da: als ich n: n refeRAT gehalten hab (‘were you there when I presented my paper’), the construction da war ich — albeit with a diVerent intonation contour — could have been a possibly complete positive answer. But here, Nat’s Wrst unit NEE: (‘no’) has already given a negative answer. A positive answer now such as da war ich would produce an outright contradiction. And this is another reason why the fragment da war ich must be analysed as a fragment: because it does not fulWll the semantic or pragmatic projection built up in the prior unit.

250 Margret Selting

A similar case occurs in (5). As I already said above, the Wrst part of Nat’s construction dann GING das (‘then it worked’), although a possible syntactically complete unit in other contexts, with a possibly complete intonation contour as well, cannot be interpreted as possibly complete here, because it would be in contradiction to the previously built up expectation of a negated statement. Here, too, the previous talk projected a particular kind of semanticpragmatic relation to the prior talk which constrains the interpretation of talk as a complete unit as long as it does not fulWll these projections. These examples show that the completeness or non-completeness of a unit cannot be analysed with reference to only syntax and prosody: completeness or incompleteness is a context-sensitive inference about whether a construction has reached possible completion in its present semantic, pragmatic and sequential context. As I have shown, neither syntax nor prosody on their own can be used to unequivocally distinguish fragments of units and possibly complete units; in cases of discrepancy syntax overrides prosody. Further, fragments are the result of unWnished syntactic, semantic, pragmatic and prosodic projection within their sequential context. Thus, we have to conclude that it is the interaction of the syntax and the prosody of the utterance in its semantic, pragmatic and/or sequential context that makes fragments of units recognizable as such. 3.4 Recipient reactions as evidence of their interpretation of fragments Evidence for the above analysis of unWnished fragments can be gained from an analysis of the further behaviour of both the speaker and the recipient: the speaker gives evidence of his or her own interpretation of a stretch of speech as an ‘unWnished fragment’ by presenting it as a fragment through the deployment of cut-oVs and/or holding devices as well as by indeed starting a new TCU that in many cases repairs or redirects the prior utterance. As, by the turn-taking system (Sacks, SchegloV and JeVerson 1974), speakers are entitled to complete TCUs, and as recipiency tokens and other overlap is normally produced near, at or after the end of a complete TCU, most frequently an unWnished fragment will yield no response from the recipient. The recipient seems to be refraining from responding till the end or the near-end of the newly begun TCU (cf. JeVerson 1986 on ‘latency’ in overlap onset), thus displaying his or her orientation to the projected complete TCU. Those recipient responses that do occur in the data at or after unWnished fragments can be explained by factors other than the fragmentary nature of the


utterance: the mhm after da war ich in (2), line 7, can be analysed as Ida’s late response to Nat’s prior TCU NEE:. If fragments are the result of competitive incoming talk by another party, the other speaker’s continuing while the prior speaker drops out can be regarded as evidence that the interrupter understands that the former speaker is yielding the Xoor even though a TCU has not been Wnished.

4.

Conclusions: ‘Fragments of units’ and ‘units’ as epiphenomena of practices which organize talk in turns

The analysis of fragmentary units has shown that these are the result of the speaker’s non-Wnishing of her or his unit, due to the speaker’s self-interruption in order to start a new unit, to an interruption by another speaker, or to the speaker’s projecting more-to-come but then ignoring this prior projection for the sake of beginning a new unit. In some cases, the end of a fragment is signalled by the speaker’s cutting oV the given unit, but much more commonly, it is only the beginning of the new unit that is saliently contextualized and made recognizable, thus implying the abandonment of the prior, as yet unWnished unit. In general, the beginning of this new unit is signalled both syntactically and prosodically: by the recognizable beginning of a new syntactic construction and by prosodic devices such as an upstep or a downstep in pitch and/or a change in tempo. The possible completeness or non-completeness of a syntactic construction is a context-sensitive interpretation in which syntactic, semantic and/or pragmatic projection within the given sequential context is involved. On the one hand, fragments may appear with possible completion prosody. But, on the other hand, prosody does not seem to be usable against syntax: if possibly complete syntactic constructions were presented as not complete by deploying prosodic unit and turn holding devices, this would not result in the unit’s interpretation as fragmentary, but in the projection of more-to-come, regardless of whether this continuation were to be displayed as a continuation of the same or as a new unit. In cases of discrepancy, syntax overrides prosody. In short: retrospectively, the fragmentary unit is reconstructable as fragmentary because syntactic, prosodic, semantic and/or pragmatic projections have been built up in the sequential context but have not been fulWlled. The interpretation of the fragmentariness of stretches of talk is thus the result of non-fulWlled context-sensitive projections. If this is true for fragments of units, it also corroborates my prior analysis

252 Margret Selting

(Selting 1998, 2000) that units cannot be deWned with respect to either syntax only or prosody only but are the result of the interaction of both in their sequential context. The interpretation of units, then, is the result of fulWlled context-sensitive projections. IdentiWcation of units results from context-sensitive inferences. They are the result of the participants’ deployment of Xexible syntactic and prosodic construction schemata, which, within the given semantic, pragmatic and sequential context, provide Xexible possible completion points of units. The points of possible completion, and thus the units themselves, are the result of diVerent kinds of projection. Syntactic projection is more far-reaching than prosodic contextualization. Yet, syntactic projection is conWned to the limits of the current unit. Prosodic contextualization is in principle the local signalling of the state of the production process. It is used to project continuation within a unit until the contextualized end of the unit, but prosody can also project beyond the current unit and into the next. In comparison to syntax and prosody, semantic, pragmatic and activity-type speciWc projections are even more far-reaching and can reach beyond units and even turns in their sequential context (Selting 1998, 2000). The interpretation of units is the product of prosodic, syntactic, semantic, pragmatic and/or activity-type speciWc projections having been built up and fulWlled within the given sequential context. The interpretation of fragments of units is the product of one or a combination of these projections having been built up but left unfulWlled before the speaker begins a new unit or relinquishes the turn. In general, syntax and prosody co-occur in the signalling of units. There is no evidence that either syntax or prosody is more important for the construction and projection of units in their sequential context. While in some instances of discrepancy in unit production, syntax overrides prosody, for the constitution of particular activity types in conversation, prosody overrides syntax and is distinctive on its own (cf. Selting 1992, 1996b; Couper-Kuhlen 1996; Günthner 1996 etc.). Both with respect to syntax and with respect to prosody, units are Xexible, expandable, revisable, situationally adaptable kinds of phenomena. Fragments of units are interpreted as ‘unWnished fragments’ because, retrospectively, after the beginning of a new unit, they become recognizable as constructions that projected some kind of continuation and completion that was not fulWlled. The projections themselves are established with respect to more far-reaching syntactic, semantic or pragmatic construction schemata, expectable activity-type speciWc action schemata, or more local prosodic con-


textualization and projection in the given sequential context. The more farreaching syntactic, semantic and pragmatic projections determine possible completion points of the unit under way. If such a projected unit is abandoned before a possible completion point is reached in the present sequential context, the result is an unWnished fragment of a unit. In its sequential context, a possible syntactic construction is thus a context-sensitive inference. The fragment can, but need not, be contextualized prosodically with, e.g., a cut-oV and, possibly, unit- and turn-holding devices such as level (or slightly rising) pitch and/or sound lengthening. But even if it is, this is not the signalling of the fragment as such but the signalling of projected unit- and turn-holding for a projected continuation. Fragments can be the result of several practices:3 (1) A speaker may selfinterrupt in order to begin a new unit. A special case of this is when a speaker changes her or his projected syntactic construction midway, thus producing an apokoinu construction as described above. (2) A speaker may drop out of his speakership and be silent or yield the Xoor to his recipient(s), thus leaving an unWnished fragment behind. In (1), which — I assume — is the most frequent case, the ‘doing beginning’ of a new unit can be analysed as the initiation of repair on the prior unWnished unit, which cancels the syntactic and prosodic projections built up in the prior fragment. This shows that the possibility of leaving an unWnished fragment behind and beginning a new unit is a local resource which cancels local syntactic and prosodic projections built up in the prior unWnished fragment; at the same time it keeps intact, i.e. only suspends, more global semantic, pragmatic, activity-type speciWc or sequential projections of being continued and fulWlled within the new and/or the following units. In the special case of the production of an apokoinu construction, the syntactic construction is changed around and re-oriented. Here, too, when a prosodic break contextualizes the ‘doing beginning of a new unit’, the new beginning can retrospectively be analysed as a repair of the previously abandoned stretch of speech, albeit one that is performed in a camouXaged and syntactically highly systematic way, leaving both parts of the construction intact albeit ‘fused’. In both these cases, the prosody prior to the prosodic break and/or the ‘doing beginning of a new unit’ may be holding or non-holding the unit and turn, with incomplete syntax as an unfulWlled more global projection overriding more local prosodic contextualization (Selting 1996a). In (2), there is simply a trail-oV and a relinquishing of the unWnished fragment, i.e. a nonholding of the unit and turn but a yielding of the turn in spite of the yet incomplete syntactic unit in the given sequential context.

254 Margret Selting

Ultimately, the fact that it is not units or fragments as such that are signalled by participants suggests that it is not the units themselves as such that are relevant for participants, but the practices related to the construction and delimitation of units in turns and of making recognizable the state of their production process within turns, i.e. the practices of beginning new units, of projecting or holding and continuing begun units with or without internal gaps or breaks, and of possibly ending units and yielding turns. This means that fragments of units, as well as units themselves, are epiphenomena of the participants’ practices of constructing, organizing and making recognizable turns at talk for recipient(s) in the interaction. This result thus supports research that aims at departing from the description of units and their properties as Wxed and static entities and that focusses more on the deconstruction and reconstruction of participants’ practices of constructing and making interpretable recognizable units and turns in interaction (Selting 1996a, cf. also Ford/Fox/Thompson 1996). Yet, as I have shown in a previous paper, these epiphenomena, i.e. units or TCUs in talk in turns, are by no means irrelevant for the internal and external organization of turns-at-talk in conversation (cf. Selting 1998, 2000). In sum, fragments are interpreted as ‘fragments’ because they are the result of built up but unfulWlled or not (yet) completed projections in their given sequential context. And as any completion of a turn presupposes the concomitant completion of the last TCU of the turn, Wnal fragments of units in a turn also always imply unWnished turns. While, however, each turn-Wnal fragment of a unit entails per implication the interpretation that the turn was left unWnished before a possible turn completion, not every turn that is left unWnished entails a fragment of a unit. We can easily imagine incomplete turns which nevertheless consist of complete TCUs, e.g. when a story is left unWnished although the speaker has not produced a fragmentary TCU before dropping out.

Notes * I am grateful to Peter Auer, Elizabeth Couper-Kuhlen and John Local for critical comments on an earlier version of this paper, which helped me to considerably revise and improve my analysis, and to Linda Paul for correcting my English. 1. When a speaker merges two constructions, as s/he does in the case of apokoinu-constructions, the Wrst part of this construction might be analysed as a fragment. But on the other hand, as two, by themselves completely grammatical constructions are merged around the


koinon in a systematic and functional way and there is no clear prosodic break signalling the start of a new unit, I would prefer to conceive of them as a special case of a possible sentence in spoken language for constructing special turns in conversation (cf. also Selting 1995b, Scheutz 1992). 2. The TCU EINmal hatt ich ihn SELBST am t am TElefon contains yet another instance of a cut-oV and new start internally: the prepositional phrase am t is cut oV, recycled to its beginning and then produced in full am TElefon — a case of self-initiated self-repair. 3. Cf. also HoVmann (1991: 99V.), who diVerentiates the procedures underlying the production of anacoluthons as ‘Ausstieg’, ‘Retraktion’ and ‘Umstieg’.

References Couper-Kuhlen, E. 1996 “Intonation and clause combining in discourse: The case of because”. Pragmatics 6 (3): 389–426. Ford, C., Fox, B. A. and Thompson, S. A. 1996 “Practices in the construction of turns: The ‘TCU’ revisited”. Pragmatics 6 (3): 427–454. French, P. and Local, J. 1983 “Turn-competitive incomings”. Journal of Pragmatics 7: 17–38. Gülich, E. and Kotschi, T. 1987 “Reformulierungshandlungen als Mittel der Textkonstitution“. In Satz, Text, sprachliche Handlung, Mortsch (ed.) [studia grammatica XXV]. 199–261. Berlin: Akademie-Verlag. Günthner, S. 1996 “From subordination to coordination? Verb-second position in German causal and concessive constructions”. Pragmatics 6 (3): 323–356. HoVmann, L. 1991 “Anakoluth und sprachliches Wissen“. Deutsche Sprache 1991 (2): 97– 119. JeVerson, G. 1974 “Error correction as an interactional resource”. Language in Society 3: 181–199. 1986 “Notes on ‘latency’ in overlap onset”. Human Studies 9: 153–183. Levinson, S. C. 1983 Pragmatics. Cambridge: Cambridge University Press. Local, J. 1992 “Continuing and restarting”. In The Contextualization of Language, P. Auer and A. di Luzio (eds.), 273–296. Amsterdam, Benjamins. Local, J. and Kelly, J. 1986 “Projection and ‘silences’: Notes on phonetic and conversational structure”. Human Studies 9: 185–204.

256 Margret Selting

Sacks, H., SchegloV, E. A. and JeVerson, G. 1974 “A simplest systematics for the organisation of turn-taking for conversation”. Language 50: 696–735. SchegloV, E. A. 1996 “Turn-organization: one intersection of grammar and interaction”. In Interaction and Grammar, E. Ochs, E. A. SchegloV and S. A. Thompson (eds.), 52–133. Cambridge: Cambridge University Press. SchegloV, E. A., JeVerson, G. and Sacks, H. 1977 “The preference for self-correction in the organisation of repair in conversation”. Language 53: 361–382. Scheutz, H. 1992 “Apokoinukonstruktionen. Gegenwartssprachliche Erscheinungsformen und Aspekte ihrer historischen Entwicklung“. In Dialekte im Wandel, A. Weiss (ed.), 243–264. Göppingen: Kümmerle. Selting, M. 1992 “Prosody in conversational questions”. Journal of Pragmatics 17: 315– 345. 1995a Prosodie im Gespräch. Aspekte einer interaktionalen Phonologie der Konversation. Tübingen: Niemeyer. 1995b “Der ‘mögliche Satz’ als interaktiv relevante syntaktische Kategorie“. Linguistische Berichte 158: 298–325. 1996a “On the interplay of syntax and prosody in the constitution of turnconstructional units and turns in conversation”. Pragmatics 6 (3): 357– 388. 1996b “Prosody as an activity-type distinctive signalling cue in conversation: The case of so-called ‘astonished questions’ in repair-initiation”. In Prosody in Conversation: Interactional Studies, E. Couper-Kuhlen and M. Selting (eds.), 231–270. Cambridge: Cambridge University Press. 1998 “TCUs and TRPs: the construction of units in conversational talk”. Konstanz, InLiST No. 4, June 1998. 2000 “The construction of units in conversational talk”. Language in Society 29: 477–517. (= slightly amended version of Selting 1998) Wootton, A. J. 1989 “Remarks on the methodology of conversation analysis”. In Conversation: An Interdisciplinary Perspective, D. Roger and P. Bull (eds.), 238– 258. Clevedon etc.: Multilingual Matters.


Appendix Transcription conventions Sequential structure [] =

overlap and simultaneous talk latching

Pauses (.) (..)

micropause brief pause

Other segmental conventions und=äh :, ::, ::: äh, öh, etc. ‘

assimilations within units segmental lengthening, according to duration hesitation signals, so-called ‘Wlled pauses’ cut-oV with glottal closure

Laughter so(h)o haha hehe hihi ((lacht))

laugh particles within talk laugh syllables description of laughter

Recipiency tokens hm, ja, nein, nee hm=hm, ja=a, nei=ein, nee=e ‘hm‘hm

monosyllabic signals disyllabic signals with glottal stops, usually used for negative responses

Accentuation akZENT akZENT

strong, primary accent weaker, secondary accents

Pitch upstep/downstep

to higher pitch to lower pitch

258 Margret Selting

Notation of pitch accent movements \ / – /\ \/

falling to mid rising to mid level rising-falling falling-rising

Notation of global pitch realised in the stretch of speech notated above the brackets F( ) R( ) M( ) H( ) L( ) M,F( ) H,R( ) [( )( )] ({})

falling rising mid high low falling within mid register rising within high register combined contours constituting a paratone embedded contour, e.g. for parentheses

Changes in loudness and speech rate

=forte, loud =piano, soft =lento, slow =allegro, fast

Breathing .h, .hh, .hhh h, hh, hhh

inbreath, according to duration outbreath, according to duration

Other conventions ((hustet)) (…)

para- und extralinguistic activities and events concomitant para- und extralinguistic activities and events unintelligible, according to duration

Notes on turn-construction methods in Danish and Turkish conversation Jakob Steensig University of Aarhus

The paper examines some turn-construction methods used by participants in Danish and Turkish native, everyday conversations. Three types of methods are examined in data excerpts: pragmatic, grammatical, and prosodic. The analyses suggest that these methods contribute to turn-construction in diVerent ways in the two languages: because of diVerences in grammatical structure, Danish grammar is responsible for a far-reaching projection of trajectories and possible completion points, which is not possible in Turkish. In Turkish, however, prosody comes to play a more important role in the projection of possible completion. It is further argued that turn-construction methods should be kept apart analytically; otherwise it will prove diYcult to compare methods across languages.

Introduction In interaction, people use linguistic means, including grammar, prosody, and pragmatic knowledge, in their production of talk and in the on-going interpretation of the talk of others.1 In this paper such linguistic means are termed “turn-construction methods”.2 The paper focusses on diVerences in turn-construction methods in Danish and Turkish. The starting point is an observation that in both Danish and Turkish conversations, participants seem to be able to predict points of possible turn completion before these occur, i.e. conversations in both languages rely on the feature of projectability (Sacks 1992: 40; Sacks, SchegloV and JeVerson 1974: 702; SchegloV, Ochs and Thompson 1996: 28–31). The second observation is that the methods used to do such projection cannot possibly be the same in these two languages: in Danish, grammar as a turn-construction method allows a strong projection from early on in the utterance, whereas Turkish grammar does not provide for such early projection. This implies that other methods may be used in Turkish to do the job which grammar does in Danish

260 Jakob Steensig

and, consequently, that the relationship between turn-construction methods is diVerent in the two languages. The aim of the paper is to describe the abovementioned turn-construction methods and to compare their relative importance across the two languages. After a few words on methodology, the initial observations which set oV the study will be presented. Then a preliminary deWnition of “turn-construction methods” is oVered. After this follows an analysis of the turn-construction methods used in one Danish and in one Turkish example. Finally, I try to draw conclusions concerning the diVerences between the methods used in the two languages.

Methodological preliminaries The methodology employed here tries to combine a conversation analytic insight into turn-construction (as represented in Sacks, SchegloV and JeVerson 1974) with an investigation of speciWc linguistic methods and a comparison between diVerent languages. The approach is inspired by the recent work in “interactional linguistics”, e.g., Couper-Kuhlen and Selting, eds. (1996); Ford (1993); Ford and Wagner, eds. (1996); Ochs, SchegloV and Thompson, eds. (1996); Tanaka (1999). Some of the tenets of the methodology are that the analysis should be based on authentic interactional data, and that analysts should try to use the understandings which the participants show each other as their main guide in the analysis. Furthermore, an attempt is made to understand talk as something which takes place in “real time”.

The data The data for the present study come from a corpus of Danish and Turkish everyday conversations collected at the Department of Linguistics, Aarhus University. The Danish corpus consists of 3–4 hours of everyday interaction between native speakers of Danish. The Turkish data are from a smaller corpus; it consists of approximately one hour of face-to-face conversations between native speakers of Turkish living in Denmark, supplemented with parts of a published corpus of spoken Turkish from Germany (Auer and Fearn 1992).3

Notes on turn-construction methods in Danish and Turkish conversation 261

In some cases participants were instructed to carry on a conversation. Such semi-authenticity is indicated in every case. I do not see the semi-authenticity of some of the data as a major problem because I believe that people in these conversations use many of the same turn-construction methods as in the (more) authentic data. All recordings are audio recordings, although most of the conversations are face-to-face interactions. This is a potential problem because visual cues can be important turn-construction methods (see, e.g., Goodwin 1979, 1981, 1986; Goodwin and Goodwin 1986, 1992; Heath 1992; Kendon 1981; SchegloV 1984; Streeck and Hartge 1992). However, I have judged it worthwhile starting without access to this side of the interaction, knowing that the description of the methods may have to be revised as we learn more about visual methods. The examples used in the present paper are chosen because they illustrate what I see as general tendencies in my data. No attempts were made to assess the statistical weight of the turn-construction methods described. The reliability and the validity of the Wndings will depend on their replicability and relevance in other data.

Transcription All examples were transcribed by the present author using a modiWed version of the transcription conventions developed by Gail JeVerson for conversation analysis (cf. Atkinson and Heritage, eds. 1984: ix-xvi; Ochs, SchegloV and Thompson, eds. 1996: 461–65). The appendix lists those conventions which are not part of the JeVerson system, including the abbreviations used to indicate morphemes in the interlinear glosses in Turkish excerpts. For the indication of intonation, I have used principles from the “GAT” system (Selting et al. 1998; see also the section on prosodic turn-construction methods below). The auditory description of prosodic features has been supplemented with instrumental measurements of pitch and intensity, done by means of the PRAAT phonetics analysis program.

Initial observations In both Danish and Turkish conversations, recipients are able to predict the possible completion of a turn before it actually occurs. This is shown in the following excerpts:

262 Jakob Steensig

In the Danish excerpt (1), two colleagues who work in a kindergarten are discussing which songs to sing with the children the next day: the phrases “fastelavn er mit navn” (‘carnival is my name’) and “boller op boller ned” (‘buns up buns down’) are lines from a well-known children’s carnival song: (1) [RLM:99:2] ((Danish, telephone)) 01

Inge:

Altså fastelavn er mit navn. (.)

02

Inge:

Den [tror jeg] den ka’ de hel]t= that [think I ] that know they ]whole ‘That[one I be] lieve they know ]for sure’

03

Line:

[

04

Inge:

=sikk[°ert°] sure[ ]

05

Line:

→ 06

Line:

*

Inge:

07

01 02–04 03 05–06 07

I: I: L: L: I:

fha ] ↑ste↓ lavn “carnival”

]

[Og b]oller op boll’= [and “b]uns up bun” =Det’ vist den [samme] it’s I.believe the [same ] [jha, ] yes Well “carnival is my name” that one I believe they know for sure “carnival” and “buns up bun”, it’s the same, I believe yes

In line 07, Inge shows her recognition of what Line is saying in overlap with the crucial word, samme (‘same’). Somehow she is able to predict — or at least to claim that she has predicted — what Line is about to say. Something similar happens in the Turkish excerpt (2) below. It is from the beginning of a phone call. Kemal, the caller, has just enquired about ‘his brother Nuri’, Pakize’s husband. The participants live in Germany, near the Swiss border: (2) [TRKon:2–3] 01

→ 02

*

. Pakize: Nuri abi+n+i Isviçre+ye gönder+di+m Migro+’ya. Nuri brother+2S+OBJ Switzerland+DAT send+PRT+1S Migro+DAT ‘I sent your brother Nuri to Switzerland to “Migros”’ •s Kemal: H:a↑ha Ali ¸[veri↑s ¸+e m[i Oh shop [ping+DAT [QPC ‘Oh to do [the shoppi[ng?’

03 Pakize:

[ha. [? •ö::; Çay yok+°t(u)° yes uh tea not.exist+PRT ‘yes, uh there was no tea’


01 02 03

Pakize: I sent your brother Nuri to Switzerland, to Migros4 Kemal: Aha, to do the shopping? Pakize: Yes, uh there was no tea

In line 03, Pakize responds to what Kemal is saying (in line 02) in overlap with and in the middle of the crucial word, alF¸sveri¸s (‘shopping’). Somehow Pakize can Wgure out what Kemal is in the midst of saying and respond appropriately. The question is now, how can the recipients know this? If we look at the pragmatic methods used, we can see one basis for this: In excerpt (1), line 05, Line seems to be suggesting that “boller op boller ned” is a diVerent song from “fastelavn er mit navn”. She does not complete her phrase. Instead she starts saying something (line 06) which can be heard as a repair of and a comment on what she was saying in line 05. Inge can recognize this activity and, knowing that the two lines belong to the same song, she can guess that samme (‘same’) or something similar is what Line is about to say. Excerpt (2), line 02, begins with a token, H:a↑ha, which seems to indicate that what Pakize has said in line 01 is news to Kemal. Then Kemal produces the Wrst part of the word ‘shopping’. As it has been said that Nuri has gone to the shop “Migros”, something to do with shopping may be a very likely thing to say, and this aids Pakize in guessing what Kemal is about to say before he has completed it. If we take a closer look at the grammatical processes involved, it turns out that they are rather diVerent in the two languages: In the Danish excerpt (1), line 06 begins with det (‘it’ or ‘it’s’) and an adverb, vist (‘I believe’). Already here, it is clear to the native ear that a predicative construction has been initiated and that a complement describing what det is referring to is projected. The unstressed den can be recognized as the beginning of a noun phrase (‘the’), and this, together with the pragmatic knowledge mentioned above, allows a safe guess as to what Line is saying and that she has arrived at a possibly Wnal word at this point. This is a recurrent pattern in Danish. Clauses generally begin with a pronominal or adverbial item, and then comes a Wnite verb. Usually, a quite reliable prediction can be made at this point about what elements must be produced before the grammatical unit can be completed. Here are a few examples of this from the Danish data: (3)

↑Den var sgu for nem. it was bloody too easy ‘It was too bloody easy’ [AT:94:DKMa:32] ((face-to-face))

264 Jakob Steensig

(4)

Den de::r dar he:dder Ingen °mening° that there which is.called no opinion ‘That one there which is called “no opinion”’ [AT:94:DKMa:33] ((face-to-face))

In (3), the verb, var (‘was’) projects a predicative complement; for nem (‘too easy’) is such a complement. In (4), dar (‘which’) begins a relative clause, after this comes a verb, he:dder (‘is called’), which projects a ‘name’ complement, and ingen mening is (in this context) such a complement. This pattern, with the verb placed in second position projecting a rather precise trajectory, is with minor modiWcations the general clause pattern in Danish, in both main clauses and embedded subordinate clauses. In the Turkish excerpt (2), no grammatical relations have been expressed at the moment when Pakize starts talking; the relational information comes with the suYx (+e) on AlF¸sveri¸s+e. This so-called ‘dative case’ suYx expresses a ‘purpose’ meaning (‘to’ or, ‘in order to’). The ‘question particle’, mi, makes the phrase interrogative. This is a general pattern in Turkish: elements which express important grammatical relations and give information about the overall pragmatic or syntactic function of an utterance generally come last in the clause. The next two examples illustrate other instances of this general tendency: (1) verbs occur at the end of clauses, and (2) subordinating morphemes are suYxes on verbs. On the one hand, we have clauses like the following: (5)

Daha hiç birs ¸ey ↓bil+mi+yor+uz ↑ama: … more nothing know+NEG+PRS+1P but ‘We don’t know anything more, but…’ [JS:92:TR2:4] ((face-to-face, semi-authentic))

This clause ends in a verb with a tense suYx (+yor, ‘present tense’) + a personal suYx (+uz, ‘we’). This means that the verb is Wnite and, therefore, also possibly clause Wnal. But, on the other hand, all verb stems can have subordinating suYxes instead: (6)

·hh Yani dil bil+me+dig ˘ +imiz+den; ·hh in.other.words language know+NEG+PPP+1P+ABL ‘·hh, in other words, because we do not know language(s)’ [JS:92:TR2:9] ((face-to-face, semi-authentic))

In both excerpt (5) and (6), we Wnd the verb ‘know’ preceded by a grammatical object (hiç bir¸sey (‘nothing’) in (5) and dil (‘language’) in (6)). But, whereas the suYxes on the verb in (5) made the clause Wnite and possibly Wnal, the


opposite is the case in (6). In this case, the verb stem is followed by a participle suYx (+diæg ): bil+me+diæg (‘not knowing’). This is followed by a possessive suYx (+imiz, ‘our’) and a case suYx (+den, ‘from’ or ‘because of’) giving bilmediægimizden, ‘because of our not knowing’ or ‘because we do not know’. This is a subordinate, adverbial construction which cannot constitute a possibly complete clause. Such non-Wnite and subordinate verbal constructions are frequent in the Turkish data, and they usually precede the items they relate to. The “problem” with bilmediægimizden from a turn-construction view is that the non-Wniteness only becomes apparent at the end of the verb. To sum up: in Danish, as in English, the initial elements grammatically project “the unit-type under way, and what, roughly, it will take for an instance of that unit-type to be completed” (Sacks, SchegloV and JeVerson 1974: 702). In Turkish, the initial elements do not provide such projection. The occurrence of a verb can be a signal of imminent possible completion, but verbs usually come at the end of their clauses and they only project possible completion if they turn out to be Wnite. If not, they project “more to come”. In spite of these “problems” with grammatical projection, Turkish conversationalists still seem to be able to project possible completion before it occurs (as we saw in excerpt (2) above). This could mean that other turn-construction methods are used to a greater extent, or are used diVerently, in Turkish compared to Danish. The next sections of the paper will, therefore, investigate more closely the possible interplay between the three turn-construction methods mentioned in the introduction.

Turn-construction methods, some preliminary definitions The turn-construction methods which will be described are pragmatic, grammatical, and prosodic in nature. The descriptions rely on earlier descriptions of such methods, especially Ford and Thompson (1996); Ford, Fox and Thompson (1996); Sacks, SchegloV and JeVerson (1974); Sacks (1992); SchegloV (1996); SchegloV, Ochs and Thompson (1996); Selting (1995, 1996, 2000); Steensig (1994, forthc.); Tanaka (1999). As their name indicates, the methods “construct turns”, or rather they construct “turn-constructional units” (Sacks, SchegloV and JeVerson 1974), i.e. the units which turns-at-talk consist of. The following sections will examine for all three methods how they contribute both to the step-by-step composition (in real time) of talk, and to the projection of possible completion.

266 Jakob Steensig

Pragmatic methods Conversation analysts have always stressed that turns are, basically, actions, or “adequate complete utterances” as Sacks puts it in a lecture from 1968 (Sacks 1992: 43). Even though a speciWc pragmatic method of turn-construction is not mentioned in the classic paper on turn-taking (Sacks, SchegloV and JeVerson 1974), conversation analytic literature does mention, e.g., the diVerence between “syntactic phenomen[a]” and something which is not “particularly” this (Sacks 1992: 182), the diVerence between “possibly complete actions” and “syntactic possible completion” (JeVerson 1986: 179), or “the several dimensions which together constitute possible completion — syntactic, prosodic and action/pragmatic” (SchegloV 1996: 83). Investigations into how participants recognize and react to actions lie at the heart of the conversation analytic endeavour. Action recognition works in complicated ways and there is, therefore, no simple deWnition of pragmatic projection or pragmatic completion. The approach taken here is to characterize the actions and try to see how each action makes other actions relevant. The provisional deWnition of a pragmatic unit is that it is a string of talk which others can respond to. The step-by-step analysis of the talk will, therefore, try to Wnd when (how early) and how some talk becomes “respondable-to”. This requires an analysis of the sequential context which utterances enter into.

Grammatical methods I have chosen “grammar” as a cover term for both syntax and morphology because many relations which are expressed syntactically in Western European languages are expressed morphologically in Turkish. In looking for grammatical methods I have used (1) the grammatical dependencies which are known to exist in the languages in question, and (2) Tanaka’s deWnition of “syntactic completion”: “syntactic completion will be deWned in this study as a point in a turn where no further talk is syntactically projected when considered within its prior context” (1999: 65, original emphasis). In the delimitation of “prior context”, Tanaka and the present study use Ford and Thompson’s notes on this as a guideline: We include context to the extent that it is responsible for the recoverability of reference, as in the case of answers, which do not require the full repetition of previous questions. (…) Our deWnition of syntactic completion did not, however, include intonation, nor did it include a notion of a sequentially relevant conversational action. (1996: 144)


This delimitation of context is, of course, an analytic one, which is only made in order to separate out grammatical from prosodic or pragmatic methods. It will only serve its purpose if, by using it, we can gain insight into the combination of turn-construction methods which a more participant-oriented distinction cannot give us.

Prosodic methods In studying the step-by-step composition of talk, “pitch reset” and “stress” turn out to be important features. In the analyses of the projection of possible completion, “Wnal” versus “non-Wnal” intonation and “possible last accents” Wgure most prominently. “Pitch reset” has been noted (with an initial capital letter) when some talk “sounded like a new beginning”. Subsequent pitch measurements showed that the “pitch resets” all represented pitch rises in relation to where the prior utterance ended.5 The question of “Wnal intonation” is a tricky one, especially because I have found no descriptions of intonation in conversational Danish or Turkish.6 For Ford and Thompson “Wnal intonation” equals what in the transcription system is designated by a period or a question mark, i.e. “a marked fall in pitch at the end of the intonation unit” (1996: 147) and “a marked high rise in pitch at the end of the intonation unit” (ibid.). “Non-Wnal intonation” is all other contours. The present study does not operate with “intonation units”, but, apart from that, it follows Ford and Thompson: “Final intonation” is whenever I hear a “marked fall” or a “marked high rise”. Rises which are not “marked high” (indicated with commas), and falls in pitch which are not drastic or do not drop to a low level (semicolons) are considered “non-Wnal”.7 The term “possible last accent” was introduced by Selting (1995: 201, 206, 230; 1996: 379), partly inspired by SchegloV (1987: 106f). The main point is that grammar projects trajectories with alternative possible last elements, and a clear stress (or “pitch peak”, SchegloV 1996: 84V, 1998: 238V) on such a projected element conWrms its status as a possible last element in the given turn and, thus, opens the transition relevance space. Other prosodic features, such as the ones studied by Local, Kelly and Wells (1986) as signalling “turn endings” (volume changes, tempo decrease and centralized vowel quality), are not investigated systematically here, but occasional observations concerning factors like these enter into the descriptions where possible.

268 Jakob Steensig

In the next two sections, Wrst a Danish and then a Turkish excerpt will be examined. After the presentation of each excerpt, the methods will be described for each fragment in the following order: Wrst the pragmatic, then the grammatical, and, Wnally, the prosodic method. The three descriptions will be followed by an attempt to sum up how the methods are combined in the Danish and Turkish case, respectively.

Turn-construction methods in a Danish excerpt The Danish excerpt (7) comes from a face-to-face conversation between four women recorded in a lunch break at the participants’ work place. Earlier in the conversation, Lone has talked about a place in her house where she and her husband want to put in new windows. This place has been referred to as a veranda (‘veranda’). There has been some talk about this renovation work, and Lone has said that they have decided to get professionals to do it. The analysis will focus on lines 04–07: (7) [LMP:94:DKKv:19–20] ((Danish, face-to-face, 4 participants)) 01

Lone:

02

Så ska’ det være. (1.2) Og så kommer ↓de:. then shall it be and then come they ‘Now is the time. And then they come,’ (0.7)

03

Lone:

>(Så’n så) de ↑FÅR så’n vi-< AL DEN-= (so that) they get so.that we- all that‘(So that) they get something so that we- all that-’

04

Lone:

=↑Den der ↓rum.= that there room ‘That room there,’

05

Lone:

=↑Den bli’r jo brugt til alve:hr°dens ting,° it gets y’ know used for all.kinds things ‘it gets, you know, used for all kinds of things’

06

Bente: °så’n et° p:↑ulter↓rum. a.kind.of a storage.room ‘A kind of storage room’

07

Lone:

i:Ja:h; ((smile voice)) Yes

Lone:

Og tørre ↓tøj; ((smile voice)) ‘and drying clothes’

08 09 10

(0.3)

(.)


11

Lone:

01–03

L:

04–05

L:

06 07 09 11

B: L: L: L:

Og en ↓tørre°tumblerh og,° ·hhh Så de sk↑a↓:b.h and a drying tumbler and ·hhh then those cupboards ‘and a dryer and, ·hhh then those cupboards’ Now is the time. And then they come. (So that) they get something so that we- all thatThat room there, it gets, you know, used for all kinds of things A kind of storage room Yes And drying clothes And a dryer and, ·hhh then those cupboards…

Pragmatic methods in the Danish excerpt8 In line 01, Lone re-opens the report on the renovation work and completes the talk about whether she and her husband should do the work themselves or get professionals to do it; the word, de (‘they’), refers to the people who are to do the work. It is hard to make sense of line 03, but it is probably a continuation of the report of the renovation of the house. In line 04, Lone restarts and, apparently, re-designs her report. Here she introduces a place in the house. The way she refers to it (with a demonstrative) seems to rely on recipients’ ability to locate the place with respect to what they already know, i.e. in the earlier mentioning of a ‘veranda’. This must be a re-introduction of an aspect of the report, viz., the place where the new windows are to be put in. It does not, however, seem designed to be an adequate complete utterance at the end of line 04: recipients may relevantly show their recognition of the room, but the projected role of it in the report is still unclear. If it is to become an action which others can respond to, something must be said which can relate the room to the renovation work. Line 05 provides a characterization of the room (‘used for all kinds of things’). This ties to earlier talk about the renovation, where it was said that Lone has to move ‘a lot of rubbish’ in her house. So, the characterization of the room provides a connection to the earlier topic of the renovation work in that it clariWes the relation to the ‘rubbish’ mentioned before. If the utterance does this clariWcation work, it may at this point invite recipients to display whether something has actually become clearer. Line 06 is such a recipient display of understanding. It paraphrases Lone’s characterization and provides a “formulation” of how Bente understands it. Such “formulations” make it relevant for recipients to conWrm or disconWrm

270 Jakob Steensig

the appropriateness of the “formulation” (Heritage and Watson 1979: 141V). Line 07 provides a conWrmation from Lone. In the subsequent lines (09 and Wrst part of 11) the characterization is continued and expanded, and Lone then moves on to a possible next step in her report when she introduces some ‘cupboards’ (end of line 11) which, apparently, play a role in the renovation and the moving of ‘rubbish’.9

Grammatical methods in the Danish excerpt In lines 04–07, I see the following points of possible grammatical completion: after brugt in line 05, at the end of line 05, and after lines 06 and 07. In general, utterance initial elements seem to be grammatically ambiguous. Then, as soon as the grammatical nature of the initial elements is available, it becomes possible to make quite reliable predictions about what can follow and how the turnconstructional unit can be completed. Let us look at this process in more detail. The Wrst word in line 04, den, can on its occurrence be either an independent pronoun (‘it’) or a determiner (‘that’/‘ this’ or ‘the’). Grammatical methods alone (including the relevant context) cannot determine which of these is the most likely function. The next word, der, shows that den must have been a demonstrative; the two words together mean ‘that’ or ‘that one there’. The addition of the noun, rum, results in a noun phrase (‘that room there’). At this point, however, there is not yet any “recoverable predicate” (Ford and Thompson 1996: 144; see also the deWnitions of grammatical methods above), and, therefore, something which can provide such a predicate is projected. Given the relevance of a further characterization of the room, the next item (den in line 05) must refer to the room and be the beginning of a clause about it.10 This opens the next grammatical slot of the clause, which in Danish is the Wnite verb. The verb, bli’r (‘gets’, ‘becomes’) projects a complement or a main verb. The participle brugt (‘used’) Wts in here. The construction, den der rum den bli’r jo brugt (‘that room there it gets, you know, used’), can be considered a possibly complete grammatical unit: it has a subject, ‘the room, it’, and a full predicate, ‘gets used’ and there is no further grammatical projection at this point. From a grammatical perspective, the next constituent til alve:hrdens ting (‘for all kinds of things’) is incrementally added to an already possibly complete utterance, providing a new point of possible grammatical completion at the end of line 05. This also turns out to be the actual point of speaker transition. Bente’s line 06 is grammatically possibly complete because it can be seen as


an “elliptical clause” (Ford and Thompson 1996: 143) which could be paraphrased as ‘is it a kind of storage room?’. Internally, the Wrst word, så’n, is grammatically ambiguous in this context: it can be an independent pronoun (meaning ‘that’s the way’), a “Wller” (‘kinda’, ‘like’), or a determiner indicating the beginning of a longer phrase (‘such’, ‘a kind of’). When the next word et (‘a’) is produced, it becomes clear that a noun phrase is being constructed. This projects a noun, and pulterrum (‘storage room’) is a noun that Wts in well with the characterization in line 05. Line 07, i:Ja:h (‘Yes’), is grammatically a complete lexical unit. It has the same “recoverable predicate” as line 06.

Prosodic methods in the Danish excerpt The pragmatic and the grammatical methods both project points of possible completion. Such projection of completion is less clear with the prosodic methods. Apparently, prosody contributes most importantly to turn-construction because it supports the grammatical and pragmatic methods: markedly stressed syllables (or “pitch peaks”) disambiguate the grammatical and pragmatic functions of speciWc elements, and “pitch resets” retrospectively conWrm or contradict the “unithood” of completed utterances. The following description attempts to follow the prosodic methods more closely. In line 03, Lone has cut oV her utterance on a high pitch and volume level. As she starts line 04, her pitch steps up a little more (“pitch reset”). The stress on this Wrst word (↑Den) supports the grammatical methods in that it disambiguates the grammatical function of the word den: this word is, namely, never stressed in my data when it turns out to be an article (meaning ‘the’), so prosody excludes this possibility. The pitch drops by 7 semitones (350 to 230 Hz) to ↓rum, where it reaches a low level compared to the main parts of lines 03 and 04. This is, thus, a “Wnal intonation” contour. Line 05 begins with a “pitch reset” and a stress on the Wrst word. The stressed syllable in alve:hrdens stands out because it is higher in pitch than the preceding stressed syllable and because of the non-phonemic lengthening of the vowel. This could be a marking of the relative importance of this word, and it could additionally be a “possible last accent”, in which case the transition relevance space may be open already after alve:hr°dens°. The relatively low volume on the last two syllables of this line could be a conWrmation of such an analysis in marking this as less important and possibly Wnal. There is no “Wnal intonation” at the end of this utterance.

272 Jakob Steensig

In line 06, a new person speaks, so here I rely on my subjective hearing in not indicating a “pitch reset”. The Wrst two words (°så’n et°) are spoken with low volume and with no substantial pitch Xuctuations. The lack of stress supports the grammatical determination of elements by excluding the possibility that så’n is an independent pronoun. The pitch climbs to a high peak on p:↑ulter, and then there is a drastic fall to the end of ↓rum (from 303–183 Hz, or 9 semitones), i.e. a “Wnal intonation” contour. The high pitch on p:↑ulter may have a “questioning” or “try marking” eVect (Sacks and SchegloV 1979: 18) and this may contribute to the marking of this utterance as asking for conWrmation. In line 07, the i:Ja:h; sounds as if it has a “pitch reset” (again I rely on my subjective hearing because a new person is speaking). The pitch shows a risefall contour on the word, but it does not fall to low. This may suggest “nonWnality”, and as it happens, the speaker goes on with her turn after a 0.3 second silence. In spite of this, I would not want to draw the conclusion that the prosodic projection of possible non-Wnality is crucial here. The relationship between diVerent intonation contours and the functions of response tokens is too complex to warrant any such conclusions.

Combination of methods in the Danish excerpt The three turn-construction methods examined give diVerent suggestions concerning the “products”, i.e. the possible complete units. They are summed up here and compared to the results from Ford and Thompson’s American English data (1996). Line 04 has prosodic completion without grammatical and pragmatic completion. This phenomenon is not infrequent in my Danish data, and Ford and Thompson also have examples of this (1996: 155 and 175, footnote 13). They see this as evidence that “intonational completion is in principle independent of both semantic and syntactic completion” (1996: 175, original emphasis), but they stress that “the vast majority of intonational completion points were in fact also points of both semantic and syntactic completion” (ibid.). In line 05, there is possible grammatical completion (after brugt) before any signalling of possible pragmatic or prosodic completion. This happens very frequently in my Danish data, as it does in Ford and Thompson’s (1996: 153V). At the end of line 05, there is pragmatic and grammatical completion. There are prosodic cues that support this completion (“possible last accent” followed by lowered volume), but there is no “Wnal intonation”. Points of pragmatic and grammatical completion, without clear “intonational” comple-


tion, occur frequently in my Danish data. Ford and Thompson have deWned pragmatic completion so that it presupposes intonational completion (1996: 150) and, therefore, they do not examine the possibility that “Wnal intonation” may be independent from pragmatic turn-construction methods. This is unfortunate, I believe, because such an a priori deWnitional combination of methods makes it impossible to examine the possibility that diVerent languages may employ the turn-construction methods in diVerent ways. Danish pragmatic completion certainly does not presuppose intonational completion: line 05 is not only a possibly complete pragmatic and grammatical unit, without intonational completion, but it is also treated as a possibly complete turn in that there is speaker transition at this point. In line 06, all the methods work together to create a possibly complete unit, but line 07 is again possibly complete pragmatically and grammatically without being so prosodically. In the step-by-step construction and projection process, there is an intricate interplay between the three methods examined. Pragmatic methods lay out the basic possibilities and constrain expectations within which grammatical and prosodic methods work. At the very beginning of utterances, projection is rather uncertain: a recipient will often have to wait for the second item in a unit to Wnd out what construction is being initiated. Prosody can mark beginnings as “new” and, thus, set oV units from preceding talk. And, because stress is important for the determination of grammatical roles, prosody can also help disambiguate between some of the possible grammatical functions of utterance-initial words. After the recognition of the initial elements, grammatical methods ensure a rather precise projection of possible trajectories and possible completions. Prosody (especially stress) contributes to this by helping with the identiWcation of grammatical items as they get produced. “Possible last accents” occur at points where both grammar and pragmatics have projected possible last items. Prosody, thus, contributes to the projection of up-coming possible completion through “possible last accents”, and this may be more important than “Wnal intonation”.

Turn-construction methods in a Turkish example The Turkish excerpt (8) is taken from a discussion between two Turkish women (Nuray and RFza) and a Turkish man, recorded in a Danish language

274 Jakob Steensig

school. In this fragment RFza has completed a story about how she got to talk to a Dane and found out that he was a nice person. In the excerpt, she attempts to give the upshot of the story and make it relate to an earlier topic, the cause of problems in the relationship between Turks and Danes in Denmark: (8) [JS:92:TR2:9] ((Turkish, face-to-face, 3 participants, semi-authentic)) 01

•za: Ri

02

•za: Ri

03

·hh Hani insan, (.) K:Onus ¸+tuk+ça yak- (.) ·hh you.know person talk+PPP+ADV approa‘You know, the more people talk and approa-’ •k+ça yaklas ¸+ti daHA iyi ol+uyor. approach.each.other+PPP+ADV more good become+PRS ‘approach each other the better it gets’ (0.7)

04

•za: Ri

05

•za: Ri

•si •ndaki •+yor; (.) Çünkü kars ¸i yani ANli because the.other.party so.to.say understand+PRS ‘Because the other party so to say understands’ •l °insan ol+acag •n+i •°.↓ >bil+i↑↑yo+n mu< ↓bu nAsi ˘+i

know+PRS+2S QPC this how person be+FPP+3S+OBJ ‘you know, what the person will be like’ 06

(1.8)

07

•za: Ri

08

•za: Ri

09

•za: Ri

10

Nuray: Zaten konus ¸+tug ˘ +un herkes ↑iyi; actually talk+PPP+2S everybody good ‘Actually, everybody who you talk to is nice’ •k+l[ar+i •n san+a • °dur+u(yor)° Nuray: =konus ¸+ma+di ya]banci

11

·hh Yani dil bil+me+dig ˘ +imiz+den;= ·hh in.other.words language know+NEG+PPP+1P+ABL ‘In other words, because we don’t know language(s)’ •+lar+nan= =↓bu yani danimarka+ynan d:animarkali

this in.other.words Denmark+COM Dane+PL+COM • =bizim ara+mi z+da+ki sog ˘ uk+↓luk. our between+1P+LOC+being cold+ness ‘there is coldness between Denmark the Danes and us’

talk+NEG+PPP+ [PL+2S you+DAT st]range remain+(PRS) ‘those who you [don’t talk to r]emain strangers to you’ 12

•za: Ri

01–03

R:

04 05 07 08–09 10 11 12

R: R: R: R: N: N: R:

[He:? Yeah

]

You know, the more people talk and approach each other the better it gets Because the other party so to say understands, you know, what the person will be like In other words, because we don’t know language(s), there is coldness between Denmark, the Danes, and us Actually, everybody who you talk to is nice those who you don’t talk to remain strangers to you Yeah


Pragmatic methods in the Turkish excerpt In this excerpt we get a series of evaluations of the story. Most of them have the same structure, with a summarizing statement Wrst and an evaluative term at the end. They all seem to be built to be complete when the evaluating item occurs. In lines 01–02, RFza has tried to close her story by giving a summary (suggesting that her story was about ‘talking to and approaching each other’) and an evaluation (‘the more we do of this, the better it becomes’). The evaluative item daha iyi (‘better’) makes the utterance into an evaluation. Often, such summaries and evaluations mark the possible closing of a story, and they invite recipients to show their understanding and appreciation of the story (cf. JeVerson 1978 on the sequential implicativeness of stories). In this case, however, there is no response: in line 03 there is a 0.7 second silence, and after this, RFza continues. Line 04 can, then, be seen as a pursuit of response, achieved with a speciWcation and substantiation of lines 01–02. It is a possibly complete pragmatic unit: it achieves a further speciWcation and gives would-be next speakers something more to respond to. Also in this case, the positively loaded evaluative item, the ‘understanding’, comes at the end of the unit. Again, there is no response, and the speaker goes on in line 05. Here, she Wrst adds a “recompletor” (Sacks, SchegloV and JeVerson 1974: 718), biliyon mu? (‘you know?’), but she does not wait for a response; without any delay she adds a further speciWcation of what the understanding could be (‘what the person will be like’). This is not a new evaluation; rather it is a continuation of the evaluation already produced in line 04. At the end of line 05, RFza’s turn is again possibly pragmatically complete. The lack of uptake in line 06 may make it relevant for RFza to produce something which is more “respondable-to”. The Wrst construction, yani dil bilmedi™gimizden (‘because we don’t know language(s)’) introduces a causal link. It states that the fact that ‘we don’t know language(s)’ is a cause, but the eVect has not been stated yet. Therefore, this is not a possibly complete action; it rather projects that the next utterance will describe the eVects of ‘our not knowing language(s)’. In lines 08–09, it is claimed that there is a coldness between the Danes and ‘us’ (the Turks in Denmark). This can be seen as an eVect of the cause stated in line 07, so we have a recognizable action, a cause and its eVect. It still contains an evaluation of the relationship which RFza addressed in lines 01–05, but this

276 Jakob Steensig

time the evaluation is done in negative terms (‘the coldness’). It is possible that this makes the evaluation a little more urgent to respond to than the previous less controversial statements. In line 10, Nuray responds. Line 10 is a possibly complete action; it seems to address the positive aspects of RFza’s evaluations. But in line 11, it turns out that line 10 was only the Wrst part of a contrastive construction because now Nuray counterposes the two relative constructions: konu¸stu™gun herkes (‘everybody who you talk to’) and konu¸smadFklarFn (‘the ones who you don’t talk to’). The evaluating items at the end of the two constructions, iyi (‘good’, ‘nice’) in line 10 and yabancF (‘strange(r)’) in line 11, are suYcient to make complete utterances, so duru(yor) (line 11) can be seen as incrementally added to an already possibly complete action.

Grammatical methods in the Turkish excerpt In most Turkish clauses, the grammatical role of the Wrst few words is not clear until the last words have been produced, and, furthermore, a recipient will most often have to wait till she hears the last suYxes of the very last word before she can Wnd out what kind of clause is being produced. The Wrst clause is completed in line 02. It ends with a verb, oluyor (‘becomes’), preceded by daha iyi (‘better’), i.e. here we have a predicate (‘becomes better’) where the Wnal suYx, +uyor (‘present tense’), makes the verb Wnite. The Wrst two verbs (‘talk’ and ‘approach each other’), by contrast, end in +ça which indicates that the verbs are non-Wnite and not possibly Wnal.11 The next clause, in line 04, ends with the Wnite verb anlFyor (‘understands’). Again, the suYx, +yor (‘present tense’) shows the Wniteness of the verb. Kar¸sFsFndaki (‘other party’) turns out to have been the subject. In line 05, biliyon mu (‘do you know?’) is a verb in the present tense followed by the question particle, mu. As such it constitutes a Wnite clause by itself. The next clause in line 05 ends with a participial verb, olaca™gFnF (‘his becoming’). The last suYx (+F indicating ‘object’) shows that the entire participial construction (‘what the person will be like’) must be the object of the verb from line 04, ‘understands’, and it can, thus, be seen as incrementally added to the possibly complete grammatical unit in line 04.12 In line 07, we have another participial verb, bilmedi™gimizden. We need to get to the personal suYx, +imiz+ (‘our’), and the case suYx, +den (‘from’, ‘because of’), before it becomes clear that the verb is not Wnite.13 This participial verb makes the clause-so-far into an adverbial constituent (‘because we


don’t know language(s)’) and we need to go to the end of line 09 to Wnd where its superordinate clause ends. It is the attributive suYx, +ki (‘being’), on the word before so™gukluk (‘coldness’) which shows that ‘coldness’ has an existential relationship to the rest: ‘there is coldness’. The other words in lines 09–10 specify this coldness ‘between the Danes and us’. And now we can see the relationship between the adverbial construction in line 08 and the rest: ‘because we don’t know language(s)’ — ‘ there is this coldness …’. Nuray’s lines 10 and 11 are built in much the same way as RFza’s: the predicates are iyi (‘(is) good’) and yabancF duru(yor) (‘remain strange(rs)’). And again we have non-Wnite verbs, konu¸stu™gun (‘who you talk to’) and konu¸smadFklarFn (‘those who you don’t talk to’) occurring before these predicates. Nowhere is there grammatical completion before the predicate has occurred. In all cases the crucial information about possible Wniteness comes at the very end, and nowhere is the Wniteness-indicating item grammatically projected before it occurs.

Prosodic methods in the Turkish excerpt Prosody does not seem to be central to the determination of the grammatical nature of initial elements in Turkish. But prosodic methods do important projective work through the information that they give about the status of whole contours in relation to the surroundings, and especially via the projection potential of “possible last accents”. In lines 01–02, the stress on daHA is stronger than the preceding ones, mainly because this pitch peak is accompanied by a marked increase in volume. This suggests that daHA is a “possible last accent”. After this there is “Wnal intonation” (a fall from 350 to 180 Hz, or around 11 semitones). Line 04 begins with a “pitch reset” and there is a slight pitch rise up to ANlFyor, but the marked stress on this word seems to be achieved with a louder volume rather than with high pitch. The line ends with only a slight pitch drop, so here we may have a “possible last accent”, but no “Wnal intonation”. Line 05 begins with the rapidly spoken bili↑↑yon mu, where the pitch goes up to a very high part of the speaker’s register (480 Hz), after which the rest of the line is spoken in the lower part of the speaker’s register. Both pitch and volume fall from the stressed syllable on nAsFl to the end of the line. The softness and low pitch create a contrast to the high pitched bili↑↑yon mu and may further contribute to a hearing of this as an “afterthought”, an incremen-

278 Jakob Steensig

tally added contribution conveying background (or already implied) information (see Auer 1996: 82V). In line 07, RFza speaks again in a louder voice and in the higher part of her pitch register. The pitch movements on bilmedi™gimizden seem too modest to be heard as a “possible last accent” and a pitch fall from +di™g to +den does not go down to the low part of the register, so there is no “Wnal intonation”. After this, there is a further pitch fall to bu, and the rest of line 08 is spoken with low volume and in the low part of the register. This seems to suggests a shift to a diVerent activity or a diVerent level of importance. Also the next line is spoken in the lower part of the speaker’s pitch register, but now with medium volume. The continuation of the comparatively low pitch may contribute to a hearing of this whole passage (from ↓bu in line 08 to the end of line 09) as one coherent utterance. On the last word in line 09 there is a marked pitch fall suggesting “Wnal intonation”. The prosodic structure of Nuray’s lines 10–11 resembles that of lines 01–02 and 04 above: there is stress on the participial verbs but it is less strong than the stress on the possible predicates, ↑iyi and yabancF. The tendency for “possible predicates” to have “possible last accents” and for verbal participles which do not have completion potential not to have such accents, seems to be quite a general feature in my Turkish data.

Combination of methods in the Turkish excerpt In this Turkish excerpt, there is some divergence between pragmatic, grammatical and prosodic methods concerning the projection of possible completion. In line 02, pragmatics and “possible last accent” suggest that the transition space opens already at daha iyi, but one has to wait till the end of the line for the grammatical and “Wnal intonation” projection of possible completion. In line 04 the diVerent methods agree that anlFyor is a possible last element; it carries a “possible last accent”, but there is no “Wnal intonation”. In line 05, all methods agree that there are possible completion points after both biliyon mu and olaca™gFnF. In line 07–08 all methods agree that neither dil bilmedi™gimizden nor bu yani can be the completion of the unit and they all point to possible completion at so™gukluk at the end of line 09. In lines 10–11 all methods agree that the verbal participles cannot be Wnal, and pragmatics, grammar and “possible last accents” point to possible completion after the adjectival predicates. In the Turkish data, prosodic and pragmatic methods seem to project possible up-coming completion before grammatical methods do. “Possible last


accents” occur on the last stressed syllable in Wnite clauses and possibly complete pragmatic units, and they occur a few syllables before the grammatical morphemes that show possible Wniteness. Tanaka (1999) describes turn-construction and completion in Japanese, a language which shares relevant typological features with Turkish. One of her observations is: both the syntactic nature and the interactional import of an utterance can remain essentially indeterminate until the utterance approaches completion. (…) As a consequence, the projectabilitiy of an emergent turn may be vastly limited prior to turn-completion. (Tanaka 1999: 109)

This description holds true for Turkish as well: grammatical information concerning Wniteness comes at the end of possible utterances and the same goes for the morphemes marking clausal and pragmatic functions. But as we have seen, prosody, especially “possible last accents”, may signal the pragmatic and grammatical status of an utterance earlier than this. This may mean that recipients do not need to wait for a grammatical signal of possible completion or for “Wnal intonation”, but they can rely on “possible last accents” — together with situated pragmatic knowledge — in predicting possible completion before it actually occurs.

Conclusion In this paper it has been suggested that pragmatic, grammatical and prosodic methods contribute diVerently to turn-construction in Danish and Turkish. In Danish, grammatical methods facilitate a projection from early on in an utterance, whereas in Turkish grammatical information about possible completion does not come until the very end of the clause; consequently, grammar does not have the same projection potential. It has further been suggested that prosody partly “compensates” for this, in that up-coming possible completion is signalled prosodically before it is accomplished grammatically. In both languages, one important contribution of prosody is that “possible last accents” show where the transition relevance space can open and possible completion is near. I have not yet found cases in my Turkish data where a speaker clearly marks up-coming possible completion prosodically (with a “possible last accent”) and then “contradicts” this projection by subsequently producing a non-Wnite verb. But it is important to remember that an observed correspon-

280 Jakob Steensig

dence in the use of diVerent linguistic methods does not imply automatic combination. It is conceivable that Turkish speakers may exploit the methods, and, for example, use them to retrospectively change projections. This will be something to look for in future research. In the section concerning “initial observations” above, it was noted that Turkish conversationalists are able to use the projection of completion before actual completion is reached. This is, however, only a preliminary impression. I have not examined the quality of early onsets, or whether diVerent kinds of early onsets, collaborative completions, etc. are less frequent in Turkish than in Danish. I would not be surprised if on closer inspection one could Wnd diVerences in the interaction between speakers and recipients at diVerently projected points in turn-constructional units. After all, the situation is diVerent for the participants: in Danish a speaker has committed herself to producing a certain grammatical trajectory. Such a commitment can, of course, be changed, but this calls for speciWc types of repair. In Turkish the commitment to a speciWc type of grammatical item is less clear, and it is conceivable that changing something which is prosodically, but not grammatically, projected may be easier or may call for diVerent methods. Fox, Hayashi and Jasperson (1996) examine how patterns of self-repair in English and Japanese data reXect and inXuence diVerent grammatical relations, and Tanaka (1999) shows that in Japanese grammatical structure, which is in important aspects similar to that in Turkish, favours turn-taking behaviour which diVers from that found in English interaction. More studies like these are needed in order to establish how diVerences in turn-construction methods inXuence — and are inXuenced by and exploited in — interaction. Although the results of the present study are preliminary and give rise to more questions than they answer, I believe that one methodological conclusion can be drawn: when trying to deal with the intersection between turntaking and linguistic resources, it is important to keep turn-construction methods apart. This becomes especially important as interactional linguistics expands its interests to more, and more diverse, languages. Linguistic methods may do diVerent interactional jobs in diVerent languages, and a description of both linguistic practices and of the organization of action and interaction will have to take this into account.


Notes 1. This paper is a completely revised version of a paper given at the International Pragmatics Conference in Reims, July 1998. I am very grateful to the editors of the present volume, Elizabeth Couper-Kuhlen and Margret Selting, and to Harrie Mazeland, for extremely insightful and useful comments on diVerent parts and versions of this paper. Thanks also to Mediha Can, Abdulgani ÇFtFrFkkaya, Volker Hinnenkamp and Jens Normann Jørgensen, who expended a lot of time and patience to explain their understandings of the Turkish examples to me and to guide me in my interpretation of them. A large group of cotranscribers, colleagues, consultants, and conversation participants are thanked collectively for their help. Also thanks to Althea Ryan, who corrected my English. I am, of course, solely responsible for any remaining mistakes. 2. One could also talk about turn-construction “practices”, “resources”, “patterns”, “levels” or “structures”. I prefer the term “methods” because it highlights both the procedural aspect and the systematic, structural nature of turn-construction. 3. All the Turkish data were transcribed by the author after an initial transcription had been provided by a native speaker of Turkish. 4. Migros is a shop. Apparently, it is not unusual for the participants to go shopping across the border. 5. This seems to be in line with what Couper-Kuhlen (1996) describes as “declination reset”. Other factors may be involved too (see Selting 1995: 67–70). 6. Grønnum’s studies (1985, 1986, 1988a, 1988b, 1992) are the most thorough investigations of Danish intonation, but this work is based on experimental data. Grønnum’s research shows that in Danish there are no Wnal “tonic” syllables with identiWable “Wnal” or “non-Wnal” intonation; rather the functional character of each utterance is expressed in the overall pitch contour where all stressed syllables have equal weight. Grønnum stresses, however, that the sentences she has used are “pragmatically neutral” and without emphatic stress, and this makes it diYcult to compare her data with the conversational data I am examining. Nash (1973) uses a similar type of data to examine Turkish intonation. She identiWes both local “Wnal” and “non-Wnal” patterns and more global “functional motifs”. I have been inspired by her approach, but, again, the data are diYcult to compare. 7. My use of punctuation marks is an adaptation of the “GAT” conventions (Selting et al. 1998: 101). The main diVerence is that my transcripts do not include a device for indicating “level” intonation (“gleichbleibend”, i.e. neither falling nor rising). I have followed “GAT” in making it explicit that the punctuation marks designate the pitch movement from the last accented syllable to the place where the marker is written. Thanks to Margret Selting and Elizabeth Couper-Kuhlen for their assistance and persistence in helping me become more explicit and clear on this point. Needless to say, I am solely responsible for the use I have made of the editors’ help and advice. 8. Thanks to Harrie Mazeland for invaluable suggestions for the analysis of the sequential relations in this and the next excerpt. I am, of course, responsible for the use of his advice.

282 Jakob Steensig

9. Actually, this next step is again postponed by a long side sequence concerning the length of the cupboards etc. and ending several minutes later when Lone makes a new connection between the cupboard and the renovation of the house. 10. Some grammarians refer to the use of longer phrases (such as den der rum) before and “outside” a clause as “left dislocation”: the noun phrase has been “moved to the left of the clause” and only a “pronominal copy” is left inside the clause. In Danish conversational language, however, it would be misleading to think of longer phrases as the “proper” Wrst constituents of clauses; a construction with a longer phrase outside the clause and a “pronominal copy” inside is much more frequently used. 11. By themselves, konu¸stukça and yakla¸stFkça mean ‘when talking’ and ‘when approaching each other’; the comparative daha (‘more’) introduces the comparison, which has made me translate ‘the more people talk and approach each other’. 12. Its “standard position” would have been before the Wnite verb in line 03, thus, Kar¸sFsFndaki yani nasFl insan olaca™gFni anlFyor (hypothetical). 13. The suYx, +di™g (with a mute ™g), makes the verb into a participle, but on its own it could have been the preterite tense, Wnite, suYx, +di.

References Atkinson, J. M. and Heritage, J., eds. 1984 Structures of Social Action: Studies in Conversation Analysis. Cambridge: Cambridge University Press. Auer, P. 1996 “On the prosody and syntax of turn-continuations”. In Prosody in Conversation: Interactional Studies, E. Couper-Kuhlen and M. Selting (eds.), 57–100. Cambridge: Cambridge University Press. Auer, P. and Fearn, J. 1992 Beiträge zur Fremdsprachenvermittlung: Türkische Alltagskonversationen: Eine Textsammlung zum gesprochenen Türkisch [Contributions to Foreign Language Communication: Turkish Everyday Conversations. A Text Collection of Spoken Turkish]. Konstanz: Sprachlehrinstitut der Universität Konstanz. Couper-Kuhlen, E. 1996 “Intonation and clause combining in discourse: The case of because”. Pragmatics 6: 389–426. Couper-Kuhlen, E. and Selting, M., eds. 1996 Prosody in Conversation. Cambridge: Cambridge University Press. Ford, C. 1993 Grammar in Interaction: Adverbial Clauses in American English Conversations. Cambridge: Cambridge University Press. Ford, C., Fox, B. A. and Thompson, S. 1996 “Practices in the construction of turns: The ‘TCU’ revisited”. Pragmatics


6: 427–454. Ford, C. E. and Thompson, S. A. 1996 “Interactional units in conversation: syntactic, intonational, and pragmatic resources for the management of turns”. In Interaction and Grammar, E. Ochs, E. SchegloV and S. Thompson (eds.), 134–184. Cambridge: Cambridge University Press. Ford, C. and Wagner, J., eds. 1996 Special issue on Grammar and Interaction, Pragmatics 6 (3). Fox, B. A., Hayashi, M. and Jasperson, R. 1996 “Resources and repair: a cross-linguistic study of syntax and repair”. In Interaction and Grammar, E. Ochs, E. SchegloV and S. Thompson (eds.), 185–237. Cambridge: Cambridge University Press. Goodwin, C. 1979 “The interactive construction of a sentence in natural conversation”. In Everyday Language: Studies in Ethnomethodology, Psathas, G. (ed.), 97– 121. New York: Irvington. 1981 Conversational Organization: Interaction between Speakers and Hearers. New York: Academic Press. 1986 “Gesture as a resource for the organization of mutual orientation”. Semiotica 62 (1/2): 29–49. Goodwin, C. and Goodwin, M. H. 1986 “Gesture and coparticipation in the activity of searching for a word”. Semiotica 62 (1/2): 51–75. 1992 “Assessments and the construction of context”. In Rethinking Context, A. Duranti and C. Goodwin (eds.), 147–189. Cambridge: Cambridge University Press. Grønnum, N. 1985 “Intonation and text in Standard Danish”. Folia Linguistica 17: 187–220. 1986 “Sentence intonation in textual context — supplementary data”. Journal of the Acoustical Society of America 80: 71–80. 1988a “Standard Danish intonation”. ARIPUC 22: 1–23. 1988b “Stress group patterns, focus signalling and sentence intonation in two regional Danish standard languages”. ARIPUC 22: 145–196. 1992 The Groundwork of Danish Intonation: An Introduction. Copenhagen: Museum Tusculanum. Heath, C. 1992 “Gesture’s discreet tasks: Multiple relevancies in visual conduct and in the contextualisation of language”. In The Contextualization of Language, P. Auer and A. di Luzio (eds.), 101–127. Amsterdam and Philadelphia: John Benjamins. Heritage, J. and Watson, D. R. 1979 “Formulations as conversational objects”. In Everyday Language: Studies in Ethnomethodology, G. Psathas (ed.), 123–62. Hillsdale, NJ: Erlbaum.

284 Jakob Steensig

JeVerson, G. 1978

1986 Kendon, A. 1981

“Sequential aspects of storytelling in conversation”. In Studies in the Organization of Conversational Interaction, J. Schenkein (ed.), 219–48. New York: Academic Press. “Notes on ‘latency’ in overlap onset”. Human Studies 9: 153–183.

Nonverbal Communication, Interaction and Gesture. The Hague: Mouton. Local, J., Kelly, J. and Wells, W. H. G. 1986 “Towards a phonology of conversation: turn-taking in Tyneside English”. Journal of Linguistics 22: 411–437. Nash, R. 1973 Turkish Intonation: An Instrumental Study. The Hague and Paris: Mouton. Ochs, E., SchegloV, E. and Thompson, S. A., eds. 1996 Interaction and Grammar. Cambridge: Cambridge University Press. Sacks, H. 1992 Lectures on Conversation. Vol. 2. Oxford UK and Cambridge USA: Blackwell. Sacks, H. and SchegloV, E. 1979 “Two preferences in the organization of reference to persons in conversations and their interaction”. In Everyday Language: Studies in Ethnomethodology, G. Psathas (ed.), 15–21. New York: Irvington. Sacks, H., SchegloV, E. and JeVerson, G. 1974 “A simplest systematics for the organization of turn-taking for conversation”. Language 50: 696–735. SchegloV, E. 1984 “On some gestures relation to talk”. In Structures of Social Action, J. M. Atkinson and J. Heritage (eds.), 266–96. Cambridge: Cambridge University Press. 1987 “Analysing single episodes of interaction: an exercise in conversation analysis”. Social Psychology Quarterly 50: 101–14. 1996 “Turn organization: one intersection of grammar and interaction”. In Interaction and Grammar, E. Ochs, E. SchegloV and S. Thompson (eds.), 52–133. Cambridge: Cambridge University Press. 1998 “ReXections on studying prosody in talk-in-interaction”. Language and Speech 41: 235–63. SchegloV, E., Ochs, E. and Thompson, S. A. 1996 “Introduction”. In Interaction and Grammar, E. Ochs, E. SchegloV and S. Thompson (eds.), 1–51. Cambridge: Cambridge University Press. Selting, M. 1995 Prosodie im Gespräch: Aspekte einer interaktionalen Phonologie der Konversation [Prosody in Conversation: Aspects of an Interactional Phonology of Conversation]. Tübingen: Niemeyer.


Selting, M. 1996

“On the interplay of syntax and prosody in the constitution of turnconstructional units and turns in conversation”. Pragmatics 6: 371–388. 2000 The construction of units in conversational talk. Language in Society 29: 477–517. Selting, M., Auer, P., Barden, B., Bergmann, J., Couper-Kuhlen, E., Günthner, S., Meier, C., QuasthoV, U., Schlobinski, P. and Uhmann, S. 1998 “Gesprächsanalytisches Transkriptionssystem (GAT)” Linguistische Berichte 173: 91–122. Steensig, J. 1994 “På tur i talens sprog — om syntaks og taleture i samtaler” [Touring the spoken language — on syntax and turns at talk in conversations]. NyS 18: 63–83. forthcoming Bidrag til en Interaktionel Lingvistik [Contributions to an Interactional Linguistics]. Aarhus: Aarhus University Press. Streeck, J. and Hartge, U. 1992 “Previews: Gestures at the transition place”. In The Contextualization of Language, P. Auer and A. di Luzio (eds.), 135–157. Amsterdam and Philadelphia: John Benjamins. Tanaka, H. 1999 Turn-taking in Japanese Conversation. Amsterdam and Philadelphia: John Benjamins.

Appendix The transcription of the actual talk in the text line is in italics. In addition to Gail JeVerson‘s conventions, the following conventions are used in the text line: initial capital letters indicate “pitch reset”, i.e. a new unit sounds as if it is beginning at this point, and a phonetic mark for a glottal stop (‘) is used when there is a glottal onset. In the Turkish talk, morphemes are separated by a plus sign (+). This convention is also used in the interlinear glosses. Only morphemes that are deemed of potential importance to the analysis are indicated. In the glosses and in the translation, dots are used between English words in cases where one word in Danish or Turkish has to be translated with several English words. The following abbreviations are used in the interlinear glosses for Turkish morphemes: PPP ADV

past participle suYx adverbial suYx

286 Jakob Steensig

FPP PRS PRT

1s 2S 3S 1P PL OBJ DAT LOC ABL COM QPC NEG

future participle suYx present tense suYx preterite tense suYx (past tense meaning) 1st person singular suffix (‘I’, ‘my’) 2nd person singular suYx (‘you’, ‘your’) 3rd person singular suYx (‘he’/‘ she’/‘ it’, ‘him’/‘ her’, ‘his’/‘ her’/‘ its’) 1st person plural suYx (‘we’, ‘us’, ‘our’) plural suYx object case suYx (used for deWnite objects) dative case suYx (‘to’, ‘for’) locative case suYx (‘in’, ‘on’) ablative case suYx (‘from’, ‘of’) comitative suYx (‘with’) question particle negative suYx (‘not’)

An exploration of prosody and turn projection in English conversation* Barbara A. Fox University of Colorado at Boulder

The present study explores the possibility that pitch-accented syllables which project upcoming turn completion may be phonetically diVerent from pitchaccented syllables which do not so project. This possibility was raised by SchegloV (1987b) with regard to “peak” accents. This study reports on various phonetic details of the pitch peaks of 43 utterances, which were culled from naturally-occurring American English conversations. Each utterance selected had two pitch peaks in the predicate of its syntactic structure. The goal of the study was to determine if, in fact, there were phonetic diVerences between “non-last” accented syllables (that is, those which do not project upcoming turn completion) and “last” accented syllables (that is, those which do project upcoming turn completion). A variety of phonetic details were selected, most of which are implicated in the literature on prosody in the creation of prominence. It was hypothesized that “last” accents would be phonetically more prominent than “non-last” accents. All of the Wndings except one fail to support this hypothesis. While several of the phonetic properties examined showed trends in the expected direction, only one, Duration, showed a statistically signiWcant diVerence between “non-last” and “last” accents.

1.

Introduction

Although the Sacks, SchegloV and JeVerson paper on turn-taking is now a quarter of a century old (Sacks, SchegloV and JeVerson 1974), research into the linguistic underpinnings of the proposed mechanisms for turn-taking is a fairly recent enterprise. For example, aside from Goodwin’s early work (e.g. C. Goodwin 1979, 1981), Ford and Thompson (1996) is one of the Wrst studies I know of to re-examine the SSJ turn-taking model from a more rigorously linguistic perspective. The virtual explosion of research within the last few

288 Barbara A. Fox

years on this topic (e.g. Ford, Fox and Thompson 1996a and b, to appear; Fox, Hayashi and Jasperson 1996; Hayashi, Mori and Takagi, to appear; Hayashi and Mori 1998; Lerner and Takagi 1999; Ford and Mori 1994; Tanaka 1996, 2000; Selting 2000; Couper-Kuhlen and Selting, eds. 1996; SchegloV 1996, 1998; Lerner 1991, 1996; Ochs, SchegloV and Thompson, eds. 1996) is thus testimony to the growing awareness among linguists that turn-taking is both organized by and organizing of linguistic practices and hence deserving of linguists’ attention. Nevertheless, in spite of this growing attention to turn-taking (and other facets of conversational organization), there remain large gaps in our understanding of exactly how turn-taking is achieved. The need for further linguistic research is especially evident in the area of prosody and turn-taking. While quite a bit has been done on the prosody of turn-Wnal syllables, or even Wnal words (e.g. Duncan 1974; Duncan and Fiske 1977; Wells and Peppe 1996; Lehiste 1979; Local, Wells and Sebba 1985; Local, Kelly and Wells 1986; Berkovits 1984a, b), very little has been done on how prosody might project upcoming turn completion from earlier in the turn. Perhaps not surprisingly, it is SchegloV (cf. SchegloV 1987a and b, 1996, 1998) who has made the only proposal I know of with regard to turn-taking and prosody before the last syllables of the turn. SchegloV (1987a) proposes the following relationship between prosody and turn-taking in discussing the last accented syllable in the utterance he’s about the only good regular out there: The relevance of a pitch peak of this sort — but mark, not of all pitch peaks — is that it marks the enhanced likelihood that the next possible completion of the turn constructional unit will be an actually intended turn completion. That is, the developing grammatical structure of an utterance in the course of its production is potentially compatible with alternative points of possible completion. Pitch peaks, and their suppression, are one means by which speakers can indicate which syntactically possible completions are built to be completions on this occasion and which not. A pitch peak can thus project intended turn completion at the next grammatically possible completion point. (1987b, transcribed from Xerox Parc lecture)

When asked during the question and answer period after this public lecture why the pitch peak on good doesn’t also project in this way, SchegloV said he didn’t know. Clearly, if SchegloV is correct, some pitch peaks project upcoming completion and others do not. But we do not yet know what distinguishes the former from the latter. SchegloV’s proposal thus opens up a fascinating arena of research which

Prosody and turn projection in English 289

very much requires contributions from linguists. The larger arena asks: How are prosodic practices in a given language implicated in turn projection? The more narrow focus of SchegloV’s proposal asks: What distinguishes pitch peaks (a term which will be discussed more fully below) which project upcoming turn completion from pitch peaks which do not so project? It is the goal of the present paper to begin an exploration of the more narrowly focused question. This question could have been approached in a number of diVerent ways. In the present paper I have adopted one particular approach to the question, one which naturally arises from the particular utterance for which SchegloV’s proposal was oVered. This approach compares multiple pitch peaks within a single utterance to determine if there are phonetic diVerences between what we can for now call the “non-projecting” pitch peak(s) and the “projecting” pitch peak. Other approaches to the question would have yielded diVerent insights. For example, it would have been possible to make a collection of utterances with a single pitch peak which appears to project upcoming turn completion at the next place of grammatical completion, and contrast those cases phonetically with cases in which the single pitch peak does not appear to project upcoming completion (in spite of being at or near the end of the turn). Or, it would have been possible to make a collection of utterances in which each exhibits a pitch peak that the recipient analyzes as projecting upcoming turn completion but where the speaker displays that analysis to be incorrect; a phonetic study of such pitch peaks could then be compared to pitch peaks for which the recipient displays an understanding of “no upcoming completion.” Contrasting pitch peaks with other types of accented syllables could also be done. The current study focuses exclusively on prosody in order to address a fundamental, and obviously naive, hypothesis, which is that the phonetic realization of a syllable indexes projection by itself, without regard to the syntactic, semantic, or sequential position in which that syllable is placed. That the study Wnds little evidence to support this hypothesis — and, indeed, may be compatible with quite a diVerent hypothesis regarding the relationship of prosody to projection — is hardly surprising, given what we know about the complex constellation of practices involved in turn projection (see, for example, Ford, Fox and Thompson 1996a). Nonetheless, there were two motivations for exploring this hypothesis in detail: Wrst, there is a great deal of literature in the Weld of intonation which holds that there are two kinds of accented syllables, nuclear and pre-nuclear accents — which as I will discuss below may be seen to bear a relationship to “projecting” and “non-projecting”

290 Barbara A. Fox

accents — and that these nuclear and pre-nuclear accents exhibit distinct phonetic realizations. This claim would seem to strongly support the possibility that speakers can index through prosody alone whether an accented syllable is designed to be heard as the last accented syllable of the turn or as not the last accented syllable of the turn. Second, it was necessary to explore the role of prosody by itself so that the next step can be taken, which is to explore the phonetic realization of syllables in diVerent syntactic, semantic, and sequential locations. As we will see, the results of the study suggest that phonetic realizations of accents seem to show too much variation to index projection by themselves. Rather, they are very likely heard in and through the syntactic, semantic, and sequential positions in which they occur. For example, a very prominent syllable may not be heard as being a “last” accent (thereby projecting upcoming completion of the turn) if it occurs in a position which is not syntactically or semantically possibly complete. In such cases, the accent may not be “interrogated” (SchegloV, pc) for whether it is doing projection or not, which leaves that particular prosodic movement free to accomplish other work (showing contrast, doing repair, etc.). The present study asks the following question: Are “last” and “non-last” pitch peaks phonetically distinct (in particular, are “last” pitch peaks more prominent than “non-last” pitch peaks?)? That is, can a recipient tell from hearing the phonetic details of a particular pitch peak that that syllable will not be the last accented syllable in the turn? Or that it will be the last accented syllable in the turn? The motivation for asking the question in terms of “last” and “non-last” (versus, for example, “projecting” and “non-projecting”) will be discussed in detail in Section 2. Let us now turn to developing the questions a bit more thoroughly. 1.1

Pitch peaks and turn projection

SchegloV’s original proposal for the relationship between pitch peaks and turn projection was developed in connection with the analysis of a particular utterance. It may therefore be helpful here to consider that utterance in more detail (the utterance in question is given below in bold face): (1) Curt: Mike:

Dz he go out there pretty regular? (some turns omitted) Generally evry Saturdee.

Prosody and turn projection in English 291

Phyllis: He wins js about evry Saturday too. ==> Curt: He- he’s about the only regulartookyuu-< ] ‘Tokyu-’

5

(.)

6

Akiyo: animek [kusu san. ] ‘Animekkusu.’

7

Takie:

8

Akiyo: U: [:N.] ‘YEAH.’

9

Takie:

10

Kaori: =nani sore. what that ‘What’s that?’

11

animek [kusu Takie: tookyuu Tokyu Animekkusu ‘Tokyu Animekkusu…’

12

Akiyo:

13

Takie: = [kaisha.] company ‘Company.’ [ ] Kaori: = [°a::::: ]::.°= oh ‘°Oh:::::::°’

14

[ ] [>ANIMEKKU ]SU< ‘>ANIMEKKUSU is well formed if and only if there is a string C such that the string < OC or R > is well formed, where C is a completion of the constituent directly dominating the last element of O (or is to be deleted if the last element is itself a connective such as or or and) (Levelt 1989: 486).

The main idea behind this rule is that there is a syntactic parallelism between repairs and coordinations. This idea is not only syntactically but also functionally intriguing, because the coordinating conjunction oder (‘or’) also serves as a repair initiator in everyday German conversation: (16) Kurzsichtig (short-sighted) →

01

B:

02

C:

[da war imma ] nur einsfünf oder einssiebefünf [‘there was always ] only one five or one seven five [ aba ich ] ‘but I’ ] [

(17) China 31 01

02 →

03

04

T:

und die REden anscheinend noch en portuGIEsisch, (0.8) ‘and they apparently still speak a kind of Portuguese’ das verwAndt is: ‘that is related to’ oder das: (.) ja (.) äh- das halt Eher DEM portugiesisch ähnelt, (0.3) ‘or that (.) yeah (.) uh- that resembles more the Portuguese’ vom SIEBzehnten JahrHUNdert; ‘of the seventeenth century’

Syntax of same-sentence self-repair 387

Together with the qualification that the completion C can be dropped if the phrase containing the repairable has been completely realized before the initiation of repair was started, well-formed self-repairs correspond to well-formed oder-coordinations. As example (18) shows, the Well-Formedness Rule for Repairs (see Levelt 1983: 84) generates a considerable amount of variation for the accomplishment of self-repair:

(18) The ball rolls to the left side, uh

 (a)  (b)  (c)  (d)  (e)  (f) 

corner left corner the left corner to the left corner rolls to the left corner the ball rolls to the left corner

According to Levelt’s rule, all the variants of (18) should be equally wellformed self-repairs, because if the repair particle uh is replaced by the coordinating conjunction or, the results are well-formed coordinations. But Levelt’s predictions do not correspond either to intuition-based judgements (cf. e.g. 15', which corresponds to 18f) or to the self-repairs that participants actually perform in German everyday interaction. Real-time data, as in (4) and (9) to (15), overwhelmingly exhibit a scope of recycling as in Figure 1. The scope of recycling demonstrated in Figure 1 is determined by a (sentence) grammatical category — the functional head of the repairable. After selfinitiation, the actual accomplishment of self-repair (including the “bit of talk”

initi ation of repair repairable head

der arbeitet jeden morgen bis- (.) he works every morning till

s y ntactic loop

Figure 1. Scope of recycling

accomplishment

jeden tach bis um elf every day till eleven

388 Susanne Uhmann

which is repeated in the syntactic loop) systematically and regularly starts with the functional head immediately c-commanding the repairable. We can therefore state a Head Rule for self-repairs in German. The Head Rule is formulated as a preference rule (according to Hawkins 1994: 77ff), in order not to exclude the possibility that it can be violated in naturally occurring episodes of verbal interaction. But violations should be less acceptable and less frequent than self-repairs that respect the Head Rule. Head Rule Self-repairs are preferred if the accomplishment of repair starts with the repetition of the functional head which immediately c-commands the repairable. Some brief comments may be helpful for the clarification of the concepts ‘functional head’ and ’immediately c-command’. The assumption that all phrases are endocentric, which means that they are projections of some head element, is a central principle in X-bar theory (see e.g. Haegeman 1994). Functional heads (see Figure 2) are those non-complex constituents in head position which (simplifying somewhat) carry grammatical meaning, i.e. the preposition in PP (prepositional phrase) or the determiner in a nominal phrase, which therefore actually is a DP (determiner phrase). As Figure 2 shows, all functional heads also have an obligatory complement and an optional specifier: XP

Spec

X

0

X (head)

YP (complement)

Figure 2. Structural position of functional heads

Applied, for example, to PPs and DPs, this generalized structure results in the following more specific structures (in these cases the Spec-Position is normally empty):


PP

DP

S pec

S pec

D

D0 (head)

NP Adj

Ø

ein a

blaues blue

P

P0 (head)

N

N

dreieck triangle

Ø

in in

Hamburg Hamburg

Figure 3. Structure of DPs and PPs

The definition of c-command is: X c-commands Y if Y is (a part of) a sister constituent of X; a functional head X c-commands Y immediately if there is no intervening functional head that c-commands Y. For example, in Figure 3 D0, the indefinite article (ein), is the functional head that immediately c-commands N (Dreieck), because the latter constituent is part of NP, the sister of D0, and there is no closer functional head c-commanding it. The Head Rule enables us to explain why the self-repair in the constructed example (15') is ill-formed: the repair does not start with the functional head (in italics) which immediately c-commands the repairable: (15') 01 →

X:

02

die dame die mir damals die schecks gegeben hat- äh ‘the lady who gave me the checks at the time uh’ die dame die mir damals die barschecks gegeben hat ‘the lady who gave me the checks for cash at the time’

In the same way, (18a, b, e, f) are excluded. (Cases like (18c, d) are discussed below.) The Head Rule describes self-repairs like excerpts (4) and (9) to (15) adequately, but it still has some shortcomings. These come to light if we look at example (8) — repeated here for convenience: (8)

Hundertfünfzig 8 01

X:

aber: irgendwie (.) die Tatsache daß man WEISS, ‘but somehow (.) the fact that you know

390 Susanne Uhmann

02 →

dass es fürs exAmen is (.) ‘that it is for the exam’ verdirbt einem:: =sel- sämtliche spoils INDEF. ART. DAT. col- complete ‘spoils all your pleasure’

03

freude pleasure

In contrast to the self-repairs discussed so far, in excerpts (8) and also (19) to (22) the repairable itself is a functional head (shown in italics in the German lines): (19) (from Hoffmann 1991: 108) →

01

X:

also der kinder warn na- die kinder warn natürlich n ansatzpunkt ‘well the children were of- the children were of course a starting point’

(20) China 41 01

T:

02 03 →

04

ich muß dir heut abend auch noch die adresse durchgeben ‘I’ll have to give you the address too this evening’ ich hab die verschlampt ‘I‘ve lost it’ aber die britta in freiburg hat sie ‘but Britta in Freiburg has it’ die hol (0.3) ich- ruf ich heut abend an (0.47) ‘I’ll get (0.3) I’ll give her a call this evening’

(21) Aktienberatung →

01

M:

aber SON wErt liegt nat- legt natürlich am ersten märz ZU. ‘but such a stock lies of- goes up of course on March 1st’

(22) (from Hoffmann 1991: 103) 01 →

02

X:

äh diese medikamente ‘uh these medicines’ die machen- die verändern die atmung und das sprechen etwas ‘they make- they change your breathing and your speaking a bit’

Let’s start with a closer syntactic examination and an interlinear glossing of (19). The repairable is the determiner der (D0, the functional head of the DP) (see also Figure 3). The self-repair does not start with the functional head immediately c-commanding the repairable — as the Head Rule predicts — but with the ‘new’ functional head. As the Spec-position is empty in these DPs, speakers have no choice and start their self-repair with the functional head, the corrected determiner, itself. The same holds for example (8). In the remaining cases (20)–(22) the repairable is the head of a complementizer-phrase (CP). Again the syntactic loop does not start as predicted with the head immediately c-commanding the repairable. But here, speakers obvi-


CP

C

S pec C0

DP

IP

D

S pec

D0 (head) N (19)

Ø

der

kinder warn children were die kinder warn ART. Nom.Pl.N. children were ART. Gen.Pl.N.

naofnatürlich n ansatzpunkt of course a starting point

Figure 4. Syntactic structure of (19) CP

C

S pec

DP

(20)

die REL.PRO.

(21)

(22)

C0 (head)

hol get ruf call

IP

ichIich heut abend an I this evening

liegt nason Wert such a stock lies oflegt natürlich am ersten märz zu goes of course on first march up die machenREL.PRO. makedie verändern die Atmung und das S prechen etwas the breathing and the articulation a bit REL.PRO. change

Figure 5. Syntactic structure of a complementizer phrase (CP)

392 Susanne Uhmann

ously have and use options: they can either start the syntactic loop within the CP at C0, the functional head of the phrase containing the repairable, as in (20) and (21); or the constituent in the Spec-position (whose presence is syntactically dependent on and governed by its head)21 can also be included into the syntactic loop, as in (22). This option is possible because the Spec-position is filled in verb-second sentences.22 If these observations are correct, a first step towards a descriptively adequate account of the Head Rule will be its extension and subdivision depending on whether the repairable itself is a functional head or not. Extended Head Rule (provisional) (a) If the repairable is not a functional head, the accomplishment of repair preferably starts with the repetition of the functional head which immediately c-commands the repairable. (b) If the repairable is itself a functional head, the accomplishment of repair preferably starts with the ‘new’ functional head (or optionally with its specifier). Before making further refinements, I have to return — as indicated above — to Levelt’s examples (18c, d), which are repeated here for convenience: (18) The ball rolls to the left side, uh

  

(c) the left corner (d) to the left corner

Part (a) of the provisional Extended Head Rule (the repairable is not a functional head itself) predicts that participants in natural conversation will prefer to start their self-repairs with the functional head that immediately c-commands the repairable. That would make (18c) the preferred self-repair, because it is the determiner (i.e. the definite article) that exhibits these syntactic properties. But the analysis of naturally occurring data shows that, at least in German, participants prefer (d) to (c) — as in the following examples (24) and (25) — and thus do not follow the provisional Extended Head Rule: (24) Seglerinnen 01 02 03 →

04

A:

des schEInt aber in frankreich Üblich zu sein. ‘but that seems to be usual in France’ die sind völlig un(-)selbständig ‘they are completely dependent’ des war ja: (.) im GRUNde genommen auch bei der (1.0) äh (.) ‘it was basically the same thing with (1.0) uh (.)’ bei der niCOLE der fall ne? ‘with Nicole wasn’t it’


(25) Interview D5 01 →

I:

02

03

G:

.h das sin als- so so ETwa in der alt‘those are th- roughly in the old-‘ in der gleichen weise so die allgemeine strukTUR the general structure ‘in the same way like [von den interviews ] [of the interviews‘ ] [mhm mhm]

In (24) and (25) the repairables are part of a more complex PP than in examples (9) and (10) (including a DP instead of only N), and the self-repairs do not follow the relevant part (a) of the Extended Head Rule in that they do not start with D0, the functional head that immediately c-commands the repairable, but with P0, the preposition which is the head of the PP: PP

P

Spec

0

P (head)

DP (complement) D

Spec

0

D (head)

(24)

Ø

bei with

Ø

Ø

in in in in

(1.0) äh (.)

ART.

der

bei with (25)

der

NP

ART.

Ø

der the der the

nicole Nicole altoldgleichen weise same way

Figure 6. Syntactic structure of (24) and (25)

Yet these self-repairs do not sound odd or ill-formed. Moreover, no self-repair was found in my data which starts the syntactic loop with D° if the repairable is

394 Susanne Uhmann

part of such a complex PP. Although this might be due to the limitations of my data, the fact remains that participants regularly start with P° in these cases. The data in (26) and (27) may give a clue as to how these findings can be incorporated into the Extended Head Rule: (26) Interview D5 01 →

I:

02

das hast du bestimmt auch schon gemerkt ‘you have probably noticed that too’ bei dem (.) beim JÜRgen, PREP. ART. PREP.+CLITIC Jürgen ‘with with Jürgen’

(27) China 19

→

01

I:

02 03

T: I:

04

und [ in=ner ] ZUG- (0.57) and [PREP.+ CLITIC ] train ] train’ ‘and [ in a [kutsusin teitä .tch Well I (.) >would like to invite you(PL) meiän tupaantuliaisiin, to our housewarming ((party)),

2 3

S:

.mh A↑↑haa:, .mh I ↑↑see:,

4

T:

.hh Kaheskymmenesyheksäs päivä lauantaina, .nh hh .hh On the twenty ninth on a Saturday,

((57 lines omitted: talk about the date of the party; Susanna rejects the invitation on the basis of the unsuitability of the date. Then:)) 62

S:

↑Mitä sä sanoit. Kaheskymmenesyhe:ksäs,= ↑What did you say. Twentyni:nth,=

63

T:

=Nihh, =Yes,h (1.0)

64 65

S:

Ei se oo ens viikonloppu?, It isn’t the next weekend?,

66

T:

Ei se ookkaa nyt mut se on sitä seuraava,hh It isn’t now but it is the one following it,hh

67

S:

Voi-da-an me tul-la. can-PAS-4 we come-INF We can come.

68

T:→ Voi-tte-ko, can-2PL-Q Can you,

Simple answers to polar questions 413

69

S:⇒ Joo::.

70

T:

↑Ai ku kiva. ↑Oh how nice.

The epistemic function of joo here is to conWrm something that is in some way already shared by the participants: it conWrms that a positive response was given, that an acceptance of an invitation to a house-warming party was indeed produced (see SchegloV 1990). In this case, the activity needs to be continued after the joo response: a receipt of the acceptance is a relevant action by Tiina, and in line 70 she produces one. The V-interrogative to which the particle joo responds does not need to be a repetition of the prior talk. In example (5), the question has been built as an inference from the prior talk. Marja has been telling Eeva about a fellow worker who has had her last day at work. The question at line 5 takes up a new aspect of the topic by asking about the future of the fellow worker. (5) [Eeva/Dog: 6] 1

M:

Voi: hitto et se on ollu vähä tip:pa silmäs Oh he:ll how she has been so tea:rful been ollu joka päivä täälä ja halattu on miljoona every day here and we’ve been hugging a million

2

E:

kertaa päi: [vässä ja tapute ]ltu toisiamme ja, times a da:y and patting each other and, [ ] ] [Joo::, ] [Yea::,

5

E:→ .mhh

Meinaa-k se nyt jää-hä kotti-i intend-Q it now stay-INF home-ILL .mhh Is she gonna stay at home then.

6

M:⇒ Joo::, Joo::,

(.)

7 8

sitte. then

M:

Ja gradua rupee sit tekee ja .mhhh And then she begins to work on her gradu

9

rupee haastattelemaa ((master’s thesis)) and .mhhh begins to interview

10

itsemurhan tehneitä °Rovaniemellä°.hh those who have committed suicide °in Rovaniemi°.hh

11

E:

Joo::,

12

M:

Tai et ei:: itse-murha-n teh-ne-i-tä vaan --I mean not those who’ve committed suicide but ---

Through the utterance-Wnal adverb sitte ‘then’ the question at line 5 is marked

414 Marja-Leena Sorjonen

as an inference, the accuracy of which is being asked about. The question gets a joo response, followed by another joo as an intonation unit of its own. These joo responses agree with the function of the question: a conWrmation rather than a piece of new information is provided.

Apparent counterexamples The use of joo as a response to V-interrogatives that encode an assumption that the questioner already knows something about the issue may be deployed in contexts in which the question does not encode such an assumption. In such cases, joo may be heard to imply that the questioner in fact has the information she or he has asked for. Example (6) is a case in point. In line 31, Pekka makes an announcement of a visit. (6) [Pekka/Trip to Lapland: 16–17] 1

P:

↑Me alettii tekeen veroilmotuksii justiin ↑We just started to do the taxes

2

mut täss ois niinku vähä koht kahvitauon but it would be like soon a place for a coffee

3

paikka onks teillä kahvii siellä. break do you ((PL)) have coffee there.

4

S:

(0.6)

5 6

O:::n.

P:

Venttaas vähä. ((huoneeseen:)) Virve Wait a second. ((to the room:)) Virve shall käydääks me- käydään kahvil välillä. we go- let’s go and have a coffee in between.

7

((23 lines omitted P talking with S’s little daughter)) ↑No me< käydään pistäytymäs siälä. ↑Well we< will drop by.

31

P:

32

S:→ >L:ähe-tte-k-s te kohta puolii.⇐ leave-2PL-Q-CLI you.PL soon >Will you be leaving soon.⇐

33

P:Þ

=#tJoo:.#

34

S:

↑tSe:lvä juttu. ↑Alright.

The announcement at line 31 makes relevant a display of acceptance or nonacceptance by the recipient. Instead, Simo (line 32) produces a V-interrogative that requests information concerning the time of the visit. This question functions as a pre-second (SchegloV 1990): it requests information needed for


oVering the second pair part to the announcement. The interrogative gets a joo response (line 33), whereafter the questioner responds to the announcement (line 34). In this case, by doing joo, the answerer can be understood to imply that the question at line 32 requested information which its speaker already had, and to provide a conWrmation. Notice that in his earlier talk, in lines 1–3 and 7, Pekka can be heard to imply an immediate visit. In example (7), on the other hand, we have a case in which a repeat is given as a response to a question that builds on the prior talk by the coparticipant. This is an environment in which we would expect to Wnd joo as the response. The participants are here talking about what to do about Veera’s broken bicycle. (7) [Veera/Visit: 6–7] 1

S:

↑Tai sit sä käyt niinku ihan huo:ltoasemal ↑Or then you go like just to a se:rvice shop and

vaan mainitsemassa. Jos ne osais korjata. just mention ((it)). If they could repair ((it)).

2 3

V:

Nii joo. Yeah right.

4

S:

Koska kylhän ne< (.) autojen huoltoasemal Because they surely y’know< (.) understand about ymmärtää pyörienki päälle. bicycles too at a car service shop.

5 6

V:

Ai au:tojen. Oh ca:rs.

7

S:

Nii:.

8

V:

↑Ai jaa:. ↑Oh I see:.

9

S:

10

V:

.hh Tai [voi olla .hh Or [it can be [ [(-)(-)

11

S:

sä mitää menetä jos sä käyt kysymässä.= don’t y’know lose anything if you go and ask.=

12

V:

=£Nii::,£ .hh Mä aattel-i-n semmos-i-a on-han PRT I think-PST-1 such-PL-PAR is-CLI =£Yea::,£ .hh I was thinking about like there’re

13

] et ym:märtää. Tai ethä ] that they un:derstand. Or you ] ]

nii-t semmos-i-i pyör-i-en-ki they-PAR such-PL-PAR bicycle-PL-GEN-CLI y’know like also bicycle


16

huolto-a [sem-i-i#. service-station-PL-PAR service shops. [ S:→ [On-ko, [is-Q [Are there? V:⇒ On,

17

S:

14

15

No mee semmoselle. Well go to a such ((one)).

The V-interrogative at line 15 (‘Are there’) repeats the Wnite verb of the prior turn in which Veera explained what she thought Sari was referring to with the compound noun huoltoasema ‘service shop’ at line 1. In her turn, Veera oVers an alternative understanding of kinds of service shops. It is this counterinforming that gets a V-interrogative as its response at line 15. As against, for example, a turn that treats the prior turn as news (e.g. ai jaa ‘oh I see’), the question interprets the prior turn as one that has provided unexpected information, and information the accuracy of which is not immediately clear. To this question, Veera responds with a repetition at line 16. By selecting a repeat rather than the particle joo, Veera is analyzing Sari’s prior turn as if it asked for the information for the Wrst time. The repeat thus analyzes the prior repeat as one that has the status of a request for information. Thus the recipients can use a certain way of responding outside of its prototypical sequential context in order to recast and transform the prior question by the speaker. A repeat as an answer to a question that repeats the verb of the prior turn can be used to index that the information its speaker already provided needs to be reinforced. The particle joo, on the other hand, can be used to suggest that what the coparticipant asked is something he in fact has access to.

Answers to non-V-interrogative polar questions When we move from V-interrogatives to questions with a diVerent design, the range of possible ways of giving a positive response becomes broader. In addition to a repeat and the particle joo, a response can also be given through the particle niin. Furthermore, it is the particle niin that forms a basic agreeing response in these environments. Non-V-interrogative formatted questions can be constructed in a range of diVerent ways: they can be focused interrogatives, declaratives or diVerent


kinds of non-clausal utterances (see e.g. Bolinger 1957; Weber 1993; Heritage and Roth 1995; Sorjonen 1996, 1997 and the literature quoted there). By contrast to V-interrogatives, they all presuppose, in one way or another, the existence of some event or state of aVairs but check either the entire proposition or some part of it.

Niin as a response: conWrmation The particle niin is a prototypical response to non-V-interrogative polar questions. Its meaning in this context includes two axes. First, it claims to ratify as shared knowledge something that has already been shared by the participants in some way: it oVers a conWrmation. Second, this epistemic work is associated with a larger sequence and/or activity that is continuation relevant. In example (8), niin responds to a non-clausal utterance: to a turn with which the co-participant repeats a key element of the niin speaker’s prior turn. Eeva has told her brother Seppo that their mother has called and mentioned a letter sent to Seppo by the tax oYce. Eeva has tried to make Seppo guess the contents of the letter but he has been reluctant. He is now Wnally asking about the letter: (8) [Eeva/Tax return: 5–6] 1

S:

2

E:

No mitäs siel [lä muka. ] Well what do you say there is then. ] [ ] [ .hh Sinä oot saa ]nu t[ you be-2 get-PPC ?back [ .hh You’ve got ]ten bsemmose-n tuhat kaks-satta-a takas. like-GEN thousand two-hundred-PAR back about one thousand two hundred back.

3

(0.8)

4 5

S:→ Taka:si.= Ba:ck.=

6

E:⇒ =Nii:.

(1.1)

7 8

S:

9

E:

10

E:

ö Nyt kyllä nyt valehtelet [jotta vipa:t ]taa. er Now surely now you’re lying in your throa:t. [↑No ei::. ] [↑Well no::. ]

E(h):n valehtele niin se äiti sano. .h I’m no:t lyi:ng that’s what mother said. .h


(0.8)

11 12

S:

13

E:

Ihanko oike [e. Is it really so. [ [↑No iha oikeesti mutta sitte se [↑Well it’s really so but then she

14

käski sanoo että .hhh ei nyt pie sitte told me to tell you that .hhh you shouldn’t now

15

panna haisemaan kaikkia siellä Helsingissä. throw away all of it there in Helsinki.

In line 5, after a substantial silence, Seppo receives Eeva’s report of the contents of the letter with a repetition of the last element in Eeva’s turn. The repetition is responded to by Eeva with niin in line 6. With it she oVers a conWrmation, and by not continuing her talk she makes relevant a further receipt of the news by Seppo, for example, an evaluation of the news as good or bad. After another substantial silence, Seppo produces a turn which displays disbelief toward the informing. Talk about the accuracy of the informing still continues until Seppo Wnally provides an evaluation (data not shown). In example (9) from the same call, niin is provided as a response to a declarative question: to an utterance that brings up a B-event (Labov and Fanshel 1977: 100f.) or a ‘type 2 knowable’ (Pomerantz 1980), that is, information to which the recipient has privileged access. The utterance is in the second person plural form and it makes a claim about the niin speaker together with some other people. It is Saturday and Eeva has called Reijo to reach her brother Seppo, who is visiting Reijo in the same city where Eeva lives. Just before line 1 Seppo has said that he has not called Eeva yet, since he has left his phone book at home. (9) [Eeva/Tax return: 2] 1 2

S: ⇒ Joo:.< No hyvä eipähän nyt tarvii kottii ⇒ Joo:.< Well good now we don’t have to

rimputella.= tinkle ((=call)) home.=

3

oo-tta tul-lu E: → =eJoo:.hh .h< No mi-tä-s työ PRT PRT what-PAR-CLI you.PL be-2PL come-PPC at you ((PL)) have arrived =eYea::hh .h< Well wh wha

4

→ torstai-na jo. h .h[h Thursday-ESS already on Thursday already. h .h[h [ torstai ilta-na, S: ⇒ [↓Nii:n [PRT Thursady evening-ESS [↓Nii:n on Thursday evening,

5


6

E:

hh Joo:, hh Yeah?

7

S:

Yheksän jäläkkeen olttiin tiällä. After nine we were here.

8

E:

.hh No mitä ootte puuhannu.hh .hh Well what have you been doing.hh

Niin conWrms the assumption encoded by the question. However, the speaker proceeds, without any break, to an elaboration that speciWes and corrects the time reference in Eeva’s turn. In so doing he is responding to and resisting the possibly accusatory character of Eeva’s turn (cf. torstaina jo ‘on Thursday already’ — the call is made on Saturday and Seppo is leaving the city on the following day). As a positive response to non-V-interrogative formed questions the particle niin claims to remove the uncertainty the question expresses by providing a conWrmation. Its epistemic task is thus to change the state of knowledge of the coparticipant so that what the coparticipant has assumed or inferred now becomes ratiWed shared knowledge. In this respect its function bears a similarity to that of joo as a response to V-interrogatives discussed above. This kind of epistemic task is typically interwoven with certain kinds of activity contexts: the question and the niin conWrmation are part of a larger stretch of talk which is still incomplete, and niin makes relevant the continuation of that larger unit of talk. I have shown above niin as a response to a non-clausal and to a declarative question. Although the database contains only a few instances of focused interrogatives, they also get the particle niin as the positive response (e.g. Keskiviikkona-ko ne tulis. (Wednesday-Q they come) ‘Is it on Wednesday they would come?’ — Niin.).

Repetition: upgraded conWrmation The particle niin does not exhaust the means for responding positively to a non-V-interrogative question. Another response type used to that end is repetition. However, in contrast to cases in which the question is a V-interrogative, recipients only seldom respond to non-V-interrogatives with a repetition in the database. Furthermore, repetition has a function of its own when used as a response to a non-V-interrogative formed question. It often indexes upgraded conWrmation as compared to the particle niin.5 Example (10) from the same call as examples (8)–(9) contains an instance of repetition. At line 2, Eeva responds to Seppo’s reciprocal how-are-you


question by telling about what she is doing at the moment. After that she takes an inbreath (.hhh in line 2) and overlapping it, Seppo (line 3) repeats the last content word of Eeva’s talk so far, the temporal adverb ‘now’ but uses its reduced, one-syllable form (see Couper-Kuhlen 1996): (10) [Eeva/Tax return: 3–4] °mth° ö Mitäs sinä oot, °mth° er What have you, ((been doing/planned to do))

1

S:

2

No minä oo-n tässä tö-i:-ssä nytte ja .hh[h ] be-1 here.in work-PL-INE now and PRT I Well I’m here at wo:rk now and .hh[h ] [ ] S: → [Nyt.] [Now.]

3 4

5

E:

E: ⇒ Nyt. Ja [sitte tuota] mee-n (.) illa-lla now and [then PRT ] go-1 evening-ADE Now. And [then well I’m gonna go (.) [ ] ] S: [.Jaa [.I see ]

6

E:

#Soili-n luo ja# .hhh= 1nameF-GEN to and to #Soili’s in the evening and# .hhh=

7

S:

8

E:

=Joo [:. ] =Yea [:. ] [ ] [ja ja sit]te meinasin huomenna–[and and then tomorrow I was gonna — -

With the repetition, Seppo treats the time expression in Eeva’s turn as a source of trouble or in need of an elaboration (SchegloV et al. 1977). At line 4, Eeva responds to Seppo’s turn with a repetition. With it, she conWrms what Seppo has said as part of her own prior turn. The repetition in line 4 seems to function in the same way as the particle niin: it oVers a conWrmation. But a repeat and niin have also diVerences as responses. The particle niin is inherently a responsive linguistic expression: in its usages as a turn and utterance of its own it typically responds to the prior verbal action by another speaker. By choosing the particle niin as the means of action, the speaker indexes that what is being done is done as a response, as a second. A repetition as a means of conWrming is, of course, responsive, too. But in cases like example (10), even though the speaker uses the reduced form used by the recipient, the repetition echoes two prior utterances: the prior turn by the coparticipant and its speaker’s own prior utterance at line 2. As compared to niin, by echoing her


own prior utterance the speaker ties her response more clearly to her own prior turn and implies strong commitment to what she had said. The conWrming repetition at line 4 is followed by talk by the same speaker. She now continues her report from the point onward where it was before Seppo’s turn in line 3. By moving forward, instead of elaborating on the issue taken up by the coparticipant, the speaker can be heard to underscore the selfevidentiality of what was taken up by the coparticipant, that is, to emphasize that there is nothing special in being at work on a Saturday.

Particle joo: departure from some main line of talk The particle joo is yet another way of responding to a non-V-interrogative question. Like niin and the repetition, joo provides a conWrmation. However, it diVers from niin in one respect: it treats what the prior speaker took up in her or his utterance as a departure from some main line of talk and thus as closure relevant. It is unlike a repetition in that its conWrmation value is “normal” and not upgraded. In the following example Marja (line 8) responds with the particle joo to a repetition of a part of her prior turn. The extract comes from the beginning of a phone call. After Eeva’s response to her how-are-you question, Marja begins a report of her day in line 1: (11) [Eeva/Dog: 1–2] 1

M:

he

Nii::n .ihhh £Voi: sentä mie täälä tö-i-ssä PRT PRT PRT I here work-PL-INE he Yea::h .ihhh £O:h go:d I’m here at work oo be.1

2

#ja m and #and m

[mm# [mm# [ [↑↑Töis:sä. [At ↑↑wo:rk.

3

E: →

4

M: ⇒ >Joo: k’le tänään tulee tommone< superpitkä päivä >Joo: listen it’s gonna be a< super long day

5

ku ensi mä olin siel seminaaris ja sit tulin< as I first attended the seminar and then I came
me ol-t-i-in --PRT NEG PRT it is again then when we be-PAS-PST-4 Joo: No: but that was when we were --- [DETAILING OF THE TRIP TO VAASA FOLLOWS]

In line 3, Veke Wrst receives Kake’s report of a second possible obstacle as informative and then (line 5) treats the reported plan as extraordinary (oho ‘oh boy’). He then continues his response with an utterance which formulates the obstacles reported as indicative of a change in Kake’s life. This utterance carries, by virtue of the third person reference form mies ‘man’, a jocular tone (see Sacks 1992: 157–162). From that utterance Veke moves to ask for the reason for Kake’s behavior (lines 6 and 8). Being asked in a context in which the recipient has reported two competing activities instead of producing a second pair-part to an invitation, this question can be heard as throwing suspicion on the accounts oVered. However, the preceding jocular utterance makes it possible to hear the question as a continuation of the light tone of talk. In line 9, Kake begins an answer to the question with an utterance that he is building into a predicate nominal clause. When he reaches a point after which the next word would form a possible completion of the utterance and oVer a reason for his situation, Kake takes a long inbreath. In the course of this inbreath, Veke (line 10) oVers a collaborative completion of Kake’s utterance (see Sacks 1992: 144–149, 651–655; Lerner 1991). The oVered completion liikkuva ammatti ‘a job on the road’ sustains the jocular mode in Veke’s


utterance in lines 5–6: it formulates Kake’s non-work-related weekend activities as ones that are caused by the type of occupation he has. Through this completion, he cancels the possible seriousness Kake might have heard in his question. In line 11, Kake responds with the particle joo. The particle joo oVers a conWrmation. By selecting this device the speaker treats the prior turn and the current turn as a departure from the main line of talk. In this context, the relevant departure is from a serious action to a nonserious one (see Drew 1987; SchegloV 1987). The scope of the departure is displayed by the following utterance to which Kake moves from the conWrmation. With it, he cancels the oVered completion, through the negation, as a relevant and serious description. He then moves to a detailing of the reason for going to Vaasa, thereby answering Veke’s question in lines 6 and 8 but treating it as a question about the basis of the trip, and not as one about the basis of the two reasons for a possible non-attendance at the party. In cases such as (11) and (12), providing conWrmation through the particle joo seems to orient to the fact that the issue taken up by the coparticipant departs from the main line of talk by the speaker, and joo projects a closing down of discussion about the issue.

Conclusion The following table summarizes the relationship between question types and response types. A plus sign indicates that the response type is a prototypical way of responding to the question. A plus sign in brackets shows that the response type is used in the context but is not the central means in it. A minus sign indicates that the response type is not used in the context: Table. Relationship between response forms and question types. Question type niin information-seeking V-interrogatives checking V-interrogatives non-V-interrogatives

– – +

Response type joo repeat (+) + (+)

+ (+) (+)

A prototypical way of responding to a V-interrogative is to repeat the verb in the question. A repetition treats the question as a request for information, and it implies the continuation of the activity. When the V-interrogative encodes


an assumption that the information is already shared to some extent by the participants, a central response is the particle joo. Joo oVers a conWrmation. And when the question is non-V-interrogative, the prototypical response is the particle niin, which also gives a conWrmation. Hence each of these three response forms has a distinctive place in the Finnish answering system. All of these devices cue the speaker as the person who is in the position to provide the knowledge requested by the other, conWrming the epistemic authority of the speaker over the issue. A repeat and the particle joo can additionally be used in environments other than their base environment for special purposes. By selecting a repetition or the particle joo the speaker can, in addition to giving a response that fulWls what the question sequentially implicates (an aYrmation or a conWrmation), also address some other aspects in the prior question. These include the epistemic status or action characteristics or the kind of action the question accomplishes in the larger activity. Thus as a response to a V-interrogative that is built as a request for information, joo can suggest that the questioner already has the information. And conversely, a repeat as an answer to a question that seeks conWrmation can index that the information needs to be reinforced. As a response to a non-V-interrogative question, repeats can be deployed to oVer an upgraded conWrmation which, in turn, can be used as a vehicle to resist the co-participant’s prior action, for example, the co-participant’s treatment of prior talk as problematic or newsworthy. The particle joo as a conWrming device appears to be associated with contexts in which the prior question forms a departure from some main line of talk. In these contexts, in addition to providing a conWrmation, the particle joo suggests closure of the departure. In some cases, however, the kinds of functional diVerences between the response forms I have suggested seem to be fuzzy. It is especially some usages of the particle joo that do not clearly diVer from usages of the particle niin. It may well be that as a historical newcomer, the particle joo is a device which is currently Xoating from one context to another. Furthermore, there may also be diVerences among speakers in the use of the three response forms. It is, however, still possible to distinguish core environments for each of these response forms and it is by reference to an analysis of these basic environments that on-going change in the functions of response forms can be studied. The current database does not allow extensive analysis of individual speakers but there are speakers (e.g. Veke and Eeva in the examples of this paper) from whom there is enough data to see the range of responses they deploy, as they make use of the three diVerent ways of responding to yes-no questions.


A repetition of the verb in the question as a central response to a Vinterrogative is motivated by the structure of the question. In V-interrogatives the question morpheme, the enclitic particle -ko ~ -kö, is suYxed to the Wnite verb of the question. The morpheme points at the target of the question and in this fashion singles out the Wnite verb to be used in the response. When the particle joo, a loan word from Swedish, became part of the Finnish response system, it found its place as a device which underscores the structural and functional diVerences between diVerent types of V-interrogatives as to their sequential placement and their design. The ungrammaticality of niin as a response to V-interrogatives can be explained by its history: niin is originally a nominal element, a demonstrative which has anaphoric usages. This history can be seen in our cases, too: niin is not used as an answer to questions that deploy the morphosyntactic devices in the language for building up a question, the enclitic particle -ko ~ -kö, when the clitic is suYxed to the Wnite verb and the entire proposition is questioned. In all other cases, that is, as a response to non-V-interrogative questions and to focused interrogatives (see Section 2.1), it appears to be the basic form of response: it gives a response to questions that assume that some state of aVairs prevails but there is still uncertainty to the extent that a conWrmation is needed. In that way, the anaphoric basis of niin is present when it is used as a response to polar questions. However, the cataphoric aspect of pronouns is also contained in its usages: it projects a continuation of the activity. In their three-way division of answering systems, Sadock and Zwicky (1985) seem to count as questions only V-interrogatives that function as requests for information. In this respect one could say that Finnish deploys an echo system. But this kind of typology gives only a partial picture of the answering system. As we have seen, speakers, instead of repeating the verb in the question, can respond to a V-interrogative that requests information with the particle joo, in order to comment on the epistemic implications of the question. In other words, with respect to V-interrogatives Finnish exhibits a mixed system. And when we also take into account other types of questions, the character of the Finnish system as a mixed one becomes even more clear: Finnish makes use both of echo responses and particles. In the introduction to this paper, I suggested that very little is known about the answering systems in languages, especially the division of labor between diVerent minimal response types in a given language. I also pointed out that in the treatments of answering systems considerations of interactional factors are to large extent absent. This paper is an attempt to Wll these gaps and to


demonstrate how using data from real-life interactions and taking into account the interactional character of utterances can allow us to diVerentiate seemingly similar responses from each other and to come to grips with the dynamics of the answering system of a language.

Notes * I am grateful to Elizabeth Couper-Kuhlen, Auli Hakulinen and Margret Selting for their helpful comments on this paper. 1. While Raymond concentrates on yes-no interrogatives, I will also analyze responses to non-interrogative questions. 2. The choice between the two allomorphs is conditioned by vowel harmony. If the stem of the word contains back vowels the back vowel variant -ko is chosen; if it has no back vowels the form chosen is the front vowel variant -kö. 3. The clitic -s shows up especially in interrogatives and imperatives. It is said to soften the utterance but there are no extensive empirical studies on its use. 4. The verb form voidaan in 68 is the so-called passive form, which has to a great extent replaced the Wrst person plural verb form (voimme) in the spoken language. 5. I am only discussing partial repeats here. In cases in which the recipient repeats all of a clausal prior utterance by the speaker, the repetition appears to function as what SchegloV (1996) has termed ‘allusion conWrmation’ (see Sorjonen 1997, 2001).

References

Bolinger, D. 1957

Interrogative Structures of American English. Publication of the American Dialect Society 28. Alabama: University of Alabama Press. Couper-Kuhlen, E. 1996 “The prosody of repetition: on quoting and mimicry”. In Prosody in Conversation, E. Couper-Kuhlen and M. Selting (eds.), 366–405. Cambridge: Cambridge University Press. Drew, P. 1987 “Po-faced receipts of teases”. Linguistics 25: 219–253. Hakulinen, L. 1961 The Structure and Development of Finnish Language. Indiana University Publications, Uralic and Altaic Series 3. Bloomington: Indiana University.


Heritage, J. and Roth, A. 1995 “Grammar and institution: questions and questioning in the broadcast news interview.” Research on Language and Social Interaction 28 (1): 1– 60. Jefferson, G. 1972 “Side sequences”. In Studies in Social Interaction, D. Sudnow (ed.), 294– 338. New York: Free Press. Jones, B.M. 1999 The Welsh Answering System. Berlin: Mouton de Gruyter. Karlsson, F. 1987 Finnish Grammar. Helsinki: Werner Söderström. Labov, W. and Fanshel, D. 1977 Therapeutic Discourse. New York: Academic Press. Lerner, G. 1991 “On the syntax of sentences in progress”. Language in Society, 20 (3): 441–458. Nykysuomen sanakirja 1992 [1951–1961] [Dictionary of Modern Finnish.] 13th edition. Porvoo: WSOY. Ochs, E. 1988 Culture and Language Development. Language Acquisition and Language Socialization in a Samoan Village. Cambridge: Cambridge University Press. Pomerantz, A. 1980 “Telling my side: “limited access” as a “fishing” device”. Sociological Inquiry, 50 (3&4): 186–198. Pope, E. N. 1976 Questions and Answers in English. Janua Linguarum. Series Practica, 226. The Hague: Mouton. Quirk, R., Greenbaum, S., Leech, G. and Svartvik, J. 1972 A Grammar of Contemporary English. London: Longman. Raevaara, L. 1993 Kysyminen toimintana. Kysymys-vastaus -vieruspareista arkikeskustelussa. [Questioning as action. On question-answer adjacency pairs in everyday conversation.] Unpublished licentiate’s thesis. Department of Finnish language, University of Helsinki. Raymond, G. 2000 The Structure of Responding: type-conforming and nonconforming responses to yes/no type interrogatives. Ph.D dissertation, University of California, Los Angeles. Sacks, H. 1992 Lectures on Conversation. Volume I. G. Jefferson (ed.). Cambridge, Mass.: Basil Blackwell.


Sadock, J. M. and Zwicky, A. M. 1985 “Speech act distinctions in syntax”. In Language Typology and Syntactic Description, Volume I, Clause structure, T. Shopen (ed.), 155–196. Cambridge: Cambridge University Press. Schegloff, E. 1972 [1968] “Sequencing in conversational openings”. In Directions in Sociolinguistics, J. J. Gumperz and D. Hymes (eds.), 346–380. New York: Holt Rinehart and Winston. 1987 “Some sources of misunderstanding in talk-in-interaction”. Linguistics 25: 201–218. 1990 “On the organization of sequences as a source of “coherence” in talk-ininteraction”. In Conversational Organization and its Development, B. Dorval (ed.), 51–77. Norwood, NJ: Ablex. 1995 Sequence organization. Manuscript. University of California, Los Angeles. 1996 “Confirming allusions: toward an empirical account of action”. American Journal of Sociology 102: 1, 161–216. Schegloff, E., Jefferson, G. and Sacks, H. 1977 “The preference for self-correction in the organization of repair in conversation”. Language 53: 361–382. Selting, M. 1996 “Prosody as an activity-type distinctive signalling cue in conversation: the case of so-called ‘astonished questions’ in repair initiation”. In Prosody in Conversation, E. Couper-Kuhlen and M. Selting (eds.), 231–70. Cambridge: Cambridge University Press. Sorjonen, M.-L. 1996 “On repeats and responses in Finnish conversations”. In Interaction and Grammar, E. Ochs, E. Schegloff and S. Thompson (eds.), 277–327. Cambridge: Cambridge University Press. 1997 Recipient activities: particles nii(n) and joo as responses in Finnish conversations. Unpublished Ph.D dissertation. University of California, Los Angeles. 2001 Responding in Conversation. A Study of Response Particles in Finnish. Amsterdam: John Benjamins. Vähämäki, B. 1994 Mastering Finnish. New York: Hippocrene Books. Vilkuna, M. 1989 Free Word Order in Finnish: its syntax and discourse functions. Helsinki: Finnish Literature Society. Weber, E. 1993 Varieties of Questions in English Conversation. Amsterdam: John Benjamins.


Appendix Key to the glossing and transcription symbols Abbreviations used in glossing are: Case ablative adessive essive genitive illative inessive partitive

Abbreviation abl ade ess gen ill ine par

Approximate meaning ‘from’ ‘at, on’ ‘as’ possession ‘into’ ‘in’ partitiveness

Other abbreviations used: cli inf pas pl ppc prt

clitic infinitive passive plural past participle particle

pst q 1namef 1 2 4

past tense interrogative first name, female 1st person ending 2nd person ending passive person ending

Transcription symbols . , ? ?, ↑ ↓ _ : < su[ ] (.) (0.5) = >< °° £

falling intonation level intonation rising intonation slightly rising intonation rise in pitch fall in pitch emphasis is indicated by underlining (eg. cat) lengthening of sound head of an arrow pointing left at the end of a word: the word is finished abruptly but is not done as a clear cut off; glottal stops are indicated this way dash indicates a cut-off of a word utterances starting simultaneously point where overlapping talk stops micropause: 0.2 seconds or less silences timed in tenths of a second no silence between two adjacent utterances talk inside is done with a pace faster than the surrounding talk talk inside is done with a pace slower than the surrounding talk degree signs indicate a passage quieter than the surrounding talk smile voice


# @ hh .hh .joo (h) () (-) (( ))

creaky voice animated voice the letter h (or several of them) indicates an audible aspiration a period + the letter h (or several of them) indicates an audible inhalation a period in front of a word: the word is said with inbreath h in brackets within a word indicates aspiration, often laughter item in doubt a word in doubt a comment by the transcriptionist

List of contributors

Juliette Corrin Department of Human Communication Science University College London London WC1N 1PG United Kingdom Elizabeth Couper-Kuhlen Department of Linguistics University of Konstanz 78463 Konstanz Germany

Mike Huiskes Utrecht Institute for Linguistics / OTS University of Utrecht 3512 JK Utrecht Netherlands Kyu-hyun Kim Department of English Education Kyung Hee University Seoul (130–701) Korea

Cecilia E. Ford Department of English University of Wisconsin-Madison Madison, Wisconsin 53711 USA

Harrie Mazeland Centre for Language and Cognition Groningen (CLGG) University of Groningen 9700 AS Groningen Netherlands

Barbara A. Fox Department of Linguistics University of Colorado Boulder, Colorado 80309 USA

Hannes Scheutz Institut für Germanistik University of Salzburg 5020 Salzburg Austria

Auli Hakulinen Department of Finnish Language University of Helsinki 00014 Helsinki Finland Makoto Hayashi Department of East Asian Languages and Cultures University of Illinois at UrbanaChampaign Urbana, Illinois 61801 USA Marja-Liisa Helasvuo Department of Finnish Fennicum 20014 University of Turku Finland

Margret Selting Institut für Germanistik University of Postdam 14415 Potsdam Germany Marja-Leena Sorjonen Research Institute for the Languages of Finland 00500 Helsinki Finland Jakob Steensig Department of Linguistics University of Aarhus 8000 Aarhus C Denmark

434 List of contributors

Hiroko Tanaka Department of Sociology University of Essex Colchester CO4 3SQ United Kingdom Susanne Uhmann FB4 Sprach- und Literaturwissenschaft Bergische Universität-Gesamthochschule Wuppertal 42097 Wuppertal Germany

Bill Wells Department of Human Communication Sciences University of Sheffield Sheffield S10 2TA United Kingdom

Index of terms

A accent 13, 14, 247, 289–308, 310, 311 last ~ 267, 271–273, 277–279, 293, 295, 297–301, 303, 307, 308 terminal ~ 298, 302, 304 adjacency pair 6, 149, 195, 320, 323, 324, 326, 327, 336, 375 afterthought 119, 130, 277, 347 aligning 66, 67, 69, 163, 176, 177, 179, 180, 184–186, 190, 193, 195 amplitude 14, 295, 297, 304, 305, 311 answering pre-emptive ~ 320, 323, 326, 327, 339 ~ systems 16, 405, 406, 426 assessment 35, 37, 66, 82, 91, 96, 178, 185, 186, 195, 320 B because 112–129, 131–136, 142, 155, 238, 246, 255, 265, 275–277, 282 C causal ~ adverbial 111, 129, 133, 134 ~ clause 4, 111, 130, 131, 135 ~ relationship 11, 111, 119, 122, 123, 126, 127, 133, 134, 136, 137, 275 causality 112, 116, 122, 133, 135 child language 199, 201, 204, 220 clause combining 111 clitic 394, 408, 426 co-construction 7, 8, 35–37, 43 co-responding 327, 332 cognitive process 81, 89, 91, 95, 98, 99, 103, 104, 129 ~ shift, see cognitive process collaborative completion 8, 280, 423 conjunction 10, 11, 35, 41, 111, 122,

127, 136, 141, 142, 146, 165, 166, 177, 386, 387, 400, 401 connective 141, 142, 144, 146, 164, 166, 167, 186, 366, 386, see also conjunction constituency 29–31, 38, 44, 401 construction format 141, 161, 163–165 context interactional ~ see sequential context ~ of use 16, 148 ~ sensitivity 5, 13, 17, 251–253 sequential ~ 13, 17, 51, 58, 86, 142, 166, 167, 190, 193, 229–232, 235, 239, 243, 249–254, 266, 319, 416 contextualization cue 13, 33, 240, 241 conversation analysis (CA) 2, 34, 54–56, 84, 222, 231, 261, 345, 346, 367, 374– 376, 385, 397 conversational implicature 127, 164 D demonstrative 269, 270, 319, 348, 349, 351, 353–358, 361, 367–369, 406, 426 deviant case 229–232 discourse ~ particle 32, 44, 127, 171, 193 ~ topic 118, 131, 132 duration 14, 212, 219, 221, 287, 295, 297, 298, 304, 306, 308, 309 E elaboration 10, 15, 51, 52, 54, 56, 57, 60– 73, 151, 154, 155, 161, 345, 346, 349, 350, 364, 365, 368, 419, 420, 422 pursuit of ~ 52 epiphenomena 229, 251, 254 epistemic ~ task 419 ~ weil 116, 118, 126, 134 Extended Head Rule 392–397

436 Index of terms

eye gaze 12, 199, 207–209, 220 F fragments of units 12, 229–232, 243–245, 250–252, 254 free NP 25, 37, 39 functional head 15, 16, 373, 374, 387– 395, 397–399 G gesture 199, 208, 209, 219, 220, 306 grammar 1, 6, 12–14, 28, 34, 51, 53, 54, 56, 61, 74, 81, 83, 85, 104, 204, 206, 211, 219, 220, 259, 266, 267, 273, 278, 279, 306, 307, 311, 317, 318, 336, 337, 339, 345, 365, 373–376, 397, 398, 405 sentence ~ 171, 377, 380, 385

M maar 11, 141–145, 147, 148, 151, 154, 161, 163–166 minimal answer 173, 175 morphology 1, 8, 16, 18, 266, 397 N negation 10, 51, 52, 68–74, 239 affiliative ~ 68, 69 disaffiliative ~ 69–72 new start 231, 233, 255 next-turn repair-initiator (NTRI) 350, 351, 367 no 52, 58, 60, 62–73, 106, 171, 172, 176, 186, 264, 322, 349 No Gap Rule 396

H Head Rule 388–392

O objection 143 other-repair 67, 378, 379

I increment 39, 345, 351, 360, 364, 365 indexical expression 5, 15, 346, 350, 351, 354, 357 prospective ~ 355 insert-expansion 149 interrogative 16, 173, 264, 406, 408, 409, 412–417, 419, 421, 424–427 interruption 241, 251, 354 intersubjectivity 345, 346, 354, 358, 363, 364, 366 intonation 2, 8, 95, 230, 261, 289 cohesive vs. non-cohesive ~ 114, 117–120, 124, 134 ~ completion point 218, 294 ~ contour 200, 204, 241–244, 246, 248–250 falling vs. rising ~ 115, 117–120, 124, 129, 341 final vs. non-final ~ 13, 27, 267, 271–273, 277–279, 281 ~ unit 9, 26, 29–31, 37, 41, 42, 44, 172, 267, 294, 295, 414; see also ~ contour

P participation framework 72, 73, 159 peak alignment 14, 291, 298, 302–304, 311 phrasal unit 15, 345–347, 349–355, 357, 360, 364–366, 368 pitch 6, 12, 114, 205, 211–221, 223, 234– 249, 251, 253, 261, 271, 272, 277, 278, 281, 295–302, 305–311, 422 ~ accent 14, 218, 240, 247, 248, 287, 295 change in ~ 14, 297, 298, 306, 308 ~ peak 13, 267, 271, 287–295, 301, 307, 310, 311 ~ reset 267, 271, 272, 277, 285 play 75, 199, 201, 203–207, 210, 211, 216–219, 220, 223 polysemy 11, 165 positionally sensitive grammar 337 possible completion 6, 53, 56, 58, 59, 61, 63, 71, 75, 167, 171, 193, 230, 233, 243, 244, 250–254, 259, 261, 265–267, 271, 273, 278, 279, 288, 294, 296, 306, 308, 309, 321, 322, 350, 360, 380, 423

Index of terms 437

post-completion stance markers 172, 195 post-expansion 149, 336 postposition 14, 105, 317–320, 322, 324, 326–329, 331, 332, 335–341, 350, 361 practices 3, 8, 9, 11–18, 51, 53, 55, 56, 74, 75, 159, 165, 229, 231, 249, 251, 253, 254, 280, 281, 288, 289, 294, 306, 405 pragmatics 1, 13, 218, 273, 278, 306 pre-expansion 149 pre-second 414 preposing 124 presequence 178, 181 productional feature 95 projectability 36, 37, 55, 56, 104, 259, 337, 338 projection 12, 58, 104, 289, 306; see also projectability prosodic break 236–240, 248, 249, 253, 255 prosody 1, 5, 8, 13, 26, 27, 30, 35, 36, 51, 53, 55, 56, 58, 61, 95, 199, 212, 217, 221, 222, 229–232, 234, 235, 242–253, 259, 271, 273, 277, 279, 287–290, 294, 305–309, 311, 411 proverbs 187, 194, 195 Q question declarative ~ 418, 419 and-prefaced ~ 166 non-interrogative ~ 408, 427 quoted speech 37, 89, 91, 94, 106 R reassurance 177, 178, 193 recipient ~ reaction 250 ~ design 346, 363–365, 368 recycling 6, 7, 15–17, 145, 160, 161, 164– 166, 234, 373, 374, 385, 387, 396–398 redirection of talk 90 reference form 340, 357, 367, 423 locally subsequent ~ 167, 351 reference position locally initial ~ 167, 351

locally subsequent ~ 167 relinquishing 241, 253 repair 4, 6–11, 15–18, 25, 26, 31, 36, 43, 44, 63, 64, 67, 81, 82, 86, 89, 90, 92, 94–96, 98, 104–106, 141, 143–152, 161, 164, 171, 176, 202, 253, 263, 280, 290, 307, 309, 345–347, 350–365, 367– 369, 374, 377–399 accomplishment of ~ 378, 388, 392, 394 ~ initiation 15, 98, 143, 149–151, 253, 322, 346, 354, 355, 359, 361, 364, 373, 378, 382–384, 387, 396– 400 other-initiated ~ 15, 345–347, 350, 351, 353, 354, 357, 360, 361, 363– 365, 368, 378 self-initiated ~ 378 ~ sequence 15, 36, 120, 145, 146, 155, 326, 345–347, 350, 351, 354, 360, 361, 365 ~ syntax 382 topically-steering ~ initiation 151 transition-space ~ 350, 365 repetition 16, 93, 178, 266, 326, 328, 329, 341, 351, 385, 388, 392, 395, 405, 407, 409–411, 413, 416, 418–422, 424–427 request for information 173, 416, 424, 425 resuming 11, 124, 144, 147, 148, 155, 156, 159, 164–166, 340 return 65, 102, 141, 143, 144, 147, 148, 150–155, 159, 163–167, 175, 236 right dislocation 347 S self-repair 6–9, 15, 18, 81, 86, 106, 167, 177, 247, 255, 280, 359–361, 364, 368, 377, 378, 380–382, 385–391, 395–398, 400, 401 same-sentence ~ 15, 373, 374, 382 sequence 5, 10–12, 15–17, 31, 34, 35, 37, 38, 41–43, 51, 52, 55, 56, 60, 61, 64–68, 70, 71, 73–75, 111, 126, 131, 137, 145– 152, 155, 159, 161, 163–165, 172, 180,

438 Index of terms

181, 184, 189, 199, 201–203, 205–217, 219–223, 282, 326, 327, 336, 338, 340, 341, 345, 346, 350–352, 358, 363, 365, 367, 369, 411, 417 ~ expansion 141, 149, 151, 164 non-successive ~ 213, 214 side ~ 148, 150, 151, 282, 326, 327 sequential ~ context, see context, sequential ~ reasoning 165 skip-connecting 155–157, 159, 160 slope 14, 247, 291, 310 step down from 297, 301, 302, 306 stress 8, 45, 178, 267, 271–273, 277, 278, 281, 310, 340 successive utterance sequence 201, 206, 212–214, 218, 219 syntactic ~ completion 9, 32, 33, 38, 44, 218, 242, 266, 272, 373, 381 ~ loop 388, 390–392, 396 ~ unit 25, 44, 217, 253 syntax 1, 5–7, 12, 13, 15, 17, 18, 25–28, 30, 33, 35, 36, 74, 104, 199, 200, 212, 218–222, 229, 231, 232, 234, 243, 244, 246, 249–253, 266, 306, 309, 373–375, 382, 385, 395–400 T TCU 5, 10, 13, 19, 53, 54, 57–65, 68–73, 75, 222, 229–231, 241, 242, 244, 250, 251, 254, 255, 310, 373, 382, 385, 397, 398 ~ sentence 398 topic 11, 15, 41, 60, 61, 64–67, 93, 94, 102, 118, 121, 123–125, 130–132, 141, 148, 150–152, 154, 155, 161–165, 167, 173, 175, 182, 186, 194, 202, 207, 208, 211, 222, 269, 274, 324, 326, 327, 345, 346, 349–351, 353, 356–358, 361, 363– 366, 368, 413; see also discourse topic ~ closing 141 ~ shift 151, 324, 366 step-wise ~ shift 327 topical talk 154

transition relevance place 6, 202, 205, 221, 400 trouble-source turn 15, 89, 98, 99, 345, 346, 350–355, 357–361, 363, 365, 367– 369 turn 4–8, 10–18, 30, 32–34, 36, 39, 51– 56, 58–68, 70–75, 81–83, 85, 86, 89– 93, 95, 96, 98–105, 111, 120, 126–131, 137, 141–147, 149, 151, 155, 156, 158– 164, 166, 167, 171, 176–178, 180–187, 189–195, 199, 202–212, 214–223, 229, 230, 235, 241–246, 248–255, 259–261, 265–268, 270–273, 275, 279–281, 287– 290, 292–298, 304–309, 322, 328, 329, 332, 338, 345–347, 350–355, 357–369, 373–375, 380–385, 394, 395, 397–400, 405, 406, 408, 411, 412, 416–422, 424, 425 ~ continuation 199, 210, 212, 220, 221 ~ holding 56, 243, 244, 251 multi-unit ~ 54, 58, 74, 75, 160 ~ projection 7, 12, 13, 17, 51, 53, 74, 215, 222, 287, 289, 290, 292–294, 305–309 ~ taking 53, 56 ~ transition 33, 36, 53, 206 ~ yielding 248; see also relinquishing tying 167 long-distance ~ 160 type-2 knowable 418 V V-interrogative 16, 409, 412–414, 416, 417, 419, 421, 424–426 non-~ 416, 417, 419, 421, 425, 426 verb-final 112–115, 117, 118, 121, 122, 130–132, 137, 401 verb-second 112–115, 123–128, 130–132, 135, 137, 401 weil 10, 11, 17, 111–137, 238, 246, 247 ~ clauses 10, 11, 130, 131 word order 10, 16, 31, 83, 86, 111, 112, 340, 341, 365, 397, 401, 422

In the series STUDIES IN DISCOURSE AND GRAMMAR (SiDaG) the following titles have been published: 1. GELUYKENS, Ronald: From Discourse Process to Grammatical Construction: On Left-Dislocation in English. Amsterdam/Philadelphia, 1992. 2. IWASAKI, Shoichi: Subjectivity in Grammar and Discourse: Theoretical Considerations and a Case Study of Japanese Spoken Discourse. Amsterdam/Philadelphia, 1993. 3. WEBER, Elizabeth G.: Varieties of Questions in English Conversation. Amsterdam/ Philadelphia, 1993. 4. DOWNING, Pamela: Numerical Classifier Systems: The Case of Japanese. Amsterdam/Philadelphia, 1996. 5. TAO, Hongyin: Units in Mandarin Conversation: Prosody, Discourse, and Grammar. Amsterdam/Philadelphia, 1996. 6. DORGELOH, Heidrun: Inversion in Modern English: Form and function. Amsterdam/ Philadelphia, 1997. 7. LAURY, Ritva: Demonstratives in Interaction. The emergence of a definite article in Finnish. Amsterdam/Philadelphia, 1997. 8. MORI, Junko: Negotiating Agreement and Disagreement in Japanese. Connective expressions and turn construction. Amsterdam/Philadelphia, 1999. 9. HELASVUO, Marja-Liisa: Syntax in the making: The emergence of syntactic units in Finnish conversation. Amsterdam/Philadelpia, 2001. 10. SELTING, Margret and Elizabeth COUPER-KUHLEN (eds.): Studies in Interactional Linguistics. Amsterdam/Philadelpia, 2001.

Studies in Interactional Linguistics (Studies in Discourse & Grammar)

Studies in Interactional Linguistics (Studies in Discourse & Grammar)

Discourse and Identity (Studies in Interactional Sociolinguistics)

Discourse Strategies (Studies in Interactional Sociolinguistics)

Prosody in Conversation: Interactional Studies (Studies in Interactional Sociolinguistics)

Discourse in Ritual Studies

Laughter in Interaction (Studies in Interactional Sociolinguistics)

Mental Spaces in Grammar: Conditional Constructions (Cambridge Studies in Linguistics)

Gender and Discourse (Sage Studies in Discourse)

Gender and Politeness (Studies in Interactional Sociolinguistics)

Ergativity (Cambridge Studies in Linguistics)

Discourse Studies (Sage Benchmarks Discourse Studies)

Chinese Discourse LE (Routledge Studies in Asian Linguistics)

Switch-Reference and Discourse Representation (Cambridge Studies in Linguistics)

Switch-Reference and Discourse Representation (Cambridge Studies in Linguistics)

Discourse in Ritual Studies (Empirical Studies in Theology)

Grammar And Inference in Conversation: Identifying Clause Structure in Spoken Javanese (Studies in Discourse and Grammar)

Connectivity in Grammar and Discourse (Hamburg Studies on Multilingualism)

Derivations in Minimalism (Cambridge Studies in Linguistics)

Italian Syntax and Universal Grammar (Cambridge Studies in Linguistics)

Grammar in Interaction: Adverbial Clauses in American English Conversations (Studies in Interactional Sociolinguistics)

Keyness in Texts (Studies in Corpus Linguistics)

Ellipsis in Comparatives (Studies in Generative Grammar)

Talking Voices: Repetition, Dialogue, and Imagery in Conversational Discourse (Studies in Interactional Sociolinguistics)

The Morality of Knowledge in Conversation (Studies in Interactional Sociolinguistics)

Discourse In The Professions: Perspectives From Corpus Linguistics (Studies in Corpus Linguistics, SCL 16)

Mathematical Methods in Linguistics (Studies in Linguistics and Philosophy)

Mathematical Methods in Linguistics (Studies in Linguistics and Philosophy)

Grammar and Interaction: Pivots in German Conversation (Studies in Discourse and Grammar)

Point of View and Grammar: Structural Patterns of Subjectivity in American English Conversation (Studies in Discourse & Grammar)

Studies on Grammaticalization (Trends in Linguistics. Studies and Monographs)

Studies in Interactional Linguistics (Studies in Discourse & Grammar)