Subordination in Conversation: A Cross-Linguistic Perspective (Studies in Language and Social Interaction)

Subordination in Conversation Studies in Language and Social Interaction (SLSI) Studies in Language and Social Intera...

Author: Ritva Laury | Ryoko Suzuki

45 downloads 954 Views 1MB 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

Subordination in Conversation

Studies in Language and Social Interaction (SLSI) Studies in Language and Social Interaction is a series which continues the tradition of Studies in Discourse and Grammar, but with a new focus. It aims to provide a forum for research on grammar, understood broadly, in its natural home environment, spoken interaction. The assumption underlying the series is that the study of language as it is actually used in social interaction provides the foundation for understanding how the patterns and regularities we think of as grammar emerge from everyday communicative needs. The editors welcome language-related research from a range of different methodological traditions, including conversation analysis, interactional linguistics, and discourse-functional linguistics.

Editors Sandra A. Thompson

University of California, Santa Barbara

Elizabeth Couper-Kuhlen University of Helsinki

Editorial Board Peter Auer

University of Freiburg

Paul Drew

York University

Cecilia E. Ford

University of Wisconsin

Barbara A. Fox

University of Colorado

Marja-Liisa Helasvuo University of Turku

K.K. Luke

Hong Kong University/Nanyang Technological University

Volume 24 Subordination in Conversation. A cross-linguistic perspective Edited by Ritva Laury and Ryoko Suzuki

Subordination in Conversation A cross-linguistic perspective Edited by

Ritva Laury University of Helsinki

Ryoko Suzuki Keio University

John Benjamins Publishing Company Amsterdamâ•›/â•›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 Subordination in conversation : a cross-linguistic perspective / edited by Ritva Laury, Ryoko Suzuki. p. cm. (Studies in Language and Social Interaction, issn 1879-3983 ; v. 24) Includes bibliographical references and index. 1. Grammar, Comparative and general--Subordinate constructions. 2. Conversation analysis. I. Laury, Ritva. II. Suzuki, Ryoko. P294.S78â•…â•… 2011 415--dc22 isbn 978 90 272 2634 1 (Hb ; alk. paper) isbn 978 90 272 8696 3 (Eb)

2011005989

© 2011 – 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

Table of contents List of contributors Introduction Ritva Laury and Ryoko Suzuki N be that-constructions in everyday German conversation: A reanalysis of ‘die Sache ist/das Ding ist’ (‘the thing is’)-clauses as projector phrases Susanne Günthner

vii 1

11

Interrogative “complements” and question design in Estonian Leelo Keevallik

37

Syntactic and actional characteristics of Finnish että-clauses Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen

69

Clause-combining and the sequencing of actions: Projector constructions in French talk-in-interaction Simona Pekarek Doehler

103

A note on the emergence of quotative constructions in Japanese conversation Ryoko Suzuki

149

Clines of subordination – constructions with the German ‘complement-taking predicate’ glauben Wolfgang Imo

165

Are kara ‘because’-clauses causal subordinate clauses in present-day Japanese? Yuko Higashiizumi

191



Subordination in Conversation

Teyuuka and I mean as pragmatic parentheticals in Japanese and English Ritva Laury and Shigeko Okamoto

209

Name index Subject index

239 243

List of contributors Susanne Günthner Westfälische Wilhelms-Universität Germanistisches Institut Abteilung Sprachwissenschaft Stein-Haus, Hindenburgplatz 34 D- 48143 Münster Germany [email protected] Yuko Higashiizumi Tokyo Gakugei University International Student Exchange Center 4-1-1 Nukuikitamachi, Koganei-shi Tokyo 184-8501 Japan [email protected] Wolfgang Imo Westfälische Wilhelms-Universität Germanistisches Institut Abteilung Sprachwissenschaft Stein-Haus, Hindenburgplatz 34 D- 48143 Münster Germany [email protected] Leelo Keevallik Department of Modern Languages Box 636 Uppsala University 751 26 Uppsala Sweden [email protected]

Aino Koivisto Department of Finnish, Finno-Ugrian and Scandinavian Studies P.O. Box 3 (Fabianinkatu 33) 00014 University of Helsinki Finland [email protected] Ritva Laury Department of Finnish, Finno-Ugrian and Scandinavian Studies P.O. Box 3 (Fabianinkatu 33) 00014 University of Helsinki Finland [email protected] Shigeko Okamoto Language Program 218 Cowell College 1156 High Street University of California Santa Cruz, CA 95064 USA [email protected] Simona Pekarek Doehler Institut des Sciences du Langage et de la Communication FLSH, Université de Neuchâtel Espace Louis Agassiz 1 CH-2000 Neuchâtel Switzerland [email protected]

 Subordination in Conversation

Eeva-Leena Seppänen Department of Finnish, Finno-Ugrian and Scandinavian Studies P.O. Box 3 (Fabianinkatu 33) 00014 University of Helsinki Finland [email protected]

Ryoko Suzuki Faculty of Economics Keio University 4-1-1 Hiyoshi, Kohoku-ku Yokohama 223-8521 Japan [email protected]

Introduction Ritva Laury and Ryoko Suzuki

University of Helsinki and Keio University

Functional explanations for language phenomena The approach taken in this volume of articles can be broadly characterized as a functional one. One of the basic goals of functional linguistics has been the search for functional explanations for linguistic regularities. This search has taken such forms as the early work of iconicity (Haiman 1985), where researchers were looking for isomorphism between language structures and their meaning and function. In another functional approach, Emergent Grammar (Hopper 1987), grammatical patterns are seen to emerge as usage patterns in discourse. In the research on grammaticalization (e.g., Traugott & Heine 1991, Bybee 2007), new grammatical categories are seen as a result of crystallization of originally relatively free combinations into fixed grammatical patterns through frequency in language use. In the most recent functional paradigm, Interactional Linguistics, grammar is seen as a resource for the organization of social life (Ochs, Schegloff & Thompson 1996; Selting & Couper-Kuhlen 2001). The volume is based on a collection of papers presented in the panel ‘Subordination in conversation – crosslinguistic analyses of form-function matches’, organized by the editors of this volume at the International Pragmatics Conference in 2007 (Göteborg, Sweden). The papers can most clearly be seen as belonging to the last three strands of research, in the sense that in this volume, we consider grammatical patterns in a variety of languages in terms of how they arise both synchronically and diachronically out of recurrent practices in conversation, and also in terms of how they relate to the actional and sequential patterns in conversation, or, more broadly, in terms of their interactional functions. The focus of this book – ‘Subordination’ The main theme of this volume is subordination in conversation. Subordination has been the focus of much recent interest in functional linguistics (e.g., Thompson



Ritva Laury and Ryoko Suzuki

2002, Englebretson 2003, Couper-Kuhlen & Thompson 2008, Laury (ed.) 2008). Traditionally, subordination has been defined in syntactic terms: subordinate clauses are “grammatically dependent on another clause or some element in another clause” (Thompson, Longacre & Hwang 2007: 237–238). The subordinate clause has been seen as an argument in the other (its main, or matrix) clause (complement clauses), as a modifier of an NP in it (relative clause) or as a modifier of the verb or the whole clause (adverbial clause). However, this traditional definition has been challenged by researchers. Some linguists (e.g. Matthiessen & Thompson 1988) have viewed subordination as a continuum – in this approach, relative clauses and complement clauses are viewed as embedded within their matrix clause, while adverbial clauses are hypotactically linked to their main clause. It has also been proposed that even within one clause type, there may be a continuum of subordination (e.g. König & van der Auwera 1988; Givón 1990; Günthner 1996, 1999; Tao & McCarthy 2001), and that certain clause types which have been considered subordinate in particular languages may not actually be best described as such (e.g. Dixon 1995; Thompson 2002; Englebretson 2003; Laury 2006). It has also been proposed that many biclausal combinations actually involve projection rather than subordination (e.g. Auer & Günthner 2005; Hopper 2004; Hopper & Thompson 2008, and Thompson 2002). Subordination has also been recently discussed within Cognitive Linguistics, yet another functional approach. Cognitive linguists have suggested that subordination can be viewed conceptually as the distinction between figure (main clause) and ground (subordinate clause) (Talmy 2000: 324). Similar to this is Langacker’s (1991: 436; 2008) description of the subordinate clause as the one whose profile is overridden by the main clause; in this approach, the subordinate clause lacks an independent profile while that of the main clause imposes its own profile over the whole sentence. (For a useful summary of various approaches to subordination, see Cristofaro 2003: 22–29). Some of the papers in this volume propose yet another direction of research. They approach ‘subordination’ in terms of social action, taking what participants are doing with their talk in conversation into account. We could think of one action forming a part of another (main activity vs. subordinate action), so that subordinated verbal actions would be ones which are included within larger activities, or dependent on, or supportive of, other actions, so that subordination would be a matter of inclusion (for a discussion of actions embedded within other actions, see Lerner and Kitzinger 2007; Etelämäki 2009). We could also look at the relation in terms of focality. For example, Auer (2005: 9) suggests that ‘pres’ (i.e., prefatory activities in conversation, such as story prefaces, preinvitations, preannouncements, etc., which foreshadow another action)1 “project by noticeably being 1.

On pres, see Schegloff 1980, 2007.

Introduction

subsidiary (subordinated) to some focal activity”. Here Auer seems to be talking more about what is central to the activity vs. what is less central, and also suggesting that activities which serve to foreshadow or project actions (or activities) may be considered to be subordinated to those activities. Thompson (2002) considers the action that a turn is performing to be the part of the turn that is sequentially implicative; she specifically claims that “conversationalists are normally oriented to the material in the complements and independent clauses (154)”, rather than to the content in the matrix clause, and therefore argues against the treatment of complements as ‘subordinate/not main.’ However, the ways in which syntactic subordination correlates with the actional and sequential structure of conversation is still not well understood. Several of the papers in this volume deal with that topic. Other papers in this volume involve more classic themes such as the grammaticalization of particles and the grammaticalization of discourse markers from conjunctions and main clauses. All of the papers consider linguistic form as emergent from recurrent practices engaged in by participants in conversation. Therefore, this volume contains two related functional approaches to subordination: interactional and emergent approaches. All the contributions critically examine central syntactic notions in interclausal relations and their relevance to the description of clause combining in conversational language, to the structure of conversation, and to the interactional functions of language. Contents of the volume The articles contributed by Günthner, Keevallik, Koivisto, Laury & Seppänen, Pekarek Doehler, and Suzuki address, in different ways, the relationship between action and grammatical subordination. Günthner’s contribution concerns the German die Sache ist/das Ding ist- ‘the thing is’ construction and its use as a projective fragment in conversation. She shows that the traditional treatment of this construction in reference grammars does not adequately account for its use in spoken German. In conversation, speakers use the construction to project an upcoming syntagma as a focal point. What follows die Sache ist/das Ding ist is often not clausal in nature; that is, the construction is not by nature a matrix clause followed by a subordinate clause, as it is described in grammars. What follows the construction in conversation can in fact be a nonintegrated main clause, or it can even be a longer stretch of discourse. Günthner points out that the nature of this construction follows the usual tendencies noted in grammaticalization, namely the degrading of a construction in pragmatic weight, interactional function, and also in terms of syntactic features, as the construction becomes routinized and formulaic and skewed toward certain types of communicative functions.





Ritva Laury and Ryoko Suzuki

The article by Keevallik deals with Estonian indirect questions and their status as complements. She shows that indirect questions are not treated as subordinate by participants in Estonian conversation, but rather are interactionally prominent and responded to in the next turn. This is reflected in their grammar in that they also have features of main clause direct questions. Similarly to Günthner, Keevallik also suggests that the ‘main’ clauses associated with indirect questions have been grammaticized into epistemic particles which function to project and design questions in a sequentially and interpersonally sensitive way. Thus, the papers by both Günthner and Keevallik suggest that in the clause combinations they examine, the initial element does not serve as a main clause, but rather serves to project an upcoming segment, which may be clausal or nonclausal in form. Both papers show that what participants orient to in the upcoming conversation, and what therefore seems central in the interaction is what follows the projector phrase, not the projector phrase itself (cf. Thompson 2002). Therefore, as suggested by Auer (2005: 9) for prefatory activities, the projector phrases could perhaps be considered “subsidiary (subordinated) to some focal activity”, namely the action carried out by the main clause which follows. The paper by Koivisto, Laury & Seppänen takes up a similar issue, but from a different angle. The topic of their article is the role of complement-taking constructions and particles in the organization of conversation. They discuss the use of the Finnish että, which functions both as a complementizer and as an initial and final particle. They show that utterances which follow the most frequently occurring types of CTPs involving että, as well as utterances starting and ending with the particle että, are not syntactically subordinate, and since their content is clearly oriented to by the participants in the interaction, they are not subordinate on the level of action either. However, the authors suggest, differing from e.g., Thompson (2002), that the CTP phrases (as well as the particle että) are not actionally subordinate to the complement clauses or other utterances they are associated with, since they can also be shown to be oriented to by the participants. Instead, the authors suggest that the particles and the CTP phrases can be understood to perform a separate action in the service of the action done by the utterance they accompany, similarly to referential actions described by Lerner & Kitzinger (2007). Pekarek Doehler takes up the question of certain French constructions which have traditionally been analyzed as biclausal, namely the je veux dire ‘I want to say’ + complement construction, the il y a ‘there is’ presentational cleft, and the pseudocleft-construction. Pekarek Doehler shows that the initial projector fragment in these constructions serves to fit together stretches of talk and in that way functions to organize complex turns. The constructions differ in that je veux dire has a retrospective orientation, as it functions to extend a turn, prolonging some previous action and presenting a stretch of talk as a modification of preceding talk, while

Introduction

the il y a-cleft is more exclusively prospectively oriented and opens a new segment of talk. The pseudocleft, in turn, also projects forward to a complex segment, and frames it in certain ways so that that the second part is oriented to by co-participants – and hence interpretable – as accomplishing an action that is projected by the first part. The talk that follows these initial projector fragments varies in length and scope, but the talk and its eventual boundary is related to both the nature of the projector and its contextual embedding. Pekarek Doehler suggests that the projector fragment in this way projects both turn-shape and turn-type. She further proposes that the initial projector fragment can thus be interpreted as part of a turn-constructional device that is instrumental for floor-holding, turn-extension, and the construction of complex turns. In this sense, Pekarek Doehler’s point is similar to that made by Günthner and Keevallik. Suzuki’s paper discusses the diachronic development of constructions involving the quotative complementizer tte in Japanese. The sequence of [reported speech + tte + verb of saying] is usually considered to be typical, but in natural conversation, tte also occurs at the end of clausal utterances without any main verb. Suzuki examines the earliest tte tokens in her data, and finds that tokens of utterance-final tte without any main verb following it can be found in three interactional contexts: (1) introducing reported speech as the 2nd pair part of a question – answer sequence; (2) introducing reported speech with an explicit reference to the original speaker; (3) introducing immediate repetition of the prior speaker’s utterance. At this point, we would not be able to directly link utterance-final uses of tte in the early data and tte used as the utterance-final pragmatic particles in today’s natural conversation. However, utterance-final tte in the early data, which implements several interactional practices with specific communicative intent, i.e., offering information, clarification, asking for elaboration etc., may feed into the emergence of utterance-final tte as an independent syntactic construction. The papers by Imo, Higashiizumi, and Laury & Okamoto, which form the second part of the volume, are also concerned with the use of grammatical forms in conversation, but focus more on issues of the development and continuum nature of grammatical categories. Imo’s paper concerns constructions with the German complement-taking predicate glauben. Imo shows that in his spoken German corpus, the CTPs involving glauben form a cline with clear matrix clauses at one end, and with uses as adverbs or modal particles at the other end. He shows that when the clause with ich glaube ‘I think’ is initial,and the clause following it has subordinate grammar, the glauben-clause is relatively asserted and its profile overrides that of the clause that follows. On the other hand, when a clause-initial ich glaube is followed by a main clause, it functions as a pragmatic particle or discourse marker. In clause-final and clause-medial use, the matrix-clause character of glaub(e) ich is further





Ritva Laury and Ryoko Suzuki

lessened. In these positions, the expression has become nearly fully recategorized. As also noted by Auer & Günthner (2005), in these positions, it has lost syntactic powers of projection, and has become morphologically and phonologically reduced and prosodically downscaled and integrated. Higashiizumi’s paper focuses on Japanese kara ‘because’-clauses from synchronic and diachronic perspectives. This study, like some other recent studies of complex clause constructions, rejects the traditional dichotomous approach of subordination vs. coordination, and suggests that a three-way continuum of parataxis-hypotaxis-subordination (Hopper and Traugott 2003) would be more useful in describing the evolution of kara-clauses. However, she finds the direction of change of the kara-clause goes the opposite direction from what has earlier been suggested; starting from subordination to parataxis, and furthermore, to independent kara-clause without any main clause. Before 1900, the kara-clause was used as a subordinate clause, expressing real-world causality, and was followed by a main clause. Then, from the 1900s onward, one finds an increase of paratactic kara-clause constructions, expressing epistemic or speech-act causality. Accordingly, the clausal order becomes more flexible, so we find an increasing use of kara-clauses following a main clause (‘the inverted order’). Independent kara-clauses, whose interpretation relies on surrounding discourse, also show an increase in the 20th century. In her paper, Higashiizumi demonstrates the complex relationship between syntax and semantic/pragmatic functions of kara-clauses. Laury & Okamoto’s paper compares the use of the English I mean and the Japanese teyuuka. Each of these constructions has been extensively studied separately, but have not been compared with respect to their interactional functions. Laury & Okamoto show that although the expressions have quite different syntactic and semantic origins, they have come to be used in similar ways in conversation. The function of both expressions can be characterized in terms of self-monitoring: they indicate that a preceding utterance is somehow inadequate or in need of modification, and they may project another utterance which is intended as a modification of that preceding utterance. The article also shows that teyuuka may be further along in its path of grammaticalization than I mean, while I mean may have more of an interpersonal function than teyuuka. On the other hand, teyuuka functions on a more metalinguistic level. This may pose a problem for the theory of grammaticalization, since it has often been assumed that items move from subjective to intersubjective functions as they become further grammaticalized. The paper also discusses the contribution of the original syntax and semantics of the expressions, and the syntax of Japanese and English in general, to the developmental paths and current use of these expressions.

Introduction

Contribution to subordination research and future directions This volume adds to the existing literature which has questioned the standard assumptions regarding subordination in the languages of the world. Several papers (in particular Günthner, Keevallik, Imo, Laury & Okamoto) suggest that the tendency for complement-taking constructions to become grammaticalized into formulaic fragments (e.g. Thompson & Mulac 1991, Auer 1996, Hopper 2001, Thompson 2002, and others) is widespread in a range of genetically and geographically unrelated languages. The papers by Higashiizumi, Günthner, Imo, Keevallik, and Suzuki also discuss the changes that subordinate clauses undergo so that they increasingly develop main clause–like syntax: while Higashiizumi and Suzuki view this change from a diachronic perspective, Günthner, Imo, and Keevallik base their observations on synchronic data. Suzuki and Koivisto, Laury & Seppänen discuss a related process, the development of complementizers into pragmatic particles. Another related question has to do with biclausal constructions in general; as Günthner as well as Pekarek Doehler point out in their paper, and as also suggested by Hopper and Thompson (2008), many constructions normally thought to be biclausal, such as pseudo-clefts, actually consist of a formulaic fragment which serves to project, and, as Pekarek Doehler shows, tie the utterances they occur with to ongoing talk, framing them in certain ways and affording the construction of complex turns. A novel topic taken up in several papers is the connection between grammatical subordination and the structure of conversation. These papers, notably the contributions by Günthner, Keevallik, Koivisto, Laury & Seppänen, and Pekarek Doehler add to the claims made in earlier research (e.g. Thompson 2002) regarding the orientation of participants to the matter expressed in the part of the utterance normally considered grammatically subordinate; such utterances, responded to and otherwise oriented to by participants, may also show grammatical features of main clauses in languages with distinctive main clause and subordinate clause syntax. Several of the papers also touch upon the functions of formulaic projecting phrases and particles. Keevallik suggests that the Estonian particles in her data serve to deal with interpersonal and sequential matters in conversation, while Pekarek Doehler shows that the French projecting formulae she discusses in her paper are devices specializing in contextualizing the utterance which follows in terms of turn type and turn shape. Koivisto, Laury & Seppänen also show that the Finnish projecting formula and particle which they discuss in their paper does important work in organizing the participant framework in conversation (Goffman 1981), and question whether the work done by these elements should be considered actionally subordinate, since they can also be shown to be oriented to by the participants.





Ritva Laury and Ryoko Suzuki

A challenge to our quest to relate grammatical subordination to the actional and sequential structure of conversation is that is not yet quite clear just how the link between the syntactic level and the actional level in general should be characterized. The work on this topic, while highly exciting and important, is only in its beginning stages. Traditional views of syntactic subordination have been challenged, and this volume continues that work, but the vocabulary to talk about the types of relations we actually see in conversation is still developing. There are still only scattered references in the literature to subordination on the level of action. In fact, our volume is one of the very first attempts to link together the two concepts, subordination and action. We expect to see significant work on this topic in the future, along the lines pointed to in the articles in the volume. References Auer, Peter. 1996. “On the prosody and syntax of turn-continuations.” In Prosody in Conversation, Elizabeth Couper-Kuhlen and Margaret Selting (eds), 51–100. Cambridge: Cambridge University Press. Auer, Peter. 2005. “Projection in interaction and projection in grammar.” Text 25–1, 7–36. Auer, Peter and Günthner, Susanne. 2005. “Die Entstehung von Diskursmarkern im Deutschen – ein Fall von Grammatikalisierung?” In Grammatikalisierung im Deutschen, Thorsten Leuschner and Tanja Mortelmans (eds), 335–362. Berlin: de Gruyter. Bybee, Joan. 2007. Frequency of Use and the Organization of Language. Oxford: Oxford University Press. Cristofaro, Sonia. 2003. Subordination. Oxford: Oxford University Press. Couper-Kuhlen, Elizabeth and Thompson, Sandra A. 2008. “On assessing situations and events in conversation: Extraposition and its relatives.” Discourse Studies 10(4): 443–467. Dixon, R.M.W. 1995. “Complement clauses and complementation strategies.” In Grammar and Meaning: Essays in Honor of Sir John Lyons, F.R. Palmer (ed.), 175–220. Cambridge: Cambridge University Press. Englebretson, Robert. 2003. Searching for Structure: The Problem of Complementation in Colloquial Indonesian Conversation. Amsterdam: John Benjamins. Etelämäki, Marja. 2009. “Activities within activities: left dislocations with relative clauses.” Paper given in the symposium ‘Aspects of clause combining: The linking of actions in speech and writing.’ Oct 29–31, University of Helsinki. Givón, Talmy. 1990. Syntax. Amsterdam: John Benjamins. Goffman, Erving. 1981. Forms of Talk. Oxford: Basil Blackwell. Günthner, Susanne. 1996. “From subordination to coordination? Verb-second position in German causal and concessive constructions.” Pragmatics 6(3): 323–356. Günthner, Susanne. 1999. “Wenn-Sätze im Vor-Vorfeld: Ihre Formen und Funktionen in der gesprochenen Sprache”. Deutsche Sprache 3: 209–235. Haiman, John (ed.). 1985. Iconicity in Syntax: Proceedings of a Symposium on Iconicity in Syntax, Stanford, June 24–26, 1983. Amsterdam: John Benjamins.

Introduction

Hopper, Paul. 1987. “Emergent grammar.” Berkeley Linguistic Society 13: 139–157. Hopper, Paul. 2001. “Grammatical constructions and their discourse origins. Prototype or family resemblance?” In Applied Cognitive Linguistics I: Theory and Language Acquisition, Martin Pütz, Susanne Niemeier and René Dirven (eds), 109–130. Berlin: Mouton de Gruyter. Hopper, Paul. 2004. “The openness of grammatical constructions.” Chicago Linguistic Society 40: 239–256. Hopper, Paul and Traugott, Elizabeth C. 2003. Grammaticalization, (2nd ed.). Cambridge: Cambridge University Press. Hopper, Paul and Thompson, Sandra A. 2008. “Projectability and clause combining in interaction.” In Crosslinguistic Studies of Clause Combining: the Multifunctionality of Conjunctions, Ritva Laury (ed.), 99–124. Amsterdam: John Benjamins. König, Ekkehard, and van der Auwera, Johan. 1988. “Clause integration in German and Dutch conditional, concessive conditional, and concessive clauses.” In Clause Linking in Grammar and Discourse, John Haiman and Sandra A. Thompson (eds), 101–133. Amsterdam: John Benjamins. Langacker, Ronald W. 1991. Foundation of Cognitive Grammar. Stanford: Stanford University Press. Langacker, Ronald W. 2008. Cognitive Grammar: A Basic Introduction. Oxford: Oxford University Press. Laury, Ritva. 2006. “Oblique mentions of human referents in Finnish conversation: The effects of prominence in discourse and grammar”. In Grammar from the Human Perspective, Marja-Liisa Helasvuo and Lyle Campbell (eds), 153–171, Amsterdam: John Benjamins. Laury, Ritva (ed.). 2008. Crosslinguistic Studies of Clause Combining: The Multifunctionality of Conjunctions. Amsterdam: John Benjamins. Lerner, Gene H. and Kitzinger, Celia. 2007. “Extraction and aggregation in the repair of individual and collective self-reference.” Discourse Studies 9(4): 526–557. Matthiessen, Christian and Thompson, Sandra A. 1988. “The Structure of Discourse and ‘Subrodination’.” In Clause Combining in Grammar and Discourse. John Haiman and Sandra A. Thompson (eds), 275–329. Amsterdam: John Benjamins. Ochs, Elinor, Schegloff, Emanuel A. and Thompson, Sandra A. (eds). 1996. Interaction and Grammar. Cambridge: Cambridge University Press. Schegloff, Emanuel A. 1980. “Preliminaries to Preliminaries: ‘Can I ask you a question’.” Sociological Inquiry 50 (3–4): 104–152. Schegloff, Emanuel A. 2007. Sequence Organization in Interaction: A Primer in Conversation Analysis I. Cambridge: Cambridge University Press. Selting, Margaret and Couper-Kuhlen, Elizabeth. 2001. Studies in Interactional Linguistics. Amsterdam: John Benjamins. Talmy, Leonard. 2000. Toward a Cognitive Semantics, Volume I: Concept Structuring Systems, iviii, 1–565. Cambridge, MA: MIT Press. Tao, Hongyin, and McCarthy, Michael J. 2001. “Understanding non-restrictive which-clauses in spoken English, which is not an easy thing.” Language Sciences 23: 651–677. Thompson, Sandra A. 2002. “‘Object complements’ and conversation: towards a realistic account.” Studies in Language 26(1): 125–164. Thompson, Sandra A. and Mulac, Anthony. 1991. “A quantitative perspective on the grammaticization of epistemic parentheticals in English. In: Approaches to Grammaticalization Vol. 2, Elizabeth C. Traugott and Bernd Heine (eds), 313–329. Amsterdam: John Benjamins.





Ritva Laury and Ryoko Suzuki Thompson, Sandra A., Longacre, Robert E. and Hwang, Shin Ja. 2007. “Adverbial Clauses.” In Language Typology and Syntactic Description (2nd ed.), Timothy Shopen (ed.), 237–300. Cambridge: Cambridge University Press. Traugott, Elizabeth C. and Heine, Bernd (eds). 1991. Approaches to Grammaticalization (Vols. 1 and 2). Amsterdam: John Benjamins Publishing Company.

N be that-constructions in everyday German conversation A reanalysis of ‘die Sache ist/das Ding ist’ (‘the thing is’)-clauses as projector phrases* Susanne Günthner Based on a corpus of conversational German, I will show that the standard view of N be that-constructions (e.g. ‘the thing is/the point is...’) as [matrix clause + subordinated complement clause] cannot be supported by actual data from spoken interactions. Instead of a subordinated complement clause following the matrix clause, speakers reanalyse ‘die Sache ist’ (‘part A’) as a projector phrase, anticipating the upcoming core message (‘segment B’). Part B (‘the complement clause’), which overwhelmingly occurs with no complementizer ‘dass’ (‘that’), can take various forms in spoken interactions: it can be realized as a subordinate clause, a main clause, a complex clause and even as a larger discourse segment. In all these cases, it is no longer a conceptual element of segment A’s proposition. Instead of the traditionally assumed matrix clause vs. subordinate clause relation, the ‘matrix clause’ (‘die Sache ist’ syntagma) is reanalysed as a projector phrase, focussing recipients’ attention on the following segment, which expresses the core message. Thus, the ‘complement’ overrides the ‘matrix clause’. This functional upgrading of the ‘complement clause’ (part B) tends to be accompanied by formal indications of prosodic and syntactic independence.

1. Introduction In the search for modelling a grammar of spoken language, which includes issues of ‘prefabricated parts’ and ‘constructional schemas’, work within Interactional Linguistics has increasingly turned toward ‘usage-based’ variants of ‘Construction * This paper is based on my presentation at 10th International Pragmatics Conference (Göteburg, 8–13 July 2007). I am grateful to Ritva Laury and Ryoko Suzuki for helpful comments. Thanks to Lisa Roebuck for checking the English.



Susanne Günthner

Grammar’ (Ono/Thompson 1995; Thompson 2002a; Hopper 2004; 2008; Auer 2005a; b; 2006, Couper-Kuhlen/Thompson 2006; Günthner 2007; 2008a; b; c; 2009a, b, Günthner/Imo 2006; Günthner/Bücker 2009; Günthner/Hopper 2010; Imo 2006; 2007; Deppermann 2007; 2009; Birkner 2008; Deppermann/Elstermann 2008). Its non-modular, holistic perspective on language, its interrelation of form and function of linguistic units, its inclusion of pragmatic, discourse-functional and cognitive aspects in analysing linguistic constructions, as well as its assumption, that grammatical structures grow out of communicative actions, make Construction Grammar rather attractive for a praxis-oriented perspective on linguistic phenomena. However, even though the call for a usage-based perspective has often been emphasized in studies of Construction Grammar, so far, there are very few studies that analyse grammatical constructions in authentic, everyday interactions (Fried/Östman 2005; Günthner 2009a). Even so-called ‘usage-based’ approaches (Goldberg 1995; 1996; Lambrecht 2001; Croft 2001) neglect questions concerning sequential, dialogical, as well as genre- and activity-related aspects in the use of constructions. Instead, most studies within Construction Grammar are still oriented toward decontextualized examples based on written language, without inquiring into the emergence of constructions in the process of interaction (Günthner/Imo 2006; Deppermann 2007; Imo 2007; Birkner 2008; Günthner/Bücker 2009; Günthner 2010). This study is a contribution toward an understanding of linguistic structure and function by studying grammar ‘at work’ (Schegloff 1996). Instead of taking grammatical constructions for granted, I shall inquire into their nature by looking at their interactive emergence in everyday usage. On the basis of an empirical analysis of ‘die Sache ist/das Ding ist,...’-patterns (‘the thing is’), I will study form and function of this complex, bipartite nominalconstruction in spoken German, and address issues of matrix clauses vs. subordinate clauses in spoken interaction. The analysis is based on a corpus of 91 everyday interactions (30 to 180 minutes in length), collected during the years 1989–2006 in different parts of Germany. They include informal face-to-face interactions among friends and family members, office hours at university, genetic counselling sessions, radio phone-in programs, as well as data from the TV series ‘Big Brother’.

N be that-constructions in everyday German conversation

2. Uses of N be that-constructions in German1 In Construction Grammar, complex sentences such as ‘the thing/point/problem is that...’ are referred to as N be that-constructions (Schmid 2001).2 They consist of an initial noun phrase headed by an abstract, metapragmatical noun (‘thing, point, problem...’) functioning as a subject, a form of the copula BE and a THATclause syntactically functioning as subject complement [abstract noun, copula, that-clause]:3 ‘die Sache’ COPULA (‘ist’) → complement clause (‘dass...’ (‘that’)) ‘das Ding’ In German reference grammars, the pattern at issue is mainly treated as a biclausal construction, consisting of a matrix clause followed by a subject complement clause: [matrix clause + complement clause].

The matrix clause is not a fully-fledged syntactic ‘gestalt’, as its verb (the copula) requires a further constituent. This constituent is produced in the following complement clause, introduced by a subjunction such as ‘dass’ (‘that’). The matrix clause (‘die Sache ist/das Ding ist’) is treated as the main clause; the following complement clause is considered as subordinate – indicated by the subjunction ‘dass’ as well as by verb-final constituent order (a grammatical feature of subordinate clauses in German). In general, complement clauses are considered to be ... prototypical instances of subordination; [...] A subordinate clause is then describable as one whose profile is overridden by that of a main clause. [...] In a typical complement clause construction, the two clauses combine directly and the main clause is clearly the profile determinant: ‘I know she left’ designates the process of knowing, not of leaving. (Langacker 1991: 436)

1.

Cf. also Günthner (2008a; c).

2. Constructions are treated as form-function pairings whose structural and semantic properties cannot (entirely) be accounted for in terms of other properties of the grammar: “Together, the matrix clause and the RC form a constructional unit whose global meaning is not equal to the sum of the meanings of the parts.” (Lambrecht 2001: 469) 3. In his corpus-based study of N be that-constructions in written English, Schmid (2001) argues that the ten nouns that were found to occur most frequently in this construction are “problem, thing, truth, fact, trouble, point, result, view, reason, idea”. Schmid’s observations are based on data of written English and, thus, are not automatically transferable to German interactional data.





Susanne Günthner

In this article, however, I shall argue that categories such as ‘matrix clause’ and ‘complement clause’ are problematic when it comes to N be that-constructions in actual spoken German: neither can the N be-part (‘die Sache ist/das Ding ist’) be treated as a matrix clause holding the relevant information for the following discourse, nor is the following syntagma (i.e. the complement clause) formally and conceptually subordinate to the preceding clause. Or to put it in Langacker’s (1991: 435–7) terms: Neither can the N be-part be treated as the clause holding the main event, nor is the following part construed in the perspective of the event expressed in the N be-clause.4 In looking at ways in which interactants use N be that-constructions in spoken language, we realize that the instantiation of the ‘complement clause’ dissolves into various construction formats. Some of these can no longer be treated as subordinate. Instead, we observe a downgrading of the ‘matrix clause’ on behalf of the syntagma that follows; it tends to be reanalysed as a projecting phrase (Hopper 2005; 2006; Günthner 2008a; b; Günthner/Hopper 2010), building up a projecting space, and contextualizing ‘more to come’. In the following, I shall introduce the various formats the ‘complement clause’ in ‘die Sache ist/das Ding ist’-constructions can take and will show how a grammatical construction changes its nature when it is examined from the perspective of everyday usage in spoken interaction. 2.1

‘die Sache ist/das Ding ist’-utterances followed by a subordinate clause introduced with the subjunctor ‘dass’5

Some ‘die Sache ist/das Ding ist’-constructions in my data reveal the canonical structure described in German reference grammars: they are complex clauses with the first clause consisting of an initial NP (‘die Sache’, ‘das Ding’, etc.) and the copula ‘ist’ (in present tense); the following clause being introduced by the complementizer ‘dass’ and showing subordinate clause word order (i.e. final positioning of the finite verb).6 4. Cf. also Cristofaro (2003: 2). 5. This study will concentrate on ‘die Sache ist/das Ding ist’-patterns, without taking semantically specific nouns, such as ‘Frage’ (‘question’), ‘Problem’ (‘problem’), etc. into account. I.e. I will only consider abstract head nouns. Cf. also Aijmer (2007) for uses of the English construction ‘the fact is that’. In contrast to English ‘ (the) fact is that’-phrases (‘fact is...’) or German ‘Tatsache ist’ (‘fact is’), I find no cases of ‘die Sache ist/das Ding ist’ with the article (‘die’ or ‘das’) missing. 6. German, which has verb-second as its basic word order in simple and main clauses, requires final position of the finite verb in subordinate clauses. Thus, complement clauses introduced by the subjunction ‘dass’ (that) – according to German grammar – display verb-final order. German thus provides a clear signal for the grammatical incorporation of one clause into another.

N be that-constructions in everyday German conversation

The following transcript stems from a talk during a university office hour. Elke, a lecturer, has just proposed to her student Birte that – instead of starting to work on a new topic – she would be better off to write her dissertation about the same topic she explored in her Master’s thesis. (Up to this interaction Elke has assumed that Birte plans to get a Ph. D. in General Linguistics.):7 PROMOTION 1 Elke: 2 3 Birte: 4 5 6 Elke: 7 Birte:

und dann auch vie- vielleicht, (.) lieber DAS thema. das DING ist aber auchdass ich in der germanIStik promoVIEren will. (0.5) [mhm] [und] deshalb ein germanIStisches THEma brauch.

PH.D. DISSERTATION 1 Elke: and then also per- perhaps, (.) 2 THIS topic would be better. 3 Birte: but the THING is also 4 that I want to get my PH D in german. 5 (0.5) 6 Elke: [mhm] 7 Birte: [and] that’s why I need a topic within germanistics. In response to Elke’s proposal (lines 1–2), Birte produces a rejection, introduced by “das DING ist aber auch-” (‘but the THING is also-’). Already the opposition marker “aber auch” (‘but also’) foreshadows an upcoming disagreement. The matrix clause “das DING ist aber auch-” opens a projection space, which delays her main argument: that she wants to write her Ph. D. thesis in German Studies (and not in General Linguistics) and, thus, needs a new topic. Only with the closing of the second syntagma, is the syntactic ‘gestalt’ (Auer 2005) complete. In line 4, Birte’s turn comes to a point of possible syntactic, prosodic and interactional completion. However, following a brief pause (l. 5) and overlapping with Elke’s response, Birte continues her turn with a dependent clause (l. 7), introduced by the conjunction “und” (‘and’). With its verb-final positioning, this clause ties back to the complement clause in line 4 and functions as a ‘glue on’ (Couper-Kuhlen/Ono 2007: 521ff.), used to expand a prior TCU following a prosodic break.

7. The transcription is based on GAT (Gesprächsanalytische Transkriptionskonventionen). For GAT-conventions see the appendix.





Susanne Günthner

The construction, thus, consists of two parts: Part A: the matrix clause Part B: the complement clause part A

part B

N + copula-clause das DING ist aber auchbut the THING is also-

‘dass’-clause dass ich in der germanIStik promoVIEren will. that I want to get my PH D in german. (...)

This bipartite construction reveals a complex syntactic gestalt: The first component (part A) is incomplete as the predicative element is lacking. Thus, the instantiation of part A opens a projection space which – due to the open argument position – is only closed after the production of the expected constituent, and thus, the identification of the open element. Even though the two parts are syntactically integrated by means of subordinate verb order in the complement clause, they are realized in two independent prosodic contours. In uttering part A, the speaker delays the focal point (part B) so as to give it more salience and attract the recipient’s attention to it.8 The next segment is also taken from an office hour consultation at the university. Karl, Anni’s lecturer, proposes that during winter term she attend his colloquium: KOLLOQUIUM: ANNI-KARL 2005 1 Anni: ich komm auf JEden fall zu Ih[nen.] 2 Karl: [hm.] 3 (-) 4 Anni: das ding wÄre, 5 dass ich im oktOber noch ein prAktikum in ENGland hab, 6 (-) un- erst im noVEMber teilnehmen kann. 7 Karl: hm (-) 8 Karl: w- wann genAU sind sie dann wieder [HIER?] COLLOQUIUM: ANNI-KARL 2005 1 Anni: I will definitely come to y[ou.] 2 Karl: [hm.] 3 (-) 4 Anni: the thing would be, 5 that in october I still have a practical training period in england, 8. Cf. Schmid (2001: 1535) and Miller/Weinert (1998: 243) who treat N be that-constructions and ‘the thing is’-constructions as ‘focus constructions’.

N be that-constructions in everyday German conversation 

6 7 Karl: 8 Karl:

(-) an- I can only take part in november. hm (-) w- when exactly are you back [here?]

The matrix clause “das ding wÄre,”9 (l. 4) opens a projection span which is only closed when the expected component is being provided. By announcing and delaying the kernel of the message, the speaker not only holds the floor but also increases the salience of the information in the second component. part A

part B

N + copula-clause das ding wÄre, the thing would be,

‘dass’-clause dass ich im oktOber noch ein prAktikum in ENGland hab, (-) un- erst im noVEMber teilnehmen kann. that in october I still have a practical training period in england, (-) an- I can only take part in november.

In both transcripts PH. D. DISSERTATION as well as COLLOQUIUM, the ‘die Sache ist/das Ding ist’-constructions are used to introduce a face-threatening action: In PH. D. DISSERTATION, Birte states that she no longer intends to get a Ph. D. in General Linguistics but in German Studies and thus rejects Elke’s advice. In COLLOQUIUM, Anni provides the reason, why she cannot attend the proposed colloquium in time. Thus, in both cases speakers express disagreeing stances, and at the same time they set up a new (somewhat problematic) focus in the interactional proceeding. In marking their subsequent talk as ‘focal’, they prepare their co-participants to align themselves to the upcoming information. In my data, participants use ‘die Sache ist/das Ding ist...’-constructions to break the interactional continuity and to establish a somewhat different (often problematic, face-threatening) context. Furthermore, in both ‘die Sache ist/das Ding ist...’-constructions, part B shows typical features of syntactic integration into the matrix clause ‘die Sache ist/das Ding ist’. It contains the subjunction ‘dass’ (‘that’) as well as verb-final constituent order (i.e. subordinate verb order). However, in spite of their syntactic integration, the two clauses are prosodically independent; i.e. they are realized in two separate prosodic units. At the same time, the matrix clause (part A) is not a self-contained ‘turn construction unit’ (TCU); it cannot stand on its own, as the predicative element is lacking. It foreshadows what is going to come next, and, thus, enables the recipients to anticipate the following components. In PH. D. DISSERTATION and COLLOQUIUM, it projects a subject clause to follow, and attaches itself to the 9. This is the only example in my data, with the copula in the subjunctive: “das ding wÄre,” (‘the thing would be’).



Susanne Günthner

subsequent syntagma (part B). The opened projection only comes to close once the expected component is provided. The construction at hand is used for a rhetorical deferral of the focal – often face-threatening – point. Furthermore, the main predication, i.e. the interactively relevant information, is not presented within the matrix clause but in the subordinate complement clause. It is the subordinate clause which provides the ‘kernel’ of the utterance, and thus, the information which remains relevant in the following turns: In PH. D. DISSERTATION, the participants keep on talking about an adequate topic within germanistics (lines 7ff.); in COLLOQUIUM, they carry on talking about the length of Anni’s practical training (lines 8ff.). Studies within Conversation Analysis (Schegloff 1980) show that interactants frequently make use of ‘pres’ (‘pre-sequences’) in order to open up a conversational space for the focal utterance. Story prefaces, ‘pres’ to invitations, ‘pres’ to rejections, etc., they all share the function of foreshadowing an action (by the same speaker). Frequently, these subsidiary activities project something potentially face-threatening. Even though part A in ‘die Sache ist/das Ding ist’-constructions occurs in the same turn as the following part (part B),10 it shares some of these features: interactionally, these matrix clauses are no longer “profile determinant” (Langacker 1991: 436), but are subsidiary activities which are used to foreshadow the focal activity;11 i.e. the profile of the matrix clause is overridden by the following syntagma, as the interactants typically orient to the complement clause rather than the main clause (Thompson 2002b: 131ff.).12 Thus, on the basis of conversational data, we can treat the construction at hand, which builds up expectations concerning its continuation and thus, projects a certain kind of gestalt-closure, as a projective construction (Hopper 2004; 2006; Günthner 2006; 2007; 2008b; c; Hopper/Thompson 2008; Günthner/Hopper 2010). As Auer (2005b: 8ff.) points out: “projection is the adumbration (foreshadowing) of an action by a segment of talk.” Projective constructions share the following features (Hopper 2006): – they mark a break in an on-going interaction, – they may – for various reasons – delay the delivery of an important point, 10. Cf. also Pekarek Doehler (in this volume). 11. Cf. also Laury & Okamoto (in this volume) on the “complement-taking predicates “tte yuu ka” and “I mean” in Japanese and English. As the authors argue, in spoken discourse these Japanese and English constructions have become set phrases which no longer function as main clauses. 12. Cf. also Keevallik (in this volume) for similar results concerning interrogative complements as independent clauses in Estonian interactions. Cf. also Aijmer (2007) concerning English ‘the fact is (that)’-constructions.

N be that-constructions in everyday German conversation 

– they may signal an especially strong focus on a point of argument, – they may hold on the floor and forestall a predicted interruption, – they may provide a cognitive breathing space for formulating the next utterance in a maximally effective way, – they may project ‘more to come’ and thus permit participants to negotiate the future course of an interaction. In everyday talk, speakers make use of various types of projective constructions such as pseudoclefts (Hopper 2001; 2004; Günthner 2006; Günthner/Hopper 2010), extrapositions (Couper-Kuhlen/Thompson 2006; Günthner 2009a), matrix-complement-clauses (Imo 2006; 2007), formulas of thematization (Altmann 1981; Zifonun et al. 1997), ‘Complement-Taking-Predicate Phrases’ (Thompson 2002b), ‘hanging topics’ (Selting 1993; Scheutz 1997; Altmann 1981), ‘es ist so’ (‘it’s like that’)-constructions (Auer 2006) etc. These constructions are connected within a taxonomic network of related constructions with structural and functional overlaps (Croft 2001: 25). They all, for various reasons, delay the delivery of a significant segment of talk. 2.2

‘die Sache ist/das Ding ist’-utterances followed by a main clause

In the data at hand, we frequently encounter ‘die Sache ist/das Ding ist’-constructions with main clauses filling the complement part; i.e. part B is no longer introduced by the subjunction ‘dass’, instead part B follows in an asyndetic form and shows the word order of an independent sentence (i.e. verb-second positioning): ‘die Sache’ COPULA (‘ist’) → main clause ‘das Ding’ In the following segment, Sven tells his fellow student Tanja about his professor who refuses to give him credit for having attended a seminar: PHILOSOPHIE-SCHEIN 21 Tanja: dann wÜrd ich auch nich mehr (.) 22 zu dem PROF gehen, (-) 23 und ihn auch nich als PRÜFer NEHmen. 24 Sven: ne. < MACH ich auch [NICH.>] 25 Tanja: [mhm] 26 Sven: die sache is; 27 er will mir nich MAL den ↑SCHEIN anerKENnen; (.) 28 weil er sagt,



Susanne Günthner

29 es wäre maniPU[liert.] 30 Tanja: [mhm]

CREDIT IN PHILOSOPHY 21 Tanja: in that case I also wouldn’t go anymore (.) 22 to that professor, (-) 23 nor would I pick him as your supervisor. 24 Sven: no. < I won’t do [that.>] 25 Tanja: [mhm] 26 Sven: the thing is; 27 he does not even want to give me ↑CREDIT for the course; (.) 28 cause he says, 29 it would be manipu[lated.] 30 Tanja: [mhm] The beginning of this bipartite ‘die Sache ist’-construction corresponds with the canonical form above. The “die sache is;”-part (l. 26) projects by virtue of being an incomplete fragment, which has no predicative element. However instead of a subordinate complement clause, the matrix clause is followed by a syntactically as well as prosodically independent clause, displaying ‘main clause order’ (with the finite verb (“will”) in verb-second position): The syntagma “er will mir nich MAL den ↑SCHEIN anerKENnen,” (‘he does not even want to give me ↑CREDIT for the course’) (l. 27) – incrementally followed by a causal clause (“weil er sagt, es wäre maniPU[liert.]”; ‘cause he says, it would be manipu[lated.]’) – shows no sign of adhering whatsoever to rules of syntactic embedding; i.e. it is not integrated hypotactically into a complex sentence and, thus, has to be treated as topologically non-subordinate.13 Here, the bipartite construction consists of a juxtaposition of a matrix clause and a main clause (which lost its dependent clause characteristics; i.e. the initial complementizer ‘dass’ and the verb-final positioning): part A

part B

N + copula-clause die sache is; the thing is;

main clause er will mir nich MAL den ↑SCHEIN anerKENnen; he does not even want to give me ↑CREDIT for the course

As in the preceding examples, the matrix clause functions as a projector phrase indicating ‘more to come’ and, thus, foreshadowing the focal activity. Instead of a 13. Cf. Matthiessen/Thompson (1988) for a discussion of the squishiness of “subordination”. Also Cristofaro (2003) on different notions of subordination. Cf. also Laury/Thompson (2008: ix ff.)

N be that-constructions in everyday German conversation 

syntactic and conceptual dependence of the complement clause on the matrix clause, we are faced with the opposite case: The projecting matrix clause, which cannot stand on its own, now becomes syntactically, semantically, as well as interactionally dependent on the following syntagma, which holds the expected, relevant information, and which is realized as an independent clause showing no sign of subordination. Thus, the present construction can no longer be assigned to the schema [matrix clause + complement clause], as the ‘matrix clause’ no longer – neither on the syntactic nor on the interactional level – functions as the “profile determiner” (Langacker 1991: 436). In the next transcript, stemming again from a university office hour interaction, professor P and her student S are talking about possible topics for S’s oral exams. P suggests that S works on visual poetry and proposes her to focus on ‘labyrinth poems’. S, however, rejects this advice and points out that she plans to concentrate on Calvino instead:14 SPRECHSTUNDENTRANSKRIPT Nr. 21 (MEER et al. SPRECHSTUNDEN) 296 P: also irgendwie so inhaltlich. 297 (...) 298 S: das wär (auch was.) 299 P: mhm, 300 S: mhm, also ehm; 301 die sache is, 302 jetz hab ich also mit calvino angefangen; (-) 303 jetz so vom zeitplan her; ehm (2,0) 304 weiß nich wie lang das dann halt dauern wird, OFFICE HOUR TRANSCRIPT No. 21 (MEER et al. OFFICE HOURS) 296 P: well somehow from the content. 297 (...) 298 S: that would be (something.) 299 P: mhm, 300 S: mhm, well ehm; 301 the thing is, 302 I’ve started to read calvino; (-) 303 now concerning the time schedule; ehm (2,0) 304 I am not sure how long it will take, With the “die sache is,”-phrase S introduces a somewhat problematic issue: a rejection of the professor’s suggestion. Again, the “die sache is,”-syntagma does not 14. Thanks to Dorothee Meer for providing this example.



Susanne Günthner

represent a self-contained utterance but syntactically projects a subject clause to follow. The ‘complement clause’ (“jetz hab ich also mit calvino angefangen;”, ‘I’ve started to read calvino;’ (l. 302)), which is attached asyndetically, is a full-fledged and complete German sentence, with no formal sign of syntactic dependency (i.e. no subjunction, no final positioning of the verb). This syntactic independence is supported prosodically. part A

part B

N + copula-clause die sache is, the thing is,

main clause jetz hab ich also mit calvino angefangen; I’ve started to read calvino;

From a topological perspective, the matrix clauses in OFFICE HOUR and CREDIT IN PHILOSOPHY occupy the position of the pre-front field in German. (In German there is no fixed order of S and O, as in an SVO language. Instead, the topology of the standard declarative clause is defined by the position of the finite (and non-finite) parts of the verb. Together they form the so-called ‘sentence brace’.)15 The general function of the pre-front field in spoken German is to frame (often metapragmatically) the subsequent utterance (Auer 1999): pre-front field

front field

left verbal brace

middle field

right verbal brace

die sache is;

er

will

anerKENnen,

die sache is,

jetz

hab

mir nich MAL den ↑SCHEIN ich also mit calvino

end field

angefangen;

Semi-fixed phrases positioned in the pre-front field often function as framing devices or metapragmatic cues projecting a continuation (Auer/Günthner 2005). The connection between the ‘die Sache ist/das Ding ist’-phrase – positioned in the pre-front field – and the following syntagma (part B) is rather loose: First, each part has its own prosodic design, second, part B can stand on its own without being ungrammatical. Thus, due to the loss of the elements indicating grammatical incorporation (i.e. no subjunction, no final positioning of the verb) as well as the related consequence that part B is a full-fledged and complete sentence, the grammatical cohesion between the two parts is strongly reduced.

15. Cf. Auer (1996) for a detailed syntactic account of the pre-front field in spoken German. Cf. also Günthner/Imo (2004) and Auer/Günthner (2005).

N be that-constructions in everyday German conversation 

The matrix clause, positioned in the pre-front field is reanalysed as a framing device, projecting the upcoming stretch of discourse and marking metapragmatic information.16 The focal information (i.e. the information which is framed as foregrounded and is, furthermore, treated as focal in the subsequent talk) is positioned in the complement clause. This bi-part division between a framing element (positioned in the pre-front field) (part A) and a following, syntactically somewhat independent clause (part B) shows formal and functional parallels with other elements in spoken German which are regularly positioned in the pre-front field, such as ‘topic formulas’ (Altmann 1981: 82ff.; Selting 1993; Auer 1997; Scheutz 1997; Schwitalla 1997/2003; Zifonun et al. 1997: 524ff.) and discourse markers (Gohl/Günthner 1999; Günthner 1999; Auer/Günthner 2005). In all these cases, the element positioned in the prefront field is (generally) prosodically independent,17 furthermore, it represents a syntactically non complete, non-turn-constitutive unit, which is only complete with the following syntagma. Such reanalyses of matrix clauses as framing elements (Auer/Günthner 2005; Günthner/Imo 2004; Günthner 2008a; c; Rehbein 2003), which lead to an interactional (and syntactic) downgrading of the matrix part, are rather frequent in spoken language (Auer 1998). As Thompson/Mulac (1991) and Thompson (2002b) show, complement-taking constructions in spoken English also tend to be reanalysed as epistemic/evaluative/evidential phrases.18 Thompson (2002b) furthermore argues that what is traditionally considered to be a subordinate complement clause can be shown not to be ‘semantically’ subordinate; instead the complement taking phrase has been weakened. Diessel/Tomasello’s (2001) study of ‘finite complement clauses’ also reveals that the ‘CTV-clause’ (‘Complement Taking Verb-Clause’) in spoken English mainly functions as a discourse pragmatic phrase, whereas the main proposition is expressed by the ‘COMP-clause’: .... rather than being viewed as a conceptual element of the CTV-clause proposition (as in the assertive use), the COMP-clause expresses the main proposition, which the hearer interprets with the help of the CTV-clause. (Diessel/Tomasello

2001: 106).

16. Cf. Edmondson (1981: 155), who treats ‘the point is’ and ‘the thing is’ as “underscorers”; i.e. “message-oriented devices” which “draw special attention to the following, preceding, or ongoing communicative act”. 17. Cf. Auer (1996; 2005a; 2006); Auer/Günthner (2005); Günthner (2008b). 18. Cf. also Laury/Okamoto (in this volume) on complement-taking predicates in Japanese and English. Cf. Keevallik (this volume) on interrogative complements in Estonian interaction.



Susanne Günthner

Thus, if we take interactional uses of these ‘matrix-complement’-combinations seriously, we are confronted with what Aijmer (2007: 38) calls a “usage-based attack on the classical subordination analysis”, as the analyses reverse “the hierarchical relationship between the clause containing that and the clause containing the content (the ‘complement clause’)” (Aijmer 2007: 38). 2.3. ‘die Sache ist/das Ding ist’-utterances followed by a complex stretch of discourse Even though the ‘die Sache ist/das Ding ist’-syntagma can be left out without the utterance becoming ungrammatical, it still has important interactional functions: On the one hand, it anticipates the following syntagma and guides the recipients’ attention to the focal – often face-threatening – information; on the other hand it ensures the speaker the opportunity to finish her/his turn. This floor-holding function is especially important in the case of longer discourse segments. Thus, it is not surprising that ‘die Sache ist/das Ding ist’-constructions are often used to introduce complex arguments and information which extend several turn construction units (TCUs): ‘die Sache’ COPULA (‘ist’) → longer stretch of discourse ‘das Ding’ The following segment is taken from an interaction between Olga, a patient who suffers from panic attacks, and her friend Eva: PANIKATTACKEN: OLGA-EVA 21 Olga: es hat mich SEHR v- viel überWINdung ge[kOstet;] 22 Eva: [un-] 23 Eva: hm? 24 Olga: d- das ding is hAlt; (-) 25 < is nunma so;> 26 Eva: hm 27 Olga: wenn = du dat EIN.MA hAst, 28 dat LÄSST dich NICH (meh)los. 29 ECHT. NICH. PANIC ATTACKS: OLGA-EVA 21 Olga: I really had to force [myself;] 22 Eva: [un-] 23 Eva: hm?

N be that-constructions in everyday German conversation 

24 Olga: 25 26 Eva: 27 Olga: 28 29

th- the thing is; (-) < it’s like that;> hm once you have had it, you can’t escape it. REALLY. NO.

Olga describes the difficulties she had driving her car after suffering from a panic attack. Following Eva’s encouraging minimal response (l. 23), Olga starts with a general statement about panic attacks (l. 24ff.). “d- das ding is hAlt; (-)” (‘th- the thing is; (-)’) 19 opens a projection span, which is delayed for the sake of a parenthetical insertion “< is nunma so;>” (‘< it’s like that;> ‘) (l. 25) as well as Eva’s minimal response token (l. 26). part A

part B

N + copula-clause d- das ding is hAlt; (-)

complex segment +PARENTHESIS+ wenn = du dat EIN.MA hAst, dat LÄSST dich NICH (mehr)los. + parenthesis+ once you have had it, you can’t escape it.

th- the thing is;

The insertion of a parenthesis as well as a minimal response token indicate that opened projections can be deferred and still remain active across the inserted material (Auer 2005a). Thus, in using a ‘die Sache ist/das Ding ist’-construction, speakers can exploit its function of delaying or stretching the main point; the ‘die Sache ist/das Ding ist’-construction can provide cognitive and interactional space for thinking through claims. Furthermore, the construction allows speakers to keep their turn even in cases in which co-participants provide insertions – as in PANIC ATTACKS. In line 27, Olga finally finishes her construction by adding a conditional clause (“wenn = du dat EIN.MA hAst, dat LÄSST dich NICH (mehr)los.”; ‘once you have had it, you can’t escape it.’).

19. In ‘die Sache ist/das Ding ist’-constructions we frequently find modal particles and adverbs (such as ‘halt’, ‘nämlich’, ‘natürlich’, ‘aber’ etc.), which are used to contextualize speakers’ argumentative direction (such as disagreement) or to back the validity of following argument. In PANIC ATTACKS: OLGA-EVA the modal particle “halt” (“d- das ding is hAlt;”) enforces the validity of the following sententious maxim.



Susanne Günthner

The fact that recipients wait with their turn taking until the gestalt has been closed, indicates that participants orient to this constructional pattern. In the next example, the ‘matrix clause’ is followed by a complex chunk of discourse, a narrative sequence. Hilde and Hanna are complaining about the ‘impertinence of some doctors’. Hanna states that in cases in which doctors know their patients personally, they appear to be much more committed: KRANKENHAUS-ERFAHRUNG 32 Hanna: das kann natürlich sein, 33 dass er sich einfach MEHR engagiert, 34 wenn er DICH kennt. mhm. 35 Hilde: DU (.) die sache is die, (-) 36 eines morgens kam der oberarzt mit ner schwester rein, 37 un- die sAgten mir, 38 SO: (.) SIE kommen jetzt gleich innen OP zur ausschAbung. 39 ich (.) so (.) 40 WIE? 41 gestern abend sagte man mir noch, 42 alles sei in Ordnung! 43 < dann stellte sich RAUS,> 44 < die hatten MICH verWECHS[ELT!]> 45 Hanna: [BOH] 46 Hanna: un[GLAUBlich.] 47 Hilde: [echt.] ich wa: so FETTICH. EXPERIENCE IN THE HOSPITAL 32 Hanna: it might be of course, 33 that he is more thorough committed, 34 if he knows YOU personally. mhm. 35 Hilde: YOU (.) the thing is such, (-) 36 one morning the assistant medical director came in with a nurse, 37 an- they told me, 38 SO: (.) YOU will have to go into the operating room soon to have a D&C. 39 I (.) like (.) 40 WHAT? 41 last night they told me, 42 everything was okay! 43 < then it turned OUT,> 44 < they had misTAKEN[ME for someone else!]> 45 Hanna: [BOH]

N be that-constructions in everyday German conversation 

46 Hanna: un[BELIEVable.] 47 Hilde: [really.] I was: so UPSET.

After Hanna mentioned that it is advantageous to know one’s doctor personally, Hilde starts her turn with an emphatically marked vocative: “DU” (‘you’). She then introduces a “die sache is die,”-syntagma, which includes the cataphoric element “die” (‘the, such’), line 35. In my data, cataphoric elements are often used in cases in which the ‘die Sache ist/das Ding ist’-syntagma introduces complex stretches of discourse.20 In order to back her assessment concerning the impertinence of some doctors, Hilde reconstructs her past experience – in the form of an exemplary complaint-story (Günthner 2000). Again, the “die Sache ist”-phrase (l. 35) functions not only to open up a projection span indicating ‘more to come’; but it also works as a floor-holding device in order to construct a longer, sequentially complex turn, a “big package” (Sacks 1964–68/1992). Thus, long projections are not restricted to written texts, but also in spoken interactions participants are able to deal with long stretching complex sequences. In general there is a tendency in spoken interactions to omit the markers of subordination (the subjunction ‘dass’ as well as final positioning of the finite verb) in cases in which complex segments form the second part of the ‘die Sache ist/das Ding ist’-construction. In such cases, the grammatical tying between the two parts is reduced. 3. Conclusions As the analysis of ‘die Sache ist/das Ding ist’-patterns shows, the postulated N be that-schema with its syntactic format [matrix clause + complement clause] is insufficient when it comes to actual usage in spoken interaction. Speakers make use of ‘die Sache ist/das Ding ist’-constructions in order to project an upcoming syntagma as the focal point.21 However, the gestalt of the syntagma to come (part B) is underdetermined; i.e. no precise predictions can be made about the specific kind of syntactic structure that is going to emerge. Instead of producing part B as i. an integrated subordinate clause with verb final positioning – introduced by the complementizer ‘dass’: 20. However, even in cases, in which the ‘die Sache ist/das Ding ist’-phrase is followed by a complex stretch of discourse, the cataphoric element “die” is not necessary (cf. the transcript PANIC ATTACKS: OLGA-EVA). The projective force of part A (‘die Sache ist/das Ding ist’) is strong enough to cover a complex discourse segment. 21. Cf. Schmid (2001) and Aijmer (2007) for similar observations concerning “the fact/truth/ problem/trouble is that”-formulas.



Susanne Günthner

‘die Sache’ COPULA (‘ist’) → complement clause (‘dass...’ (‘that’)) ‘das Ding’ Speakers can use other formats as well: ii. part B can be realized as a syntactically non-integrated main clause (verb second positioning): ‘die Sache’ COPULA (‘ist’) → main clause ‘das Ding’ Furthermore, iii. it can consist of a complex construction or even a longer stretch of discourse: ‘die Sache’ COPULA (‘ist’) → longer stretch of discourse ‘das Ding’ In case of (i), part B is formally (by means of the complementizer ‘dass’ and verb final positioning) – but no longer conceptually – subordinate to the matrix clause. Even though the presence of ‘dass’ in combination with final positioning of the finite verb indicates syntactic subordination, the main message and central activity is no longer placed in the ‘matrix clause’ but in the ‘subordinate’ part. Thus, the grammatical coding neither reflects the information structure nor the interactive functions. In cases (ii) and (iii), the ‘matrix clause’ (‘die Sache ist/das Ding ist’) moves into the pre-front field and comes close to a metapragmatic framing device. At the same time, the ‘complement clause’ looses its dependent features, such as the subjunction ‘dass’ and the verb-final positioning. Altogether, we can observe how the ‘matrix clause’ has developed into a framing device, assigning the main information to the following part. Thus, the construction at hand lines up with certain tendencies in grammaticalization: In reanalysing a matrix clause as a framing device (i.e. a ‘projector phrase’), it is degraded in its pragmatic weight, its interactive function, as well as its syntactic features in favour of the following syntagma (i.e. the former ‘complement clause’) (Auer 1998; Günthner/Imo 2004; Auer/Günthner 2005; Günthner 2008b; Günthner/Hopper 2010). This increase in relevance on the side of the ‘complement clause’ confligates with syntactic subordination, and thus, leads to an elimination of subordination markers (i.e. loss of the complementizer, loss of verb final positioning, as well as loss of prosodic dependence). Instead of a syntactic and conceptual dependence of the ‘complement clause’ on the ‘matrix clause’, we are now faced with the opposite case: The seeming ‘matrix clause’ is syntactically,

N be that-constructions in everyday German conversation 

semantically, as well as pragmatically and interactionally dependent on the following syntagma, which holds the focal information.22 In line with this reanalysis of matrix clauses as projector phrases, these phrases become routinized and formulaic23 and are highly skewed toward certain communicative functions: They project and defer an upcoming segment of discourse as ‘focal’, and thus draw the recipients’ attention to the following segment of discourse and prepare them to align themselves to the upcoming information. It is not surprising, though, that speakers use these constructions in face-threatening (disagreements, providing sensitive information...) as well as in argumentative contexts: They delay the main point, so as to provide interactional space to contextualize a dispreferred and/or complex utterance (Pomerantz 1984) and to work out how to present their position in as non-assailable and persuasive a way as possible. Furthermore, uses of these nominal-constructions in everyday spoken interactions reveal that constructions cannot be conceived as fixed entities, but as emergent products of interactional practices. They turn out to be less schematic and more flexible in actual spoken usage than the way they are portrayed in reference grammars and linguistic studies based on written sentences. In accordance with Hopper/Thompson (2008: 119), my data support the view that biclausal sentences of literate language represent the ‘normativized’ version of the patterns of spoken language: The constructions of written language manifest greater compactness and a higher degree of syntactic integration than the more fragmentary and paratactic arrangement of spoken utterances. The counterpart in spoken conversation of the strict syntactic biclausality that we take as the norm in grammatical studies is a more or less formulaic segment of speech that serves to project an upcoming region of discourse combined with a subsequent stretch of text of indeterminate length that responds to the projection.

To conclude, I would like to rise the question: What does this study teach us about subordination-coordination? Traditionally, a matrix clause and its complement clause are seen as a biclausal structure, with the complement clause being subordinate to the matrix clause, which carries the main information. However, in studying everyday conversation, it becomes evident that formats involving grammatical phrases such as those introduced here cannot be satisfactorily explained on the assumption of a biclausal syntactic construction. The ‘die Sache ist/das Ding 22. Cf. also Verhagen (2001) and Aijmer (2007). 23. Part of the routinization of “die Sache ist/das Ding ist”-patterns is that the NP in my data is always in the singular, and the copula is mainly restricted to “ist” (present tense). There is only one example with a subjunctive use of the copula “wäre”. Cf. Aijmer (2007) for similar observations concerning the English “the fact is that”-constructions.



Susanne Günthner

ist’-construction turns out to be much more than a mere matrix-complementclause when it comes to the ways in which this construction is employed in everyday German interaction. The ‘main clause’ does not hold main information, instead it is transformed into a framing device projecting the main information which is produced in the ‘subordinate clause’. This study, once again, problematizes the prevailing assumptions of syntactic subordination as well as dichotomy between subordination and coordination (Haiman/Thompson 1984; Matthiessen/Thompson 1988; Günthner 1996; 1999; Auer 1997). The re-interpretation of a matrix clause as a pragmatic phrase, however, is not a singular case, but lines up with observations from other interactional studies of so called ‘matrix-’ and ‘complement’-clauses: Various patterns that have been treated as consisting of a main and a subordinate clause can also be seen as constructions consisting of a pragmatic phrase (i.e. a discourse marker, a ‘Complement Taking Verb-Clause’, a projector phrase, etc.) and a syntactically rather independent segment of discourse to follow (Thompson/Mulac 1991; Auer 1996; 1997; 1998; Rehbein 2003; Günthner/Imo 2004; Auer/Günthner 2005; Günthner 2008a; b; c; Hopper/Thompson 2008; Imo 2006; 2007; Aijmer 2007; Günthner/Hopper 2010). Instead of a biclausal sentence pattern consisting of a matrix- and complement-clause we are confronted with a bipartite construction, whereby the first part functions as a projector phrase and the second part can take various patterns ranging from syntactic subordination to longer stretch of discourse: ‘die Sache’ COPULA (‘ist’) → complement clause ('dass...')/ main clause/ longer stretch of discourse ‘das Ding’ Our data not only challenge the traditional idea concerning the subordinate status of complement clauses, but once again reveal that traditional written-based analysis turns out inadequate for describing linguistic features used in spoken interaction. References Aijmer, Karin. 2007. “The interface between grammar and discourse: The fact is that.” In Connectives as discourse landmarks, Agnès Celle and Ruth Huart (eds), 54–72. Amsterdam/ Philadelphia: John Benjamins. Altmann, Hans. 1981. Formen der ‘Herausstellung’ im Deutschen. Rechtsversetzung, Linksversetzung, Freies Thema und verwandte Konstruktionen. Tübingen: Niemeyer. Auer, Peter. 1996. “The pre-front field position in spoken German and its relevance as a grammaticalization position.” Pragmatics 6 (3): 295–322.

N be that-constructions in everyday German conversation Auer, Peter. 1997. “Formen und Funktionen der Vor-Vorfeldbesetzung im gesprochenen Deutsch.” In Syntax des gesprochenen Deutsch, Peter Schlobinski (ed.), 55–92. Opladen: Westdeutscher Verlag. Auer, Peter. 1998. “Zwischen Parataxe und Hypotaxe: ‘abhängige Hauptsätze’ im Gesprochenen und Geschriebenen Deutsch.” Zeitschrift für germanistische Linguistik 26: 284–307. Auer, Peter. 2005a. “Syntax als Prozess.” InLiSt (Interaction and Linguistic Structures) 41 (http://www.uni-potsdam.de/u/inlist/issues/41/index.htm). Auer, Peter. 2005b. “Projection in interaction and projection in grammar.” Text 25 (1): 7–36. Auer, Peter. 2006. “Construction Grammar meets Conversation: Einige Überlegungen am Beispiel von ‘so’-Konstruktionen.” In Konstruktionen in der Interaktion, Susanne Günthner and Wolfgang Imo (eds), 291–314. Berlin: de Gruyter. Auer, Peter. 2009. “On-line syntax. Thoughts on the temporality of spoken language.” Language Sciences 31: 1–12. Auer, Peter and Günthner, Susanne. 2005. Die Entstehung von Diskursmarkern im Deutschen – ein Fall von Grammatikalisierung?” In Grammatikalisierung im Deutschen, Thorsten Leuschner and Tanja Mortelmans (eds), 335–362. Berlin: de Gruyter. Birkner, Karin. 2008. Relativ(satz)konstruktionen im gesprochenen Deutsch: Syntaktische, prosodische, semantische und pragmatische Aspekte. Berlin/New York: de Gruyter. Couper-Kuhlen, Elizabeth and Thompson, Sandra A. 2006. “You know, it’s funny: Eine Neubetrachtung der ‘Extraposition’ im Englischen.” In: Konstruktionen in der Interaktion, Susanne Günthner and Wolfgang Imo (eds), 23–59. Berlin: de Gruyter. Couper-Kuhlen, Elizabeth and Ono, Tsuyoshi. 2007. “‘Incrementing’ in conversation. A comparision of practices in English, German and Japanese.” Pragmatics 17 (4): 513–552. Cristofaro, Sonia. 2003. Subordination. Oxford: Oxford University Press. Croft, William. 2001. Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford: Oxford University Press. Deppermann, Arnulf. 2007. Grammatik und Semantik aus gesprächsanalytischer Sicht. Berlin: de Gruyter. Deppermann, Arnulf and Elstermann, Mechthild. 2008. “Lexikalische Bedeutung oder Konstruktionsbedeutungen? Eine Untersuchung am Beispiel von Konstruktionen mit verstehen.” In Konstruktionsgrammatik II. Von der Konstruktion zur Grammatik, Anatol Stefanowitsch and Kerstin Fischer (eds), 130–134. Tübingen: Stauffenburg Verlag. Deppermann, Arnulf. 2009. “Verstehensdefizit als Antwortverpflichtung: Interaktionale Eigenschaften der Modalpartikel ‘denn’ in Fragen.” In Grammatik im Gespräch. Konstruktionen der Selbst- und Fremdpositionierung, Susanne Günthner and Jörg Bücker (eds), 23–56. Berlin/New York: de Gruyter. Diessel, Holger and Tomasello, Michael. 2001. “The acquisition of finite complement clauses in English: a corpus-based analysis.” Cognitive Linguistics, 12 (2): 97–141. Edmondson, Willis. 1981. Spoken Discourse. A model for analysis. London/New York: Longman. Fried, Miriam and Östman Jan-Ola. 2005. “Construction Grammar and spoken language: The case of pragmatic particles.” Journal of Pragmatics 37: 1752–1778. Goldberg, Adele E. 1995. Constructions. A Construction grammar approach to argument structure. Chicago: University of Chicago Press. Goldberg, Adele E. 1996. “Construction Grammar.” In Concise Encyclopedia of Syntactic Theories, Keith Brown and Jim Miller (eds), 68–71. Oxford: Pergamon. Gohl, Christine and Günthner, Susanne. 1999. “Grammatikalisierung von weil als Diskursmarker in der gesprochenen Sprache.” Zeitschrift für Sprachwissenschaft 18 (1): 39–75.





Susanne Günthner Günthner, Susanne. 1996. “From subordination to coordination? Verb-second position in German causal and concessive constructions.” Pragmatics 6 (3): 323–370. Günthner, Susanne. 1999. “Wenn-Sätze im Vor-Vorfeld: Ihre Formen und Funktionen in der gesprochenen Sprache.” Deutsche Sprache 3: 209–235. Günthner, Susanne. 2000. Vorwurfsaktivitäten in der Alltagsinteraktion. Grammatische, prosodische, rhetorisch-stilistische und interaktive Verfahren bei der Konstitution kommunikativer Muster und Gattungen. Tübingen: Niemeyer. Günthner, Susanne. 2006. “‘Was ihn trieb, war vor allem Wanderlust’ (Hesse: Narziss und Goldmund). Pseudocleft-Konstruktionen im Deutschen.” In Konstruktionen in der Interaktion, Susanne Günthner and Imo Wolfgang (eds), 59–90. Berlin: de Gruyter. Günthner, Susanne. 2007. “Brauchen wir eine Theorie der gesprochenen Sprache? Und: wie kann sie aussehen? – Ein Plädoyer für eine praxisorientierte Grammatiktheorie.” GIDI (Grammatik in der Interaktion) Arbeitspapier Nr. 6. (http://noam.uni-muenster.de/gidi/). Günthner, Susanne. 2008a. “‘Die Sache ist’: Eine Projektorkonstruktion im gesprochenen Deutsch.” Zeitschrift für Sprachwissenschaft 27 (1): 39–72. Günthner, Susanne. 2008b. “Projektorkonstruktionen im Gespräch: Pseudoclefts, die Sache istKonstruktionen und Extrapositionen mit es.” Gesprächsforschung. Online-Zeitschrift zur verbalen Interaktion 9: 86–114. (http://www.gespraechsforschung-ozs.de/heft2008/gaguenthner.pdf) Günthner, Susanne. 2008c. “Die ‘N be that’-Konstruktion (‘die Sache/das Ding ist...’) im gesprochenen Deutsch – eine interaktionale Perspektive auf Konstruktionen im Gebrauch.” In Von der Konstruktion zur Grammatik, Anatol Stefanowitsch and Kerstin Fischer (eds), 157–178. Tübingen: Stauffenburg. Günthner, Susanne. 2009a. “Extrapositionen mit es im gesprochenen Deutsch.” Zeitschrift für germanistische Linguistik (ZGL) 37: 15–47. Günthner, Susanne. 2009b. “‘Adjektiv + dass-Satz’-Konstruktionen als kommunikative Ressourcen der Positionierung.” In Grammatik im Gespräch. Konstruktionen der Selbst- und Fremdpositionierung, Susanne Günthner and Jörg Bücker (eds), 149–184. Berlin/New York: de Gruyter, Günthner, Susanne. 2010. “Konstruktionen in der kommunikativen Praxis – zur Notwendigkeit einer interaktionalen Anreicherung konstruktionsgrammatischer Ansätze.” Zeitschrift für Germanistische Linguistik (ZGL) 37 (3): 402–426. Günthner, Susanne and Imo, Wolfgang. 2004. “Die Reanalyse von Matrixsätzen als Diskursmarker: ICH MEIN-Konstruktionen im gesprochenen Deutsch.” In Jahrbuch der Ungarischen Germanistik 2003, M. Orosz and A. Herzog (eds), 181–216. Budapest/Bonn: DAAD. Günthner, Susanne und Imo, Wolfgang. 2006. Konstruktionen in der Interaktion. Berlin: de Gruyter. Günthner, Susanne and Bücker, Jörg. 2009.Grammatik im Gespräch. Konstruktionen der Selbstund Fremdpositionierung. Berlin/New York: de Gruyter. Günthner, Susanne and Hopper, Paul. 2010. “Zeitlichkeit & sprachliche Strukturen: Pseudoclefts im Englischen und Deutschen.” Gesprächsforschung. Online-Zeitschrift zur verbalen Interaktion. (http://www.gespraechsforschung-ozs.de/heft2010/heft2010.htm): 1–28. Haiman, John and Thompson, Sandra A. 1984. “‘Subordination’ in Universal Grammar.” In Proceedings of the Tenth Annual Meeting of the Berkeley Linguistics Society, Brugman, C. and Macaulay, M. (eds), 510–523. Berkeley: Berkeley Linguistics Society.

N be that-constructions in everyday German conversation  Hopper, Paul. 2001. “Grammatical constructions and their discourse origins: prototype or family resemblance?” In Applied Cognitive Linguistics I: Theory and Language Acquisition, Martin Pütz, Susanne Niemeier and René Dirven (eds), 109–129. Berlin: de Gruyter. Hopper, Paul. 2004. “The Openness of Grammatical Constructions.” 40th Annual Meeting of the Chicago Linguistic Society, April 15, 2004. Manuscript. Hopper, Paul. 2005. “Biclausal Constructions and Emergent Grammar.” Talk delivered at the Department of Germanistik at the Westfälischen Wilhelms-Universität Münster, April, 2005. Hopper, Paul. 2006. “Time and Grammar.”. Talk delivered at the Department of Germanistik at the Westfälischen Wilhelms-Universität Münster, November, 2006. Hopper, Paul. 2008. “Die Bedeutsamkeit der mündlichen Interaktion für die Linguistik: Die Pseudocleft-Konstruktion im Englischen.” In Konstruktionsgrammatik II. Von der Konstruktion zur Grammatik, Anatol Stefanowitsch and Kerstin Fischer (eds), 179–189. Tübingen: Stauffenburg Verlag. Hopper, Paul and Thompson, Sandra A. 2008. “Projectability and Clause Combining in Interaction.” In Crosslinguistic Studies of Clause Combining. The multifunctionality of conjunctions, Ritva Laury (ed.), 99–123. Amsterdam/New York: John Benjamins, Imo, Wolfgang. 2006. “A Construction Grammar Approach to the Phrase I mean in Spoken English.”, GIDI (Grammatik in der Interaktion) Arbeitspapier Nr. 4. (http://noam.uni-muenster.de/gidi/). Imo, Wolfgang. 2007. Construction Grammar und Gesprochene-Sprache-Forschung: Konstruktionen mit zehn matrixsatzfähigen Verben im gesprochenen Deutsch. Tübingen, Niemeyer. Lambrecht, Knud. 2001. “A Framework for the Analysis of Cleft Constructions” Linguistics 39 (3): 463–516. Langacker, Ronald. 1991. Foundations of Cognitive Grammar. Vol. 2. Descriptive Applications. Stanford: Stanford University Press. Laury, Ritva and Thompson, Sandra A. 2008. “Introduction”. In Crosslinguistic Studies of Clause Combining. The multifunctionality of conjunctions, Ritva Laury (ed.), 9–14. Amsterdam/ New York: John Benjamins, Matthiessen, Christian and Thompson, Sandra A. 1988. “The Structure of Discourse and ‘Subordination’.” In Clause Combining in Grammar and Discourse, John Haiman and Sandra A. Thompson (eds), 275–330. Amsterdam/Philadelphia: John Benjamins. Miller, Jim and Weinert, Regina. 1998. Spontaneous Spoken Language: Syntax and Discourse. Oxford: Oxford University Press. Ono, Tsuyoshi and Thompson, Sandra A. 1995. “What can Conversation Tell us about Syntax?” In Alternative Linguistics: Descriptive and Theoretical Modes, P.W. Davis (ed.), 213–271.Amsterdam: Benjamins. Pomerantz, Anita. 1984. “Agreeing and Disagreeing with Assessments: Some Features of Preferred/Dispreferred Turn Shapes.” In Structures of social action: Studies in conversation analysis, J.M. Atkinson and J. Heritage (eds), 57–101. Cambridge, Cambridge University Press. Rehbein, Jochen. 2003. “Matrix-Konstruktionen in Diskurs und Text.” Zeitschrift für Interkulturellen Fremdsprachenunterricht 8: 252–276. Sacks, Harvey. 1964–68/1992. Lectures on Conversation. Vol. I. Cambridge/Mass.: Blackwell. Schegloff, Emanuel A. 1980. “Preliminaries to preliminaries: ‘can I ask you a question?”’ Sociological Inquiry: 104–152. Schegloff, Emanuel A. 1996. “Turn organization: One Direction for Inquiry into Grammar and Interaction.” In Interaction and Grammar, Elinor Ochs, Emanuel A., Schegloff and Sandra A. Thompson, (eds), 52–133. Cambridge: Cambridge University Press.



Susanne Günthner Scheutz, Hannes. 1997. “Satzinitiale Voranstellungen im gesprochenen Deutsch als Mittel der Themensteuerung und Referenzkonstitution.” In Syntax des gesprochenen Deutsch, Peter Schlobinski (ed.), 27–54. Opladen: Westdeutscher Verlag, Schmid, Hans-Jörg. 2001. “‘Presupposition can be bluff ’: How abstract nouns can be used as presupposition triggers.” Journal of Pragmatics, 33 (10): 1529–1552. Schwitalla, Johannes. 1997/2003. Gesprochenes Deutsch. Eine Einführung, Berlin: Erich Schmidt Verlag. Selting, Margret. 1993. “Voranstellungen vor den Satz.” Zeitschrift für germanistische Linguistik 21: 291–319. Thompson, Sandra A. 2002a. Constructions and Conversation. Manuscript. University of California at Santa Barbara. Thompson, Sandra A. 2002b. “‘Object complements’ and conversation toward a realistic account.” Studies in Language 26 (1): 126–163. Thompson, Sandra A. and Mulac, Anthony. 1991. “A quantitative perspective on the grammaticization of epistemic parentheticals in English. In Approaches to Grammaticalization Vol. 2, Elizabeth C. Traugott and Bernd Heine (eds), 313–329. Amsterdam: John Benjamins. Verhagen, Arie. 2001. “Subordination and discourse segmentation revisited, or: Why matrix clauses may be more dependent than complements. In Text representation. Linguistic and psychological aspects, Ted Sanders, Josst Schilperoord and Wilbert Spooren (eds), 337–357. Amsterdam: John Benjamins. Zifonun, Gisela et al. 1997. Grammatik der deutschen Sprache. Band 1–3. Berlin/New York: de Gruyter.

Appendix: Transcription conventions (based on GAT) [ ] overlap [ ] = latching of new turns or single units (.) micro-pause (-), (–), (—) short, middle or long pauses of ca. 0.25–0.75 seconds, up to ca. 1 second (2.0) estimated pause of more than ca. 1 second (2.85) measured pause (measured to hundredths of a second) and = uh slurring within units :,::,::: lengthening, according to duration uh, ah, etc. hesitation signals, so-called “filled pauses” ‘ glottal stop so(h)o laughter particles during speech haha hoho heehee syllabic laughing ((laughing)) description of laughter ? high rise , rise to mid

N be that-constructions in everyday German conversation 

- level pitch ; fall to mid . low fall ((cough)) paralinguistic and non-linguistic actions and events <> accompanying paralinguistic and non-linguistic actions and events over a stretch of speech <<surprised> > interpretive comments over a stretch of speech (â•…â•…â•…â•… ) unintelligible passage, according to its duration (such) presumed wording al(s)o presumed sound or syllable (such/which) possible alternatives ((...)) omission of text specific line in the transcript which is referred to in the text ACcent primary or main accent Accent secondary accent !AC!cent extra strong accent `SO fall ´SO rise – SO level ^SO rise-fall `´SO fall-rise ↑` small pitch step-up to the peak of the accented syllable ↓´ small pitch step-down to the bottom of the accented syllable ↑`SO or ↑´SO conspicuously high or low pitch step-up or down to the peak or the bottom of the accented syllable < > forte, loud < > fortissimo, very loud <

> piano, soft <

> pianissimo, very soft < > allegro, fast < > lento, slow < > crescendo, becoming louder < > diminuendo, becoming softer ACcent primary or main accent Accent secondary accent !AC!cent extra strong accent `SO fall ´SO rise – SO level ^SO rise-fall



Susanne Günthner

`´SO fall-rise ↑` small pitch step-up to the peak of the accented syllable ↓´ small pitch step-down to the bottom of the accented syllable ↑`SO or ↓´SO conspicuously high or low pitch step-up or down to the peak or the bottom of the accented syllable < > forte, loud < > fortissimo, very loud <

> piano, soft <

> pianissimo, very soft < > allegro, fast < > lento, slow < > crescendo, becoming louder < > diminuendo, becoming softer < > accelerando, becoming faster < > rallentando, becoming slower .h, .hh, .hhh in breathe, according to duration h, hh, hhh out breathe, according to duration

Interrogative “complements” and question design in Estonian Leelo Keevallik

Research Associate, Uppsala University Some interrogative subject and object complement clauses are not treated as subordinate in Estonian interaction. They are interactionally profiled, as participants answer them as questions. Grammatically, they behave like independent clauses, displaying inversion and the turn-final question particle vä/ve. The main clauses considered in the chapter, ütle/öelge ‘say!’, räägi ‘talk/tell!’, ei tea ‘not know’, and uvitav ‘interesting’, instead function as (epistemic) particles projecting and designing questions in a sequentially and interpersonally sensitive way. Keywords: interrogative complements, indirect questions, interaction, epistemics, pragmatic particles

Introduction Grammar comes into being in the everyday usage of language. It is created, negotiated, and changed in encounters between the speakers of the language (Bybee & Hopper 2001, Hopper 1987, 1998, Schegloff 1996). When engaging in conversation, participants analyze each others’ contributions and regularly display their understanding of what was said in subsequent turns at talk. These displayed understandings are available to the analyst, who by focusing on sequences of turns can make empirically based claims on how the speakers in a certain language community deploy and interpret different grammatical devices. This proof procedure has been established as one of the cornerstones of conversation analysis (Sacks, Schegloff & Jefferson 1974: 728–729). Studying the core grammatical structures, such as clause combining and complementation, in interaction is therefore a great opportunity as well as a necessity for an adequate description of a language. The aim of this study is to look at how speakers of Estonian use clauses that have been analyzed as interrogative complements in Estonian grammar and as indirect questions in some other traditions (e.g. Payne 1997: 316). The clauses start



Leelo Keevallik

with interrogative pronouns or question particles, and they have been described as dependent clauses that follow complement taking predicates (CTPs) or as predicate adjective clauses (Erelt et al. 1993: 107–108; 289–291). The paper focuses on three prototypical CTPs, namely verbs of speaking and knowing, and one adjective. The choice is made purely on the basis of frequency, as these were the only recurrent formats of the relevant kind in the data. The three CTPs are ütle/öelge, the imperative of ‘say’, räägi, the imperative of ‘tell/talk’, and (ei) tea ‘(NEG) know’; the predicate adjective is uvitav ‘interesting’. In interactional Estonian all the four items are regularly used together with interrogative clauses. The paper argues that these interrogative clauses are not treated as subordinate complements by the speakers of Estonian, i.e. they are not cognitively, interactionally or prosodically backgrounded, or grammatically dependent. They do not display any syntactic features that would make them incapable to stand in isolation, as they involve finite verb forms and do not involve subordinate conjunctions. Instead, the clauses function as independent clauses, while the “main clauses” are used as epistemic particles and devices of structuring conversation. Their contribution is content-wise subsidiary and at a different structural level in comparison with what the interrogative clause accomplishes. There is only one process that is cognitively profiled (Langacker 1991) in the compound syntactic unit, and that is expressed in the interrogative clause. All of the formats are implemented in the social action that can most generally be characterized as asking a question. The paper will take a systematic look on question design that involves these short “main clauses” and “complements”. A number of recent studies have questioned the existence of complementation in different languages (Englebretson 2003, Laury 2006, Thompson 2002, Thompson and Mulac 1991). The speakers in actual conversations have been shown to treat the CTP clauses as evidential/epistemic/evaluative phrases that modify the upcoming clause (Thompson 2002). Therefore, the CTP clauses cannot be analyzed as being main clauses in regard to the complements. In contrast to the CTPs that have been described in the literature so far, the imperatives studied in the current paper relate to turn-taking and conversation structure rather than modify the upcoming interrogative clause. The paper thus contributes to the literature on reanalyzing the relationship between complements and main clauses, but brings in a new dimension of sequencing of actions as a crucial factor in understanding the functions of the “main clauses”. Second, it will argue that the CTPs attend to the assumed epistemic state of the conversation partner, thus working on the intersubjective level rather than phrasing the subjective attitude of the speaker. It is customary to analyze the following sentences as including a main or matrix clause and an interrogative complement clause (Erelt et al. 1993: 281–282, 289–292). The idiomatic English translations of the CTPs are best approximations

Interrogative “complements” and question design 

but since there is no word-by-word counterpart to any of these items, they will be left untranslated in the rest of the article. Transcription and glossing conventions can be found at the end of the article. (1) Ütle mis ma tegema pean. say;imp;2sg what I do:sup must:1sg ‘Tell me what I have to do’ (2) Räägi kuidas sul läks täna. tell;imp;2sg how you:ads go:imf;3sg today ‘Tell me how it went today’ (3) Ei tea mis kellast doktor Laane vastu võtab. neg know what time:elt doctor name receive:3sg ‘I wonder when doctor Laane will be seeing patients’ (4) Uvitav miks ma talle meeldin. interesting why I she:all please:1sg ‘I wonder why she likes me.’ In cases (1–3), the verb is transitive, needing an object, and the following clause is thus analyzed as an object complement clause. In Example (4) the adjective (possibly an elliptical predicate adjective clause) is followed by a subject complement clause. The traditional analysis is as follows. Main clause Ütle ‘tell me’ Räägi ‘tell me’ Ei tea ‘I wonder’ Uvitav ‘I wonder’

Object or subject complement mis ma tegema pean ‘what I have to do’ kuidas sul läks täna ‘how it went today’ mis kellast doctor Laane vastu võtab ’when doctor Laane will be seeing patients’ miks ma talle meeldin ‘why she likes me’

Using interactional data, the paper discusses the semantic, syntactic and prosodic features of these short ‘main clauses’ in relation to the following interrogative clauses. The first question is, which part is profiled by the participants in terms of the achieved action. And the second one, related and no less crucial, is how the two parts figure in the interactive tasks of turn construction, sequence organization and interpersonal epistemics. The data The data comes mainly from two contemporary spoken language corpora. The first one was collected and transcribed by myself in 1997–1998 and consists of



Leelo Keevallik

naturally occurring telemarketing calls as well as everyday calls between family members, relatives, friends, and colleagues (henceforth the LK corpus). It includes about 103,Â€000 words. The Tartu corpus is publicly available and constantly growing, consisting of about 700, 000 words at the moment. The data there comes from a variety of settings, including face-to-face conversations (henceforth the TA corpus). Internet has been used as a supplementary source. It is nowadays the largest collection of informal and interactional Estonian. My own native speaker intuition has provided the guarantee that the examples from there are indeed idiomatic. Each example is provided with a code showing its origin (LK, TA, NET respectively). Hundreds of relevant examples can of course be found on the internet. However, the analysis here depends on a close study of the evolving interactional sequences, as the immediate reaction to the format by another participant is crucial. The main weight of the analysis is therefore on the conversational data. As can be seen in Table (1), there are two imperative forms to take into account, as number is marked in verbs in Estonian. The singular forms are ütle and räägi, and the plural ones öelge (with a morphological variant ütelge which occurs four times in TA) and rääkige. The latter does not occur with interrogative clauses in the corpora due to contextual differences discussed below. The choice between singular and plural may depend on the number of recipients. More often, though, the forms are chosen according to social distance between the speakers and the formality of the situation. Plural implies distance and deference. In negation, however, Estonian verb forms take neither person nor number endings, which is why ei tea can be applied to any person or number. The numbers in the table show that out of the three CTPs the imperative plural of ‘say’ is most often used with interrogative clauses and the imperative of ‘tell, talk’ is least frequent in this collocation. Qualitatively, however, they work in a similar way. We will first look at these two verbs when followed by an interrogative clause. Table 1.â•‡ General frequencies of the items in the spoken language corpora

ütle ‘say!’ öelge ‘say:PL!’ räägi ‘tell!’ rääkige ‘tell:PL!’ (ei) tea ‘(NEG) know’ uvitav ‘interesting’

LK

TA

Total

16 50 10 â•⁄ 0 13 â•⁄ 4

22 55 + 4 (ütelge) 13 â•⁄ 0 25 17

â•⁄ 38 109 â•⁄ 23 â•⁄â•⁄ 0 â•⁄ 38 â•⁄ 21

Interrogative “complements” and question design 

Projecting actions 1: The imperative of ‘say’ A typical example of ütle/öelge comes from the very beginning of a telephone call. In Example (5) the caller inquires about when a person will arrive at his job. The inquiry, which also represents the reason for the call, is preceded by öelge. In this particular sequential environment, öelge alone cannot constitute a complete turn, as it does not “recognizably implement an action” (Schegloff 1996: 59). It is pragmatically incomplete and therefore heard by the participants as projecting additional talk. The next speaker is expected to wait until the projected unit, a question, has been produced. Throughout the examples of the paper, the “main clause” will be bolded and the “complement clause” italicized. (5) 1 K: rahvamuuseum, Mirja Kaarep kuuleb, folk.museum name name listen:3sg ‘Folk museum, Mirja Kaarep here.’ 2 I: tervist. ‘Hi!’ 3 K: tervist, ‘Hi!’ 4 I: palun öelge (0.8) mis kell: omme Siivelt please tell:imp;2pl what time tomorrow name ‘Please öelge (0.8) at what time will Siivelt’ 5 Alo tööle tuleb. name work:all come:3sg ‘Alo arrive at work tomorrow.’ 6 K: ta tuleb (.) kella kaheksaks tavaliselt, he come:3sg clock:gen eight:tra usually ‘He usually arrives at eight.’

(LK)

Öelge projects a continuation of the turn, which is witnessed by the interlocutor’s silence during the pause in line (4). If palun öelge ‘please tell (me)’ had been a full turn construction unit on its own, the speaker K could have responded in this space. The projection achieved by öelge is partly actional, as it projects a specific further action, and partly grammatical, as it projects an object. The two types of projection are, of course, interdependent (Auer 2005), as interaction is the primordial locus of the emergence and implementation of grammar. As a strong projector of a same-turn continuation, öelge is a device of turn-design.



Leelo Keevallik

In terms of the semantic/pragmatic meaning, the imperative of ‘say’ explicitly orders the recipient to say something. It thereby makes relevant a response, which is indeed provided in line (6). Epistemically, öelge as a straightforward imperative marks certainty that the recipient will have the answer to the upcoming question. The content of what the interlocutor is expected to ‘say’ is yet to come, as öelge does not convey anything about the semantics of what is coming up. Instead, it projects the action type that will be carried out in the ongoing turn, asking a question. The action itself is yet to be provided. Crucially, Estonian clauses do not regularly reveal their dependency status via word order or auxiliaries. It is only marginally possible to construct inversion yes/ no questions that will be discussed below. The clause in Example (5) thus has two alternative translations, shown as (a) and (b). (lines 4–5, Example 5) mis kell omme Siivelt Alo tööle tuleb (? clause) what time tomorrow name name work:all come:3sg a. ‘when Siivelt Alo will arrive at work tomorrow’ (dependent clause) b. ‘When will Siivelt Alo arrive at work tomorrow?’ (independent clause) Below, variant b will be used in all the translations, but the reader is requested to always keep variant a in mind, as for the speakers of Estonian both variants are equally possible. Characteristically, öelge in Example (5) initiates the topic of reason-for-call and in institutional calls regularly collocates with the polite palun ‘please’. In fact, the plural form öelge hardly ever occurs without palun1. Palun öelge, and especially öelge palun seems to be a recurrent way of initiating a question and the first topic in an institutional call. All the telemarketers in the LK corpus use the formulaic öelge palun or palun öelge to initiate the first and only topic of marketing calls, which is why the overall frequency of öelge is very high (50 cases in all). The TA corpus also includes numerous cases of öelge (59 in all). A large share of the instances there come from call centers (33), which may reflect the high epistemic expectation that the recipient will have the answer. A typical example follows. (6) 1 V: infotelefon Ker[sti tere,] information.line name hi ‘Information line, Kersti, hi!’ 1. Interestingly, palun itself has grammaticalized into a pragmatic particle from a verb form ‘I beg’ and it also occurs as a question preface, as in palun kuhu te elistate ‘please/excuse me, where are you calling?’ (Keevallik 2003: 140–153).

Interrogative “complements” and question design 

2 H:

[tere palun] öelg(e) hi please tell:imp;2pl ‘Hi, please öelge’

3 kuskohas Pärnus on ilusalong Dione. where:ins name:ins is beauty.parlor name where is the beauty parlor Dione in Pärnu? 4 V: üks hetk, one moment ‘One moment’

(TA)

Thus, öelge + palun + question is a format that is used in institutional single-task calls to initiate the first and often the only topic of the call. This kind of so-called question frames have been claimed to be characteristic of institutional discourse, at least what concerns their Swedish counterparts (Lindholm & Lindström 2003: 52–57). However, ütle/öelge is not restricted to institutional calls. Example (7) demonstrates identical usage in a non-institutional call between members of the same housing unit. As these speakers address each other in singular, the verb occurs in singular. (7) 1 H: =tere Alo. kuule Veiko Tohter on siin. hi name kuule name name is here ‘Hi Alo, listen, Veiko Tohter here.’ 2 I: noh, ‘Yeah’ 3 H: .hh kule ütle palun kudas hh edeneb kule tell;imp;2sg please how progress:3sg ‘Listen ütle please how is it going’ 4 meil see hh erastamine meil seal – we:ads this privatization we:ads there ‘with the privatization there?’ 5 I: ei edene, ei edene.= neg progress neg progress ‘It’s not.’

(LK)

In non-institutional conversations, ütle frequently follows the informal turn-initial kule, originally kuule ‘listen:IMP:2SG’, which is regularly used to present the reason for the call in Estonian and several adjacent cultures (Hakulinen et al. 2003). While kule does not put constraints on the grammatical nature of the upcoming syntactic units (Keevallik 2003: 53–74), ütle projects an interrogative clause, a



Leelo Keevallik

question for the interlocutor to be responded to. Both regularly initiate reason-forthe-call turns, thus displaying a certain amount of overlap in function. However, the usage of ütle/öelge is not restricted to the beginning of a conversation. It may be used more widely to initiate and pursue alternative topics. In the middle of an ongoing conversation, öelge or ütle is used to project a question that initiates a new (sub)topic. In Example (8), the speaker E is describing some logistical matters of a Christmas performance, when A starts competing for the turn with an information request concerning the length of the performance. This competing topic is initiated with an ütle-question. (8) 1 E: [ja] sis: ee, me and then we And then we 2 [lähme karjaste juurde sisse.] go:1pl shepherd:pl;gen to;ill in;ill go in to the ‘shepherds.’ 3 A: [ütle mitu minutit Ene ] say;imp;2sg many minute:prt name ‘Ütle Ene how many minutes’ 4 see kokku tuli präägu. this together come:imf;3sg now ’the whole thing is now?’ 5 E: h ee, eks ta tuleb umbes pool tundi. eks it come:3sg about half hour;prt ‘Well it should be about half an hour.’

(LK)

A’s question in lines (3–4) is an initiation of a longer sequence concerning the problem that too few people would be able to see the show, as the space for the audience is limited. On the other hand, the show is planned to run several times, which is why its length is of importance. The ütle-initiated turn thus initiates a whole new topical and action sequence. It breaks the ongoing logistical explanation by E and makes an answer by her a next relevant action. Ütle initiates a new local trajectory for the conversation. The speaker A initiates the turn and the topic in overlap, without reacting to the prior turn. Being an item that only defines the nature of the upcoming turn without giving away any of its semantic content, ütle is a suitable turn-claiming device to be produced in overlap. It need not be properly heard. Ütle is a device for the next speaker to escape the sequential context and contingencies of the prior turn. Every turn is inescapably produced in the given context and ütle addresses this problem as a displacement marker that locally accomplishes a shift from the

Interrogative “complements” and question design 

relevant next action to an initiation of something new. It has been argued that imperatives generally claim noteworthiness and can therefore develop into this kind of turn-claiming items (Waltereit 2002). It seems, though, that any item that is regularly turn-initial, may also function as a turn-claiming item, including some imperatives. In all of the examples with the imperative of ‘say’ + interrogative clause, the recipient of the turn responds by answering the question. This proves that the interrogative clause is profiled in conversation. It constitutes the main focus for the participants that attend to the interrogative. Subsequent action by the recipient is the criterion used by Sandra Thompson (2002: 131–134) to demonstrate that the “complement clause” is profiled in interaction. She argues that claims on which clause is profiled in a grammatical structure need to refer to the course of actions that the participants are engaged in. Thompson demonstrates that in conversational data the profiles of this type of “complement clause” regularly override those of the “main clause”, because the recipients of these turns typically orient to the “complement clause” in their subsequent actions. Accordingly, when the Estonian speakers regularly orient to the interrogative clause as making relevant an answer, it is interactionally profiled and overrides the “main clause”. In the case of ütle/öelge, however, the main clause also gets a response, as in it the interlocutor is literally urged to say something and she indeed does. At the same time, the imperative is quite redundant – a question alone would have accomplished exactly the same task of pursuing an answer to the question. This is one reason for looking closer into when and why these questions are designed with the preface. The general pattern that emerges for ütle/öelge-question is that they are not evenly distributed across all the data. As has become clear in the examples above, they are specifically used to initiate a new topic and sequence, including the initiation of the first topic of telephone calls. Their function is to create structure in the conversation. The imperative CTP preface marks a distinct action boundary in interaction, making salient certain transitions between larger chunks of talk. Besides the distributional pattern, there are also grammatical and turn-constructional reasons for considering ütle/öelge a question preface rather than a main clause. For example, while projecting the action type of the upcoming turn, it does not have to be adjacent to the question. The interrogative clause can be produced further down the turn after an intervening segment, which should be impossible in case ütle/öelge was truly a main clause, as CTPs precede their complements in Estonian. Excerpt (9) illustrates this point. In this telemarketing call, the segment that immediately follows öelge palun ‘tell me please’ is not a question but a statement. After öelge palun the telemarketer introduces some background information to the question he is about to pose. At the same time, öelge palun still projects a question. There is no final prosody at the end of the statement about the



Leelo Keevallik

complementary subscription, and the turn is not pragmatically complete either. Accordingly, the interlocutor refrains from taking the turn until the question has been produced. (9) 1 M: [.h] öelge palun te tegite tell:imp;2pl please you:pl do:imf:2pl ‘Öelge please, you made an introductory’ 2 tutvumistellimuse Liivi introductory.subscription:gen name ‘subscription for Liivi Linnaleht,’ 3 Linnalehele, kas leht käib name:all ques paper go:3sg ‘do you receive the paper’ 4 teil präegu. you:pl:ads now ‘at the moment?’ 5 K: jaa tänan, käib,= yeah thank.you go:3sg ‘Yeah I do, thank you’

(LK)

Interlocutors treat ütle/öelge as projecting a request for information. Other units may come in between but the speakers do not treat their completion as a transition relevance place, i.e. a place where the other speaker could start speaking. The turninitial imperative of ‘say’ can thus be used to initiate longer multi-unit turns ending in an information request. This interactional structure, where a segment projects an action which is then delayed by another unit, has been described by Schegloff (1980). Although the projective segments discussed by him are clausal, such as let me ask you this or can I ask you a question, the Estonian single word ütle/öelge functions in a similar way with the exception that the segment itself never elicits a recipient response (a go-ahead). Furthermore, the upcoming side-unit is rarely responded to by the recipient. The whole format is thus much more concise in the case of Estonian ütle/öelge which seems to be a device for turn-design rather than an entire action sequence. However, just as described by Schegloff (1980: 114) for English, and Lindholm and Lindström (2003: 50) for Swedish, the side-units following the projective segments introduce the referent. In the above example, the referent is the subscription to the newspaper that the upcoming question concerns. The side-unit after the initial projection of a question is produced and heard as being in the service of the projected action. Its beginning latches prosodically smoothly to the preface and the falling pitch at its end is not interpreted as a place where the next speaker can take the turn.

Interrogative “complements” and question design 

The preface ütle/öelge is a means of showing at the earliest possible point in the turn that an information request is coming up. It projects a question which does not have to occur adjacent to it and cannot therefore be considered its complement. Additional evidence of the autonomy of the CTP and the interrogative clause can be found in cases where ütle/öelge occurs within a particle chain. In these cases the CTP is not adjacent to the interrogative clause either. In Example (10), ütle is used turn-initially among other pragmatic particles. The question it projects is produced after the particle chain including kuule, approx. ‘listen’, ütle, aga ‘but’, and näiteks ‘for example’. (10) 1 M: kuule ütle aga, (.) näiteks mis kuule say;imp;2sg but example:tra what ‘Listen ütle but, (.) for example what’ 2 emotsioonid sind valdaksid, kui sa emotion:pl you:prt fill:cond:3pl if you ‘emotions would you have if you were’ 3 oleksid juba vanaema. be:cond:2sg already grandmother ‘a grandmother already?’

(TA)

Finally, there is word order evidence that ütle/öelge is a question preface rather than a main clause. There is one optional syntactic difference between independent and dependent interrogative clauses in Estonian and that is inversion in yes/ no questions. Inversion cannot be used in dependent interrogative clauses, while it is occasionally used in main clauses. And it occurs after ütle/öelge, as in Example (11). This is the grammatical evidence that the interrogative clause used after ütle should be analyzed as an independent clause and not as a complement. (11) 1 E: halloo. Ene kuuleb. hello name listen:3sg ‘Hello, Ene listening.’ 2 H: tere:. Klaarika siinpool. hi name here ‘Hi! Klaarika here.’ 3 E: mhmh? ‘Uhuh’ 4 H: .h ütle on sul Kristeli: telefon. say;imp;2sg is you:ads name:gen phone ‘Ütle do you have Kristel’s phone number?’ 5

(0.7)

Leelo Keevallik

6 H: on tal üldse telefoni. is she:ads üldse phone:prt ‘Does she have a phone?’

(LK)

There are thus two types of syntactic evidence that the interrogative clauses projected by ütle/öelge are not subordinate or dependent. First, they may be non-adjacent to the CTP and second, inversion may be used. The reasons for the preface implementation are rather interactional, concerning turn-design and sequencing of actions in a conversation. In addition to the syntactic evidence, prosodic features suggest that ütle/öelge is not profiled for the participants. It is often incorporated into the following intonation unit as an enclitic element. Occasionally, it carries some amount of stress or a high onset as is characteristic of reason-for-the-call turns (Couper-Kuhlen 2001) but it does not stand out as the most heavily stressed item in the turn or utterance. Instead, it is often produced quicker than the upcoming question and could therefore be heard as subsidiary to it2. The pitch peak is regularly on another item, as shown in Figure 1. In summary, there is hardly any reason to analyze the interrogative clauses following ütle/öelge as subordinate complements. Ütle/öelge indeed projects a question but since it is not necessarily adjacent to the interrogative clause, it cannot be considered a CTP. Rather, it structures the conversation by making explicit the initiations of new topics, functioning like a particle. In fact, it is not uncommon for an

Pitch (Hz)



300

200 150

0

1.85737 Time (s)

Ü t l e o n s u l K r i s t e l i

telefon

Figure 1.â•‡ The pitch contour of Example (11), line (4)

2. In a similar way, the corresponding Swedish question-projecting phrase jag undrar ‘I wonder’ has been shown to have subordinate prosody with no stress and higher tempo than the surrounding units (Lindström 2002: 65).

Interrogative “complements” and question design 

imperative matrix clause to diachronically develop into a particle (Brinton 2001). Ütle/öelge provides a metapragmatic frame anticipating the following segment, as has been shown for e.g. N-be-that-constructions, pseudo-clefts and extrapolations in German (Günthner 2008, this volume; Hopper & Thompson 2008). At the same time, the CTP ütle/öelge may be part of an entire clause, when projecting the turn type and structuring conversation in the above ways. Formats such as ütle mulle seda ‘tell me this’ and ütle mulle niisugust asja ‘tell me such a thing’ occur, where the cataphoric pronoun seda ‘this’ and proadjective niisugust ‘such’ further highlight the projecting nature of the clause. This means that the verb-only format is but one device among a number of functionally identical semantically relatively empty formats for question projection, some of which are more prominent by virtue of their length. However, there is one positional feature that is idiosyncratic to the verb-only format. It is marginally possible to use at least the singular ütle at the end of the turn, after the interrogative clause (a single case in the corpora). (12) 1 R: kule, mis pääval teil vaheaeg kule which day:ads you:pl:ads vacation ‘Listen, when will your school vacation’ 2 akkab koolis ütle. start:3sg school:ins say;imp;2sg ‘start ütle?’ 3 E: mm, (0.2) ma arvan et üheksateiskümnendast. I think:1sg that nineteenth:elt ‘Um, (0.2) on the nineteenth, I think.’

(LK)

Importantly, this is a topic initiation similar to all the other cases of ütle/öelge usage presented earlier. The topical break is also marked by the turn-initial kule, approx. ‘listen’. Like in the cases described above, ütle is here prosodically part of the same unit as the interrogative clause. The intonation does not fall before the turn-final ütle. Also, the interrogative receives an answer. Sequentially and compositionally, this format is identical to the ones described above. However, the turnfinal ütle is heard as an insisting item, reinforcing the question. Rather than projecting the question, it is here retroactively pushing for a response. It is a more pronounced epistemic device than its turn-initial relative, because of this insisting force. The mobility of ütle in a turn underscores its nature of a particle, since an imperative CTP can never occur after the complement clause in Estonian. Ütle/ öelge is not a main clause but an interactional and interpersonal element of question design, a particle that cannot take complements. It is indeed important for the participants, but at a different organizational level as compared to the interrogative clause that sets the action trajectory for the next turn.



Leelo Keevallik

Projecting actions 2: The imperative of ‘tell, talk’ The second person imperative form of räägi ‘tell, talk’ is used very similarly to the imperative of ‘say’ in conversation. It is somewhat less frequent, and in contrast to ‘say’, the plural form räägite is not used with interrogative clauses in the available corpora. This is not a coincidence, as the format of imperative of ‘tell, talk’ + interrogative clause is characteristic of informal style and the verb can thus be expected to occur in singular only. In everyday interaction, the choice between the two verbs is to some extent idiomatic. Similarly to ütle/öelge, the turn-initial räägi initiates a new or the very first topic in conversation. Produced immediately after a self-introduction or a greeting, it cannot constitute a turn on its own, because it does not provide a complete action. It is therefore understood as projecting a continuation by the same speaker, specifically showing that an information request is coming up. Example (13) is a case in point, where after the call-taker’s response to the summons in line (1) the caller initiates an information request with räägi immediately after the greeting. (13) 1 P: jaa, ‘Yeah.’ 2 V: no tere. räägi kuidas sul no hi tell;imp;2sg how you:ads ‘Hi räägi how did (it)’ 3 läks täna. go;imf;3sg today ‘go today?’ 4

(2.0)

(LK)

It turns out, however, that P is not the person that the caller hoped would pick up the phone. Therefore there is a long pause after the elicitation and no response to the question. In its literal meaning, ‘telling, talking’ implies a longer account from the interlocutor than ‘saying’. This may be reflected in the above example with an open question. However, räägi is also used to initiate polar questions with potentially simple short answers identically to ütle/öelge, as happens in Example (14). (14) 1 H: /—/ kesse kuuleb seal. who listen:3sg there ‘Who is listening?’ 2 V: Üllar? name ’Üllar.’

Interrogative “complements” and question design

3 H: Üllar oled jah? (.) kule räägi name be:2sg yeah kule tell;imp;2sg ‘You’re Üllar. Listen, räägi’ 4 kas sul vanaema ka kodus ques you:ads granny too home:ins ‘is your granny at home’ 5 [on vä.] is ques ‘too?’ 6 V: [jaa,] ma kutsun kohe. yeah I call:1sg at.once ‘Yeah, I’ll call her right away.’

(TA)

A yes/no interrogative basically makes relevant a one-word answer, as far as the grammar is concerned. The action it implements, however, may be varied. In the above case, the question is a request for another speaker, which is complied with in line (6). Nevertheless, the ‘telling’ elicited by the räägi-initiated question is very short, which demonstrates the semantic bleaching of räägi. This is characteristic of the grammaticalization process of particles. Similarly to the ütle/öelge pattern, there are both syntactic and prosodic indications to suggest that räägi + interrogative clause cannot be analyzed as main clause + complement. Syntactically, räägi may be implemented before interrogative clauses with inversion, whose analysis as complements is impossible. (15) 1 M: -lo 2 K: tere. ‘Hi!’ 3 M: tere, ‘Hi!’ 4 K: kule räägi on sul see: m, (.) mm kule tell;imp;2sg is you:ads this ‘Listen räägi, do you have the’ 5 selle ambaarsti see, kaart seal this;gen dentist:gen this card there ‘dentist card somewhere’ 6 kuskil käepärast. ks sa ütleks mulle somewhere at.hand ques you say:cond i:all ‘there at hand. Would you tell me’



Leelo Keevallik

7 selle telefoninumbri mis seal peal on. this;gen phone.number:gen that there on is ‘the phone number that’s on it.’ 8

(5.0)

9 M: vot ei ole. vot neg be ‘I don’t’

(LK)

Prosodically, the item is latched to the upcoming prosodic unit, similarly to ütle/ öelge. Even when there is some amount of stress on räägi, it never constitutes a prosodic unit on its own and always projects a continuation. There is no stress on räägi in the above Example (15) and the highest pitch is on the first word of the interrogative clause, as shown in Figure 2. The prosodic features of räägi suggest that the speakers do not necessarily treat it as the profiled main clause but as something subsidiary to the interrogative clause.

Pitch (Hz)



300

200 150

0

Time (s)

5

Räägi on sulsee ambaars tikaart sealkuskil käepärast

Figure 2.â•‡ The pitch contour of Example (15), lines (4–6)

Similarly to ütle/öelge, räägi can be applied to project a question further down the turn. The question does not have to be adjacent to the projective item itself. Also, it can be part of a longer clausal format, such as räägi seda ‘tell (me) this’, which is functionally identical with the short format. However, only the latter can appear turn-finally, as shown in Example (16). (16) 1 P: jaa ma kuulen, yeah I listen:1sg ‘Yeah, listening’ 2 E: tere Pille, Ene siin. hi name name here ‘Hi Pille, Ene’s here.’

Interrogative “complements” and question design 

3 P: tere Ene. ‘Hi Ene’ 4 E: no kuidas läks. no how go:imf;3sg ‘How did (it) go?’ 5

(0.7)

6 P: ma just jõudsime. I just arrive:imf:1pl ‘I/we just came in.’ 7 E: mhmh, <=no kuidas läks räägi. uhuh no how go:imf;3sg tell;imp;2sg ‘Uhuh, how did (it) go räägi!’ 8 P: täitsa normaalselt läks. /—/ quite normally go:imf;3sg ‘Quite okay.’

(LK)

In this call, räägi is used while trying to initiate the first topic of the call, as was shown above with the turn-initial usage (Example (15)), albeit in this case it is the second attempt. While the imperative to talk always implies that the recipient of the turn is able to provide the necessary information, the epistemic connotation seems to be even stronger in turn-final usage. It restates that an answer is strongly expected after the question has already been produced. The insisting character of the format is in this particular case (the only one of its kind in the corpora) used to re-launch the question that was already posed in line (4) but did not receive an answer. The mobility of räägi within the turn and in relation to the interrogative clause suggests that they are independent units and that there is no subordinating relationship between the two. As a particle, räägi does not have the ability to take object complements and it does not define any grammatical feature in the interrogative clause. Consequently, the interrogative clauses following and projected by räägi are to be considered independent. The format räägi + interrogative clause usually elicits a telling from the interlocutor. It is regularly used in topic-initiations, as a means of overtly marking the topical structure of the conversation. It sometimes implies longer accounts than ütle/öelge and is more characteristic of informal talk. Turn-initially, both of these items project the type of action that the turn will be doing, namely asking a question. Neither of them carries any semantic information about the content of the upcoming unit, apart from conveying the assumption that the recipient will be able to answer the question. They project the action of information request in the current turn, while not yet doing the action itself. This is characteristic of question



Leelo Keevallik

frames cross-linguistically. For example, Swedish question frames, such as får jag fråga ‘can I ask’, jag tänkte fråga ‘I thought I’d ask’ and jag undrar ‘I wonder’ function in the same way (Lindholm & Lindström 2003: 41)3. At the same time, all of these formats also project what the next speaker is expected to do after the current turn has been terminated, namely, to answer the question. They thus help the speakers to coordinate their actions in time, which is a basic challenge for participants in interaction. Marking intersubjective uncertainty: The negation of ‘know’ The third very common CTP that is regularly used before interrogative clauses in conversational Estonian is (ei) tea ‘don’t know’. Similarly to the above particles, (ei) tea projects more talk by the current speaker. Consider the following Example (17) from a call to the reception of a medical center. (17) 1 V: registratuur tere ‘Reception, hi!’ 2 H: tere. (.) ei tea mis kellast hi neg know what clock:elt ‘Hi, ei tea when will’ 3 doktor Laane vastu võtab. doctor name receive:3sg ‘Doctor Laane be seeing patients?’ 4 V: e kella neljateistkümnest seitsmeteistkümneni. clock fourteen:elt seventeen:ter ‘From two to five p.m..’

(TA)

The turn in lines (2–3) involves a projective ei tea and an interrogative clause. It is treated as a question by its recipient, as is evidenced by his answer in line (4). Formally, however, ei tea ‘don’t know’ + interrogative clause constitute a statement about speaker knowledge. In translation, that would mean ‘(I/you) don’t know when doctor Laane will be seeing patients’. Indeed, the longer patterns with pronouns, te ei tea ‘you:PL don’t know’ and sa ei tea ‘you:SG don’t know’ also occur in identical function, and the shorter version (ei) tea is likely to have developed from these in the process of grammaticalization (Keevallik 2006). In its shortest form, 3. Estonians generally tolerate more imperatives than Swedes, who use more interrogatives and modals to formulate e.g. directives (Tulviste et al. 2002). The two imperative patterns in this study may in fact reflect the Estonian direct speaking style.

Interrogative “complements” and question design 

only tea ‘know’ is used. In Example (18) tea + interrogative clause is interpreted as a question by M, who answers it in lines (7–9) after an insertion sequence. (18) 1 R: tea kas sul vanaema juba know ques you:ads grandmother already ‘Tea did your grandmother receive’ 2 sai sünnipäevakaardi kätte. get:imf;3sg birthday.card:gen hand;ill the birthday card already? 3 M: millal sa posti panid. when you post;ill put:imf:2sg ‘When did you post it?’ 4 R: laupäeva ommikul siis kui Saturday;gen morning:ads then when ‘On Saturday morning when we went’ 5 sides käisime. post.office:ins go:imf:1pl to the post office. 6 (1.5) 7 M: täna kätte then she must:cond today hand;ill ‘then she should have received’ 9 saand olema:. get:ppt be:sup ‘it today.’

(TA)

In the above Examples (17–18), the interrogative clause is treated as implementing the main action in the turn. The recipients respond to the interrogative clause and not to the CTP clause that makes a statement about knowledge. In other words, the participants analyzed the “complement” to be profiled. This is an argument for not considering ei tea the main clause in the above examples. Instead, (ei) tea functions as an epistemic expressing the speaker’s stance. This stance stems from its original meaning of ‘not knowing’ something. (Ei) tea-questions are in fact regularly implemented in cases when the recipient of the question is quite likely to lack the knowledge to respond to the question. There is thus an interesting conflict between the assumption that the recipient is not likely know and asking anyway, in



Leelo Keevallik

case she does. Example (19) comes from a conversation between two sisters who discuss whether a third sister would like to join them at their father’s birthday. The recipient of the question does not necessarily have this knowledge, just like the recipient in Example (18) above is quite likely not to have information about whether his grandmother has already got her birthday card. This assumed uncertainty by the recipient is expressed in the (ei) tea-preface to the question. (19) 1 E: jah, ei tea kas Viiu tahab ka yeah neg know ques name want:3sg too ‘Yeah, ei tea does Viiu want to’ 2 tulla ve. come:inf ques ‘come too?’ 3 M: no, v:ist ikka tahab. no probably ikka want:3sg ‘She probably does.’

(LK)

As it turns out, responses to questions in Examples (18) and (19) are accordingly epistemically marked. The ‘should’ in Example (18) and ‘probably’ in Example (19) mark epistemic uncertainty. Their producers thus confirm the prior speaker’s assumption that they do not have definite knowledge of the matter asked about. The question preface (ei) tea marks uncertainty, especially the assumed lack of knowledge by the recipient of the projected question. In Example (17), however, the strategy is employed for politeness reasons. It is less imposing to request information while at least formally leaving open the possibility that the recipient may not have the knowledge. In this way, the (ei) tea-format also deals with the appropriateness of the question posed to the current recipient. Questions are by definition addressed to knowing recipients and (ei) tea marks that this condition may not be fulfilled. Implying that the question may not be properly addressed eases the burden of answering for the recipient and diminishes the imposition on her autonomy. Besides, avoiding person reference with items such as (ei) tea helps to conceal who the implied non-knowing subject is and thereby minimizes the threat to the recipient’s social face. The negation ei tea in Estonian can be applied to all persons and numbers, since verbs are not conjugated in the negative form. In the above examples, it can be anybody who lacks the knowledge, not necessarily the speaker or the hearer. Although the pattern has most probably developed from sa ei tea ‘you don’t know’, the insinuation of recipient’s lack of knowledge (Keevallik 2006: 179–181), the question preface (ei) tea does not make this explicit. This distinguishes the Estonian (ei) tea from pragmatically similar question prefaces in other languages, such as the Swedish jag undrar ‘I wonder’, which underlines speaker’s

Interrogative “complements” and question design 

lack of knowledge (Lindström 2002: 64–65), and the English I wonder4. All of the items claim lack of knowledge but the Swedish and English ones mention an agent. It is clear who the non-knowing subject is. In contrast, personal reference is avoided in the Estonian format, as it is on a wide scale in the Estonian speech community (Erelt 1990, Keevallik 1999: 130–131, Keevallik 2006: 189–191). The (ei) teaformat enables the speaker to pose the question in a generic impersonal manner and eases the contextual constraints on answering it in the next turn. Interactionally, this contrasts with the imperative-prefaced questions that assumed recipient knowledge and obliged them to answer. Indeed, the questions are sometimes treated as rhetorical (Keevallik 2006: 186–189). To summarize, there is interactional evidence in participant behavior that the interrogative clause is the profiled clause, independent and attended to by the next speaker. At the same time, (ei) tea is an epistemically downgrading preface that achieves certain interactional ends. The independence of the interrogative clause is further supported by grammatical evidence. First, a dependent interrogative clause cannot be terminated with the turn-final question particle ve/vä ‘or’, as the ei tea-prefaced interrogative clause does in Example (19). Second, inversion cannot be used in dependent clauses, but occurs in (ei) tea-prefaced questions, as shown in Example (20). (20) Ei tea on mul õigus või mitte neg know is i:ads right or not ‘Ei tea am I right or not?’

(NET)

These cases demonstrate that the interrogative clauses following (ei) tea are not syntactically dependent. Additional evidence can be provided by showing that ei tea (i.e. with the negation word) is mobile in the turn, thus behaving like an epistemic phrase. It does not have to occur utterance-initially, as epistemic stance may be added after the modified unit has been terminated. This kind of mobility has, for example, been shown for the English epistemic phrase I guess (Thompson & Mulac 1991). The turn-final usage of the epistemic ei tea is demonstrated in Example (21), where the telemarketer terminates his question with it, adding the epistemic stance after the question has been terminated. (21) 1 M: a millal ta tuleb ei tea. but when she come:3sg neg know ‘But when (will) she come back ei tea?’ 2 P: .h vot mai tea. /—/ vot i:neg know 4. In requests, the English I wonder addresses the entitlement of the speaker to have the request granted (Curl & Drew 2008). This does not seem to be an issue with the preface (ei) tea.



Leelo Keevallik

‘I really don’t know.’

(LK)

The epistemic ei tea can furthermore occur in the middle of the interrogative clause. (22) 1 Millest see ei tea tuleb et mõnel what:elt this neg know come:3sg that some:ads ‘What it ei tea depends on that some get (it)’ 2 läheb tööle ja mõnel mitte go:3sg work:all and some:ads neg ‘to work and some not?’

(NET)

In these cases, it is impossible to argue that the interrogative clause is in any way subordinated to ei tea. Rather, ei tea is embedded in or added to the interrogative clause for the purposes of expressing an epistemic stance. Even prosodically, (ei) tea is usually non-prominent with regard to the interrogative clause and incorporated into the preceding, following, or ongoing intonation unit. Occasionally it displays a certain amount of stress, as in Example (17) but it still forms a single coherent prosodic contour together with the interrogative clause. Unfortunately the quality of these examples does not allow drawing pitch curves. In contemporary informal Estonian, the original negative verb form (ei) tea has become a mobile epistemic device for question design, the aim of which is to mark the posed question as potentially unanswerable by the current recipient. As such, it can also be implemented for politeness purposes. The speakers treat the format (ei) tea + interrogative clause as a modified question that makes relevant an answer, albeit contingent upon availability of knowledge. While the interrogative clause is thereby analyzed as profiled and autonomous, (ei) tea attends to the interpersonal dimensions of its production in the current situation for the present participants. Marking topical continuity across sequences: The adjective ‘interesting’ The last format under discussion, the assessment word uvitav5 + interrogative clause, is different from the previous ones because it does not contain a verb as one of the two components. Its first part can be seen as an elliptical predicate adjective clause ‘it is interesting’, while the second half constitutes the subject clause, containing the information about what the interesting thing is. In the following example, ‘interesting’ precedes an interrogative clause inquiring about an aspect of what has just been reported by another participant. In lines (1–2), M quotes his grandmother. 5. In standard language, the word is huvitav but in colloquial usage, word-initial h is customarily dropped in Estonian.

Interrogative “complements” and question design 

(23) 1 M: nimodi ta koguaeg ju mõtleb name:ads like.this she always ju think:3sg ’she thinks like this all the time and sighs’ 3 ja ohkab ja räägib teistele ka. and sigh:3sg and tell:3sg others:all too ‘and tells everybody else too.’ 4 R: @ 5 M: päriselt. (.) ma juba tian oma vanaema. really I juba know:1sg own granny ‘Really. (.) I know my granny.’ 6 R: jah, ’Yeah?’ 7 M: jaa, ’Yeah.’ 8

(2.5)

9 R: uvitav miks ma talle meeldin. interesting why I she:all please:1sg ‘Uvitav why does she like me.’ 10 M: noh, (0.8) sa oled ju: (0.5) nii tore. noh you be:2sg ju so friendly ‘Well, (.) you are (0.5) so friendly.’

(TA)

The assessment uvitav in line (9) could be retrospective, commenting on what M just reported and thus a pragmatically complete turn. However, it is produced as part of the upcoming turn, projecting an expansion on what exactly is ‘interesting’ (see Couper-Kuhlen and Thompson (2008) on retrospective and prospective assessments and “extraposition” in English). It could be followed by a declarative as well as an interrogative clause, so it is not unambiguously projecting a question, like the above verbs did. Before looking closer at the format uvitav + interrogative clause, let us look at a more general syntactic pattern that it represents. The pattern concerns a number of other assessment words, such as hea ‘good’, kahju ‘pity’, and tore ‘great’. These words regularly occur without the copula and any correlate word before clauses that explicate what is assessed by them. There is a distinct pattern in Estonian, in which assessment words are used with what are formally complement clauses,

 Leelo Keevallik

e.g. initiated by the complementizer et. The words are more adequately analyzed as stance-taking devices in regard to the upcoming clause (Keevallik 2010). It has been argued that a similar pattern in German is a result of a grammaticalization of the respective full matrix clauses involving the copula (Auer 1996: 315–320). In the process, original reactive adverbials and one adjective gut ‘good’ have become projective utterance-initial items in conversation. According to this account, the pattern is to be considered elliptical, at least historically. In contrast with the pattern described by Auer, Estonian pattern involves a considerable number of different adjectives (Keevallik 2010). Examples (24–26) display cases of declarative subject clauses used in combination with assessment words, an adjective (24) and an adverbial (25) that very frequently occur in this syntactic pattern. (24) H: selge no igal juhul hea et sa oled veel clear no anyway good that you be:2sg still ‘Okay. Anyway, (it is) good that you are still’ reedel Tartus. Friday:ads name:ins ’in Tartu on Friday.’ (25) P: kahju et sa ei saa tulla. pity that you neg can come:inf ’(What a) pity that you cannot come.’ (26) 1 M: .hh % e uvitav et ma:: % enne interesting that I earlier ’(It’s) interesting that I checked’ 2 neid Rootsi pileteid these:prt Sweden;gen ticket:pl:prt ‘these tickets to Sweden earlier’ 3 kontrollisin. siis öeldi et kõik check:imf:1sg then say:ims:imf that everything ‘(and I) was told that everything’ 4 on nagu korras.= is like in.order ‘was alright.’

(LK)

Utterances (24–25) constitute assessments on some reported state-of-affairs, where the upcoming complement clause contains the information on what the positive or negative assessment concerns. The word uvitav is different in this regard. Although it is inherently a positive evaluation, it regularly prefaces counter-arguments, and thus implies negativity. Example (26) is a response to a travel agent

Interrogative “complements” and question design 

who claims not to have received money for some tickets. Instead of receiving the news about the unpaid ticket, the speaker puts forward a claim that she has been informed of the contrary and prefaces it with ‘interesting’. In this way, turn-initial ‘interesting’ displays that some aspect of what has been said by the interlocutor is problematic and the aspect itself is explained in the following clause. In terms of the relative strength of the counter-argument, ‘interesting’ functions as a hedge. A claim that something is merely ‘interesting’ reduces the confrontational power of the upcoming utterance. In any case, none of the above assessment words besides uvitav can take an interrogative clause as a subject complement. There is thus a general pattern of combining assessment words with clauses in Estonian but only the ambivalent semantics of uvitav allows it to be combined with interrogative clauses, which will therefore be the focus of this subsection. Similarly to the other patterns with interrogative clauses, the format with uvitav also does specific interactional work. The first example in (23), uvitav miks ma talle meeldin, approx. ‘I wonder why she likes me’, represents a typical case of its use. The utterance is produced in a position where the previous sequence has been terminated. Meelis (M) has reported that his granny likes his wife (R). R receives this information with a laughter syllable in line (4), which is apparently understood as expressing disbelief, because after it M reinforces his claim. After another round of incredulity by R and reinforcement by M, the sequence is closed down. The uvitav-prefaced question after a long pause initiates a new sequence. At the same time, the question relates to the topic of the previous segment of talk, the claim that M’s granny likes R. The two most distinct sequential features of uvitavprefaced questions is that they initiate new action sequences (like many questions do) and that they relate to an earlier topic, often the topic that was just discussed in the prior sequence. Example (27) presents another case. The speakers have been gossiping about a guy Erki who has abandoned his family. In lines (1–2), V provides new information about Erki’s background. This is received with a surprise token after which V provides the source of this information. In lines (7–8) H initiates a new sequence by asking an uvitav-prefaced question, suggesting a generalization on what V has just reported. (27) 1 V: et Erki isa olevat ka Erki maha ET name;gen dad be:quo too name;gen behind ’Erki’s dad had also left him behind’ 2 omal ajal jätnud. own:ads time:ads leave:ppt ‘back in his days.’



Leelo Keevallik

3

(2.0)

4 H: issand jumal. ’Good god!’ 5 V: sellest oli nüüd juttu. (0.8) Sille that:elt was now talk;prt name ’(we) talked about that. (0.8) Sille’ 6 nagu [mainis.] like mention:imf:3sg ’mentioned (it)’ 7 H:

[uvitav] kas inimesed tõesti interesting ques people really ‘Uvitav do people really behave like this’

8 käituvad nimodi mallide järgi. behave:3pl like.this pattern:pl;gen according ’according to the patterns.’ 9 V: no vat ei tea jah, no vat neg know yeah ‘Well, I don’t know.’

(TA)

Grammatically, the utterance in lines (7–8) is formatted as a statement about something being interesting. It is not literally formatted as a question and therefore should not make relevant an answer. Nevertheless, the uvitav-prefaced interrogative clauses are regularly followed by answers to the interrogatives. This happens in both Examples (23) and (27) above. This demonstrates that the speakers indeed treat these utterances as asking a question. They never treat them as delivering an assessment (at least in the current corpora), which would make a second assessment or an agreement/disagreement the next relevant action. Thus, the interrogative clause interactionally overrides the importance of the turn-initial assessment. At the same time, the grammatical fact that the uvitav-prefaced utterances are technically statements weakens the default assumption of a question that a recipient would have an answer to it. Uvitav therefore functions as an epistemic device, just like (ei) tea, to mark the speaker stance of uncertainty about whether the answer is available for the recipient(s). In addition, uvitav shows that the upcoming question is topically tied to prior talk, which is probably a reflection of its backward-looking evaluative role6. Uvitav-prefaced questions focus on a new aspect of 6. Even the rest of the assessment items presented in Examples (24–25) function similarly as topical bridges. All the words participating in the pattern can constitute retrospective assessments on their own (Keevallik 2010).

Interrogative “complements” and question design 

what has already been talked about. Just like the above verbs/particles, uvitav organizes the sequential flow of conversation. As an epistemic device, uvitav is not restricted to the utterance-initial position. There are altogether seven examples in the corpora in which it occurs turnfinally, which makes it the most frequent turn-final question frame studied here. Even in turn-final position, it displays no features of being a clause on its own. It is a pure epistemic, which can characteristically be added after the interrogative clause that it modifies. A case is shown in Example (28). (28) A: mismoodi see värk kinni käib uvitav:? how this thingy closed go:3sg interesting ’How to shut down this thingy uvitav?’

(TA)

It is also possible to implement uvitav in the middle of the interrogative clause. (29) millal ta ministritööd huvitav teeb? when he minister.job:prt interesting do:3sg ‘When does he uvitav do the minister’s job?’

(NET)

As in all the other patterns reported in this study, uvitav + interrogative clause cannot be analyzed in terms of the interrogative clause being subordinated to or dependent on the assessment term. Uvitav functions as an epistemic device that is mobile in terms of its placement with regard to the interrogative clause. Further grammatical evidence of the independence of the interrogative clause after ‘interesting’ is provided by the turn-final question particle vä/ve ‘or’ that does not occur with dependent clauses, while it does occur at the end of interrogative clauses after uvitav. (30) 1 K: uita ks see oli mingi väga vana interesting ques this was some very old ’Uvitav was it a very old’ 2 etendus vä. performance ques ‘performance?’

(TA)

Prosodically, uvitav is latched to the interrogative clause. Even though it may carry a certain amount of stress and start in high pitch, it never displays a final contour other than in turn-final cases. Characteristically to particularized items, uvitav may be considerably reduced phonologically, as shown in Example (30). Interactionally and grammatically, the interrogative clause following, preceding, or surrounding uvitav is profiled and independent. Uvitav-prefaced questions initiate topically tied new sequences, which introduce a new aspect on



Leelo Keevallik

what has just been talked about. While they also display epistemic uncertainty about the availability of the response, their use displays the speaker’s understanding of where and how the turn is appearing in the sequence of conversational events and what its status is in regard to the assumed knowledge state of the recipient. Conclusion Based on conversational data, this study shows that a number of interrogative clauses that have been analyzed as subordinate complements are treated as profiled by the participants. The interrogative clauses discussed above display clear features of being independent, while the ‘main clauses’ behave as particles that structure conversation and express epistemic matters. To demonstrate this, interactional, syntactic and prosodic features were discussed. Interactionally, recipients of the turns involving ütle/öelge, räägi, (ei) tea, uvitav + interrogative clause treat them as questions by answering the interrogative clause. Syntactically, the interrogative clauses may involve inversion and the turnfinal question particle vä/ve, which is impossible in dependent clauses in Estonian. Content-wise, the relevant information is provided in the interrogative clause. Furthermore, ütle/öelge and räägi do not necessarily appear adjacent to the interrogative clause. There is thus hardly any grammatical reason for considering the following interrogative clauses subordinate complements. Prosodically, the CTP clauses were often produced as enclitics incorporated into a larger intonation unit, without carrying the most prominent stress themselves. The paper suggests a different analysis for the clauses presented in the introduction and discussed through the paper. (Epistemic) particle Independent clause Ütle mis ma tegema pean ‘what do I have to do?’ Räägi kuidas sul läks täna ‘how did it go today?’ Ei tea mis kellast doctor Laane vastu võtab ’when will doctor Laane be seeing patients?’ Uvitav miks ma talle meeldin ‘why does she like me?’ Thus, the notion of subordination, insofar as it is a relationship between two clauses, is not relevant in the current patterns. Nevertheless the analysis suggests that even turn-beginnings consisting of fully fledged clauses, such as oskad sa öelda ‘can you tell me’ or ega te ei tea ‘I suppose you don’t know’, could be treated as action-projection formats. Instead of seeing the relationship between these formats

Interrogative “complements” and question design 

and the following interrogative clauses as a case of morphosyntactic subordination or a reflection of hierarchical cognitive profiling, we might have to analyze it in terms of temporality. Formats like these orient the speakers in the real-time sequencing of actions and deal with interpersonal face needs. What kind of advantages are achieved by early action-projection in the turn and by taking care of interpersonal relations before the main action is produced could be a subject of another study. As was shown above, all of the focused items are devices of question formation in Estonian. As turn- and utterance-initial particles with particular prosodic features (non-final contours), they partake in projection in interaction. One of the functions of at least ütle, räägi, and (ei) tea is to project that a question is coming up and thus define early on in the turn that it is going to be about information request. While showing the action type of the turn and thus even what the next relevant action should be, they do not yet do the action and do not reveal anything about its content. Projection in itself helps the speakers to coordinate their actions. Designing the questions not only by producing an interrogative but also by adding a pragmatic particle enables the speaker to define and display the nature and sequential placement of the contribution for current participants. The question prefaces discussed above deal with the action type of the turn and the positioning of it within the whole interactive sequence. Especially the particles ütle, räägi, and uvitav display to the participant(s) the understanding of where the upcoming utterance belongs in a sequence of actions. Topic-initiations, including reasons for the call can be formatted with ütle and räägi, while continuations are formulated with uvitav. All of the devices described here also do intersubjective work by suggesting the level of certainty of recipient knowledge. The imperatives display strong assumption that the recipient knows the answer, also turn-finally, while ei tea and uvitav show that the availability of relevant knowledge among the participants is uncertain. They contextualize the interrogative clauses in terms of the respective knowledge states of the participants and ease the contextual pressure for the recipient to produce an answer as the next turn. These formats that may indeed look like prototypical subordinating constructions, when written down, are actually combinations of interrogative clauses and particles specialized in dealing with sequential and interpersonal matters for current participants in conversation.



Leelo Keevallik

References Auer, Peter. 1996. “The pre-front field in spoken German and its relevance as a grammaticalization position.” Pragmatics 6(3): 295–322. Auer, Peter. 2005. “Projection in interaction and projection in grammar.” Text 25(1): 7–36. Brinton, Laurel J. 2001. “From matrix clause to pragmatic marker: the history of look-forms.” Journal of Historical Pragmatics 2(2): 177–199. Bybee, Joan and Hopper, Paul. 2001. “Introduction to frequency and the emergence of linguistic structure.” In Frequency and the Emergence of Linguistic Structure [Typological Studies in Language 45], Joan Bybee and Paul Hopper (eds), 1–24. Amsterdam, Philadelphia: John Benjamins. Couper-Kuhlen, Elizabeth. 2001. “Interactional prosody: High onsets in reason-for-the-call turns.” Language in Society 30(1): 29–53. Couper-Kuhlen, Elizabeth and Thompson, Sandra. 2008. “On assessing situations and events in conversation: ‘extraposition’ and its relatives”. Discourse Studies 10(4): 443–467. Curl, Traci and Drew, Paul. 2008. “Contingency and action: A comparison of two forms of requesting.” Research on Language and Social Interaction 41(2): 129–153. Englebretson, Robert. 2003. Searching for structure: The problem of complementation in colloquial Indonesian conversation. Amsterdam, Philadelphia: John Benjamins. Erelt, Mati. 1990. “Kõneleja ja kuulaja kaudse väljendamise võimalusi eesti keeles.” Keel ja Kirjandus 38(1): 35–39. Erelt, Mati, Kasik, Reet, Metslang, Helle, Rajandi, Heino, Ross, Kristiina, Saari, Henn, Tael, Kaja and Vare, Silvi. 1993, 1995. Eesti keele grammatika I–II (Estonian grammar I–II). Tallinn: Eesti Teaduste Akadeemia Eesti Keele Instituut. Günthner, Susanne. 2008. “Projective constructions in German interactions – an analysis of pseudo-clefts, N-be-that-constructions and extraposition.” Talk given at LISO, Santa Barbara, March 7th Hakulinen, Auli, Keevallik Eriksson, Leelo and Lindström, Jan. 2003. “Kuule, kule, hördu – projicerande praktiker i finska, estniska och svenska samtal.” In Grammatik och Samtal: Studier till minne av Mats Eriksson [Skrifter utgivna av institutionen för nordiska språk vid Uppsala universitet 63], Bengt Nordberg, Leelo Keevallik Eriksson, Kerstin Thelander and Mats Thelander (eds), 199–218. Uppsala: Uppsala universitet. Hopper, Paul. 1987. “Emergent grammar.” Berkeley Linguistics Society: Papers of the Thirteenth Annual Meeting: 139–157. Hopper, Paul. 1998. Emergent grammar. In The New Psychology of Language: cognitive and functional approaches to language structure, Michael Tomasello (ed.), 155–175. Mahwah NJ: Lawrence Erlbaum. Hopper, Paul and Thompson, Sandra A. 2008. “Projectability and Clause Combining in Interaction.” In Crosslinguistic Studies of Clause Combining: The Multifunctionality of Conjunctions, Ritva Laury (ed.), 99–124. Amsterdam, Philadelphia: John Benjamins. Keevallik, Leelo. 1999. “The use and abuse of singular and plural address forms in Estonian.” International Journal of the Sociology of Language 139: 125–144. Keevallik, Leelo. 2003. From interaction to grammar: Estonian finite verb forms in conversation [Acta Universitatis Upsaliensis. Studia Uralica Upsaliensia 34]. Uppsala: Uppsala university. Keevallik, Leelo. 2006. “From discourse pattern to epistemic marker: Estonian ei tea ‘don’t know’.” Nordic Journal of Linguistics 29(2): 173–200.

Interrogative “complements” and question design 

Keevallik, Leelo. 2010. “Hinnangu grammatikast.” [On the grammar of assessments]. In Keele Rajad Pühendusteos professor Helle Metslangi 60. sünnipäevaks, Ilona Tragel (ed.) Eesti ja soome-ugri keeleteaduse ajakiri 2: 147–161. Langacker, Ronald W. 1991. Foundations of Cognitive Grammar. Vol II: Descriptive Applications. Stanford: Stanford University Press. Laury, Ritva. 2006. “On Subordination, Finnish-Style: Questioning the Category of Finite Clausal Complements in Spoken Finnish.” SKY Journal of Linguistics 19: 310–321. Lindholm, Camilla and Lindström, Jan. 2003. “Får jag fråga? Frågeramar i institutionell interaktion.” Språk & Stil NF 13: 35–64. Lindström, Jan. 2002. “Från satsschema till turschema? Förfältet i fokus.” Språk & Stil NF 11: 25–80. Matthiessen, Christian and Thompson, Sandra A. 1988. “The structure of discourse and ‘subordination’.” In Clause combining in grammar and discourse, John Haiman and Sandra A. Thompson (eds), 275–329. Amsterdam, Philadelphia: John Benjamins. Payne, Thomas E. 1997. Describing Morphosyntax: A Guide for Field Linguists. Cambridge: Cambridge University Press. Sacks, Harvey, Schegloff, Emanuel A. and Jefferson, Gail. 1974. “A simplest systematics for the organization of turn-taking for conversation.” Language 50(4): 696–735. Schegloff, Emanuel A. 1980. “Preliminaries to preliminaries: “Can I ask you a question?”” Sociological Inquiry 50(3–4): 104–152. Schegloff, Emanuel A. 1996. “Turn organization: One intersection of grammar and interaction.” In Interaction and Grammar, Elinor Ochs, Emanuel A. Schegloff and Sandra A. Thompson (eds), 52–133. Cambridge: Cambridge University Press. Thompson, Sandra A. 2002. ““Object complements” and conversation. Towards a realistic account.” Studies in Language 26(1): 125–164. Thompson, Sandra A. and Mulac, Anthony. 1991. “A Quantitative perspective on the grammaticalization of epistemic parentheticals in English.” In Approaches to Grammaticalization, Vol. II. Focus on Types of Grammatical Markers [Typological Studies in Language 19], Elizabeth C. Traugott and Bernd Heine (eds), 313–339. Amsterdam, Philadelphia: John Benjamins. Tulviste, Tiia, Mizera, Luule, de Geer, Boel and Tryggvason, Marja-Terttu. 2002. “Regulatory comments as tools of family socialization: A comparison of Estonian, Swedish and Finnish mealtime interaction.” Language in Society 5 (31): 655–678. Waltereit, Richard. 2002. “Imperatives, interruption in conversation and the rise of discourse particles: A study of Italian guarda.” Linguistics 40(5): 987–1010.

Transcription and glossing conventions underlining - [ ] = <= (0.5) (.) :

emphasis truncation overlapped talk latching of turns next unit initiated hastily pause length in tenths of a second micropause lengthening of a sound



Leelo Keevallik

@ laughter syllable <@ smile @> smiling quality quotation quality <slow> slower talk .hh breathing in hh breathing out mts lip smacking % glottal sound /—/ something has been left out from the same turn in the example (parenthesis) item not present in the Estonian original . falling intonation at the end of the intonation unit ? rising intonation at the end of the intonation unit , non-final intonation at the end of the intonation unit – unfinished intonation unit abl ablative ads adessive all allative cond conditional elt elative gen genitive ill illative imf imperfect imp imperative ims impersonal inf infinitive ins inessive neg negation pl plural ppt past participle prt partitive ques question particle quo quotative sg singular sup supine ter terminative tra translative other capital letters untranslatable particle 1,2,3 person

Syntactic and actional characteristics of Finnish että-clauses Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen University of Helsinki

This article examines the Finnish että as a complementizer and as an initial and final particle in conversation. The article shows that että is not a subordinator either syntactically or actionally. Instead, the particle että, as well as the formulaic phrases (complement-taking constructions) it is a part of, implement actions in service of the main action underway. They manage the ongoing talk strategically by regulating the participant framework, and are oriented to by the participants. Therefore, neither the CTP että is a part of, nor the particle että, are actionally subordinate to the talk following että. However, when että functions as a turn-final particle, the turn in which it appears can be seen as a dependent action from the perspective of some previous activity.

1. Introduction A central assumption in functional linguistics has been that grammars of human languages both reflect and are shaped by the uses to which languages are put by their users. Although this assumption is shared by most functional linguists, exactly how it should be understood is still very much an open question. One difficult issue has been the nature of subordination. Traditionally, subordination has been defined syntactically in terms of embedding, one clause being part of another, and/or on the basis of morphosyntactic marking. These criteria have been shown to be inadequate in various ways (e.g. Haiman & Thompson 1984, Matthiessen & Thompson 1988). As an alternative, within functional linguistics, semantic criteria such as relative prominence (e.g. Langacker 1991, Cristofaro 2003) have been suggested by cognitive linguists. On the other hand, interactional linguists have proposed that the definition of subordination should take into account how clause types are used in conversation and how speakers orient to them (e.g. Thompson 2002).



Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen

In this article, we wish to contribute to the on-going discussion of the nature of clause combining by connecting the concept of subordination to the sequential analysis of conversation. To this end, we will examine utterances containing the Finnish et(tä), which has traditionally been considered a subordinating conjunction, or more exactly, a complementizer1, but is also used as an utterance-initial and -final particle in contexts where there is no associated complement-taking construction (see Raussi 1992, Seppänen & Laury 2007, Laury & Seppänen 2008, Koivisto 2006 for detailed discussion). Syntactically, the constructions että appears with are often independent, but sequentially, and in terms of content, they are dependent on what has gone on just prior, while they also project more to come. Even when että joins two syntactic constructions as a complementizer, we suggest that the material following että (the ‘complement clause’) cannot be considered syntactically subordinate to the clause preceding it (the ‘main’ or ‘matrix clause’; see Seppänen & Laury 2007, Laury & Seppänen 2008). In our data, the talk following että, the ‘complement’ clause, also turns out to be more central, and thus not subordinate, in terms of content than the ‘main’ clause preceding it. At the same time, the clause preceding että, which consists of a speech act verb or a mental verb together with its subject (the complementtaking predicate -phrase, or CTP-phrase, cf. Thompson 2002) performs crucial tasks in the management of the talk in terms of action projection and the organization of participation. In terms of the on-going main activity, therefore, the CTP-phrase and the talk following että have different functions, and the question of which one of them is relatively more prominent or more subordinate is connected to which functions are oriented to and treated as central by participants in interaction. The type of action which että implements ties together its uses as a complementizer and as an initial and final particle. Through the use of että, its speaker reorganizes the participant framework of the conversation. By the term ‘participation framework’ we mean the roles of the participants in interaction to the upcoming utterance, to “the words said”, in terms of speakership and recipiency (Goffman 1981; Goodwin 1981, 1984, Goodwin & Goodwin 1990). The simplest example of this may be quoted speech. By using että initially, the current speaker indicates that she is quoting an earlier utterance; that is, että indexes that she is only acting as an animator of the utterance, is only in the role of producing it 1. Finnish grammars classify että as a conjunction (Fi. konjunktio). In the latest comprehensive grammar of Finnish (ISK, Hakulinen et al. 2004, §819) että is called a general conjunction (Fi. yleiskonjunktio), and conjunctions are in turn included in the larger class of particles. The term ‘complementizer’ is not used in Finnish linguistics, although the description of että in grammars shows that it primarily functions as a complementizer (see, e.g. Seppänen & Laury 2007, Laury & Seppänen 2008). We use the term complementizer for että here, because our article is intended for a wider audience, not just for Finnish scholars.

Syntactic and actional characteristics of Finnish että-clauses 

aloud, while someone else is the author, the one who “has selected the sentiments that are being expressed and the words in which they are encoded” or the principal, who “is committed to what the words say” (Goffman 1981: 144; for Finnish, see Seppänen 1998). – While it manages the participation framework, että thus also makes relevant, or links to, what has been said earlier in the same conversation or elsewhere (the original utterance, such as the speech that is now being quoted); in that sense, että is profoundly dialogic (Bakhtin 1987 [1953]), since it imports another voice to the the conversation, the voice of the author of the quoted utterance. We suggest below that the use of että in conversation to organize the participation framework implements an action oriented to by the participants. The three use types of että can be summarized as follows. – When että functions as a complementizer, that is, when it occurs with a CTPphrase such as se sano (mulle) että ‘s/he said (to me) that’, it normally specifies the source (the author) and, often, also the recipient (addressee) of the earlier talk it makes relevant (see Laury & Seppänen 2008 for further details), as well as indexing the roles of the current participants toward the words said as, for example, an animator, thus accomplishing in significant ways a change in the context of the utterance. The CTP-phrase also functions to project more talk to come, thus also regulating turn taking, and indexes the talk to come as being of a certain type, such as a paraphrase, summary or candidate understanding of the prior text it makes relevant. – In the initial particle use, että functions alone to organize the participant framework. In that use, että by itself indexes that the talk that follows is a paraphrase, summary, candidate understanding or upshot of earlier talk (Raussi 1992; Seppänen & Laury 2007, Laury & Seppänen 2008). – In final use, että makes relevant what has gone on in prior talk by projecting an upshot of the turn in which it appears, but leaving it to be inferred based on prior material. It also functions as a turn transitional device, indicating readiness to give up the turn, and thus, like the other two use types, also organizes participation. What all three use contexts of että share in common, then, is that in all of them, että functions metacommunicatively to organize the participant framework of the conversation. In all three uses, että both links backward and projects forward. It links backward by making relevant or evoking some prior talk, and it projects forward by indicating that what follows että is a paraphrase, a candidate understanding, or an upshot of the prior talk. In the complementizer and initial-particle use, the paraphrase, candidate understanding, or upshot is expressed in the talk following että. In its use as a final particle, että leaves the projected upshot to be



Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen

inferred. What precedes että, the prior text it makes relevant, functions as a frame of interpretation for what follows it. 2. Data Our data come from the Finnish conversation archives housed in the Department of Finnish, Finno-Ugrian and Scandinavian Studies at the University of Helsinki. These data are made up of audiotapes of ordinary multiparty face-to-face conversations and telephone calls; some of the data are also videotaped. For the last section of the paper, data from the Research Institute for the Languages of Finland were also analyzed; these data include institutional encounters. The data were transcribed according to the conventions commonly used in conversation analytical studies, adapted for Finnish. A list of the symbols used is in Appendix 1. 3. The use of että with complement-taking predicates In her influential 2002 paper, Thompson suggested that many English CTP-phrases, such as I think or I mean, do not function as proper main clauses but function as “epistemic, evidential and evaluative fragments”. She also claimed that it is the complement clause, not the CTP-phrase, which constitutes the main action by the speaker and which is oriented to by the participants in the interaction. This might lead one to conclude that it is the complement clause, and not the CTP, which is really the ‘main’ clause in terms of social action. Thompson refers to Langacker’s (1991: 436ff) definition of a subordinate clause as “‘one whose profile is overridden by that of the main clause’, where ‘profile’ refers to the ‘relative prominence accorded to various substructures’” (cf. Cristofaro 2003: 33). Langacker considers complement clauses prototypical instances of subordination and gives as an example the sentence I know that she left which, according to Langacker, “designates the process of knowing, not leaving “. Thompson (2002: 131) suggests, however, that “claims about what is ‘profiled’ or ‘prominent’ can only be validated by providing empirical evidence. Convincing evidence comes from building a case from an examination of what the speaker of the utterance in question is doing with that utterance. This means that claims for the relative ‘prominence’ of any part of a given utterance need to be based on the interactional actions that utterance is performing in its local situated context (...). In other words, to prove what is ‘profiled’, we need to look at the way in which grammar is organized with reference to particular courses of action that the participants are engaged in doing.” Thompson suggests that in this sense, in her

Syntactic and actional characteristics of Finnish että-clauses 

English conversational data, “it’s the CTP whose ‘profile’ is ‘overridden’ in complement utterances (2002: ibid.)”. 2 We very much agree with Thompson’s suggestion that what is considered prominent is an empirical question, and in our data, it is indeed the content of the material following the CTP-phrase which is oriented to by the participants in the sense of being subsequently maintained in the conversation, and the profile of the CTP would, in that sense, then, be overridden. However, the question of what is actionally prominent is not straightforward. In our view, this cannot be determined only on the basis of the topic or content of the talk underway. Rather than functioning in terms of content, CTP-phrases seem to be “playing a key role in the strategic management of the current talk” (Hopper & Thompson 2008: 114) both in terms of projection, “the adumbration (foreshadowing) of an action by a segment of talk” (Auer 2005: 8–10), as well as the expression of an epistemic, evidential, or evaluative stance toward the upcoming segment of talk (e.g. Thompson 2002; Kärkkäinen 2004; see also Östman 1981). It is our view that the strategic management of talk is no less central in terms of conversational action than topical continuity or other content phenomena. In this section of the paper, we will focus on the function of CTPs with että and their role in managing the interaction strategically by organizing the participant framework, by projecting more talk to come, and by indexing the nature of that talk as being of a certain type. In conversational data, most of the uses of että where it could be analyzed as a complementizer occur after speech act verbs and mental verbs. The three most frequent CTPs followed by että in our data were sanoa ‘say’ which accounted for 27 uses (nearly a third of all complementizer uses), ajatella ‘think’ with 15 uses and kysyä / kysellä ‘to ask’ with 6 uses, as shown in Table 1 below. Together these three verbs accounted for two thirds of the uses of että in CTP-phrases (48/73, 66%). Table 1.â•‡ Distribution of verbs in CTPs in our data Verb sanoa ‘say’ ajatella ‘think’ kysyä ‘ask’ Other verbs Total

Number

Percentage

27 15 â•⁄ 6 25 73

â•⁄ 37% â•⁄ 21% â•⁄â•⁄ 8% â•⁄ 34% 100%

2. In a recent paper, Langacker seems to have conceded Thompson’s point, proposing that the profile of the matrix clause can be overridden in certain clause combinations (2008: 146), and that ‘in discourse terms ... a main clause is one whose profile is maintained as such, and a subordinate clause one whose profile is overridden (2008: 146)’.



Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen

Next, we will illustrate the use of että as a complementizer based on examples in which it appears with these three verbs. Our first example illustrates the use of että with sanoa ‘to say’. In its use with sanoa, että cleaves prosodically to the verb, and together they form a formulaic phrase, or “fragment” (Hopper 2001; Thompson 2002) which is often unstressed and phonologically reduced (s’noet) (Laury 2006, Seppänen & Laury 2007, Laury & Seppänen 2008). As noted above, these types of constructions, as has been pointed out for frequently occurring CTPs in other languages (e.g. Thompson 2002; Scheibman 2002; Kärkkäinen 2004; Hopper & Thompson 2008; Keevallik, this volume; Imo, this volume; Pekarek Doehler, this volume), do not function as main clauses but rather as epistemic/ evidential particles, discourse markers, fragments, or small constructions which index the speaker’s stance or footing toward what it being said. The construction also serves as a projector phrase (cf. Günthner, this volume; Pekarek Doehler, this volume; Hopper & Thompson 2008), projecting more talk to follow and, as we have mentioned above, indexing the type of talk that will follow. Example (1) is taken from a telephone conversation in which two women, Irma and Ella, are negotiating Irma’s ride to an event. Just prior to the excerpt below, Ella has offered to call someone who could give Irma a ride, and Irma has suggested that she could even ride her bicycle if the weather is good, and if not, she could catch a ride with a certain person. She then mentions another occasion when this person has offered her a ride. There is a short exchange about which event that was. After this, there is a short pause (line 59). (1) Pappila SG143 A 02 59 (0.8) 60 Irma: e:i mikäs se oli: (0.5) .hh (.)↑ nii hänen kanssa.(.) neg what 3sg be-pst so 3sg-gen with no: what was it. yeah with her 61 62 Ella:

hänen kanssa ↑niin. [tulin Järvelään sillo. .h nii 3sg-gen with so came-1sg j.-ill then so with her yeah I came to Järvelä then. so [joo-o Yeah

63 Irma: hän sano että .h hänen kanssa .h aina pääsee 3sg say-pst that 3sg-gen with always can.go she said that with her (one) can always go 64

pääset millo hyvänsä ku vaan #e# soitat nii can.go-2sg any.time when only call-2sg so you can come any time when you just call so

Syntactic and actional characteristics of Finnish että-clauses 

65

aina hänen kans pääset jos hän< sillai always 3sg-gen with can.go-2sg if 3sg dem.adv come with her if she like that

66

mennee ja noin nii [(–) go and dem.adv so goes and like that so

67 Ella:

[no kato sehä on hyvä .h= ptc ptc 3sg-clt be good oh well that’s good.

68 Irma: hän ol erittäin ystävälli[nen. 3sg be-pst very friendly She was very kind. It is clear that the topical focus of this exchange is on Irma’s ability to get a ride, and the open offer she has received (lines 63–65), and not the fact that the person making the offer has said something (cf. Thompson 2002). Ella’s response (line 67) to Irma’s turn appears to concern the nature of the offer: the verb on ‘is’ is in the present tense, and thus seems to evaluate the offer, which carries into the present time, and not the call, which is reported in the past tense in line 63, sano ‘said’. Irma’s own evaluative comment in line 68 can also be understood to be prompted by the content of the offer, not its epistemic/evidential status. This example shows that the material following että, lines 63–66, is topically the focus of the conversation: what the participants are talking about is expressed in the talk following että, which includes information pertinent to the reason for the call (getting a ride). Syntactically, what follows että is not strictly clausal in nature, but rather a lengthy report of what has been said3 during an earlier conversation between Irma and the person referred to as hän4 ‘s/he’ in this excerpt, alternating between indirect and direct quotes, loosely bound together with the hesitation marker #e# and the conjunctions ku ‘when, because’, nii ‘so’, jos ‘if ’ and ja ‘and’. Here, then, we are not dealing with a biclausal combination consisting of a main clause and a subordinate clause, but rather a projector phrase followed by an indeterminate stretch of talk (cf. Hopper & Thompson 2008; Günthner, this volume; Pekarek Doehler, this

3. We do not, of course, suggest that quoted speech is an exact rendering of earlier talk (see, e.g. Tannen 2007). 4. Hän is the third person pronoun used in standard Finnish for human referents. In spoken non-standard language, se is the most common anaphoric pronoun for both humans and nonhumans. In most Finnish dialects, hän is used logophorically to refer to the original speaker of a quote within quoted speech (see, e.g. Laitinen 2005).



Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen

volume.) Irma’s turn in fact fails to achieve a clear prosodic closure; it is overlapped at the end with Ella’s turn. The ‘matrix’ clause hän sano että is formulaic, as noted above. The function of this formulaic phrase is metacommunicative: it functions on the level of footing, expressing the speaker’s stance towards what is being said: it expresses that Irma is not the author of the words which follow, only an animator (Goffman 1981: 144). We take it that this changes the context of the talk, how it is to be understood, in a fundamental, though entirely commonplace way. In everyday terms, the phrase expresses that what follows is a quote. Thus Irma brings to the conversation from outside the current situation another voice besides her own. The use of the third person and the outside voice facilitates the current interaction from Ella’s perspective and affects her role as a participant, since the arranging of the ride is now presented as not being her responsibility, but rather a matter of an offer made to Irma by a third person. At the same time, it also frees Irma from a possible imposition on Ella; she has her own resources, and it is not necessary for Ella to make calls in order to secure a ride for Irma. Thus the formulaic reporting phrase, through regulating footing by attributing the upcoming words to someone other than the current speaker, also has a facilitating function in terms of what the interaction is focused on, the activity of working out how Irma will get to an event. In this way, it functions to manage the interaction strategically (cf. Hopper & Thompson 2008: 114). And, as already noted, it also manages the interaction by regulating turn taking since it projects more talk to come. Thus it simultaneously functions to project a continuation of Ella’s turn and to index the nature of her turn as a quote, which, as we have argued above, is crucial to the negotiation. It should also be noted that although the material following että in this example contains the topically relevant portion of Irma’s turn, the epistemic status of the reported talk is oriented to in subsequent talk as well. It is crucially important to the negotiation, the main activity underway, that Irma has received another offer for a ride from someone else, since this preempts Ella’s offer to call someone to find Irma a ride. And Irma’s turn also orients to the fact that the offer was made by a third party, whose kindness she evaluates. To summarize, we have seen that the use of että with sanoa ‘to say’ consists of a formulaic phrase which has an evidential, metacommunicative function, and which projects more talk of a certain nature to follow; it plays a crucial role interactionally in facilitating the main action, and it is also oriented to by the participants in the subsequent conversation. Rather than functioning as a ‘main clause’, it projects more talk to come. The stretch of talk following this phrase is often not strictly clausal in nature (see also Laury & Seppänen 2008: 162; cf. Hopper 2004, Hopper and Thompson 2008 for English).

Syntactic and actional characteristics of Finnish että-clauses 

We would like to suggest that the use of että with a CTP-phrase be viewed as implementing an independent action. It fundamentally transforms the context and facilitates the current activity sequence, in a manner parallel to what is suggested for the use of switches in terms of reference by Lerner & Kitzinger: they suggest that such switches in personal pronouns are “constitutive of specific actions” while they are “situated within an ongoing course of action (2007: 550)”. Here, the arranging of a ride for Irma is the main activity underway. We have argued that the use of the CTP-phrase se sano että constitutes a specific action and has a crucial role in facilitating the ongoing activity. The next example illustrates the use of että with ajatella ‘to think’ and with kysyä ‘to ask’. Ajatella ‘to think’ appears to function in conversational language, not so much as a device for reporting the internal thought world of someone, but rather as a way of making suggestions, especially in an interactionally sensitive context. Example (2) comes from a telephone conversation between a mother and a daughter. At the beginning of the call, Anna, the mother, reports feeling dizzy due to the hot weather, and there is a short exchange about that. She then asks when her daughter Mari and her family are coming for a visit.

(2) [Sg 141 A06 Palaudutaan]

6 Anna: S:immost juuh mitäs teitille .hh koskas tulette, dem-adj-prt ptc wh-clt 2pl-all when-clt come-2pl Like that yeah how are you .hh when are you coming, 7 Mari: No (.) me ei oikee o (.) meh- tavan mukaan ptc 1pl neg really be 1pl custom-gen according Well as usual we haven’t 8

ei oikeen päätetty m’t me aateltiin neg really decide-p.pple but 1pl think-pst-pass5 really decided but we were just thinking

9

just kysyä kans sitä että (.) et6 onks tota just ask also 3sg-prt että et be-q ptc about asking whether (you) have um

10 (.) o- onks sulla esimes lauantaiaamu be-q 2sg-ade for.example saturday-morning you have for example Saturday morning

5. The passive has been grammaticized into the first person plural verb form in spoken Finnish (see, for example, Shore 1986). 6. Et is a short form of että.



Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen

11

hirvee ohjelmoitu ku (0.2) niinku (.) terribly program-p.pple because ptc terribly programmed because like

12

me aateltiin sitä et sekin 1pl think-pst-pass 3sg-prt et 3sg-clt we were thinking that

13

ois aina h- semmonen pieni houkutus be-cond always dem-adj small temptation it would also always be a bit of a temptation

14

(.) tulla jo perjantaina (0.2) sitte jo (.) come-1inf already Friday then already to come on Friday already

15

että tota noin ni (.) jos siel on vähän niinku (.) että ptc ptc ptc if there be little ptc that um like if there are a bit like

16

hoitajia lauantaiaamuna caregiver-pl-prt saturday.morning-ess caregivers Saturday-morning

17

mut ooksä hirveen kiireinen. but be-q-2sg terribly busy but are you terribly busy.

18 Anna: .mthhh no,(.) mä oon sillai kiireinen että, ptc ptc 1sg be-1sg such-adv busy että Well, I am busy in such a way that 18

(0.2) mun (>täytys<) nää siivoushommat tehrä 1sg-gen must-cond dem cleaning-job-pl do-1inf I should do the cleaning stuff

19

ja sitten,lei#poo# ja semmosta kun mä en and then bake-1infand such because 1sg neg-1sg and then bake and such because I haven’t

20

oo jaksanu nyt tehrä mittään be be.able-p.pple now do-1inf anything had the energy to do anything

21

kun mä en es tiä tuleeks se #piispa# because 1sg neg even know come-q-clt det bishop because I don’t even know if the bishop is coming

Syntactic and actional characteristics of Finnish että-clauses 

22

tänne aamu#palalle# °ja°? dem.adv morning-snack and here for breakfast and

(0.2)

23 Anna: sil[lai, .hhhhh dem-adv like that,

]

24 Mari: [aijaa:, ai tuleeks se] ptc ptc come-q-clt 3sg oh, I see, is he coming 25

lauan[taina vai,] saturday-ess or on Saturday or,

26 Anna: [hhh]hhh no sunnuntaina mutta ptc sunday-ess but Well on Sunday but 27 täytyyhän sitä sitte (0.2) tehrä, must-clt 3sg-prt then do-1inf it will have to be done then, 28 Mari: nii= ptc yeah 29 Anna: =mul on huomenna koulupäivä, .hh 1sg-ade be tomorrow school-day I have a school day tomorrow, Anna’s turn in line 6 is a direct question and makes relevant an answer as the next appropriate action. However, instead of providing a direct answer to Anna’s question, Mari expresses that she cannot answer the question, since the decision has not been made yet. The main part of Mari’s turn consists of a carefully worded suggestion that she and her family might come on Friday, in case there might be childcare available for them on Saturday morning. It is clear in this context that by hoitajia ‘caregivers’ in line 16 Mari is indirectly referring to Anna, since her conditional clause regarding the availability of caregivers is preceded by an indirect question in lines 10–11 about her mother’s planned activities on Saturday morning, and directly followed by a question in line 17 about whether she is busy. Thus, in her turn in lines 7–17, Mari counters Anna’s question with another question, expressing that her answer to the question in line 6 is negotiable, and depends on



Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen

Anna’s answer to the question Mari presents in line 17, mut ooksä hirveen kiireinen ‘are you terribly busy’. We suggest that the action implemented in Mari’s turn is a request: she is here (indirectly) asking her mother to babysit. Let us now focus on the two CTPs including että in lines 8–9 and line 12. It is unlikely that me aateltiin (just kysyä kans sitä että) ‘we were thinking (of asking just about it that)’ in lines 8–9 and me aateltiin sitä et ‘we were thinking (about it) that’ in line 12 really function to report Mari’s and her husband’s internal thoughts about asking or about being tempted to come. Rather, Mari presents her implied request for babysitting in such a way that it is not a direct question or a bald statement. What Mari is doing is interactionally sensitive: she is requesting a favor (babysitting) from her mother. Mari’s request involves a possible imposition (the implication is, ‘could you babysit for us on Saturday morning?’), and she formats the action she is performing as being dispreferred; the ajatella ‘think’ clauses have the function of delaying the action of requesting, and may already be orienting to a negative answer. At the same time, it is noteworthy that the ajatella ‘think’ verbs are in first person plural in both CTPs. This imports another voice to the conversation, giving Mari only partial responsibility for what she is about to say (cf. Lerner & Kitzinger 2007), since she is only partly an author of the talk that follows, and is acting only as an animator for the part of her husband. In other words, the “main” clauses containing a CTP, a cognitive verb, are managing the participation framework. While we might say that the talk that follows että is where the content the participants are focusing on is expressed, the ajatella ‘think’ clauses offer Anna a frame of interpretation and a possible way of taking a stance toward the implied request. Thus, both the CTP-phrase and the talk that follows, be it clausal in format as in lines 9–11, or a longer stretch of talk more loosely bound together as in lines 13–16, have their own functions, and implement separate actions. The CTPs manage the participation framework (‘we were thinking’), while the talk that follows them actually performs the action of requesting (‘are you busy because we were thinking we might come on Friday if there is babysitting available on Saturday morning’). For the participants, neither the CTP-phrases nor the post-että material override the other in Langacker’s (1991) terms. Participants orient to both the CTP and the material after that, the presentation of the possible plan to come on Friday and the possible availability of babysitting, as well as the eventual formulation of the request for babysitting as an (syntactically independent, coordinated) polar question about whether Anna is busy. So extract (2) shows that the ajatella ‘think’-clauses implement separate actions which are engaged in by Mari in order to accomplish a request. In her turn, the CTPs in lines 8–9 and 12 are interactionally relevant, and their role is central in the strategic management of the action (Hopper & Thompson 2008) of

Syntactic and actional characteristics of Finnish että-clauses 

requesting in terms of the participation framework (it is presented as an idea or possible plan attributed to both Mari and her husband) and in terms the formulation of the request also as far as timing is concerned, since it delays a dispreferred action. In contrast, the focus of the matter under negotiation is expressed in the segment of talk which follows että, which expresses the possibility of coming on Friday and the possibility of there being babysitting available on Saturday. That is, we are suggesting that in terms of content, the participants orient to the timing of Mari’s and her family’s arrival, the material expressed in the talk following että, and to Anna’s activities on Saturday. But just as in Example (1), the participants also orient to the CTP-phrases. The CTP-phrases regulate the tone of the interaction in that they express the participants’ orientation to each other. Although Mari’s main point, and the main action, is not to express that she and her husband have thought about something, the part of the utterance expressing this is interactionally relevant. Whether there is any evidence for claiming that the clauses that follow että are subordinate syntactically is an interesting question here. As we have suggested, at least some of the reporting phrases preceding että are grammaticized into formulaic phrases or fragments, but it could be argued that the CTPs involving the verbs ajatella ‘think’ and kysyä ‘ask’ are not as formulaic as CTPs involving sanoa ‘to say’. The verbs ajatella ‘think’ and kysyä ‘to ask’, as mental and speech act verbs, are ordinarily classified as complement-taking constructions (but see Munro 1982 on the low transitivity of ‘say’ verbs), and the combination of a CTP and että could be said to project talk (another clausal or other phrase expressing the content of the thought or the question) to follow. However, in Example (2), both of these verbs have their complement ‘slots’ already filled with sitä ‘3SG-PRT, it’, which functions as a cataphoric or support pronoun here. Presumably a verb which already has a pronoun complement cannot take another complement7. In that sense, then, the syntactically subordinate nature of what follows että is questionable. The third että, in line 15, is more clearly a particle use8; it is not preceded by any complement-taking construction and has the typical summarizing or paraphrasing function of the particle-että. It also links back to the fact that Mari is

7. The pronoun complement of the CTP in the main clause has traditionally been considered a ‘support pronoun’, allowing for an analysis of the following clause as a complement as well, However, see Lindén (1957) for a slightly different analysis. See also Leino 1999 on the use of se before että in written Finnish. 8. However, as in Example (1), the second että in line 9 could also be analyzed as being linked to the same CTP as the clause preceding it, making what follows the second use of ajatella a stretch of talk which is not just clausal but rather multi-clausal in nature; the borderline between the conjunction and particle use of että is not discrete (see Laury & Seppänen 2008).



Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen

still in the role of animator as well as an author. This use of että is discussed in Section 4 below. In this section, we have argued that even when että functions in its canonical role as a complementizer, it does not initiate a subordinate clause. The CTP-phrase, rather than functioning as a proper main or matrix clause, functions to project more talk to come and frames that talk epistemically or evidentially, implementing thereby an independent action. When a speaker expresses that she is not the author or the principal of the words she is uttering (see Goffman 1981), the listener is also not cast in the role of the recipient of the animated talk. The phrase hän sano et ‘s/he said that’ projects reported speech, and gives the current speaker the role of a recipient in the reported situation; thus the current recipient is detached from the utterance to the role of a bystander. In the same way the phrase me aateltiin sitä et ‘we were thinking that’ has a detaching function, since through it, the speaker places her subsequent utterance in some other situation in the past instead of stating it in the current situation (cf. Routarinne 2005: 91). In our view, the CTP-phrases in our data have a crucial function in strategically managing the action the participants are engaged in, and can also be shown to be oriented to by them. The main focus in terms of the content of the current action is expressed in the segment of talk that follows. We suggest that the CTP-phrase implements an independent action in service of the on-going main activity oriented to by the participants. 4. Että as an utterance-initial particle This same metacommunicative character, the feature of commenting on what is being said and the shift in participation roles, is also central when että occurs in syntactic positions where it cannot be considered a complementizer; that is, as the first or last element in a syntactically and prosodically independent unit. If it is, as we think, this metacommunicative nature which is central in the meaning of the particle että, we can ask how that can be connected to the idea of subordination in terms of action. The clearest cases of että as the first element in a syntacticallly and prosodically independent unit are those in which että occurs in turn-initial position. Let us consider next an example of a case of a turn beginning with että. Here Liisa has been telling Tyyne and Raija (lines 1–15) about how she, her husband Keijo and some friends were building a summer cottage in Lapland, and taking a nap after lunch. The main point of the story is the description of how one of the friends, Pena, was startled when Keijo suddenly began to snore very loudly. The description is produced by describing Pena’s movements, se varmasti kaks

Syntactic and actional characteristics of Finnish että-clauses 

metrii hyppäs ‘he certainly jumped two meters’ and ascribing to Pena a prosodically heavily marked utterance nyt se kuolee ‘he’s dying now’ (lines 13–15).

(3) Christmas coffee [sg S15]

01 Liisa: >mutet< j- joham mä< (.) me siel Lapis but-et already 1sg 1pl there l-ine well but have I< (.) we were there in 02

oltii be-pass-pst Lapland

03 Tyyne: °mm° 04 Liisa: niin siis mökkiä a- a: alotettii ni .hh ptc ptc cottage-prt begin-pass-pst ptc st- starting to build the summer-cottage and 05

>s’tte ku< me oltii syöty then when 1pl be-pass-pst eat-pass-pst.pple .hh >then when<we had eaten then

06

ni päivällä ni J:arkko ja a: Keijo ptc day-ade ptc 1nameM and 1nameM in the day-time then Jarkko and a: Keijo

07

nukku m: niillä< l- leveillä sleep-pst-3pl dem-ade broad-ade were sleeping on m: those br- broad

08

puupenkeillä ja wood-bench-pl-ade and wooden benches and

09 Tyyne: °mm° 10 Liisa: >ja tota< Pena meni siihe l:a:ttialle and ptc 1nameM went dem-ill floor-ade >and uhm Pena streched himself on the 11

pitkälleej ja (0.3) ja tää kuulek long-ade-poss and and this.one listen floor and (0.3) and he (=Keijo) imagine when

12

ku se yks kaks (.) krh ku sä when 3sg one two when 2sg he suddenly (.) krh when you (=Keijo)



Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen

13

aloit. (.)siis Pena kuules se varmast began-2sg ptc 1nameM listen 3sg certainly began. (.) I mean Pena you see he surely

14

kaks metrii hyppäs two meter-prt jump-pst jumped two meters high

15

[nyt se kuolee?, hehehe hehe now 3sg die [he’s dying now?, hehehe hehe

16 Keijo: [no:: e:i nyt sentääm mutta tota pikkusen well not ptc ptc but ptc a little [well:: no:t quite like that but a little 17 Raija: nii:n, ptc ye:ah, 18 Liisa: >kuule se< s:e on< se listen 3sg 3sg is 3sg >you see that< th:at is< that 19

[on hirveetä. is terrible [is terrible.

20→Tyyne: [että pelästy tosissaan niin, ptc be.frightened-pst really so [että he really was so frightened, 21 Liisa: nii. se todella pelästy että: et miten< ptc 3sg really frighten-pst comp comp how yeah. he really was frightened that what< 22

nyt on käyny. miten tolle how dem-all now be-sg3 happen-pst.pple what has happened now to that guy (=Keijo)

Tyyne’s turn in line 20 begins with että. Syntactically, it is not possible to consider this clause subordinated: there is no preceding structure which could be analyzed as its ‘main’ or ‘matrix’ clause. Tyyne’s previous contribution has been a silent mm in line 9, which is a complete turn by itself. Neither do any of the turns by the other participants project an että-clause as a completion. Yet it is clear that että acts as a linking element, a textual connector marking coherence: specifically, it marks the relation of Tyyne’s turn to the previous talk in terms of action and

Syntactic and actional characteristics of Finnish että-clauses 

sequential organization. Could we say that the connection it creates here is subordinating by nature or not? What kind of metacommunicative commenting does it perform? This is a story-telling sequence. Story-telling is typically collaborative action; the role of the recipients is central in marking and elaborating the main point of the story (cf. Sacks 1974: 347; Schegloff 1982: 73]). Here Keijo, who is the main character of Liisa’s story, takes his role as a co-teller in line 16 and gives his own description of Pena’s movements. (Cf. Sacks [1971] 1992: 437–443; C. Goodwin 1979, 1981: 149–166, 1984; Seppänen 1996 regarding the role of knowing recipients in story-telling.) After this, Liisa does not continue the telling, but evaluates the situation she has been describing (lines 18–19). Sequentially, Tyyne’s turn is thus in a position where it is relevant for the story-recipients to express their stance towards the story. Tyyne’s turn, että pelästy tosissaan niin ‘että he really was so frightened’ gives a candidate interpretation of the main point of Liisa’s story: she names the situation, which Liisa has described through reported speech and action, as “being frightened”. It seems to be the turn-initial että which marks the turn as an interpretation of earlier speech: if Tyyne had said se pelästy tosissaan niin, ‘he really was so frightened’ the interpretation would be that she is reporting her own observations of the situation, in the same way as Keijo is doing in line 16. The function of the particle että, together with the backwards-connecting adverb tosissaan ‘really’ and the anaphoric demonstrative niin ‘so’, is to mark the relation of the current turn to the previous turn and to guide the recipients in taking it as a re-wording of something which the previous speaker has said, not as the current speaker’s own take on the situation. It tells the recipients how to contextualize Tyyne’s turn. As Tyyne is giving a candidate understanding of Liisa’s story, she sequentially projects Liisa to either confirm or repair it. Thus että is here a metacommunicative element, which on the one hand connects the turn backwards by marking it as a reactive turn, and makes the story-teller, Liisa, the principal (Goffman 1981; see above) of this reactive turn. On the other hand, että has a forward-projecting function: because Liisa is the principal of the turn Tyyne is animating, the relevant next action in the conversation is Liisa’s confirmation or repair of this turn. In lines 21–22 Liisa begins to act in the way she is projected to: she confirms Tyyne’s interpretation with the response particle nii and by repeating Tyyne’s wording of the situation, se todella pelästy ‘he really was frightened’. Tyyne’s turn forms an important part of the story-telling sequence which the participants are formulating together. Tyyne’s contribution is to explicate in one word, pelästy ‘was frightened’, the situation about which Liisa has given a lengthy description. After Tyyne’s contribution, the participants can go on discussing this situation by using her words. However, as a markedly reactive turn, Tyyne’s turn



Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen

here does not initiate a new sequence, and in fact, että-initiated turns in general are not sequence-initiating. The että-initiated turns seem to project backwards. Thus, even if, in terms of the on-going activity, they are not subordinated, they could be seen as not being independent either. The että-initiated turns perform types of actions (here, a candidate interpretation of a story) at a sequential point where such an action is relevant (here, after the conclusion of a story). In this section of the paper, we have seen that the turn-initial että, like the ‘complementizer’ uses of että, ties its clause to preceding talk, making relevant some prior talk, and indexing that the turn it initiates is a paraphrase of that prior talk. Both uses of että have the function of regulating the participant roles in interaction. In the initial particle use, että carries alone the regulating function. In terms of content, what follows the particle että is a topical continuation of what has been said just prior. Both project backward and move forward the sequence of which they are part. 5. The turn-final että So far, we have been concerned with uses of että as a conjunction and as turn/utterance-initial particle. However, että does not always function as a link between clauses or turns and thus project continuation by the same speaker (see also Koivisto 2006; in preparation).That is, että can also occur as the last element in a turn. In these cases, että can be characterized as “final” in two ways: it occurs at the end of a prosodic unit (cf. Hakulinen 1989: 117–118; Seppänen & Laury 2007: 556, see also Example 1) and also at the end of the whole turn, functioning as a final particle (cf. Mulder & Thompson 2008 on final but)9. By turn-final we mean that että can form a possible point for turn-transition. In these cases että typically occurs with syntactic constructions which do not include complement-taking predicates or other complement-taking constructions. The turns ending with että also perform recognizable actions in recurrent sequential environments. Prosodically, however, these turns do not always sound “final” – at least not in terms of intonation. Että-final turns that are followed by turn transition may have either slightly falling (marked with ;), falling (marked with .) or slightly rising intonation (marked with ?,). This variation indicates that final pitch is not a distinctive feature in terms of turn-holding and turn-yielding. This observation is consistent with 9. In this article, we are using the term “final particle” instead of “sentence final particle” (see, e.g. Okamoto 1995; Okamoto & Ono 2008). With “final particle” we refer to particles that occur in turn-final (or, prosodic unit -final) position. Hence, we consider utterances and turns to be the basic units of conversational talk, and determine “finality” with respect to the interactional factors. (Cf. Mulder & Thompson 2008.)

Syntactic and actional characteristics of Finnish että-clauses 

earlier studies done on final or stand-alone conjunctions in English (see Local & Kelly 1986: 195–196; Local & Walker 2005: 126). However, certain phonetic features often occur with final conjunctions, such as creak, breathiness, whisper and audible out-breath. Ogden (2004) showed that in Finnish, the above-mentioned features (i.e. non-modal voice quality) indicate transition relevance. Interestingly, he also notes that this applies to turn-final conjunctions as well (ibid. 47–51; see also Ogden 2001: 143). The following example illustrates the use of että at the end of a possibly complete turn. In this case, the same speaker continues after a pause. However, the analysis of the syntactic structure of the utterance will show that it does not project a complement clause to follow. Also in terms of the sequential placement in the on-going action, the turn can be analyzed as possibly complete. Prosodically, however, the case is more complex. The example comes from a conversation between a female hairdresser and her male client. In the beginning of the sequence, the hairdresser starts a new topic by asking a question (“topic proffer”, see e.g. Schegloff 2007: 167–180). She asks whether the client is planning to attend the local rock festival (line 2).

(4) [Kotus, Hairdressers’, Eastern Finland T1208] ((H = hairdresser, C = client, both in their mid-twenties))

01

((43 sec))

02 H:

meinasitko Ilosaarirok↑kiim mennä? be.going.to-2sg-q event.name-ill go-inf are you going to go to the Ilosaari rock festival?

03 C:→ .mt (.) no en oov vielä suunni#tellu että#?, ptc neg-1sg be yet plan-p.pple että .mt (.) well I haven’t planned it yet että?, 04

(1.4)

05 C:

°kyllähän° siellä tietysti olis iham (0.2) ptc-clt dem.adv of.course be-cond fairly surely there would be of course fairly (0.2)

06

hyviäki (0.4) esiintyjiä. good-pl-cli performer-pl-prt good (0.4) performers too.

07

(1.0)

08 H:

mie taas #en ikinä# niistä esiintyjistä 1sg again neg-1sg ever 3pl-ela performer-pl-ela I for my part don’t ever care about those performers



Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen

09

niiv välitä mut siel so care but dem.adv that much but I think that there has always

10

[on miu]sta hirmu hyvä me:ininki=ai[na ollu.] be-3 1sg-ela terribly good atmosphere always be-p.pple been a terribly good atmosphere = always there.

11 C:

[mm. ] ptc well yes it has.

[no o:n. ] ptc be-3

In line 3, the client provides an että-final answer to the hairdresser’s question. The turn seems syntactically incomplete: the transitive verb suunnitella (‘to plan’) occurs without an object. But with respect to the turn’s sequential status as an answer to a question, the turn is complete: it relies heavily on the syntactic structure of the preceding question. In addition, the predicate verb suunnitella ‘plan’ is not a typical CTP – at least it does not occur in any of the complementizer-type että-instances of our data. Thus, the syntactic-sequential analysis shows that että does not function as a sign of an inferred complement clause and in this way, the että-final clause does not (necessarily) project more to come. However, in terms of the ongoing action, the turn’s completeness is not crystal clear. The turn functions as a negative answer to a question that is proffering a new topic. In spite of its negative formulation, the answer leaves the stance towards the question somewhat open: the client has not yet made plans regarding the rock festival. This openness (he doesn’t actually say yes or no) in the content of the answer may orient the hairdresser to expect an elaboration, even if it is not syntactically projected. After the että-final answer, there is relatively long pause (line 4). According to our interpretation, että here creates a place for turn transition. That is, että works as a particle that guides the hairdresser to infer that the client does not really have much to say on the topic and in this way, displays readiness to give up the turn. There is also prosodic evidence to support this analysis: although the turn has a slightly rising intonation, it also has some turn-final creak, which indicates transition relevance (Ogden 2001, 2004). 10 When the hairdresser remains silent, the client is free to go on (cf. Sacks, Schegloff & Jefferson 1974: 704 on turn-taking 10. The slightly rising intonation is not uncommon in että-final turns. It is unclear what it means, though: there is very little research on conversational prosody in Finnish. Routarinne (2003) shows that high final rise can be used to mark structural junctures in narratives. In this case it projects more talk but also invites some display of recipiency. Further, it can be used in non-final list-items (ibid. 177). Ogden (2004: 56), on the other hand, has noticed that rising intonation may co-occur with creak, indicating transition relevance. He concludes that intonation and voice quality are potentially separate systems in Finnish, serving different functions (ibid. 57).

Syntactic and actional characteristics of Finnish että-clauses 

rules). He does so by elaborating his answer in lines 5–6. This elaboration presents an argument for the positive alternative (going to the rock festival), but keeps it on a general level (kyllähän, tietysti ‘surely’, ‘of course’). By making this concession, the client orients to the function of the question as a topic-proffer and also to his role as a participant who is being invited to develop the topic further. However, the elaboration constitutes a new action, not a projected continuation: the continuation does not fit to the preceding utterance syntactically (cf. Auer 1992: 48; 1996: 60; Ford, Fox & Thompson 2002; cf. Mulder & Thompson 2008: 184–185). In other words, the continuation on lines 5–6 cannot be seen as grammatically tied to the että-final clause at line 3. Instead, in terms of both turn transition and sequential placement, että creates a place for negotiation: who will continue and in what direction (cf. Jefferson 1983). The previous example showed that the turnfinal että can be used to leave an implication of the turn unarticulated for the recipient to infer. In fact, the inference-inviting character seems to be typical of final particles which grammaticize from complementizers or conjunctions in general (see Mulder & Thompson 2008: 186 for final but; Koivisto in preparation for Finnish että (‘that/so’), ja ‘and’ and mutta ‘but’ as final particles). The actual implication of the final että can be understood in relation to the function of the turn-initial että. As we have seen, both the initial and the complementizer use of että mark the utterance that follows as a paraphrase, summary, or an upshot of something previously said, while the final että projects an upshot of the turn but leaves it to inference. The previous material in the turn provides the basis for inference, just as in the initial and complementizer uses of että, the prior talk forms the basis for the interpretation of what follows että. This shows all three uses of että are related. The next example illustrates this function of the final että. In this case, the recipient also takes the next turn. Tero and Mikko are talking about some practicalities concerning their participation in a street basketball tournament. Prior to this extract, Mikko has asked Tero if there is a need for similar T-shirts for the team. Tero has told him that the team members can wear T-shirts of similar colour or alternatively, play without shirts. Then he brings up a possible problem: the team members don’t necessarily have similar shirts (lines 5–6).

(5) [Että115 Sg143 B03]

01 Tero: niij jos ne t:oiset pelaa sitten; .hh ilman yeah if the others play then; .hh without 02

paitoja. shirts.

03

(0.2)

 Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen

04 Tero: ja toiset pelaa paidan kaa mutta; .hhh and other-pl play shirt-gen with but and some play with a shirt but; .hhh 05

meil ei sit kyllä varmaan yhtenäistä 1pl-ade neg then ptc probably similar-prt we don’t then probably have a similar shirt

06

paitaa oo kellää. shirt-prt be anyone.ade any (of us).

07

(0.6)

08 Mikko: [(no-) ptc well 09 Tero: [tai jotaim punasta ehkä vois or something-acc red-prt maybe can-cond or maybe we could take something red 10

ottaa mukaan. take along with us.

11

->onks sulla punasia; teepaitoja. be-q-clt 2sg-ade red-pl-prt T-shirt-pl-prt do you have red; T-shirts.

12

(0.2)

13 Mikko:=>n:o mull_ on yks semmonen punaraidallinen ptc 1sg-ade be one dem.adj red-stripe-adj w:ell I have one like with red stripes 14

ainaki se on aika punanen että?, että at.least 3sg be pretty red at least it is pretty red että?,

15 Tero: nii:n >kato ku< [mul o ainaki teepaita ptc ptc when 1sg-ade be at.least T-shirt Well, look, because I have a T-shirt that’s 16 Mikko: [tse 17 Tero: punane ja; tolla on kans joku pelipaita red and dem-ade be also some game-shirt red and; that (guy) also has some team shirt

Syntactic and actional characteristics of Finnish että-clauses 

18

punanen tolla; red dem-ade that’s red that (guy),

19

(0.2)

20 Mikko: j:oo. ptc yeah. In lines 9–10, Tero suggests that the team members could bring something red with them. Then he asks whether Mikko has red T-shirts (line 11). In comparison to the topic-proffer question about the rock concert in the previous example, this question seems to be a genuine request for information. Syntactically, it is a polar question which projects either an affirmative or a negative answer as a next relevant action. However, Mikko provides an että-final answer which takes a form of two full clauses (lines 13–14) (cf. Hakulinen 2001). Again, the syntax of the ettäfinal turn does not project a grammatical ‘complement’ to follow; rather että functions as a particle that guides the interpretation of the turn. In his answer, Mikko states that he has a shirt with red stripes which is aika punanen ‘pretty red’. This is not an unambiguously affirmative or negative answer but rather falls between the two polarities. This is actually typical of että-final answers in general. When occurring at the end of an answer, että calls for the recipient’s understanding for the fact that the speaker is unable to provide a straightforward answer. Moreover, että invites the recipient to infer the relevance of the answer in its context on the basis of what has been said. In Example (5), että seems to be exposing the answer to the recipient’s judgment: it is up to Tero to decide whether the described shirt with red stripes qualifies as a red shirt. That is, the final että makes the status of the turn as a positive answer to the question negotiable and leaves the responsibility of its treatment to the recipient. What is interesting is that the turn carries no prosodic or phonetic signals for transition relevance. However, the recipient treats the turn as complete by taking the turn smoothly. In both of the previous examples, the final että does not grammatically project a complement clause to follow but rather works as a particle that leaves some implication of the turn to inference. In terms of the ongoing action, the final että guides the interpretation of the turn and creates expectations for the next relevant action. The two previous examples were representative of a typical sequential context for että-final turns, answers to questions. In these cases, it seems clear that the final että does not have anything to do with syntactic subordination. However, through the analysis of the next example, we will consider the possibility of interpreting the että-final turn as subordinate in terms of the on-going action and sequential development. In this example, the että-final turn links back to previous context by



Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen

providing an explanation for something said in the beginning of the sequence. The contribution of the final että is to mark this connection and thereby invoke the prior context (on backward linking as a property of final particles in general, see Koivisto in preparation). That is, että ties its turn (and its interpretation) to some prior activity that has been going on earlier in the sequence and indexes the turn’s relevance to it (cf. Raymond 2004: 204 on stand-alone so). In this example, Ella has called her friend Aune to ask for her help in organizing a coffee serving after a church concert on Saturday evening (see lines 1–7).

(6) [Että62 Sg142 B06]

01 Ella: .mthHH mut ↑tiäksää mitäh::. [kunm- mä kyselisin but you know what::. I would ask 02 Aune: [°ni°, what 03 Ella:

sulta nyt semmosta asiaa ettäh; .hh että olisiksä you such a thing that; that would you have been

04

pystyny auttamaan meittiä tommosessa kun ens, able to help us in such a thing as next

05

lauantai-iltana kirkon kesäjuhlien yhteydessä Saturday evening at the church’s summer festival

06

on semmoset, (.) .hhh iltakonsertin jälkeen there’s such, (.) after the evening concert

07

Pihatuvalla ↓kah:vit.h at Pihatupa a coffee serving.

08 Aune: no kun: mä: (.) tuossa (0.2) kuulin noin: tolta well I: just heard from 09

P:irjetalta että olis: olis tehtaviä hommia Pirjetta that there would have been assignments

10

ollu mut mä: en oo p:aikalla mää< m- lähen jobs but I won’t be around I will leave for a

11

piänelle ma:t[kalle. little trip.

12 Ella: 13

[↑soittiks Pirjetta ↑did Pirjetta

£j(h)o[(h) hhh] heh heh£ alr(h)eady call you heh heh ((side sequence omitted))

Syntactic and actional characteristics of Finnish että-clauses 

[ai se:m↑mosii. oh things like that.

38 Ella: 39

just juu. nii. nii. tää on niin semmonen (.) that’s right. right. right. this is such a

40

laajah, [.hh laaja-alanen asia £he he£] exten- extensive issue [niin; (on). mone]n:laista. yes; (it is). all kinds of things.

41 Aune: 42

et mul on tosiaan pieni, so I indeed have a little,

43

(0.3)

44 Ella:

ai sä [meet matkalle.] oh you are leaving for a trip.

45 Aune: 46 Ella: 47 Aune:

[piäni a little

rei:]ssuh.= trip.

= sehän on hauskaa. ohh how nice. [joo. yeah.

48 Aune:→#en mää noin tota:# (0.8) pitkään oo pois. neg-1sg 1sg ptc ptc long be away I won’t be away for a long time. 49

=mut oon kumminki (noi) viikol#lopun# sitte but be-1sg however those week-end-acc then =but I will be however

50

↓#poi[:s# että?, away että away for the weekend että?,

51 Ella:

[↑ju:st ↓juu.= I see.

52 Aune: =mm-m.= 52 Ella:

=.hh no sehän on mukava virkistävä kun on well that’s nice refreshing that you have

53

tämmöneh, (0.5) tämmö[nen ede]ssä. this kind, (0.5) this kind of thing ahead.



Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen

54 Aune:

[ku en mä o] I haven’t been

55 Aune: yhtään (noita) missään# (0.5) missään at all anywhere been anywhere in fact. 56

>o ollukka< mull_oli >niinku< lapsia täss#ä# (0.2) I had like children here that I took care

57

hoidossa kans kesä, (.) kesäloman aikana °ja°.h of also during summer, (.) summer vacation and.

In response to the request, Aune says that she has already heard about the work offer (lines 8–9). In the same turn, she also reports that she won’t be available as she is getting away for a “little trip”. This report can be seen to function as a rejection, which is a dispreferred response (Drew 1984). However, Ella doesn’t address this issue at this point but asks – with surprise in her voice – if Pirjetta (a mutual acquaintance) has already called Aune. This question launches a sequence in which the interactants solve some misunderstandings regarding the information Aune has received from other people (data not shown). After the closure of this side sequence, Aune gets back to her announcement about the trip (et mul on tosiaan pieni reissu ‘so I indeed have a little trip’, lines 42, 45), which was earlier ignored. The recipient, Ella, now responds counter to expectation by making a positive evaluation regarding the matter (line 46). After this, Aune proceeds to her että-final turn (lines 48–50). In her two-clause turn, Aune begins by stating that she won’t be away for a long time and thereby expresses a concession regarding the length of the trip. The second, että-final, clause is prefaced with a constrastive particle mut (‘but’) and includes a contrastive connective kumminkin (‘however’). With these contrastive markings Aune foregrounds the upcoming utterance and marks it as a return to the central issue from her point of view (mut oon kuitenkin (noi) viikollopun sitte pois että ‘but I will be however away for the weekend so’) (cf. Schiffrin 1987: 165–177; see also Couper-Kuhlen & Thompson 2005 on retractions and revisions of an initial statement). In the light of the beginning of the whole extract, the turn is hearable as a renewed (and specified) account for rejecting the request. By referring to the exact time of her absence, the weekend, Aune implies that she won’t be available on Saturday (the date of the event). We claim that the että particle invites the recipient to interpret the turn as an account for Aune not being able to accept the offered assignment11.

11. The particle sitte (‘thus’) as a marker of inference reinforces this function: it also indicates that the explanation relates to something said earlier (see, e.g. ISK: 794; Halonen 2005).

Syntactic and actional characteristics of Finnish että-clauses 

By repeating that she will be absent at the time of the event, Aune’s turn links back to her earlier turn at lines 8–11. The function of the final että is to invite the recipient to treat the turn as a renewed account for that earlier turn. In this way the että-final turn invokes the previous rejection that was not yet acknowledged by Ella and implies readiness to bring the request–rejection sequence to a close (cf. Raymond 2004 on stand-alone so). Like the other two types of että discussed above, this että thus also links to an earlier utterance. By using the final particle että, Aune calls for some display of understanding from Ella so that the delicate topic can be closed down. That is what indeed seems to happen. Ella claims understanding at line 51 with the particles just juu, which may serve to acknowledge the issue of unavailability. After that, however, Ella turns to evaluate Aune’s holiday plans in general. In fact, the actual rejection – the dispreferred response – does not get addressed explicitly at any point of the conversation. But what can we say about the question of subordination? Our examples of the use of the particle että have demonstrated that in conversational talk, että does not function as a complementizer but has multiple tasks in different syntactic positions (see also Seppänen & Laury 2007; Laury & Seppänen 2008; cf. Keevallik 2008 on the Estonian et). However, we are suggesting that some että-final turns could be interpreted as being in the service of and dependent on some ‘main’ activity. In Example (6), the main activity (or at least the “original” activity) of the expanded sequence is the request and its rejection. When we examine the että-final turn in relation to this main activity, it is hearable as an account for the earlier dispreferred response as well as a pursuit of recognition of the earlier rejection of the request. Thus, we can say that the conversational status of the että-final turn as an account is dependent on this previous rejection. The implication of että in Example (6) is something like ‘take this (or treat this) as a reason for not being able to grant your request’. Thus the että-final turn invokes the prior context. It seems clear that in certain backward-connecting activities (namely, in explanations or justifications) että has a capacity of indicating the relevance of the turn to some earlier conversational action that has not yet been satisfactorily dealt with. It is possible to say, then, that in this kind of case, että subordinates the turn to something that has an ongoing relevance for the speaker – thus, to the main activity (cf. Raymond 2004). In this section of the paper we have seen that että can also function as a turnfinal particle, attached to turns that are syntactically complete. In these cases että does not project a subordinate clause to follow but rather implies readiness to relinquish speakership to the recipient. This job is done by projecting an unstated upshot of the turn and leaving it to the recipient to infer. When occurring at the end of an answer directly following the question, että calls for the recipient’s understanding of the fact that the speaker is unable to provide an expected answer. Että-final turn can also provide an account for an earlier action in the sequence



Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen

and thereby invoke the previous context. The function of the final että in this sequential context is to alert the recipient to the fact that this turn is related to some previous conversational action. In all contexts, final että is used to imply a certain kind of next relevant action from the recipient. In that sense, it functions metacommunicatively, regulating the interaction by managing the participant framework and turntaking, just like the other uses of että discussed in the two preceding sections of the article. 6. Discussion We hope to have shown that että appears in three distinct syntactic positions, as a complementizer within an utterance at the end of a projective phrase or CTP, as an initial particle in an utterance, and as a final particle, and that in all three uses, it functions to organize the participation framework of the conversation. In all three uses, some prior talk is being indexed as relevant. What follows että is a paraphrase, summary, or upshot of that prior talk; in its final use, the upshot of the prior talk is left for the recipient to infer. When että occurs with a projective CTP-phrase, the phrase serves as an index to what prior talk is being made relevant. In its initial particle use, että functions alone to make the prior talk relevant. In its final use, the implied upshot is to be made on the basis of the material provided in the että-final turn. Syntactically, we suggest, että is not a subordinator. In the initial use, there is no clause preceding että that could function as its main clause, while in the final use, there is no clause following että which could be subordinate to the prior clause; furthermore, the utterance preceding final että often is not a CTP or other type of construction which would syntactically project an että-complement. We argued here that even in its use with a CTP, the talk following että is not syntactically subordinate to the CTP. The most frequent types of CTP-phrases have been grammaticized or particle-ized into projective phrases, or else the CTP contains a pronominal element which functions as its complement. Furthermore, what follows että is often not clausal in nature. It is also clear that clauses (or longer stretches of talk) following että are not subordinate on the level of action, just as they are not subordinate syntactically. The talk following että is regularly oriented to by the participants on the level of content, and it carries forward the main action they are engaged in. It is what the other participants respond to, as in Example (1), or it may constitute a candidate understanding of what has been said just prior as in Example 3. However, and differing from Thompson (2002) and Langacker (2008), we suggest that in our data, the profile of the CTP is not overriden by what follows it. While CTPs containing

Syntactic and actional characteristics of Finnish että-clauses 

että, or the particle että by itself, are not subordinating, neither are they, in our view, subordinate themselves in terms of action; they manage interaction on the strategic level, by regulating the participant framework of the conversation (for example, by relegating partial responsibility for what is being said to a person who is not present in the conversation, as in Example 2), and by regulating turntaking by projecting forward to more talk to come, as in Examples 1, 2, and 3, and by indexing the type of upcoming talk as a paraphrase, summary or candidate understanding of prior talk. And the metalinguistic action performed by them is also oriented to by the participants (see, for example, the analysis of Examples 3 and 6. For these reasons, we suggest that the CTP (or, more exactly, formulaic phrases) containing että, as well as the particle uses of että, implement independent actions in service of the main activity performed. We have also shown that in certain cases in our data, the utterances preceding final että, while not syntactically subordinate in any sense, can be considered ‘subordinate’ to a prior action in terms of sequential structure. 7. Conclusions In this article, we have examined the use of the Finnish että as a linker (‘complementizer’) within an utterance and as an initial and final particle. We have suggested that in the utterance-internal use, the CTP (formulaic phrase) and the segment of talk following että turn out not to be subordinate to one another either syntactically or actionally. Instead, we suggest that the formulaic phrase containing että, as well as the particle uses of että, implement independent actions in service of the main action: they function metacommunicatively, managing the talk strategically by regulating the participant framework and the turntaking, and can also be shown to be oriented to by the participants, while the talk following että is oriented to by the participants in terms of the content and the main action. However, when että is not part of a CTP, but functions as a turn-final particle, the turn in which it appears, although syntactically independent, can be seen as a dependent activity from the perspective of some previous action. References Auer, Peter. 1992. “The Neverending Sentence: Rightward Expansion in Spoken Language”. In Studies in spoken languages: English, German, Finno-Ugric Miklós Kontra and Tamás Váradi (eds), 41–59. Budabest: Linguistics Institute, Hungarian Academy of Sciences. Auer, Peter. 2005. “Projection in interaction and projection in grammar.” Text 25 (1): 7–36.



Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen Bakhtin, M.M. 1987 [1953]. “The problem of speech genres”. In Speech Genres and Other Late Essays, C. Emerson and M. Holquist (eds), 60–102. Austin TX: University of Texas Press. Couper-Kuhlen, Elizabeth and Thompson, Sandra A. 2005. “A linguistic practice for retracting overstatements: concessive repair.” In Syntax and lexis in conversation, Auli Hakulinen and Margret Selting (eds), 257–288. Amsterdam: Benjamins. Cristofaro, Sonia. 2003. Subordination. Oxford: Oxford University Press. Drew, Paul. 1984. “Speakers’ reportings in invitation sequences.” In Structures of social action: studies in conversation analysis, J. M. Atkinson and John Heritage (eds), 129–151. Cambridge: Cambridge University Press. Ford, Cecilia, Fox, Barbara and Thompson, Sandra A. 2002. “Constituency and the grammar of turn increments”. In The language of turn and sequence, Cecilia Ford, Barbara Fox and Sandra Thompson (eds), 14–38. Oxford: Oxford University Press. Goffman, Erving. 1981. Forms of Talk. Oxford: Basil Blackwell. Goodwin, Charles. 1979. “The interactive construction of a sentence in natural conversation.” In Everyday Language: Studies in Ethnomethodology, George Psathas (ed.), 97–121. New York: Erlbaum. Goodwin, Charles. 1981. Conversational Organization: Interaction between Speakers and Hearers. New York: Academic Press. Goodwin, Charles. 1984. “Notes on story structure and the organization of participation.” In Structures of Social Action: Studies in Conversation Analysis, J. M. Atkinson and John Heritage (eds), 225–246. Cambridge: Cambridge University Press. Goodwin, Charles and Goodwin, Marjorie H. 1990. “Context, activity and participation.” In The contextualization of language, Peter Auer and Aldo di Luzo (eds), 77–99. Amsterdam: Benjamins. Haiman, John and Thompson, Sandra A. 1984. “‘Subordination’ in Universal Grammar.” In Proceedings of the Tenth Annual Meeting of the Berkeley Linguistics Society, Claudia Brugmann and Monica Macauley (eds), 510–523. Berkeley: University of California Press. Hakulinen, Auli. 1989. “Partikkelit ja muut kiteymät vuoroissa.” In Suomalaisen keskustelun keinoja I, Auli Hakulinen (ed.), 115–117. Kieli 4. Helsinki: Department of Finnish, University of Helsinki. Hakulinen, Auli. 2001. “Minimal and non-minimal answers to yes-no questions.” Pragmatics 11, 1–16. Halonen, Mia. 2005. “Mä en sit siihe sanonu mitää. Raportit sanomatta jättämisestä ja evidentiaalinen partikkeli sit(te(n)) keskustelun kertomuksissa.” Virittäjä 109: 272–298. Hopper, Paul. 2001. “Grammatical constructions and their discourse origins. Prototype or family resemblance?” In Applied Cognitive Linguistics I: Theory and Language Acquisition, Martin Pütz, Susanne Niemeier and René Dirven (eds), 109–130. Berlin: Mouton de Gruyter. Hopper, Paul. 2004. “The openness of grammatical constructions”. Chicago Linguistic Society 40, 239–256. Hopper, Paul and Thompson, Sandra A. 2008. “Projectability and clause combining in interaction.” In Crosslinguistic Studies of Clause Combining: the multifunctionality of conjunctions, Ritva Laury (ed.), 99–123. Amsterdam: Benjamins. ISK = Hakulinen, Auli, Vilkuna, Maria, Korhonen, Riitta, Koivisto, Vesa, Heinonen, Tarja Riitta, Alho, Irja. 2004. Iso suomen kielioppi. Helsinki: Suomalaisen Kirjallisuuden Seura. Jefferson, Gail. 1983. “On a failed hypothesis: ‘conjunctionals’ as overlap-vulnerable”. Tilburg papers in language and literature 28. Tilburg: Tilburg University.

Syntactic and actional characteristics of Finnish että-clauses  Keevallik, Leelo. 2008. “Conjunction and sequenced actions: The Estonian complementizer and evidential particle et.” In Crosslinguistic Studies of Clause Combining: the multifunctionality of conjunctions, Ritva Laury (ed.), Amsterdam: Benjamins. Koivisto, Aino. 2006. Että-loppuiset vuorot keskustelussa. [Turns ending with että in conversation.] Unpublished MA thesis, Department of Finnish, University of Helsinki. Koivisto, Aino. (In preparation). Konjunktiot loppuasemaisina: jatkeita vai vuoronloppuisia partikkeleita. [Conjunctions in Final Position: continuers or turn-final particles?] Ph.D. thesis, Department of Finnish, University of Helsinki. Kärkkäinen, Elise. 2004. Epistemic Stance in English Conversation: A Description of its Interactional Functions, with a Focus on I Think. Amsterdam: John Benjamins. Laitinen, Lea. 2005. “Hän, the third speech act pronoun.” In Minimal reference in Finnic: The use and interpretation of pronouns in Finnish and Estonian discourse, Ritva Laury (ed.), 75–106. Studia Fennica Linguistica 12. Helsinki: Suomalaisen Kirjallisuuden Seura. Langacker, Ronald. 1991. Foundations of Cognitive Grammar, vol. 2. Descriptive application. Stanford: Stanford University Press. Langacker, Ronald. 2008. “Subordination in cognitive grammar.” In Asymmetric Events, Barbara Lewandowska-Tomaszczyk (ed.), 137–149. Amsterdam: Benjamins. Laury, Ritva. 2006. “On Subordination, Finnish-style: Questioning the category of finite clausal complements in spoken Finnish.” In A Man of Measure. Festschrift in Honour of Fred Karlsson on his 60th Birthday, Mickael Suominen, Antti Arppe, Anu Airola, Orvokki Heinämäki, Matti Miestamo, Urho Määttä, Jussi Niemi, Kari K. Pitkänen and Kaius Sinnemäki (eds), 310–321. Turku: The Linguistic Association of Finland. Laury, Ritva and Seppänen, Eeva-Leena. 2008. “Clause combining, interaction, evidentiality, participation structure and the particle-conjunction continuum.” In Crosslinguistic Studies of Clause Combining: the multifunctionality of conjunctions, Ritva Laury (ed.), 153–178. Amsterdam: Benjamins. Leino, Jaakko. 1999. “Mitä tarkoittaa se, että? Se-pronominista subjektina ja objektina toimivan että-lauseen yhteydessä.” [What does se, että mean? About the pronoun se in connection with että-clauses functioning as subjects and objects.] Virittäjä 103: 27–51. Lerner, Gene H. and Kitzinger, Celia. 2007. “Extraction and aggregation in the repair of individual and collective self-reference.” Discourse Studies 9 (4): 526–557. Lindén, Eeva. 1957. “Eräistä sivulauseattribuuteista.” Virittäjä 61: 102–111. Local, John and Kelly, John. 1986. “Projection and ‘silences’: Notes on phonetic and conversational structure.” Human studies 9: 185–204. Local, John and Walker, Gareth. 2005. “Methodological imperatives for investigating the phonetic organization and phonological structures of spontaneous speech”. Phonetica 62: 120–130. Matthiessen, Christian and Thompson, Sandra A. 1988. “The structure of discourse and ‘subordination’.” In Clause Combining in Grammar and Discourse, John Haiman and Sandra.A. Thompson (eds), 275–329, Amsterdam: John Benjamins. Mulder, Jean andThompson, Sandra.A. 2008. “The grammaticization of but as a final particle in English conversation.” In Crosslinguistic Studies of Clause Combining: the multifunctionality of conjunctions, Ritva Laury (ed.), 179–204. Amsterdam: Benjamins. Munro, Pamela.1982. “On the transitivity of ‘say’ verbs.” In Studies in Transitivity, Paul Hopper and Sandra A. Thompson (eds). 301–318, New York: Academic Press. Ogden, Richard. 2001. “Turn transition, creak and glottal stop in Finnish talk-in-interaction”. Journal of the International Phonetic Association 31: 139–152.

 Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen Ogden, Richard. 2004. “Non-modal voice quality and turn-taking in Finnish”. In Sound Patterns in Interaction, Elizabeth Couper-Kuhlen and Cecilia E. Ford (eds), 29–62. Amsterdam: John Benjamins, Okamoto, Shigeko. 1995. “Pragmaticization of meaning in some sentence-final particles in Japanese”. In Essays in semantics and pragmatics in honor of Charles J. Fillmore, Masayoshi Shibatani and Sandra A.Thompson (eds) s. 219–249. Amsterdam: John Benjamis. Okamoto, Shigeko and Ono, Tsuyoshi. 2008. “Quotative tte in Japanese. Its multifaceted functions and degrees of ‘subordination.’ In Crosslinguistic Studies of Clause Combining: the multifunctionality of conjunctions, Ritva Laury (ed.), 205–230. Amsterdam: Benjamins. Östman, Jan-Ola. 1981. You know: A Discourse-Functional Approach. Amsterdam: John Benjamins. Raussi, Anne. 1992. Et(tä)-partikkelilla alkavat lausumat keskustelussa. Unpublished MA thesis, Department of Finnish, University of Helsinki. Raymond, Geoffrey. 2004. “Prompting Action: The Stand-Alone “So” in Ordinary Conversation.” Research on Language and Social Interaction 37: 185–218. Routarinne, Sara. 2003. Tytöt äänessä. Parenteesit ja nouseva sävelkulku kertojien vuorovaikutuskeinoina. [Girls Talking. Parentheses and Rising Intonation as Narrator’s Interactional Devices]. Helsinki: Suomalaisen Kirjallisuuden Seura. Routarinne, Sara. 2005. “Keskustelupuheen johtolauseiden kielioppia.” [On the grammar of matrix clauses in conversation.] In Referointi ja moniäänisyys, Markku Haakana and Jyrki Kalliokoski (eds), 83–113. Helsinki: Suomalaisen Kirjallisuuden Seura. Sacks, Harvey. 1974. “An analysis of the course of a joke’s telling in conversation.” In Explorations in the ethnography of speaking, R. Bauman and Joel Sherzer (eds), 337–353. Cambridge: Cambridge University Press. Sacks, Harvey. [1971] 1992. Lectures on Conversation. Volume II. Ed. by Gail Jefferson. Oxford: Basil Blackwell. Sacks, Harvey, Schegloff, Emanuel and Jefferson, Gail. 1974. “A simplest systematic for the organization of turn-taking in conversation.” Language 50: 696–735. Schegloff, Emanuel A. 1982. “Discourse as an interactional achievement: some uses of ‘uh huh’ and other things that come between sentences.” In Analyzing discourse: text and talk, Deborah Tannen (ed.), 71–93. Georgetown: Georgetown University Press. Schegloff, Emanuel. 2007. Sequence organization in interaction. A primer in conversation analysis vol. 1. Cambridge: Cambridge University Press. Scheibman, Joanne. 2002. Point of View and Grammar. Amsterdam: Benjamins. Schiffrin, Deborah. 1987. Discourse Markers. Cambridge: Cambridge University Press. Seppänen, Eeva-Leena. 1996. “Ways of referring to a knowing co-participant in Finnish conversation.” In SKY 1996. Yearbook of the Linguistic Associastion in Finland, Timo Haukioja, Marja-Liisa Helasvuo and Elise Kärkkäinen (eds), 135–176. Helsinki: The Linguistic Association of Finland. Seppänen, Eeva-Leena. 1998. Läsnäolon pronominit. Helsinki: Suomalaisen Kirjallisuuden Seura. Seppänen, Eeva-Leena and Laury, Ritva. 2007. “Complement clauses as turn continuations: the Finnish että-clause.” Pragmatics 17 (4) 553–572. Shore, Susanna. 1986. Onko suomessa passiivia. Helsinki: Suomalaisen Kirjallisuuden Seura. Tannen, Deborah. 2007. Talking Voices: Repetition, Dialogue and Imagery in Conversational Discourse. New York: Cambridge University Press, 2007. Thompson, Sandra A. 2002. “‘Object complements’ and conversation: towards a realistic account.” Studies in Language 26 (1): 125–164.

Syntactic and actional characteristics of Finnish että-clauses 

Appendix 1 Transcription symbols . , ; ?, ? ↑ ↓ en : [ ] (.) (0.5) > < < > en< a- = #e# °en° hh .hh

falling intonation level intonation slightly falling intonation slightly rising intonation rising intonation rise in pitch fall in pitch emphasis indicated by underlining lengthening of a sound utterances starting simultaneously point where overlapping talk stops micropause, less than 0.2 seconds silences timed in tenths of a second talk inside is at a faster pace than the surrounding talk talk inside is at a slower pace than the surrounding talk glottal stop cut off “latching”, i.e. no silence between two adjacent utterances creaky voice talk inside is more quietly than the surrounding talk audible exhalation audible inhalation

Appendix 2 Grammatical glosses acc ade adj adv all clt comp cond

accusative adessive adjective adverb allative clitic complementizer conditional

 Aino Koivisto, Ritva Laury and Eeva-Leena Seppänen

dem det ela ess gen ill ine inf neg pass pl poss p.pple prt pst ptc q sg 1nameM

demonstrative determiner elative essive genitive illative inessive infinitive negation passive plural possessive suffix past participle partitive past particle question singular first name male

Clause-combining and the sequencing of actions Projector constructions in French talk-in-interaction Simona Pekarek Doehler

University of Neuchâtel, Switzerland This paper extends recent discussions on projector constructions to another language (French), and to a further set of linguistic formats. I propose a reanalysis, in the light of the sequential organization of actions, of the je veux dire ‘I want to say’+ complement clause pattern, the il y a ‘there is’ presentational cleft and the pseudocleft construction. I first identify a recurrent constructional schema consisting of an initial ‘fragment’ (je veux dire, il y a NP or ce qui/ce que x) followed by a syntactically independent stretch of talk. I then show that these fragments routinely accomplish a specific social action, namely projecting upcoming talk. They are projector constructions by means of which speakers recipient-design their turns, helping recipients to monitor through the complex architecture of talk. I conclude by discussing implications regarding the on-line character of grammar.

1. Introduction1 In any verbal exchange system, participants use grammar as a central resource for the organization and coordination of actions (Ochs, Schegloff & Thompson, 1996). Although actions, turns at talk or turn construction units do not necessarily coincide with clauses, the clause has been shown to be a relevant unit of interaction (Thompson & Couper-Kuhlen, 2005): participants orient to clause structure for turn-taking (Sacks, Schegloff & Jefferson, 1974; Ford & Thompson, 1996; Selting, 1996, 2000), action projection (Auer, 2005; Hayashi, 2004) or utterance co-construction (Lerner, 1991). This raises a central question: how does the organization of clause-combining relate to the sequential organization of actions? 1. I am grateful to Arnulf Deppermann and the two editors of this volume, Ritva Laury and Ryoko Suzuki, for valuable feedback on an earlier version of this paper.

 Simona Pekarek Doehler

In the literature, clause-combining has traditionally been framed in terms of subordination and coordination, parataxis and hypotaxis. The notion of subordination is commonly understood to refer to a type of clause-combining which involves the relative dependency of one clause with regard to another (i.e. the main or matrix clause). Clause-combining of this type is typically considered to be based on the morphosyntactic marking of the subordination relation by means e.g. of subordination marker or infinitival construction. This marking provides the basis for the identification of the main clause and the subordinate clause respectively. For some time, however, dichotomous views of clause-combining – which categorically oppose coordination and subordination, or parataxis and hypotaxis – have come under critique (Auer, 1998; Günthner, 1999; Matthiessen & Thompson, 1988, inter alia), as has the focus on morphologic marking for identifying subordination (Deulofeu, 1988; Berrendonner, 2008). The organization of actions, turns and turn constructional units (TCUs) in talk-in-interaction has been discussed along different lines: it is understood in terms of the sequential placement of units within the flow of talk. Work in conversation analysis and interactional linguistics has abundantly documented the fact that relations between parts of talk are not based on mere linguistic means, but are organized on a moment-by-moment basis along the temporal unfolding of talk; they are established as a function of the sequential positions of units of talk, and their sequential organization is based on projections emanating from one unit onto a next (Auer, 2005, 2007 and 2009; Goodwin 2002; Schegloff, 1996; Selting, 2000, inter alia). The contrast between the principles that are commonly understood to underpin the combination of basic units of grammar (i.e. clauses), on the one hand, and those that are associated with the organization of basic units of talk (i.e. actions, turns, TCUs), on the other hand, raises a twofold question: – What is the nature of the relation between clause-combining in grammar and the sequencing of actions? What is the ‘fitting’ between syntactic and actional relations? – How does this relation or ‘fitting’ play into the notion of an on-line syntax (Auer, 2005, 2007, 2009), that is: a syntax which is configured through the moment-by-moment unfolding of talk, along the temporal deployment of actions? In this paper, I wish to shed light on these questions by exploring a limited range of grammatical patterns in French talk-in-interaction. Recent accounts of clause-combining in interaction, based on German and on English data, have pointed out that the initial pieces of what is commonly considered as standard bi-clausal constructions (in particular wh-clefts, extrapositions

Projector constructions in French talk-in-interaction 

and if-then clauses) often are not part of syntactically related bi-clausal patterns (Hopper, 2001, 2004; Hopper & Thompson, 2008; Günthner, 2006; Günthner & Hopper, 2010): these wh-clauses, it-clauses and if-clauses occur in talk-in-interaction as “prefabs” that project certain types of actions; they are not linked to what follows by any morphosyntactic means; they are followed by a discourse segment that can take many different shapes, ranging from a single clause to a complex, multi-unit stretch of talk. In this paper, I wish to extend these considerations to another language, French, and to other constructions. I propose a reanalysis, in the light of the sequential organization of actions, of three frequent clause-combining patterns, namely the je veux dire ‘I want to say’ + complement pattern, the il y a ‘there is’ presentational cleft and the pseudo-cleft-construction. What grammarians consider as standard forms of these is illustrated in (1) to (3):2 (1)

(sarroy) je veux dire aussi à nos anciens (..) I want say.inf also to our elderly qu’ils seront protégés. that they will-be protected ‘I want to say to our elderly people also that they will be protected’

(2)

(L1-lae1) il y a (.) un robinet qui coule. pro.3sg.m loc has a faucet that leaks ‘there is a faucet that is leaking’

(3)

(sarroy) ce qui ronge aujourd’hui l’équilibre de la pro.3sg.n pro.rel.sbj gnaws today the equilibrium of det société française ce sont les jeunes diplômés et society French pro.3sg.n are the young certified and qualifiés qui n’ ont pas accès à l’entreprise. qualified who neg have neg access to det business ‘what gnaws today at the equilibrium of French society are the certified and qualified youth that do not have access to business’

2. Some of the characteristics of these constructions are not reflected in the gloss. While the il in il y a ‘there is’ as well as the c’ ‘it’ (combined with the copula est ‘is’) in the pseudocleft are isomorphic with the third person masculine and neuter pronouns respectively, and are rendered as such in the gloss, they in fact function within these constructions as dummy pronouns. The a in ‘il y a’ there is, although formally corresponding to the third person singular of the verb avoir ‘to have’, functions within the construction as a de-semanticized construction auxiliary (cf. Blanche-Benveniste, 1989).

 Simona Pekarek Doehler

This paper sets out to empirically document the following points: – What has classically been analyzed as verb + complement clause construction (1), presentational cleft (2) and pseudocleft (3) often do not occur in talk-ininteraction as bi-clausal patterns; rather, naturally occurring talk regularly shows a configuration consisting of the initial pieces of these constructions (je veux dire, il y a NP and ce qui/ce que x – which are fragments, not full clauses) plus a syntactically unrelated (multi-unit) stretch of talk. – In all three cases, the two parts of the configuration are not syntactically combined, but are related by means of projections emanating from the initial fragment onto what follows. – In all three cases, the initial fragment functions as grammatical resource for announcing upcoming talk by the same speaker, for holding the floor and for constructing complex turns. On the basis of these observations, I argue that the je veux dire, il y a NP and ce qui x fragments recurrently found in our data are ‘projector constructions’ (see most recently Günthner & Hopper, 2010; or ‘projector phrases’, Günthner, 2006). Their nature as constructions is corroborated by the relative consistency of the lexicosyntactic and prosodic shapes of each of the three projector elements and by the recurrent interactional work they accomplish in talk-in-interaction. The data provides evidence for a sedimentation of the three constructions as discourse organizational devices. Through these points, I wish to contribute to the ongoing discussion on projections as basis for both the organization of social interaction (Auer, 2005; Hayashi, 2004, inter alia) and clause-combining (Günthner, this volume; Günthner & Hopper, 2010; Hopper & Thompson, 2008; Pekarek Doehler et al., 2010). In what follows, I first provide some background on data and on the grammatical patterns under analysis (pt. 2). I then analyze the ‘fitting together’ of stretches of talk by means of projections emanating from an initial projector construction and I show that the three constructions analyzed here are used by speakers as a resource for framing subsequent talk in a specific way and for holding the floor (pt. 3). In a second step, I focus on one further interactional job accomplished by projector constructions: they serve to organize complex turns, i.e. turns consisting of multiple TCUs (pt. 4). Results are discussed in the light of recent developments in interactional linguistics on the on-line character of grammar and the emergent nature of constructions (pts. 5 and 6). 2. Data, procedure and the grammatical patterns under discussion This paper is based on a corpus of roughly 25 hours of audio- and video-taped interactions of various kinds (focus-group discussions, TV debates, casual

Projector constructions in French talk-in-interaction 

conversation) involving adult speakers of French. Most of the data was collected in the French-speaking part of Switzerland. The data have been transcribed following transcription conventions commonly used in conversation analyses (see annex). Throughout this paper, I use the term ‘construction’ in line with usage-based approaches to grammar to refer to patterns of language use of various sizes (e.g. NP, clause, clause-combination) comprising multiple linguistic items. Constructions (see e.g. Goldberg, 1995) are (more or less) sedimented patterns for accomplishing communicative functions/actions; their meaning or function cannot be derived from the sum of their constituents; they develop and are reconfigured as a response to recurrent communicative needs. I use the term ‘configuration’ in order to account for discourse patterns that appear to be shaped ad hoc (cf. Pekarek Doehler et al., 2010), and which may include more sedimented constructions along with grammatically unspecified slots. Accordingly, I will call configurations the patterns that are initiated by projector-constructions, but whose second part may be of various morphosyntactic and lexical shape, and of various length. The following excerpts, quoted earlier, show what linguists commonly consider as standard forms of je veux dire + complement clause (1), presentational (2) and pseudocleft constructions (3) in French. (1)

(sarroy) je veux dire aussi à nos anciens (..) I want say.inf also to our elderly qu’ils seront protégés. that they will-be protected ‘I want to say to our elderly people also that they will be protected’

(2)

(L1-lae1) il y a (.) un robinet qui coule. pro.3sg.m loc has a faucet that leaks ‘there is a faucet that is leaking’

(3)

(sarroy) ce qui ronge aujourd’hui l’équilibre de la pro.3sg.n pro.rel.sbj gnaws today the equilibrium of det société française ce sont les jeunes diplômés et society French pro.3sg.n are the young certified and qualifiés qui n’ont pas accès à l’entreprise. qualified who neg have access to det business ‘what gnaws today at the equilibrium of French society are the certified and qualified youth that do not have access to business’

I will briefly discuss the characteristics of each of these constructions in order to differentiate them from the configurations analyzed here.

 Simona Pekarek Doehler

Excerpt (1), taken from the Sarkozy-Royal presidential debate, shows a sentential pattern where the verbal complex vouloir dire [modal Aux + V] appears as a complement taking predicate, followed by the subordination marker que ‘that’ introducing a complement clause. Complement clauses are considered as prototypical instances of subordination (cf. Langacker, 1991: 436sq.; but see Thompson, 2002, for a critique of complement clauses as subordinate clauses). Excerpt (2), taken from two people playing a game, shows a presentational cleft (il y a-cleft in French) build according to the typical [il y a + NP + que/qui + VP] pattern, where the relative pronoun que (for object) or qui (for subject) introduces a relative clause. According to discourse-functional accounts, the il y a cleft serves the pragmatic function of introducing a brand-new referent into discourse (e.g. Lambrecht, 1994), by placing it in post-verbal position in the il y a-clause, and then predicating something about it in the subsequent clause. The construction hence allows the separation into two clauses of the two operations of (i) introducing a new referent and (ii) predicating something about it (see Lambrecht, 1994, on the importance of this pragmatic principle for French). Excerpt (3), again taken from the Sarkozy-Royal presidential debate, shows a so-called pseudocleft or wh-cleft construction, in which the wh-clause (in French typically initiated by ce que, for direct object, or ce qui, for subject) is a free relative clause (cf. Lambrecht, 2001). This part of the construction is followed by the dummy pronoun + copula combination c’est (literally: ‘it is’, or ce sont for plural), which introduces typically a NP, an infinitive clause or a finite nominal clause (ibid.). In the literature, pseudoclefts are considered as a bi-clausal constructions corresponding to the [A c’est B] scheme, where A (i.e. the wh-clause) is a presupposition, presenting an underspecified element, and B (i.e. the post-copular phrase) is presented as specifying A. Pragmatically speaking, the pseudocleft construction is said to mark the post-copular part as focal domain, thereby introducing new or noteworthy information into discourse (see again Lambrecht, 2001, for French). The three syntactic patterns illustrated in (1) to (3), which have been discussed in the literature as three strictly different constructions, both syntactically and pragmatically, share the following two properties: a. they are bi-clausal patterns in which the two clauses are grammatically combined; b. they form a prosodic and informational unit that is clearly bounded to the right by final intonation. Several studies, mainly focusing on talk-in-interaction, have put into question the characteristics quoted under points (a) and (b) (as regards pseudo-clefts, see e.g. Hopper, 2001, for English; Müller, 2006 for French; see also Jullien, 2007, for French il y a-clefts). Some studies have evidenced, both for English and for

Projector constructions in French talk-in-interaction 

German, a tendency toward grammaticization of complement taking predicateconstructions into discourse markers or markers of epistemic stance (see Thompson, 2002; Imo, 2007; see also earlier Auer, 1996 and Mulac & Thompson, 1991). Others have argued that the wh-pieces of pseudoclefts (Hopper, 2004; Hopper & Thompson, 2008; Günthner, 2006; Günthner & Hopper, 2010) sediment as discourse organizational framing devices (the latter has also been suggested for the it-parts in extrapositions, see Couper-Kuhlen & Thompson, 2008; Hopper & Thompson, 2008). What we find speakers do in our data differs quite consistently from the canonical patterns quoted in the literature and meets in several regards the evidence provided by the quoted studies. An initial illustration is provided in (4) to (6): (4)

(FNRS D)

(5)

(E 1186) il y a des trucs pro.3sg.m loc has det.indef.pl things ça nous a servi à rien, pro.3sg.n us aux serve-pp for nothing ‘there are things they did not serve us for anything’

(6)

(J 1623 sq) ce qu’ on- eu::h on: on a fait maintenant pro.3sg.n pro.rel.obj we we we aux do-pp now j’ai une classe que je suis pour la deuxième année, I have a class that I follow for the second year ‘what we did now I have a class that I teach for the second year’

je veux dire euh (...) on a toujours un accent, I want say.inf one has always an accent ‘I mean you always have an accent’

Excerpts (4) to (6) share the following features, which I will analyze in detail in the following section: a. they are not bi-clausal patterns: there is no syntactical link tying together the two parts of the patterns (compared to the canonical forms illustrated in (1) to (3), (4) shows the absence of the subordination marker que, (5) of the relative pronoun que/qui, and (6) of the dummy pronoun + copula combination c’est; in (4), though, on a toujours un accent could be seen as the complement of the verb dire; but see pt. 3.2 infra); b. the first part (in bold) is grammatically and pragmatically incomplete: it is a fragment; this part announces some action that is then accomplished or initiated by the same speaker in the second part: in (4) the first part signals that the

 Simona Pekarek Doehler

speaker will say something; in (5) it signals that there are ‘things’ about which the speaker will say something; in (6) it indicates that there is a thing that ‘they did’ about which the speaker will say something; c. the second part of the quoted configurations, in turn, makes explicit, develops or specifies what is brought up in the first part – or at least starts to do so; d. what appears to be the second part may not be complete: the continuing intonation at the end of each excerpt suggest that there is more to come. We will see that this second part can take many different shapes and extend over a long stretch of talk. Excerpts (4) to (6) exemplify recurrent configurations in our data, consisting of an initial piece (a fragment) and a subsequent, syntactically autonomous stretch of talk of various length. In what follows, I will first (pt. 3) demonstrate that the two parts of these configurations, including the actions which they accomplish, are glued together not by morphosyntactic means of clause-combining, but by projections emanating from the initial fragment onto what follows. I will reconsider the initial je veux dire, il y a NP and ce qui/ce que x fragments in terms of grammatical formats (i.e. instruments for contingently building turns at talk and implementing actions; Thompson & Couper-Kuhlen, 2005: 483; see also Thompson, 2002), that routinely accomplish a specific social action, namely projecting upcoming talk: they are projector constructions. Doing the interactional job of announcing upcoming talk, they may be the emergent products of partially routinized interactional projection practices. In a second part of the analysis (pt. 4), I will zoom in onto a recurrent function of projector constructions, showing that speakers use projector construction as a resource for organizing complex turns.

3. Clause-combining, projection and the sequencing of actions 3.1

Projection

In this section I identify a recurrent compositional schema that consists of an initial projector construction followed by a syntactically unrelated stretch of talk. The parts of this compositional schema are tied together not through morphosyntactic means but through prosody and the local sequential implicativeness emanating from the first part onto the second. The term projector construction centrally builds on the notion of projection as defined in conversation analysis and interactional linguistics. According to Hayashi (2004: 1337),

Projector constructions in French talk-in-interaction 

‘projection refers to the feature of human conduct that prefigures possible trajectories of how an action (or a sequence of actions) might develop in the next moment, and which thereby allows interactants to negotiate and accomplish coordinated action in the subsequent course of interaction.’

Projection rests upon the temporal-sequential character of talk-in-interaction, i.e. its moment-by-moment unfolding across time, articulating ‘what comes first’ and ‘what comes next’, and including prospective effects from the first onto the next. This is true for grammatical projection at any unit-size: a determiner, for instance, projects a noun (or an adjective + noun combination) to come next, an if-clause projects a then-clause to come next (cf. Lerner, 1991), etc. It is also true for action projection in the sense that an action empirically and normatively projects an array of possible next actions, which can be accomplished by the next speaker, or by the same speaker: ‘An action (or action component) may project onto the timing slot immediately following it and make some next activity (component) expectable in this slot’ (Auer 2005: 8). Action-projection is related to ‘the fact that an individual action or part of it fore-shadows another’ (ibid.). In what follows, we will see how action-projection can ‘glue together’ parts of talk, in the absence of grammatical marking of the relation between these parts (for the role of projection in ‘fitting’ together paratactic configurations see Pekarek Doehler et al., 2010). 3.2

‘Je veux dire’

In our data, 60 out of the 65 occurrences of je veux dire (literally: ‘I want to say’) – in this very precise shape with the present tense and first person subject – are not followed by the subordination marker que, as exemplified in excerpt (4) quoted earlier (the remaining 5 are unclear cases, involving cut-offs and unfinished or revised utterances). By contrast, we do find consistent uses of subordination markers with other tenses than the present, as in (7) (7)

(FNRS B) je voulais dire qu’ après je vais pas les interroger. I want-imp say.inf that after I will neg them interrogate ‘I wanted to say that afterward I will not question them’

or in combination with second or third person pronoun, as in (8): (8)

(FNRS, A) ah tu veux dire que je comprendrais pas oh you want say.inf that I understand-cond neg ce que veut dire Hagège. that what wants say.inf Hagège ‘oh you mean that I would not understand what Hagège means’

 Simona Pekarek Doehler

This observation suggests that there is a highly sedimented dimension to the je veux dire construction that differentiates it from other uses of the combination vouloir + dire. The je veux dire construction is in fact known to be grammaticized in French as marker of epistemic stance (a hedged performative; see Jacques, 1981). In this use, it largely speaking corresponds to the English ‘I mean’ rather than to ‘I want to say’ (see below). An illustration is provided in excerpt (9) where a group of teachers talk about plurilingual education: (9) (FNRS D) 1 N de dire que c’est dangereu:x euh:: c’est vrai que: .h to say that it’s dangerous it’s true that 2

ça peu:t euh (..) ((smack)) (.) ça peut susciter that can that can trigger

3

beaucoup de craintes, many fears

4 L

oui oui yes yes

5 N hein:? part 6 L

pourtant je veux dire ces affirmations .h circulent. however I want say.inf these declarations circulate ‘I mean these declarations are in the air’

7 N oui [oui alors euh:: j’ai- j’ai bien yes yes so I ha- I have indeed 8 L 9

[c’est bien la preuve hein? donc elles sont pas that’s the proof part so they are not construites hein? made up part

At line 1, Nadine closes a long argument in which she rejects the view that it is ‘dangerous’ to expose a child to several languages, but concedes that plurilingualism can trigger fears. Lea then introduces an objection (l. 6) which, rather than rejecting the content of Nadine’s turn, appears to reject the idea that one can simply dismiss such views: they need to be dealt with, as they do circulate in public discourse. Lea’s turn (l. 4–6) implements a yes-but turn-architecture that is typical for dispreferred responses (Pomerantz, 1984): her initial oui oui (l. 4) displays alignment with Nadine, pushing disalignment further back into the turn. This moderating effect is reinforced by the subsequent je veux dire, which functions

Projector constructions in French talk-in-interaction 

here as a marker of epistemic stance. The je veux dire not only pushes the disalignment proper (although announced by means of the pourtant ‘however’) again further back into the turn, but also displays Laura’s enunicative stance, presenting ces affirmations circulent ‘these affirmations are in the air’ as her own view rather that an absolute truth – hence its hedging effect. The grammaticization of je veux dire in French as a marker of speakers’ stance or commitment is in line with Thompson’s (2002) discussion of I mean as part of a set of epistemic phrases, including I guess and I think, which introduce an independent clause (see also Okamoto & Laury, this volume). This argument goes back to Thompson & Mulac’s (1991) analysis of English I think and I guess, which undergo grammaticization resulting in the loss of the main clause status, and become epistemic parentheticals (see also Imo, this volume, for German glauben ‘think’ as epistemic qualifier). According to Thompson and Mulac’s analysis, grammatically speaking a complex clause consisting of a main verb and a complement clause has become a single main clause, modified by an epistemic expression. This is also the case for uses of je veux dire as illustrated in (9). However, in a number of significant regards, je veux dire behaves differently than I mean, I think or I guess. For one thing, its placement is highly restricted. In our data, none of the occurrences (n = 60) of je veux dire are placed in medial position within a clause or in post-clause position. Rather, je veux dire systematically precedes a clause or any other stretch of discourse to which it is related. This is significant insofar as I mean, I guess, I think used as epistemic markers can be found in initial, medial and final position (e.g. Imo, 2005). Je veux dire seems in this regard to behave more like a discourse marker, as discourse markers usually do not occur in medial position (see e.g. Auer & Günthner, 2003, on ich meine ‘I mean’ in German). In most cases in our data je veux dire functions as a projector construction that serves to organize turns and actions. This is shown in excerpts (10) to (12): (10) (I, 257) [they are talking about two contrasting definitions of bilingualism] 1 L MAI:S moi je crois qu’ en fait s- ces deux but I think that in fact these two 2 définition:s euh > °je crois°< qu’il faut pas definitions I think that one should not 3 les:: monter l’une [contre& confront them one against 4 F

[non ? no

 Simona Pekarek Doehler

5 L &[l’autre,& the other 6 N

[oui ils se rejoignent, yes they converge

7 Q

[mhm

8 L &[.h je veux dire c’est I mean it’s 9 deux sortes de bilinguismes différents? two different kinds of bilingualism 10 .h ceci dit ((turn continued)) this said (11) (I, 220) [they are talking about how to define a bilingual person] 1 N j- je suis aussi d’accord avec monsieur, I- I also agree with the gentleman 2

parce que: .h c’est clair que- (.) because it’s clear that

3

on peut avoir une aisance dans les deux lan:gues, you can be at ease in the two languages

4

= °je veux dire° eu::h (.) on peut les parler I mean you can speak them

5

régulièrement et puis être dans ce sens bilingue. regularly and then be in that sense bilingual

6

.h mais (.) il y aura toujours (..) une langue:=eh but there will always be a language

7 (.) où on arrive pas (.) à la parler that you won’t manage to speak it 8 eu:h (.) peut-être en aisance mais sans fautes, maybe easily but without mistakes 9

on arrive pas °à la parler sans fautes°. you won’t manage to speak it without mistakes

mhm

10 Q

11 N °à mon avis.° in my opinion

Projector constructions in French talk-in-interaction 

(12) (B, 405) [they are talking about conversational code-switching] 1 N alors parfois en p- (.) plein milieu d’une phrase so sometimes in the middle of a sentence 2

il change de langue. he changes the language

(1.0)

3

4

encore maintenant. still now

5 Q mhm

6 N [(à peu près.) (...) pour moi c’est incroyable. more or less to me that’s unbelievable

7 ?

[x

8 G mhm

9 P

ouais c’est que c’est c-^c’est basé sur des yeah it’s that it’s it’s based on det

10

connotations, connotations

11

=je veux dire euh: (1.4) des endroits de la maison I mean det places in the house

12

la waschküche °on disait the +laundry-room ((in German)) we used to say

13

la waschküche pour la buanderie° the +laundry-room ((in German)) for laundry-room

14

par exemple, for instance

15

(.) ben c’était pas seulement n’importe quelle well it was not just whatever

16

buanderie, c’était celle du bas. laundry-room it was the one from below

17 Q ouais yeah These three excerpts are taken from two interactions about issues of bilingualism and code-switching. In (10), Leo, who has been presented with two contrasting definitions of bilingualism, first states that it is useless to confront one definition

 Simona Pekarek Doehler

against the other (l. 1–5, note that this statement is hedged by je crois ‘I believe’). After a short in-breath, Leo introduces by means of je veux dire (l. 8) an account for his statement: the definitions under discussion describe two different types of bilingualism – and this is why you cannot simply compare them. The providing of an account can be heard as an action that is related to (or even: subjacent with regard to) the preceding affirmation. Within a similar retrospective orientation, the je veux dire in (11) introduces a specification of the speaker’s preceding point, explaining what he means by ‘to feel at ease in the two languages’. The je veux dire is pronounced in noticeably lower voice, and in quick follow-up with what precedes. Interestingly, subsequently to the je veux dire-initiated piece, the speaker returns to the preceding stretch of talk and action and displays that return by means of mais ‘but’ (l. 6). Thereby the two parts surrounding the [je veux dire + clause] segment are presented as both syntactically and pragmatically continuous: ‘you can be at ease with the two languages’ ... ‘but there will always be one language in which you will not manage to talk’. Within this organizational structure, the je veux dire-introduced segment is presented as a subjacent action, inserted into the course of a main action, and clearly delimited both at its start (by means of je veux dire, l. 4) and at its end (by means of pitch drop, l. 5, see also the in-breath and the use of a discourse marker at the point of return to the main action, l. 6). In (12), finally, Paul takes the turn (l. 9) displaying his ‘on topic’ talk by means of an initial ouais and a discourse anaphoric c’est (‘this is’), and he presents his own view on the issue under discussion: code-switching triggers connotations. He then uses je veux dire (l. 11) to introduce an exemplification for that statement. Although it initiates a new TCU, it is prosodically closely tied to the preceding talk (level intonation at the end of the preceding TCU and, again ‘rush through’). By contrast to excerpts (10) and (11), where the je veux dire is followed by one single clause, ending on final pitch drop, the second piece in excerpt (12) appears to be a multi-unit stretch of talk (l. 11–16). The three excerpts show several characteristics that are recurrent in our data: – The je veux dire, rather than marking a speaker’s stance, ties together two actions in a specific way: it projects a second action, namely that the speaker will say something, and announces this second action as retrospectively related to the preceding action. It does not specify what action is to follow, but rather creates a slot for a next action, and an expectation that that action will be related to the preceding action. The continuous intonation on the immediately preceding unit of talk (ex. 10, l. 5; ex. 11, l. 3; ex. 12, l. 8), sometimes combined with fast follow-up on that talk (ex. 11 and 12), indexically displays this strong retrospective orientation.

Projector constructions in French talk-in-interaction 

– Due to its projection property, the je veux dire can efficiently be used as a floorholding device. This is most clearly shown in excerpt (12) (but see also 11 and pt. 4.1 infra), where the je veux dire is followed by a lengthy 1.4 second pause, but no co-participant attempts to take the floor. – What follows the je veux dire is pronounced in prosodic continuity with it (level pitch, no new onset); it is projected as part of the same TCU. Based on the preceding analysis, we can sum up the properties of je veux dire as follows: – By contrast to what is considered as canonical uses of je veux dire + subordination marker + clause, the je veux dire initiated configurations documented in this section are not bi-clausal patterns; they consist of a combination of an initial projector fragment (the je veux dire) plus a stretch of talk: the two pieces are related praxeologically and prosodically rather than through morphosyntactic marking. – Je veux dire appears to have sedimented as a highly formulaic sequence, occurring in the present form, with 1st person singular subject and exclusively in the affirmative,3 very much like what Thompson (2002: 137) calls ‘re-usable fragment-schemas’ for English I think or I guess. This homogeneity testifies to a tendency observed in several recent studies of conversational data according to which some matrix clause plus subordinate clause patterns move toward discourse-marker-like element plus following utterance (e.g. Auer & Günthner, 2003; Imo, 2007; Thompson, 2002; Thompson & Mulac 1991). Accordingly, the clause following je veux dire cannot be understood as a complement of the verb. While in some cases it is used as epistemic marker, je veux dire occurs in our data mostly as a projector construction: it is a discourse-constructional element, bearing the traits of a discourse marker, which projects a slot for further talk, and forewhadows that talk as providing some elaboration on the immediately preceding talk. Due to its projection properties, it can also serve as a floor holding device. 3.3

‘Il y a+NP’ (‘there is+NP’)

Despite of the pragmatic and morphosyntactic differences as regards their ‘canonical’ counterparts, parallel observations can be made for configurations initiated by il y a ‘there is’ that resemble presentational clefts. Excerpt (13) again turns

3. In our data, we do not find any consistency across our collection of 60 cases as to whether the je veux dire is pronounced in low voice as in (11) (and sometimes rapidly) or distinctly as in (10).

 Simona Pekarek Doehler

around issues of bilingualism. One participant has just brought up a definition of the bilingual person as someone who speaks regularly two languages: (13) (FNRS A 874 sq) 1 M mais ça voudrait dire par exemple que tous les but this would mean for instance that all the 2

enseignants en italien sont bilin:gues. teachers in Italian are bilinguals

(..)

3

4 Q ouais? yeah

5

(.)

6 Q [voilà? that’s it 7 M [euh (.) il y a il y a des cas there are there are det cases 8

je mets des limites, I put det limits

mhm

9 R

10 M ((laughing))
[où où j’arrive pas& where I can’t

12 R

[non (.) là oui oui là ou je suis d’accord no there yes yes there yes I agree

13 M &à oublier que la personne elle est francophone forget that the person (she) is francophone 14 [(.) eh quand je& when I 15 Q? [ah d’accord ah okay 16 M &l’entends [au fond de la classe>& hear her in the back of the classroom 17 R

[mais (.) par exemple là- (.) oui but for example there yes

Projector constructions in French talk-in-interaction 

18 M &mais à part ça ça me plait comme définition. but except for this I like it as a definition 19 B

oui moi aussi yes me too

Excerpt (14) shows part of a discussion about the linguistic life of the perfect bilingual speaker: (14) (FNRS I) 1 N on finit par penser en français? you end up thinking in French 2

on finit par eu:h (..) par parler? you end up speaking

3

par être à l’aise? being at ease

4

mais i- il y a des choses on est marqués, but th- there are det things one is marked ‘but there are things that mark you’

5

par exemple: un numéro de téléphone euh je pourrais for instance a telephone number I could

6

jamais (.) l’avoir en tête (.) en français. never keep it in mind in French

(.)

7

8 N s’il y a un num- un numéro de téléphone if there is a a telephone number 9

vous me le dites, you tell it to me

10 .h j’ai besoin de me le rappeler:, I need to recall it 11

ben (0.5) vous me le dites en français, well you tell it to me in French

12

il entre i- il sort. it goes in it goes out

13 .h vous me le dites en allemand, il reste. you tell it to me in German it stays

14

(2.0)

 Simona Pekarek Doehler

15 Q

[est-ce que ((question word))

16 N [vous voyez c’est c’est [c’est des& you see it’s det 17 F

[ouais yeah

18 N &petites différences qui sont là qui restent. little differences that are there that remain The segments highlighted in bold in excerpts (13) and (14) appear at first sight as a mere juxtaposition of two syntactically autonomous clauses, just as in excerpt (5) quoted earlier. According to the standard interpretation of presentational clefts in French, presentational clefts involve a morphosyntactic marking of the link between the two parts. Configurations of il y a presentational constructions where the two parts are not related morphosyntactically remain largely unmentioned in the literature, but have been documented in a few studies of spontaneous speech by adult speakers and by children (Berrendonner, 2003; BlancheBenveniste et al., 1991; Jullien, 2005; Savelli et al., 2000). Under the absence of a relative pronoun, the two parts of the configurations shown in (13) and (14) are tied together by means of projections and prosody (showing no prosodic break and typically level pitch between the two parts). However, the configurations do not consist of a juxtaposition of two clauses either. The nature of the NP (cf. infra) blocks an interpretation of the il y a NP segment as an independent clause, e.g. as an existential expression (cf. Lambrecht, 1994) such as il y a aujourd’hui en France 200 accidents du travail par jour (sarroy) ‘there are currently in France 2000 accidents a day’. Also, the verb avoir ‘to have’ functions here as what Blanche-Benveniste (1989) calls an «auxiliaire de dispositif», a de-semanticized construction auxiliary. The il y a NP-part hence clearly appears as a syntactically incomplete fragment. What is common to the quoted excerpts, and recurrent in our data, is the projection emanating from the il y a NP construction. The il y a NP-part in each of the experts projects that something will be said about the ‘cases’ or the ‘things’ that are brought up in the NP; the semantic vagueness of these NPs strongly contributes to this effect (see also excerpt 5 quoted earlier). This is a significant property of the il y a NP projector constructions, as opposed to general characteristics of presentational clefts – such as il y a une sorcière qui a dessiné un chat (Lae1) ‘there is a witch who drew a cat’ – where the referential expression is said to be specific, and often human (cf. Lambrecht, 1994, for French). The NP in the il y a NP projector construction consist of semantically light elements such as ‘things’, ‘cases’, sometimes

Projector constructions in French talk-in-interaction 

called “pseudo-categorizers”, or else of generalizing elements such as ‘questions’ or ‘problems’. All of these typically occur in the plural with an indefinite determiner. This very vagueness contributes to the announcing effect of the il y a NP piece; the second part in turn specifies the ‘cases’, ‘things’, etc. that are brought up in the first part – which brings these il y a NP + clause configurations interestingly close to the pragmatics of pseudoclefts (cf. 3.4 infra). It is noteworthy that excerpts (13) and (14) show two embedded projections that can be schematized as [il y a NP → specification of NP → elaboration]: the il y a NP projects a specification of the NP, and the il y a NP + clause itself projects further elaboration on the NP so specified. This is corroborated by the level pitch at the end of the il y a NP + clause configuration, indicating a continuation of the ongoing turn. For instance in (13), the il y a NP ‘there are cases’ is first specified as ‘where I put limits’, which qualifies ‘the cases’ as something that the speaker does not accept; this is then followed by an exemplification (l. 10–16) of the ‘cases where I put limits’ consisting of a multi-unit stretch of talk. It is noteworthy that this exemplification, introduced by a further il y a NP, is demarcated from the surrounding talk by a slowing down of pace. This slowing down, together with the introductory il y a NP + clause combination, highlights the exemplification as particularly noteworthy. Finally, its end is clearly marked by a return to normal pace (l. 16/17), and to the more general conversational topic of what a bilingual person is: the conclusive marker mais à part ça opens up a generalized final comment, ça me plait comme definition ‘I like that definition’ (see Pekarek Doehler, De Stefani & Horlacher, forthcoming, for the use of left-dislocated assessments as closing initiators). Excerpt (14) shows a similar case where the il y a NP + clause, combined with the preceding discourse marker mais, is followed by a lengthy exemplification of ‘the things’. The exemplification is first framed by the statement on est marqués ‘one [i.e. people] is marked by’ (l. 4), which qualifies ‘the things’ as formatting peoples’ identities. The exemplification is then explicitly introduced by par exemple ‘for instance’, (l. 5), followed by a multi-unit stretch of talk consisting of multiple embedded paratactic configurations (for a detailed analysis of these configurations see Pekarek Doehler et al., 2010). While on a prosodic and syntactic basis, it can be argued that this stretch of talk ends with the final intonation on français at line 6, the exemplification itself is structured as a series of TCUs and comes to an end with the final intonation at line 13, which is then followed by a concluding remark in lines 16–18. Both excerpts show two embedded projections emanating from the initial il y a NP + clause configuration. The il y a NP-part announces something to be said about the ‘cases’ or the ‘things’ that it refers to. The subsequent clause specifies these cases/things (e.g. as a personality-shaping issue, or an unacceptable thing),

 Simona Pekarek Doehler

and frames the following stretch of talk as providing further elaborations on these cases/things. While the preceding excerpts showed il y a NP + clause configurations that very much resemble presentational clefts, other instances of il y a NP functioning as projector constructions appear to be harder to relate to presentational clefts because they are followed by an extended stretch of talk, comprising multiple clauses and TCUs. An example is provided in (15), where teachers talk about how to teach languages so as to favor students becoming communicatively efficient: (15) (FNRS B) 1 B mhm. (.) non mais c’est terRIble ça. no but that’s terrible

2

(..)

3 B alors il y a y a un problème là, so there is a problem there 4

moi je pense que: (.) il faut pas faire les fautes, I think that one should not make det mistakes

5

on corrige tellement les fautes, we so much correct det mistakes

6

on regarde tellement les fautes, we so much look at det mistakes

7

il faut parler correctement. you need to speak correctly

8

résultat, (..) ils sont conscients que des fautes. result they are aware only of their mistakes

(..)

9

10 B pis ils oublient que: la première chose and they forget that the foremost thing 11

c’est s’exprimer. is to express yourself

12 M mhm (..) mais c’est intéressant ((turn continues)) but that’s interesting Il y a un problème là ‘there is a problem there’ projects a slot for upcoming talk and annonces that talk as being about a problem, i.e. as specifying what the speakers presents as a problem. Lines 4 to 11 can be heard as providing that specification (including a turn-extension at lines 10–11), and appear to be oriented to as such by Marie, the inteviewer, who assesses Bruno’s presentation of the problem as

Projector constructions in French talk-in-interaction 

interesting (l. 12). We will come back to the role of il y a NP construtions in the organization of multi-unit turns in Section 4.3. In sum, then, the quoted examples document a constructional schema that shows many parallels with what we have observed for the je veux dire patterns – but also differences: it consists of an initial projector fragment followed by a syntactically independent stretch of talk, which is prosodically and praxeologically tied to the first part. Based on the syntactic and lexical consistency of the first part [il y a + indef. semantically ‘light’ NP], the initial fragment appears as highly homogenous, while the second part is much more varied. This second part can consist of a simple clause (as in ex. 13 and 14) or of a longer stretch or talk (as in ex. 15). The consistency of the il y a NP-part confers a formulaic aspect upon it, here again reminiscent of what we have observed for the even more homogenous je veux dire construction. Its syntactico-lexical consistency and its semantic vagueness clearly hint at its generalized use as a discourse organizational device: it functions as a projector construction foreshadowing further talk by the same speaker. While the je veux dire construction, due to its retrospective orientation, prolongs some previous action by announcing the subsequent talk as providing some further elaboration on immediately preceding talk (e.g. an account, an exemplification), but without further specifying what that talk will be about nor what action it accomplishes, the il y a NP fragment is more exclusively prospectively oriented. This is highlighted by the fact that it typically occurs at the start either of a new turn or of a new TCU following a complex transition relevance place (cf. Ford & Thompson, 1996) and is marked by new prosodic onset. Also, the il y a NP does not merely create a slot for upcoming talk, but it frames that talk as being about a ‘problem’, a ‘case’, a ‘thing’ brought up in the il y a NP construction itself. 3.4

Pseudocleft-like constructions

A number of parallel observations have recently been made on behalf of pseudocleft-like configurations (see Hopper 2001, 2004 and Hopper & Thompson, 2008, for English; Günthner, 2006, for German; Müller, 2006 and 2007 for French). As mentioned earlier (pt. 2), the standard French pseudo-cleft construction consists of a ce que- (for direct object) or ce qui- (for subject) initiated free relative clause (corresponding to the English wh-clause), followed by the dummy pronoun + copula combination c’est (literally: ‘it is’), which typically introduces a NP, an infinitive clause or a finite nominal clause. In our French data, however, we regularly find configurations that differ from this standard pattern, as illustrated in (16) and (17):

 Simona Pekarek Doehler

(16)

(J 1623 sq) ce qu’on- eu::h on: on a fait what we we we aux do-pp ‘what we did’

maintenant j’ai une classe que je suis now I have a class that I teach pour la deuxième année, for the second year (17) (FNRS D) ce que je voulais dire what I wanted to say c’est vrai qu’ il y a quand-même tout ce phénomène, it’s true that there is nevertheless all this phenomenon Excerpts (16) and (17) show configurations of two clause-like elements that are prosodically tied together (level pitch; no prosodic break between first and second part) but are not connected by any morphosyntactic means. Such configurations which resemble pseudoclefts have rarely been documented in the literature for French (but see Valli: 1981: 207), but have recently been accounted for as regards English and German (Günthner, 2006; Hopper & Günthner, 2010; Hopper & Thompson, 2008). These configurations bear some pragmatic and syntactic traits of classical pseudoclefts as discussed in the literature (cf. pt. 2 supra). For instance, the first part can be heard as presenting a presupposition including an underspecified referential element. By contrast, while in classical wh-clefts a specification is proffered in what follows the wh-part, in cases such as (16) and (17) it seems hard to argue that the second part specifies ‘what we did’ or ‘what I wanted to say’. Nevertheless, the first clause syntactically (and lexically) corresponds to the typical wh-part of pseudocleft constructions. Hopper & Thompson (2008) note that most of the wh-clauses in pseudoclefts in their data are lexically constrained and formulaic. A small number of verbs predominate in the wh-clause. Hopper (2004) has previously documented that the most frequent verbs in this environment in English are ‘do’, ‘happen’, ‘say’. In her French data, Blanche-Benveniste (2010) finds predominantly verbi dicendi, facendi and pensiendi. This is also what we find in the types of configurations quoted in (16) and (17). This very syntactic and lexical consistency of the initial clause confers a formulaic aspect to the first part, suggesting that this part (as opposed to the second part) is used as highly recurrent routinized format, very much like what we have observed for the two previously discussed constructions. It is a fragment, rather than a full clause, as it is visibly both syntactically and semantically incomplete.

Projector constructions in French talk-in-interaction 

Its function is to project further talk by the same speaker, and to foreshadow some particular action to be accomplished by that talk, namely a specification of what has been announced in the wh-part: in example (16), ‘what we did’ creates an expectation with the co-participants that the current speaker will specify what he did; in (17) ‘what I want to say’ creates an expectation that the speaker will specify what he wants to say. This very property is reminiscent of what has been pinpointed recently in Hopper & Thompson’s (2008), Günthner (2006) and Günthner & Hopper’s (2010) discussion of pseudoclefts in English and in German respectively (cf. pt. 4.3. infra). Hopper and Thompson (2008) note: “what is striking is that the wh-clause of the pseudocleft anticipates (or “projects”) up-coming talk by the same speaker, and, as we have seen, FRAMES that talk in terms of such categories as event, action and paraphrase” (p. 105, emphasis in the original).

In the cases quoted here, the projected talk is framed as exposing a past endeavor ‘what we did’ (16) and as presenting something the speaker ‘wants to say’ (17). The framing device hence functions as a means for action projection: it foreshadows what the speaker is going to do next. 3.5

Summary and initial discussion of the findings

Three different patterns that are documented in the literature as bi-clausal constructions regularly appear in our data as juxtapositions of two syntactically autonomous pieces, an initial projector construction plus a clause or multi-unit stretch of talk; the two parts are not tied together morphosyntactically, yet they are linked prosodically. This further corroborates the fact that grammar in talk-in-interaction is often different from what classical interpretations of grammar have documented or claimed (see the seminal statement by Ochs, Schegloff & Thompson, 1996). It also shows the crucial role of prosody in locally tying together stretches of talk where syntax does not (Selting, 1996). In the configurations under analysis, the sequentially first piece implements relatively fixed syntactic and lexical patterns, often with low prosodic pro-eminence. This point meets Hopper & Thompson’s (2008) recent observations on English and German pseudo-cleft and extraposition constructions: ‘in certain of the allegedly biclausal constructions even the first part is not really a clause but is instead a fixed sequence with limited lexical choices.’ (p. 99). While such consistency of the initial part has so far been documented in particular for pseudoclefts, we have seen that this is most strikingly the case for French je veux dire, which occurs as a projector construction only in this very combination of first person singular pronoun plus present tense. Highly limited lexical choices are also found

 Simona Pekarek Doehler

with the il y a construction comprising lexical NPs that are referentially vague, such as ‘things’, ‘issues’, ‘problems’, etc. These properties clearly show that the projector constructions discussed here bear typical traits of grammaticalization: high lexical and structural consistency, reduction of their semanticism, and sometimes low prosodic pro-eminence (cf. Traugott & Hopper, 2003). This suggests that the first parts of the configurations under analysis (i.e. the projector constructions) are produced as routinized, strongly sedimented formats while they open a space for much less constrained follow-ups. These initial parts are designed to accomplish a specific action: they announce some talk to come, thereby attracting co-participants attention to that upcoming talk and foreshadowing both turn-shape and turn-type (cf. Schegloff, 1996):4 – Je veux dire is least specific as to what is to follow. It does not specify what action is coming up, but rather creates a slot for a next action, and an expectation that that action will be tightly related to the preceding action; it frames subsequent talk as providing some kind of elaboration on the immediately preceding talk by the same speaker (e.g. an exemplification of a preceding statement, an account for a preceding affirmation). This strong retrospective orientation is reflected in the fact that the preceding talk ends on continuing intonation. – Il y a NP, although topically related to what precedes, is more straightforwardly prospective, announcing something new to come up. It typically occurs as a new TCU after a TRP marked by final pitch drop or rise, and starts with new prosodic onset. The NP in the il y a NP construction is referentially vague. Il y a NP is more specific as to what the talk coming up is about: it frames that talk as being about a ‘problem’, an ‘issue’, or a ‘thing’ brought by means of the NP in the il y a NP construction itself. That is, the lexico-semantic nature of the NP enhances the nature of the projection. – The ce que/ce qui-construction, just like the il y a NP construction, is strongly prospectively oriented, introducing a new action after a prosodically marked TRP. It tends to comprise a limited set of verbs (verbi dicendi, faciendi, pensiendi) or VPs (‘to be’ + qualifier, e.g. ‘interesting’). And it projects further talk as specifying something that is presented in the ce que/ce qui-piece: it frames that talk as being about ‘what someone said’, ‘what is interesting’, etc.). Here again, the lexico-semantic nature of the ce que/ce qui-construction enhances the projection. 4. Projector constructions and what they accomplish in talk are different from prefatory activities (such as pres or pre-pres), as the latter typically call for recipient’s acknowledgement or go-ahead signal (cf. Schegloff, 1980). Projector constructions also differ from turn-entry devices such as ‘and’ or ‘but’ or simple floor holders like ‘ehm’ in that they specify some aspect of the talk to come.

Projector constructions in French talk-in-interaction 

Based on these characteristics, all three constructions can be interpreted as projector constructions that serve as framing devices for further talk. The findings hence meet ongoing discussions on wh-pieces (Günthner, 2006; Günthner & Hopper, 2010; Hopper, 2001; 2004; Hopper & Thompson, 2010), extrapositions (CouperKuhlen & Thompson, 2008), ex-complement-taking predicate phrases such as German ich meine ‘I mean’ (Auer & Günthner, 2003; Günthner & Imo, 2004; Imo, 2007; see also Thompson, 2002) or phrases such as ‘the thing is’ (Günthner, this volume). Günthner (this volume) notes: ‘These constructions are connected with a taxonomic network of related constructions with structural and functional overlaps’ (p. 18). We have seen that the French je veux dire, il y a NP and ce que/ce quiconstructions are part of such a taxonomic network of projector constructions. But if this is the case, what is the scope of the projection? 4. Projector constructions and the construction of complex turns We find regularly in our data cases where the second part following the projector construction is long, syntactically complex, yet is treated by the participants as being related to the projector construction itself. On these grounds, I argue in this section that projector constructions are discourse-organizational devices that project a potentially open-ended slot for further talk; as such, they are one of the resources participants use for organizing complex turns. In tackling this issue, we need to look at how grammar is contextualized within the sequential organization of larger courses of practical activities. Let’s again start with an example of the je veux dire type. 4.1. ‘Je veux dire’ projector constructions in complex turns Excerpt (18) is taken from the discussion about definitions of bilingualism quoted before. Emile responds to a question by Jean as to what his opinion is about two contrasting definitions of bilingualism (bilingualism in terms of regular use vs. perfect mastery of two languages). (18) (FNRS I) 1 E [pou-] pour moi le bilingu for for me det bilingu 2

enfin moi-même je je considère pas euh well I myself I I don’t consider

3

je me considère pas en tant que bilingue? I don’t consider myself as a bilingual

 Simona Pekarek Doehler

4

parce que pour moi le bilinguisme c’était because for me det bilingualism was

5

justement comme monsieur bah just like the genleman bah

6

dès la naissance on parle euh régulièrement deux from det birth on you speak regularly two

7

langues ou plusieurs langues, languages or several languages

8 .h mai:s une fois si on ap but once if you 9

si on apprend une langue une fois plus tard, if you learn a language at a later time

10 .h on arrive jamais à la maîtriser, you will never master it 11

je veux dire euh (..) on a toujours un accent, I mean you have always an accent

12

toujours des fautes, on a you have always det mistakes

13

on a touj- (m-) (.) on a toujours quelque chose, you have alw- you have always some thing

14

=°je veux dire° on est toujours on est toujours I mean you are always you are always

15

[suisse allemand& Swiss German

16 J

[oui yes

17 E &°pour finir°. ((small laughter)) in the end 18 J

[ah oui? oh yes?

19 E [même si même si on arrive à bien parler, even if even if you manage to speak well 20

même si on arrive à: à:: .h à s’[exprimer,& even if you manage to to to express yourself

21 J

[mhm

Projector constructions in French talk-in-interaction 

22 E & donc pour moi ((lowers voice)) so for me 23

un bilinguisme c’est: euh (.) a bilingualism is

24

parler dès la naissance deux langues quoi. speaking from birth on two languages part

25 J

donc euh vous seriez plutôt eu:h so you would rather be

26

dans la définitio:n (.) une. in the definition number one

27 E °ouais si vous voulez ouais° yeah kind of yeah As part of his response to Jean’s question, Emile defines bilingualism as the use of two languages since early childhood (l. 1–7). He then adds that when you learn a language later, you will never be able to fully master it (l. 8–10). Emile’s subsequent je veux dire (l. 11) introduces an elaboration on his answer, by means of which he specifies what he meant by ‘if you learn a language later you will never master it’ (l. 9). In this sequential environment, the je veux dire projects an upcoming elaboration on the immediately preceding talk. The hesitation and the micro-pause following je veux dire, combined with the absence of recipient reaction at this point, suggest that the je veux dire also has a floor-holding effect, giving Emile time to formulate what is coming up without being interrupted by a co-participant. This analysis confirms what we have said earlier about the use of je veux dire as a projector construction with a strong retrospective-prospective orientation: je veux dire projects a subsequent segment of talk in some way as subjacent with regard to what precedes (cf. pt. 3.2 supra). This is also corroborated, in excerpt (18), by the fact that once Emile has specified what he meant, he returns to the ‘main’ action of providing an answer by means of the discourse marker donc ‘so’ plus a summary statement (l. 22), which in turn initiates a sequence closing (for the role of summary assessments in closing sequences see e.g. Drew & Holt, 1998). The summary statement is followed here by a clarification request by Jean (l. 25–26), and Emile’s response to that request (l. 27), after which a new sequence will be opened. Now, the excerpt allows us to carry the analysis one step further. The [je veux dire + elaboration] segment shows the following sequential architecture, which is recurrent in our data with this and other types of projector constructions:

 Simona Pekarek Doehler

I. A projector construction (l. 11). II. The projector construction is followed (l. 11–24) by a lengthy stretch of talk consisting of multiple TCUs. This stretch of talk is not syntactically linked to the projector construction, but its start is prosodically continuous with it. In the present case, this subsequent segment of talk is made of a complex sequence of – coordinated clauses (a juxtaposition of clauses within a three-part list: on a toujours un accent, on a toujours des fautes, on a toujou- (m-) (.) on a toujours quelque chose, followed by a second je veux dire which introduces a conclusion of the preceding list: on est toujours on est toujours (..) suisse allemande °pour finir°); – subordinated clauses (même si on arrive à bien parler, même si on arrive à s’exprimer,), which incrementally extend the turn after a first complex TRP at line 17. III. The placement of the recipient’s actions is clearly oriented to the structure of the multi-unit stretch of talk following the projector construction: recipient tokens (l. 16, 18, 21) occur after recognizable points of possible completion of the projected talk: the first of these (l. 16) is placed after the completion of the three-item list (the third element, on a toujours quelque chose ‘you always have something’, can be understood as a generalized list completor, cf. Jefferson, 1972), but comes in quite late. The second recipient reaction (l. 18) is placed at the very end what Emile presents as the gist of the preceding list. The third recipient token (l. 21) occurs in slight anticipation of a further possible point of completion of the projected action, in overlap with the end of the two incrementally added même si-clauses. The placement of these tokens shows that Jean closely monitors the architecture of Emile’s turn in a way that is oriented to the initial je veux dire projecting an elaboration of the immediately preceding talk, and the subsequent je veux dire as projecting a further elaboration embedded in that first elaboration. What we see Emile do is a way of constructing a complex turn (or at least part of it) in a recognizable manner. The projector construction serves as a resource not only for projecting a slot of talk, thereby holding the floor, and creating the expectation that that talk will in some way elaborate on the preceding talk; it also allows (together with the structure of the subsequent talk) for the recognizability, by the co-participants, of possible points where the projected elaboration can be heard as completed. The placement of recipient reactions that we have identified reveals the mutual orientation of the participants to this projection and to the unfolding architecture of the turn.

Projector constructions in French talk-in-interaction 

4.2

‘Il y a NP’ projector constructions in complex turns

Similar patternings can be found with il y a NP-initiated stretches of talk. The following excerpt is taken from a discussion among teachers about the (dis)advantages of correcting errors in second language classrooms. (19) (FNRS J 581) 1 B pendant tout notre système scolaire on a été during all our schooling system we have been 2

taxé à cause des fautes qu’on a fait, sanctioned because of the errors we made

3 L mhm

4 B .h et puis >à un certain moment< and then at a given moment 5

pour ne pas faire de fautes on ne parle pas. in order not to make errors you don’t speak

(1.2)

6

7 B xxx (.) °alors°, (0.6) là il me semble so there it seems to me 8

°il y a: (...) there is

il y a quelque chose d’important there is something important

9

à:: (.) à voir°, to to see

10

=est-ce que (..) la langue est faite ((question word)) the language is made ‘is the language made’

11

pour s’Exprimer, (.) °avec ou sans fautes°, for expressing yourself with or without mistakes

12 (0.8) ou bien est-ce que la la- la langue or does the the- the language 13

doit être correcte pour qu’on puisse need to be correct in order for you to be able to

14

qu’on puisse l’utiliser? For you to be able to use it

(.)

15

16 B ((pitch rise)) alors il me semble que dans ces so it seems to me that in these

 Simona Pekarek Doehler

17

nuances qu’on fait ((clears his throat)) nuances you make

18

il y a ce problème derrière. there is this problem behind

19 B .h oui je sais demander du pain yes I know how to ask for bread 20

mais je fais des fautes. but I make det mistakes

21 S? [mhm

22 L [mhm

23 B [alors je suis pas- (.) je suis [pas bilingue. so I am not- I am not bilingual

24 L

25 S? moi xxxxx

[xxxxxx juste

Bruno provides first a strong view on a central point of discussion (‘during the whole school years you are sanctioned for errors and finally in order to not make errors you don’t speak’, l. 1–5). He ends this statement with falling pitch, indicating a TRP. In the face of the noticeable absence of recipient reaction (see the 1.2s pause at l. 6), Bruno then announces something more to come about that point. This is done through a combination of means, including the discourse marker alors ‘so’, the là il me semble ‘there it seems to me’ (the discourse deictic là ‘there’ makes explicit a retrospective orientation of what is coming up), and the il y a-construction, produced with markedly softer voice: ... il y a quelque chose d’important à:: (.) à voir ‘there is something important to see’ (l. 8/9). The lexico-semantic nature of the complex NP in the il y a NP construction strongly contributes to the projection effect, announcing ‘something important’ to come up. While resembling the syntactic shape of the first part of a presentative construction along the standard scheme [il y a NP (qui)], the initial fragment here again bears the semantico-pragmatic traits of the initial clause in a pseudo-cleft [A c’est B], where A is presented as underspecified, and B as specifying A (see Valli, 1981, and Müller, 2006 for the resemblance between the two constructions in French). In other words, what follows is projected as specifying ‘something interesting to see’. The subsequent stretch of talk is closely tied to the projector construction, both by means of intonation (level pitch on voir, and no new pitch onset on est-ce que) and by means of phonetic latching (the est-ce que is latched to the final /r/ of voir). Again, this stretch of talk is patterned in a complex way. First a question based on two alternatives is exposed (l. 10–14), which appears to be both designed

Projector constructions in French talk-in-interaction 

and interpreted as a rhetorical question rather than a first pair part soliciting coparticipant reaction. The absence of recipient reactions at this point suggests that the talk so far, despite the TCU final (rising) intonation, is not oriented to as providing what was been projected, namely ‘something important to see’. Rather, the projected action appears to be treated as still under way. Bruno’s question is then followed by a repetition of alors il me semble ‘so it seems to me’ (l. 16), marked by new prosodic onset, which has the effect of re-launching the action projected by the initial projector fragment and indeed leads into a first recognizable completion point of that action: Bruno’s important point is recognizably made at the moment where he says that the whole thing boils down to the problem of one being able to ask for bread, even though making mistakes (l. 19/20). Here, first recipient reactions occur (l. 21/22), which are in turn followed by Bruno’s conclusive statement, introduced by alors ‘so’: alors je suis pas- (.) j’suis pas bilingue ... ‘so I am not- I am not bilingual’ (l. 23). Excerpt (19) in sum shows the following sequential pattern, which is remarkably parallel to what we have observed for excerpt (18): I. A projector construction (l. 8/9). II. A complex multi-unit stretch of talk which is not morphosyntactically tied to the projector construction, but is prosodically continuous with it (l. 9/10); this subsequent complex stretch consists of coordinated and subordinated clauses that can be read (and appear to be read by the participants) as accomplishing the projected action. III. This second segment’s first recognizable point of completion is oriented to by participants: recipient tokens occur exactly at this point, and not before. Visibly, what ties this stretch of talk together, here again, is not any morphosyntactic marking of the linkage between its clausal constituents. Rather, it is the action projection emanating form the initial projector construction and the participants’ joint orientation to it. The scope of the projector construction reaches far beyond the subsequent clause, encompassing a lengthy multi-unit stretch of talk. This scope, of course, is not defined a priori, with the mere occurrence of the projector construction, but is configured on a moment-to-moment basis, as talk unfolds across time. 4.3

‘ce qui/ce que x’ (wh-clause) projector constructions in complex turns

Quite in line with the preceding analysis of je veux dire and il y a NP, the data shows ce que/ce qui x (i.e. English wh-clause) initiated configurations where the second part does not consist of a single clause, but is shaped as a multi-unit stretch of talk. This has been documented earlier by Hopper (2004) and Hopper & Thompson

 Simona Pekarek Doehler

(2008) for English and Günthner (2006) for German (see also Günthner & Hopper, 2008). A first illustration is provided in the following excerpt: (20) (FNRS I) 1 P donc on va devoir retravailler la di so we will have to rework the di 2

la motivation différemment. det motivation differently

3 F ce qui est intéressant par rapport à what is interesting with regard to 4

ce que:: vous disiez, what you said

5 eh:hm vous parliez là de: de littérature et tout, you talked there about literature and all 6 .h moi je vois avec mes élèves qui pourtant sont au I see with my pupils who still are in 7

collège le fait de travailer par immersion, high school the fact of working by immersion

8 (.) où ils sont vraiment confrontés à where they are truely confronted with 9

des texts authentiques en allemand, det authentic texts in German

10

sur les zed des statistiques et caetera, on the ‘zeds’ of statistics et cetera

11 .h il y a cette: crain:te de l’inconnu:: there is this fear of the unknown 12 13

qui se qui c- qui disparaît petit à petit? that that that disappears step by step ((turn continued))

Fabio’s ce qui est intéressant par rapport à ce que:: vous disiez projects the upcoming talk as accomplishing a specific action, namely providing an assessment of what was interesting in Peter’s talk. However, the clause following the ce qui-piece (vous parliez là de: de literature et tout, ‘you spoke there about literature and all’, l. 5) is clearly not fulfilling this projection (and is apparently not heard as such by the coparticipants: see the absence of recipient reactions): the ‘you spoke there about literature’, rather than specifying ce qui est intéressant ... ‘what is interesting’, further frames the subsequent talk as being related to what the preceding speaker was just talking about, namely literature.

Projector constructions in French talk-in-interaction 

The right-hand boundary of the stretch of talk following the ce qui projector construction is hard to identify: what follows appears to be rather open-ended, and it is not clear what exactly in the subsequent talk presents the ‘interesting’ point foreshadowed in the projector construction. As mentioned earlier, this is in line with recent accounts of pseudoclefts in spontaneous talk. Hopper (2004) in particular, shows for English data that, frequently, ‘what follows the wh piece is a stretch of discourse of indeterminate length’ (p. 5). Parallel observations have been made by Müller (2006) for French conversational data. Accordingly, Müller considers pseudoclefts as implementing a ‘hyper-scheme’ with no discernable right-hand boundary. These structural properties of the configuration quoted in (20) highlight the projection force emanating from the ce qui piece. Hopper (2004) argues that the whparts in complex pseudocleft-like configurations (i.e. configurations with a lengthy stretch of talk following the wh-part) accomplish the function of projecting an upcoming focal stretch of discourse. These properties very much correspond to what Hopper and Thompson (2008) have found for pseudoclefts in English conversation, showing that the wh-pieces frames further talk as reporting on some action, event, and the like (cf. pt. 3.3. supra). What is interesting is that, by doing so, the projector construction helps co-participants to monitor through the architecture of the turn. For the specific configuration discussed here, based on the sequential organization of talk and its moment-by-moment deployment, and in line with the quoted studies, I would like to argue the following: the second part can be read as related to the ce qui (i.e. the wh-) part as either an open-ended or a bounded multi-unit stretch of talk due to praxeological projections emanating from the ce qui part. That is: the second part is oriented to by co-participants – and hence interpretable – as accomplishing an action that is projected by the first part and that accomplishment can be of various length. Due to the initial projection force, what follows can be read as related to that part, for instance as providing background info (as in l. 3, ex. 20), rather than as doing some random thing (see Hayashi, 2004, on this property of projection; see also Goodwin, 1996, on prospective indexicals). This property allows the initial projector construction to function as a resource for constructing complex turns or part of these. Further evidence for these properties of ce que/ce qui x projector constructions is provided in excerpt (21): (21) (FNRS J)5 1 L ça ça peu::t (0.7) ((clears throat)) that that can 5. This excerpt is part of a database shared among several members of the research team at the Centre for Applied Linguistics, University of Neuchâtel, and has previously been discussed by Müller (2007).

 Simona Pekarek Doehler

2 (..) entrer dans la motivation. be part of det motivation

3

(0.3)

4 L je dirais que (1.0) dan::s- (0.8) I’d say that in 5

ce qu’on-^eu::h on: on a fait, what we we we did

6

>maintenant j’ai une classe que je suis now I have a class that I follow

7

pour la deuxième année, for the second year

8

=°mais mêm:e ceux que je suis enbut even those who I follow in-

9

que j’ai pour la première année<,° that I have for the first year

10 .h ((slowing down)) aprè:s quelques mois, after some months 11

peut-être après le premier semester, maybe after the first term

12 (.) la langue deux est devenu:e un- (1.3) the second language has become a 13

elle fait partie des meubles. it is part of the furniture

14 L de [la de la classe. of the of the class

15 M [hm

16

(.)

17 L plus personne ne remet en cause cela. nobody questions that This excerpt, part of which was presented earlier as excerpt (6), is taken from a discussion among teachers on how they can positively influence students’ motivation for learning second languages. Laurent is talking about his immersion classes and has just provided a first example. At line 4, he introduces a second example, thereby abandoning his initial syntactic project (je dirais que (1.0) dans- (0.8) ... ‘I’d say that in ...’) and then starting, at line 5, with new prosodic onset on ce qu’on a fait ‘what we did’. This projector construction announces an upcoming report of some past doing

Projector constructions in French talk-in-interaction 

– of ‘what we did’ –, thereby foreshadowing a specification of the solution implemented in his class. However, the specification is not provided in the subsequent clause; also, that clause is not syntactically tied to the announcement segment. Rather, it is the whole stretch of talk following the initial fragment that is hearable as accomplishing the projected action. Again, the ce que piece functions as a resource for action projection, helping co-participants to monitor through the architecture of the turn. It represents one means by which the turn is recipient designed. The sequential organization of the quoted excerpt allows us to further empirically evidence the strength and effect of the projection on the subsequent talk. Interestingly, the initial segment is followed by a series of framing devices that specify the situational (‘now I have a class that I follow for the second year’) and temporal (‘after a couple of months, maybe after the first term’) conditions, leading up to the peak statement according to which ‘the second language has become ... part of the furniture’ (l. 13). Inserted in this stretch of talk is a parenthetical comment (mais même ceux ... que j’ai pour la première année ‘but even those in the first year class’, l. 8/9), clearly produced in lower voice. Duvallon & Routarinne (2005) have discussed parentheticals as resources allowing speakers to attend to additional activities, while at the same time preserving the initiated and projected sentence and/or action. This is exactly what happens here as regards action projection: the projection strength of the projector construction is highlighted by the fact that material can come in between while the relevance of some upcoming next action is being preserved. Also, due to the initial projection, what comes in between (i.e. the parenthetical and other framing devices) can be here again heard as related to the projected action (e.g. providing background info or framing that action) and not as doing some random thing (see Hayashi, 2004, and Goodwin, 1996, on this property of projection). Co-participants clearly orient to this structure as evidenced by their recipient reactions. The absence of recipient actions up to line 15 suggests that neither the framing information nor the parenthetical insert are heard as accomplishing the projected action: that is, interlocutors orient to the action as still outstanding throughout lines 6 to 14. Also, recipient reaction occurs at the first recognizable point of the gist of Laurent’s telling (marked by final intonation on meubles ‘furniture’, l. 13), in overlap with Laurent’s incremental turn-extension (l. 14), which is then followed by his formula-like concluding remark (l. 17). We then find here another instance of the recurrent structural pattern as documented in the preceding sections: I. projector construction; II. complex multi-unit stretch of talk; III. recipient token at first recognizable point of completion of the projected action.

 Simona Pekarek Doehler

This subtle coordination of speaker action and recipient reaction is based on the mutual orientation to the projection emanating from the projector construction, allowing participants to monitor through the embedded organizational structure of the subsequent talk. Like in the preceding extracts, it is this inscription of grammar in the moment-by-moment deployment of actional trajectories and the resulting projections that tie together clauses and TCUs (combined, of course, with local prosodic mappings) – to some degree independently of the syntactic marking of properties such as coordination and subordination. Despite of the structural parallels, however, this excerpt differs from the previous examples in one significant regard. In the present case, the action that is accomplished in the second segment corresponds only quite vaguely to what is announced in terms of contents in the first: Laurent does not strictly speaking report a previous doing, i.e. ce qu’on a fait ‘what we did’. Rather, the action which he accomplishes can be read as presenting the result of some such former doings, namely that ‘the language has become part of the furniture’, probably meaning that it became used naturally in the immersion classroom. Laurent here provides one example of the more general theme of the discussion, namely motivational issues regarding language learning in school. In this sense, it appears that the action projected by the projector construction does not necessarily need to be named, mentioned, made explicit by that construction. Rather, the construction draws its projection potential and the interpretability of the emanating projection from its sequential embeddedness: the subsequent action is recognizable as accomplishing the projection, as it provides an example of students’ motivation, the general theme of discussion. Based on this evidence, we can argue that in this last example the projector construction functions as a mere discourse structuring device, quite independently from its lexical-semantic content, along the lines of what we have most clearly observed for je veux dire (pt. 3.2 supra). This very loss of semantic value is here again reminiscent of classical patterns of grammaticization described in the literature (cf. 3.4 supra). Although this is rare in our data, it is still one potential of projector constructions. In sum, the functioning of the French ce que/ce qui x projector construction is in line with what Hopper (2001), in his English data, identifies as a general discourse use of the wh-clause to draw attention to an upcoming discourse element. What at first sight resembles a pseudocleft, rather than being a bi-clausal pattern, is composed of a discourse-organizational initial “pre-fab” and a multiunit stretch of talk.

Projector constructions in French talk-in-interaction 

5. Discussion: Projector constructions and the temporal unfolding of talk-in-interaction In this section, I sum up the main findings and outline some of their implications. The choice to focus in this study on what initially appeared to be three very different constructions has allowed me to demonstrate that the characteristics of projector constructions are not tied to a specific linguistic shape, but evidence a more pervasive property of how participants use grammar in interaction. These observations extend previous research on a range of constructions that can be subsumed under the notion of projector constructions (Couper-Kuhlen & Thompson, 2008; Hopper, 2001, 2004; Hopper & Thompson, 2008; Günthner, 2006, this volume; Günthner & Hopper, 2010). The preceding analyses have in particular shown the following points: A recurrent compositional schema What has been treated classically as complex or bi-clausal constructions, tied together by morphosyntactic means such as subordination markers, often occurs in talk-in-interaction as syntactically autonomous stretches of talk, fitted together by means of prosody and sequentially embedded initial projections. The resulting constructional schema runs as follows: [projector construction + syntactically independent (single or multi-unit) stretch of talk]

What we observe, then, are combinations of initial fragments of three types, je veux dire, il y a NP ‘there is NP’ and ce que/wh-pieces, plus something that follows. These are just three examples among a possibly larger inventory of related constructions. The initial projector fragment creates a slot for further talk, but leaves its syntactic trajectory unspecified. The subsequent talk is prosodically tied to the projector construction. The projector construction is oriented to by co-participants not as a complete action in itself, and not as a TCU in itself, but as projecting some further action, which then is typically acknowledged by recipient reactions at the point of its first possible completion. The second part of the constructional schema can take many different shapes. It may consist of a clause (simple or complex) or of a lengthy stretch of talk composed of multiple TCUs, as a series of (hypotactically or paratactically) linked clauses, including parenthetical inserts, sometimes without recognizable boundary to the right. In all these cases, the extent of the stretch following the projector construction is potentially open.

 Simona Pekarek Doehler

Projector constructions, recipient design and the configuration of complex turns What do projector constructions do in talk-in-interaction? The characteristics of conversational syntax allow speakers to construct complex turns by means of multiple insertions, recyclings, expansions, increments or simply multiple TCUs. Projections are a central instrument for ‘tying together things’ in mutually recognizable ways. The synchronization among participants throughout such complex turns that unfold across time is only possible because participants jointly orient toward syntactic and actional projections. These allow them to monitor through the emergent architecture of talk-in-interaction. Projector constructions are, for recipients, a central point of reference for such monitoring. Accordingly, they are for speakers a central resource by which they recipient design their talk: – Je veux dire fames subsequent talk as providing some kind of elaboration on the immediately preceding talk by the same speaker (e.g. an exemplification of a preceding statement, an account for a preceding affirmation). – Il y a NP frames upcoming talk as being about a ‘problem’, an ‘issue’, or a ‘thing’ that is brought by means of the NP in the il y a NP construction itself. – The ce qui/ce que (wh-)construction projects further talk as specifying something that is presented in the ce qui/ce que-piece, i.e. as being about ‘what someone said’, ‘what is interesting’, etc. In other words, while je veux dire merely creates a slot for further talk, and the expectation that that talk will in some sense elaborate on the immediately preceding talk by the same speaker, in the case of il y a NP and of wh-pieces, the lexicosemantic properties of these typically contribute to specify the nature of the projection. What warrants the recognizability of parts of turns as doing some sequentially relevant action, and what warrants the recognizability of the point at which the action is completed, is the contextually embedded projection emanating from the projector constructions – a projection that both announces more to come, and indicates (though only vaguely) the type of action that is being done by that more. The projector constructions can therefore be interpreted as organizational devices that are instrumental for drawing co-participant’s attention to upcoming talk, for floor-holding and for the construction of complex turns. That is not to say, though, that projector constructions (nor projections) simply determine the further course of talk. Rather they are eminently context-sensitive and non-deterministic: depending on their contextual embeddedness, they foreshadow possible trajectories of talk that are then configured on a moment-tomoment basis, as talk unfolds across time. Projection properties are not simply inherent in given syntactic or lexical patterns, but result from grammar-as-it-isinscribed-in-action.

Projector constructions in French talk-in-interaction 

Projector constructions as sedimentations from projection practices The question remains as to what the projector construction precisely is. Projector constructions show the following recurrent properties across the three forms under analysis: – they are syntactically and lexically compact, while leaving open the syntactic trajectory of their follow up; – they are not prosodically pro-eminent (not accentuated, not marked by increased volume, sometime produced with markedly lower volume); – they are semantically light or reduced; – they are syntactically and pragmatically incomplete. These features suggest grammaticization (cf. Traugott & Hopper, 2003) of the three fragments under discussion along the lines of pragmatic markers. Based on the preceding analysis, I would like to suggest that the je veux dire, il y a NP and ce que x (wh-part) formats are sedimentations resulting from highly routinized interactional projection practices (see Auer, 2005 for related argument on behalf of other grammatical phenomena); they are epiphenomena (cf. Hopper, 20046) of interactional projection routines. This leads to a reanalysis of the syntactic status of the three patterns. Je veux dire can be re-analyzed as discourse marker (see Auer & Günthner, 2003; Günthner & Imo, 2003; Imo, 2007 for similar arguments regarding German). The ce que/ce qui x pattern, although it does regularly occur as a free relative clause in standard pseudo-clefts, also regularly occurs as a discourse organizational framing element (for German and English see Günthner, 2006; Hopper, 2001; Hopper & Thompson, 2008;), and this is also the case for the il y a NP construction. Subordination in grammar and the sequential organization of actions It is interesting at this point to return to the question, brought up in the introduction to this paper, of the fitting together of subordination in grammar and the sequential organization of actions. As Thompson (2002: 130) has argued (see also Schegloff, 1996), ‘the relative ‘prominence’ of any part of a given utterance need to be based on the interactional actions that utterance is performing in its local situational context’. Canonical constructions involving je veux dire, il y a NP and ce que/ce qui x differ as to the relative syntactic (and informational) status of the two parts of the constructions: 6. ‘Grammar is an epiphenomenon of frequent combinations of constructions’ Hopper, 2004: 239. See also Thompson & Couper-Kuhlen (2005: 483): ‘Grammar is best understood as what as been ritualized from interactions’.

 Simona Pekarek Doehler

while je veux dire and il y a NP both introduce what appears on syntactic grounds as a subordinate clause (a complement clause and a relative clause respectively), in presentational clefts the ce que/ce qui x free relative clause comes first, preceding the second part. In terms of grammatical prominence, then, the canonical constructions differ as to the sequential ordering of their two parts. This ordering of constituents has in turn served as a basis for interpreting the pragmatic properties of the constructions (cf. pt. 2 supra). The different syntactic (and informationstructuring) relations between the first and the second piece in canonical forms of je veux dire + clause, presentational and pseudo-cleft constructions contrasts with what we have observed for projector constructions: the prominence of the latter is defined as a function of their sequential position as a first with regard to a next. For je veux dire, for instance, the relative prominence is inversed: what grammatically speaking (is or) used to be the subordinate clause in the canonical pattern (e.g. il est parti in je veux dire qu’il est parti) here in fact carries the main action, while je veux dire prepares the ground for that main action (for a related argument concerning clauses following the Finnish että ‘that’ see Koivisto, Laury & Seppänen, this volume). This non-correspondence between syntactic status and praxeological (i.e. action related) projections is highly significant. It clearly demonstrates that the relative importance of the clauses, stretches of talk or fragments involved is less defined by the syntactic status of the segments as subordinate or main clause in the canonical counter-parts of the constructions under analysis than by their sequential placement as a first or as a next, and by what they do in interaction based on this sequential placement: the sequentially first segment is the element of projection on which rests the recognizability of what follows as doing some specific thing. This clearly shows that the on-line sequentiality of grammar – i.e. its moment-by-moment temporal unfolding (cf. Auer, 2005, 2007 and 2009 on ‘on-line’ grammar) – is a decisive element for how segments of talk are fitted together, sometimes far beyond the reach of syntax. Emergent grammar and the sequential nature of talk But what does this tell us about the nature of grammar? As Hopper puts it, ‘everyday language is built up out of combinations of (...) prefabricated parts (...) a kind of pastiche, pasted together in an improvised way out of ready-made elements’ (Hopper, 1987: 144). This fragmentary, varying nature of grammar is corroborated by a growing body of research documenting that grammatical constructions that are well established in the literature are used in talk-in-interaction not as stable, prefabricated, more or less prototypical structures, but rather as sequentially contingent grammatical configurations. Some of these are more or less sedimented and others more flexible, variable, patterned ad hoc (Hopper, 2001, 2004; Thompson

Projector constructions in French talk-in-interaction 

& Hopper, to appear; Pekarek Doehler, in press; Pekarek Doehler & Müller, 2006). This is what Hopper (2001) calls the ‘openness’ of grammatical constructions. In this light, it is possible that the standard je veux dire + complement clause, il y a presentational clefts and pseudoclefts are the sedimented products of highly routinized projection practices. They would then simply represent the more normative versions (cf. Hopper, 2001) of a more general constructional schema consisting of an initial projector fragment plus a subsequent stretch of talk. As we have seen throughout this paper, this compositional schema is not ordered randomly, but is based on what comes first and what comes next as talk unfolds across time: the more uniform, stable, sedimented pieces are placed first, and the more flexible, variable part are placed next. The robustness of the first part is another case in point for the on-line character of grammar, where initial fragments, due to their relative compactness, can be used as resources for projecting grammatically unspecified follow-ups. Taken together, these properties highlight the eminently contingent nature of grammar (Ford, 2004), its embeddedness in (and possibly its formatting in response to) the sequential organization of talk and its moment-by-moment unfolding. 6. Conclusion From the viewpoint of classical discussions of (French) grammar, it could be argued that this paper has set out to analyze untypical occurrences of three very different constructions, both syntactically and pragmatically: complex sentences involving subordination with je veux dire (lit: ‘I want to say’), il y a NP ‘there is NP’ presentational constructions and pseudoclefts – thereby focusing on cases where the morphosyntactic marking of clause-combining is absent. Despite of the different morphosyntactic and pragmatic characteristics of the ‘canonical’ constructions, I have shown in this paper that their initial parts (je veux dire, il y a X and ce que/ce qui x) regularly occur in talk-in-interaction not as part of bi-clausal patterns, but as discourse-structuring elements: they are projector constructions. As such, they are part of a recurrent constructional schema in talk-in-interaction consisting of a projector construction followed by a (single or multi-unit) stretch of talk. I have interpreted the three constructions as formats that are routinely tied to a specific social action, namely projecting upcoming talk by the same speaker, and I have identified differences as to how exactly each of the constructions foreshadows upcoming talk. On these grounds, I have suggested that they may be the products of routinized interactional projection practices. This interpretation is in line with usage-based approaches to language as to the emergent nature of grammar,

 Simona Pekarek Doehler

understood as sedimented through locally deployed interactional routines and organized in terms of (partially sedimented and partially open) constructions. From an emic point of view – that is from the perspective of the participants’ orientation toward locally organized courses of activities –, it is interesting to ask how speakers accomplish the constructional schemas initiated by the projector constructions and how their co-participants manage to interpret them: how do participants structure complex turns in mutually recognizable ways? We have seen throughout this paper that they use the projection potential of grammar as a resource to do so: speakers start off with an initial projector fragment, sometimes re-launch it throughout the accomplishment of the projected action, and conclude the action by means of some summarizing-concluding comment; recipients in turn orient to the projections emanating from the initial construction in order to monitor through the unfolding of the turn. In this way, the projection potential of grammar serves as a resource for ordering things sequentially, for holding the floor, for gluing together parts of talk (including clauses) and for structuring the architecture of complex turns in a mutually recognizable way. It is a central part of how speakers recipient desing their turns. From the point of view of syntax, it is interesting to ask a different question: what do ‘canoncial’ syntactic constructions (here: constructions that have traditionally been treated as bi-clausal constructions) become in the light of the temporal unfolding of talk-in-interaction? What we observe speakers do in the data provides further evidence corroborating the idea that participants use grammar not as a set of pre-fabricated rules and combinatory patterns, but as a resource for action, combining “pre-fabs” with more ad-hoc configurations. Some of its parts appear to function as highly sedimented, routinized formats (such as the initial je veux dire, il y a X or ce que-parts), and others (such as the subsequent stretches of talk) are adapted step by step to the local moment-by-moment unfolding of talkin-interaction. The sequential organization of the more sedimented and the more open pieces, however, is not random: in all three types of configurations analyzed here, the initial segment is both syntactically and lexically quite consistent, and the subsequent part, though recognizable as a ‘package’, develops in many possible directions. This, together with the other evidence provided in this paper, I hope, offer an empirical contribution to a growing understanding of grammar as routinized from action, i.e. as made up of recurrent patterns of language use that are highly contingent, thoroughly temporal and formatted in response to local communicative needs.

Projector constructions in French talk-in-interaction 

References Auer, Peter. 1996. “The pre-front field position in spoken German and its relevance as a grammaticalization position.” Pragmatics 6(3): 295–322. Auer, Peter. 1998. “Zwischen Parataxe und Hypotaxe: ‘abhängige Hauptsätze’ im Gesprochenen und Geschriebenen Deutsch.” Zeitschrift für Germanistische Linguistik 2: 284–307. Auer, Peter. 2005. “Projection in interaction and projection in grammar.” Text, 25 (1), 7–36. Auer, Peter. 2007. “Syntax als Prozess.” In Gespräch als Prozess. Linguistische Aspekte der Zeitlichkeit verbaler Interaktion, Heiko Hausendorf (ed.), 95–142, Tübingen: Narr. Auer, Peter. 2009. “On-line syntax: thoughts on the temporality of spoken language.” Language Sciences, 31, 1–13. Auer, Peter and Günthner, Susanne. 2003. “Die Entstehung von Diskursmarkern im Deutschen – ein Fall von Grammatikalisierung?” InLiSt – Interaction and Linguistic Structures 38: http://www.uni-potsdam.de/u/inlist/issues/38/index.htm. Berrendonner, Alain. 2003. “Grammaire de l’écrit vs. Grammaire de l’oral: le jeu des composantes micro- et macro-syntaxiques.” In Interactions orales en contexte didactique, Alain Rabatel (ed.), 249–264. Lyon: Presses Universitaires de Lyon. Berrendonner, Alain. 2008. “L’alternance que / #. Subordination sans marqueur ou structure périodique?.” In Modèles syntaxiques, Dan Van Raemdonck (ed.), 279–298, Bern: Peter Lang. Blanche-Benveniste, Claire. 1989. “Constructions verbales ‘en incise’ et rection faible des verbes.” RSFP, 9: 53–73. Blanche-Benveniste, Claire. 2010. “Les pseudo-clivées et l’effet deux points.” In La Parataxe. Vol. 1, Marie-José Béguelin, Matthieu Avanzi and Giles Corminboeuf, (eds), 185–218. Bern: Peter Lang.. Blanche-Benveniste Claire et al. 1991. Le français parlé. Etudes grammaticales. Paris: CNRS Editions. Couper-Kuhlen, Elizabeth and Thompson, Sandra A. 2008. “On assessing situations and events in conversation: `extraposition’ and its relatives”, Discourse Studies, 10 (4), 443–467. Deulofeu, Henri-José. 1986. “Syntaxe de que en français et le problème de la subordination.” RSFP, 8: 79–104. Duvallon, Outi and Routarinne, Sara. 2005. “Parenthesis as a resource in the grammar of conversation.” In Grammar and lexis in conversation, Auli Hakulinen and Margret Selting (eds), 45–74. Amsterdam: Benjamins. Ford Cecilia E. 2004. “Contingency and units in interaction”, Discourse Studies, 6 (1): 27–52. Ford, Cecilia E. and Thompson, Sandra A. 1996. “Interactional Units in Conversation: Syntactic, Intonational, and Pragmatic Resources for the Management of Turns.” In Interaction and Grammar, Elinor Ochs, Emanuel A. Schegloff and Sandra A. Thompson (eds), 135–184. Cambridge: Cambridge University Press. Goldberg, Adele. 1995. Constructions. A construction Grammar approach to argument structure. Chicago: University of Chicago Press. Goodwin, Charles. 1996. “Transparent vision.” In Interaction and Grammar, Elinor Ochs, Emanuel A. Schegloff and Sandra A. Thompson (eds), 370–404. Cambridge: Cambridge University Press. Goodwin, Charles. 2002. “Time in action.” Current Anthropology, 43: 19–35.

 Simona Pekarek Doehler Günthner, Susanne. 2006. “Was ihn trieb, war vor allem Wanderlust” (Hesse: Narziss und Godmund): Pseudocleft-Konstruktionen im Deutschen.” In Konstruktionen in der Interaktion, Susanne Günthner and Wolfgang Imo (eds), 59–90. Berlin, New York: Mouton de Gruyter. Günthner, Susanne and Hopper, Paul. 2010. “Zeitlichkeit & sprachliche Strukturen: Pseudoclefts im Englischen und Deutschen.” Gesprächsforschung – Online-Zeitschrift zur verbalen Interaktion 11: www.gespraechsforschung-ozs.de: 1–28. Günthner, Susanne and Imo, Wolfgang. 2003. “Die Reanalyse von Matrixsätzen als Diskursmarker: ich mein-Konstruktionen im gesprochenen Deutsch”, In Jahrbuch der Ungarischen Germanistik 2003, Magdolna Orosz and Andreas Herzog (eds), 181–216. Budapest, Bonn: DAAD. Hayashi, Makoto. 2004. “Projection and grammar: notes on the ‘action-projecting’ use of the distal demonstrative are in Japanese.” Journal of pragmatics 36 (8): 1337–1374. Hopper, Paul J. 1987. “Emergent grammar.” Proceedings of the Annual Meeting of the Berkeley Linguistics Society 13: 139–157. Hopper, Paul J. 2001. “Grammatical constructions and their discourse origins: prototype or family resemblance?” In Applied cognitive linguistics I: theory and language acquisition, Martin Pütz, Susanne Niemeier and René Dirven (eds), 109–129. Berlin, New York: Mouton de Gruyter. Hopper, Paul J. 2004. “The openness of grammatical constructions.” Papers from the 40th regional meeting of the Chicago Linguistic Society 40: 153–175. Hopper, Paul J. and Thompson, Sandra A. 2008. “Projectability and Clause Combining in interaction.” In Crosslinguistic Studies of Clause Combining, Ritva Laury (ed.), 99–123. Amsterdam: Benjamins. Hopper, Paul J. & Traugott, Elizabeth C. 2003 [1993]. Grammaticalization. Cambridge: Cambridge University Press (second edition). Imo, Wolfgang 2005. “A Construction Grammar approach to the phrase “I mean” in spoken English.” InLiSt – Interaction and Linguistic Structures 42: http://www.uni-potsdam.de/u/ inlist/issues/42/index.htm Imo, Wolfgang. 2007. Construction Grammar und Gesprochene-Sprache-Forschung. Konstruktionen mit zehn matrixsatzfähigen Verben im gesprochenen Deutsch. Tübingen: Max Niemeyer Verlag. Jacques, Francis. 1981. «Je veux dire.» Philosophica, 27: 6–26. Jullien, Stéphane. 2005. “Introduction de référents topiques dans des dialogues d’adolescents dysphasiques: le cas de la construction présentative clivée.” TRANEL – Travaux Neuchâtelois de Linguistique 41: 49–68. Jullien, Stéphane. 2007. “Prosodic, syntactic and semantico-pragmatic parameters as clues for projection: the case of «il y a».” Nouveaux cahiers de linguistique française 28: 283–297. Lambrecht, Knud. 1994. Information structure and sentence form. Topic, focus, and the mental representations of discourse referents. Cambridge: Cambridge University Press. Lambrecht, Knud. 2001. “A framework for the analysis of the cleft constructions.” Linguistics 39: 463–516. Langacker, Ronald W. 1991. Foundations of cognitive grammar, vol. 2. Stanford: Stanford University Press. Lerner, Gene H. 1991. “On the syntax of sentences-in-progress.” Language in Society 20: 441–458. Matthiessen, Christian and Thompson, Sandra A. 1988. “The structure of discourse and “subordination” .” In Clause-combining in grammar and discourse, John Haiman and Sandra A. Thompson (eds), 275–333. Amsterdam: Bejamins.

Projector constructions in French talk-in-interaction  Müller, Gabriele. 2006, ms. Un procédé grammatical dans l’interaction: la pseudo-clivée. Thèse de doctorat, Université de Lausanne. Müller, Gabriele. 2007. «Entre l’unité de construction d’un tour et l’organisation de tours multiunités: la pseudo-clivée en interaction.» TRANEL – Travaux Neuchâtelois de Linguistique 47: 157–176. Schegloff, Emanuel. A., Ochs, Elinor and Thompson, Sandra A. 1996. “Introduction.” In Interaction and Grammar, Elinor Ochs Emanuel A. Schegloff and Sandra A. Thompson (eds), 1–50. Cambridge: Cambridge University Press. Pekarek Doehler, Simona. (in press). “The on-line formating of dislocated constructions in French conversation.” In Constructions: emerging and emergent, Peter Auer and Stefan Pfänder (eds). Berlin, New York: Mouton de Gruyter. Pekarek Doehler, Simona and Müller, Gabriele. 2006. “Zur Rolle von Linksherausstellungen bei der interaktiven Konstruktion von Auflistungen: Linksversetzungen und Pseudo-Clefts im gesprochenen Französischen.” In Grammatik und Interaktion. Untersuchungen zum Zusammenhang von grammatischen Strukturen und Gesprächsporzessen, Arnulf Deppermann, Reinhald Fiehler and Thomas Spranz-Fogasy (eds), 245–277. Radolfzell: Verlag für Gesprächsforschung. Pekarek Doehler, Simona, De Stefani, Elwys and Horlacher Anne-Sylvie. (forthcoming). “The grammar of closings: the use of dislocated constructions as closing initiators in French talkin-interaction.” University of Nottingham: Nottingham French Studies. Pekarek Doehler, Simona, De Stefani, Elwys, Horlacher, Anne-Sylvie, Jullien, Stéphane and Müller, Gabriele. 2010. “Configurations paratactiques et grammaire-dans-l‘interaction.” In La parataxe. Structures, marquages et exploitations discursives, Marie-José Béguelin, Mathieu Avanzi and Gilles Corminboeuf (eds), vol. 2: 387–409. Bern: Peter Lang. Pomerantz, Anita M. 1984. “Agreeing and disagreeing with assessments: some features of preferred/dispreferred turn shapes.” In Structures of social action, Maxwell J. Atkinson and John Heritage (eds), 57–439. Cambridge: Cambridge University Press. Sacks, Harvey, Schegloff, Emanuel A., Jefferson, Gail. 1974. “A simplest systematics for the organization of turn-taking for conversation.” Language 50: 696–735. Schegloff, Emanuel A. 1996. “Turn organization: One intersection of grammar and interaction. In Interaction and Grammar, Elinor Ochs, Emanuel A. Schegloff and Sandra A. Thompson (eds), 52–133. Cambridge: Cambridge University Press. Selting, Margret. 1996. “On the interplay of syntax and prosody in the constitution of turn- constructional units and turns in conversation.” Pragmatics, 6(3): 371–389. Selting, Margret. 2000. “The construction of units in conversational talk.” Language in Society, 29: 477–517. Thompson, Sandra A. 2002. “Object complements and conversation – towards a realistic account.” Studies in Language 26: 125–163. Thompson, Sandra A. and Mulac, Anthony. 1991. “A quantitative perspective on the grammaticization of epistemic parentheticals in English.” In Approaches to grammaticalization, vol. 1, Elizabeth C. Traugott and Bernd Heine (eds), 313–329. Amsterdam: Benjamins. Thompson, Sandra A. and Couper-Kuhlen, Elizabeth. 2005. “The clause as a locus of grammar and interaction.” Discourse Studies 7 (4–5): 481–505. Valli, André. 1981. «Note sur les constructions dites «pseudo-clivées» en français.» Recherches sur le français parlé 3: 195–211.

 Simona Pekarek Doehler

Symbols used in transcripts [ ] onset, and, if relevant, end of overlap = intra- and inter-turn latching & turn continuation after overlap (.) (..) (...) unmeasured (micro-)pauses up to ca. 1s (1.5) measured pauses coul- cut-off ce: lengthening of preceding sound chemin? rising intonation temps. falling intonation train, continuing intonation besoin accentuation NON louder .h in-breath °ça fait tout° soft voice >et ça ça< faster slower ((laughing)) transcriber’s comment + indicates the onset of a stretch of talk to which a comment relates Symbols used in the gloss det loc part pro pro.rel 3sg/1pl m/f/n sbj obj

determiner locative particle pronoun relative pronoun third person singular/ first person plural masculine/ feminine/ neuter subject object

A note on the emergence of quotative constructions in Japanese conversation Ryoko Suzuki

1. Introduction This short paper is an attempt to relate two functional approaches to grammar that have emerged over the past three decades: (1) the approach which delves into the relationship between interaction and grammar, e.g., Couper-Kuhlen and Ono 2007, Englebretson 2007, Ochs, Schegloff, and Thompson 1996, Selting and Couper-Kuhlen 2001; (2) the approach focusing on grammatic(al)ization e.g., Bybee 2003, Hopper and Traugott 2003, Traugott and Dasher 2005, Heine and Kuteva 2002, among many others. We will examine constructions involving tte, a quotative particle form in Japanese that frequently occurs in natural conversation. We will take historical conversational data and look at the quotative tte construction, paying close attention to the relationship between the construction and the interactional work it is doing, especially in the early stages of its occurrence. In modern Japanese, there are basically two major forms of quotatives, to which is more readily associated with higher formality, and tte, with casualness. To has been used as a quotative since Old Japanese period, whereas tte began to be documented in conversational portions of novels in early 1800s (Suzuki R. 1999a,1999b, 2007). There are at least two views in terms of the origin of tte: One view says it is a contracted form of to plus the adverbial form of the verb iu ‘say’with a connective particle te. The other view says it is a combination of to plus a connective particle te (Nihon Kokugo Daijiten (NKD)). Tte is found in conversational portions of novels as early as the 1800s as a form introducing an utterance produced elsewhere by a third person. Not just utterances, but also the content of thoughts, or even a sound or noise can be introduced using the tte form. However, introducing another’s utterance is the earliest function to emerge, the most frequent and stable, and thus the most fundamental function across time in my diachronic data (Suzuki R. 1999a,1999b, 2007). An example like the following would be representative of the contemporary use of tte:

 Ryoko Suzuki

(1)

nichiyoobi no niji goro da tte itte ta. Sunday gen two o’clock around cop comp say:prg pst ‘(She) was saying that it is around two o’clock on Sunday.’ (cited by Okamoto 1995: 239 as natural conversational data. Romanization and translation modified)

Tte follows the complement clause (nichiyoobi no niji goro da ‘It is around two o’clock on Sunday’), which represents reported speech. The reported utterance in the complement is followed by a predicate of saying (itte ta ‘(she) was saying’). Since the late 1800s, tte has gradually increased the range of its discoursepragmatic functions, and has become a very frequently-used morpheme in contemporary Japanese conversation (Suzuki R. 1999a, 1999b, 2007). The functions of tte include utterance-final uses as a pragmatic particle indicating various types of speaker attitudes, such as hearsay reporting, insistence, self-mockery, and concession (e.g., Okamoto 1995, Okamoto and Ono 2008, Suzuki R. 1999a, 1999b, 2007, Suzuki S. 1998, 2001). Example (2), an invented example, shows the utterance-final use of tte with a reportive function; the main verb ‘say’ in the English translation is not overtly expressed in Japanese. Example (3) includes two tokens of utterance-final tte as a marker of ‘insistence.’ (2) Minako, otoosan ga ima sugu koi tte. name father nom now right.away come tte ‘Minako, Dad (says) “Come right away.”’ (3)

(invented example)

hontoo da tte. ore puropoozu nanka shitena tte. true cop tte I popose such.a.thing do:neg tte ‘(It)’s true, really. (I) didn’t propose marriage (to her), really.’ (Okamoto 1995: 241 from a TV drama. Romanization and translation modified)

In comparing constructions like (1) on one hand, and (2) and (3) on the other, in terms of their syntactic and semantic-pragmatic behaviors, Okamoto draws the following conclusions: the use of tte as a ‘complementizer’, as in (1), and its use as an utterance-final pragmatic particle, as in (2) and (3), are not identical, with the latter being more pragmatically strengthened (1995: 243). In other words, she argues, a clause with the final particle tte is not a complement clause with an ellipted main clause, but rather should be treated as an independent construction with tte expressing a specific modality. At the same time, Okamoto also acknowledges the closeness in meaning between some instances of the complement clause construction and the reanalyzed independent clause construction. For example, (1) and (2) above are one such pair of examples showing this closeness, since they both introduce an utterance made

The emergence of quotative constructions in Japanese 

by someone else prior to the current speech event, with tte immediately following the reported speech. Okamoto (1995: 243) suggests that the latter use may “have evolved from elliptical expressions consisting of complements, and that these SFPs have their origin in the corresponding COMPs” (emphasis added).1 She continues by arguing that “the functions of these SFPs reflect to some extent the functions of the corresponding COMPs: ... to and tte for indicating the quotative aspect of the utterance. It is thus logical to assume that the usages of these SFPs are instances of reanalysis of morphological functions.” Having studied the development of a range of pragmatic functions of tte diachronically in my own research, I find that what Okamoto suggests above makes sense intuitively: There is a clear connection between the complementfinal tte (e.g. (1) and utterance-final uses of tte (e.g., (2) and (3)). Inspired by her study, in this paper, I will delve further into the early development of utterance-final tte clauses i.e., those which have no following main clause. What do the earlier stages of tte occurrences show us regarding the development of the independent tte clause? Would we only find full expressions (i.e., non-elliptical versions) consisting of a tte-complement and a main clause in the earlier data? We will see in later sections that at the earliest stage of development of tte2 in the 1800s, we do find instances of tte in utterance-final position without a following ‘main’ clause, with several different interactional motivations. I will suggest that the earlier data deserve careful examination, suggesting that certain types of interactional practices were presumably frequently associated with utterance-final uses of tte. 2. Background and hypothesis It has been pointed out in the grammaticization3 literature that when two clauses of certain types are combined frequently, structurally less-integrated clauses become more integrated, as shown in the continuum on Figure 1, based on Hopper and Traugott (2003: 179, 190, 204). According to the figure, combined clauses have 1. Okamoto’s SFP stands for sentence-final particles, i.e., utterance-final pragmatic particles in this paper, and her COMP stands for complementizers. 2. The earliest instances in my database come from a novel Ukiyoburo (1809) by Shikitei Sanba, which is considered to reflect the colloquial Edo (Tokyo) speech of the time. The novel consists of a comical dialogic exchanges of characters coming to a public bathhouse in Edo. NKD, one of the most comprehensive dictionary of the Japanese language, also cites the same examples. 3. There are two technical terms, grammaticalization and grammaticization, used in the field. I take them to be more or less synonymous, and choose to use grammaticization here.

 Ryoko Suzuki Subordination

Reduction to simplex

Parataxis adjacent nuclei

Hypotaxis

nucleus + margin

nucleus > 0 margin > nucleus

minimal

maximal

n/a

maximal

minimal

>0

Figure 1.â•‡ Complex clause structure over time (based on Hopper and Traugott 2003: 179, 190, 204)

relative status to one another: a nucleus refers to a clause that can stand alone, and a margin to a clause that is dependent to a nucleus. Parataxis shows minimal integration because it consists of more than one adjacent but independent clause (nucleus). Then comes hypotaxis, which exhibits an interdependent relationship consisting of one nucleus and a margin. Subordination involves maximal integration because a margin, a dependent clause, becomes part of the nucleus. Then out of subordinate, i.e., maximally unified complex constructions, develop simple clauses, “with an original verbal construction coming to be downgraded into auxiliary-like, or adverbial status (Hopper and Traugott 2003: 211)” or even disappears: The following examples are constructed Japanese clauses that correspond to each type of complex clause structures. The bold shows an independent clause (a nucleus), and the underline shows a dependent clause (a margin): parataxis yooi ga dekita minna de tabeta. preparation nom finished everyone with ate ‘The preparation (of the meal) finished. All of us ate together.’ hypotaxis yooi ga dekita node minna de tabeta. preparation nom finished so everyone with ate ‘(The meal) was ready, so all of us ate together.’ subordination yooi ga dekita no o minna de tabeta. preparation nom finished nmz acc everyone with ate ‘All of us together ate the thing (=meal) that was ready.’ reduction to simplex yooi ga dekita node. preperation nom finished so ‘(The meal) is ready so ...’4

4. There is an implication left hanging, whose interpretation is context dependent. For example, we could add the translation ‘(please come.)’ after ‘so’ as one possible interpretation here.

The emergence of quotative constructions in Japanese 

As shown in Figure 1 and the examples above, Hopper and Traugott propose that at the very final phase of the process, a subordinate clause becomes a simplex (i.e., independent) clause on its own while the original main clause disappears. In other words, after the stage of maximal integration comes the stage of reduction of the main clause (i.e., “nucleus > 0” in Figure 1), which is part of our interest here. Cross-linguistic research has found a number of cases involving the reduction of multi-clausal constructions. For example, one of the earliest studies focusing on this point is by Thompson and Mulac (1991) on “epistemic parentheticals” in modern English. They report that in English conversation, certain combinations of subject and main verb, especially I think, have become fixed phrases, occurring not only in utterance-initial positions, but also in utterance-medial and -final positions. These fixed expressions are also accompanied by phonological reduction. These findings show that the apparent “complement” clause is now treated as an independent clause, whereas the apparent “subject-plus-main-verb” combinations are going through the process of being grammaticized into unitary epistemic phrases. The case of ultimate reduction to zero is reported in Japanese, which is typologically different from English. Ohori (1995, 1997) discusses the prevalence of clauses that include markers for subordination but are not followed by a main clause, using the term “Suspended Clause Constructions (SCCs).” He introduces a wide range of stand-alone non-final clauses ending with markers such as the conditional -ba, -tara, the connectives of reason kara, node ‘because/so’, and connectives of concession kedo, noni ‘although’. He points out that these clauses occur especially in conversation, that speakers seem not to have any trouble interpreting them, and that these markers have their own pragmatic function as utterance-final discourse particles. There are many case studies focusing on some of these markers, pointing basically in the same direction (e.g., Okamoto (1995) on the complementizers no, koto, to, tte, Nakayama and Ichihashi-Nakayama (1997) on kedo ‘though’, Higashiizumi (2006) on kara ‘because’, and R. Suzuki (1999a, 1999b, 2007) on the quotative tte.) More recently, Mulder and Thompson (2006, 2008) have reported on the recent use of but in Australian and American English conversation. Although there are traditional uses of but, i.e., as a turn-initial conjunction, linking two clauses of contrasting content together, there are also instances of but in utterance-final position, with an implication left “hanging.” They also find the further change that especially in Australian English, but has become an utterance-final discourse particle, marking contrastive content, noting that the process of grammaticization of but is similar to the pathway followed by though (Barth-Weingarten & CouperKuhlen 2002). These studies support Hopper and Traugott’s claim that in clause-combining constructions, under certain interactional circumstances, one of the clauses may get

 Ryoko Suzuki

reduced, or may be omitted, leaving one independent clause. Such studies, based on natural conversational discourse, suggest that the distinction between “main clause” and “dependent/subordinate clause” may be useful as a starting point of linguistic analysis, but that this distinction is often not clear-cut in spoken discourse, especially in conversation. The papers in this volume also point in this direction. Based on Hopper and Traugott’s schematization (2003), and on Okamoto’s suggestions (1995), we can formulate the following hypothesis: the construction in (1), in which tte connects a complement and the main clause, is indeed earlier than constructions like (2) or (3), in which we find an independent clause with tte at the end, without any main clause. 3. Data My data come from conversational portions of novels reflecting colloquial language from both the early 1800s (mainly around the 1830s) and the late 1800s (mainly around the 1880s). No more than ten tokens were taken from novels written by the same author. In order to examine the structural relationship between the quotative tte complement and independent clauses with tte at the end, I culled all instances of complement-clause-final and independent-clause-final tte tokens from my data.5 4. Observation of the early occurrences of tte in the 1800s The oldest function of tte is to introduce an utterance (or thought) made previously by someone else (Suzuki R. 1999a, b). Focusing on the syntactic environment of the early instances of tte constructions from the early and late 1800s, we find four types of syntactic environments in which a tte construction is found: the tte construction followed by a main clause (i.e., tte used as a complementizer), as well as three types of apparent counter-examples to our hypothesis, i.e., utterancefinal tte constructions that are not followed by a main clause. The characteristics and distribution of each construction will be considered below. What we find in the earlier data is thus not only a complementizer tte with a main clause; the picture is in fact more complicated. 5. Since the 1930s, we also see the sequence of tte following a nominal, hence serving the topicintroducing function (Suzuki-san tte dare? ‘Who is Suzuki-san?’), or the naming function (Sanoya tte mise shitteru? ‘Do you know the shop called Sanoya?’). Therefore, those usages of tte serve noun-related functions and they are often followed by a predicate (e.g., dare? ‘Who?’, shitteru? ‘Do you know?’). In this paper, such noun-introducing tokens of tte are excluded from the analysis.

The emergence of quotative constructions in Japanese 

4.1

Tte clause as dependent clause: Quotative complement with tte, followed by a main clause

The first type is the typical textbook example of a tte construction as an embedded clause, as in Example (1) above. The tte complement clause consists of the reported utterance plus quotative tte, followed by the main clause. The original speaker of the reported utterance is either explicitly introduced, or is easily identifiable in the context. Consider Example (4):

(4) Kooichi, a male student, is talking with his friend from school, Nobuo. Kooichi asks Nobuo about the name of a male student who acted strangely in front of them a few days ago:

1 K: sora konoaida mo interj the.other.day also 2 boku ga shokudoo de chawan ga nai tte itta toki, I nom cafeteria at bowl nom neg tte say:pst when ‘(You know) when I said I couldn’t find my bowl in the cafeteria the other day,’ 3

sora mottekitekureta yatsu sa, interj bring:come:give fellow ptcl ‘It’s the guy who brought it (for me),’

4

sonotoki kimi o mite shikkei tte ittaroo then you acc look excuse me tte say:pst:tag ‘Then (he) looked at you and said “Excuse me” right?’

5 N: un wakatta aitsu wa tazaki yes see:pst that.guy top name ‘Yeah, I got it. His name is Tazaki.’

(Satsuki no koi (1888))

Line 2 is a temporal clause, within which we find the reported thought in tte construction followed by the verb of saying. Line 4 is an instance of independent clause, including the reported speech (shikkei ‘excuse me’) in tte construction, followed by the verb of saying. Therefore, the first type of construction with tte can be schematized as in (5):

(5)

Original speaker + [utterance +tte] + main clause with a ‘say’ verb (Zero=Tazaki) shikkei tte itta

In some cases, what gets introduced by tte is not the exact utterance, but rather a summary of someone’s utterance, as in Example (6) below. Bunzoo lives in his

 Ryoko Suzuki

uncle Magobei’s house and is fond of Magobei’s daughter, Osei. Osei’s friendly attitude and words confuse Bunzoo. (6–1) The night before, Bunzoo was about to confess to Osei that he loves her, saying:6 Bunzoo: oseisan, anata mo anmari da, anmari...zankoku da, name you emph terrible cop too.much cruel cop watashi ga kore... korehodomadeni... I nom this this.much ‘Osei-san, you are terrible (to do this to me). (You) are too... cruel. (You know) I (like you) so much...’ Bunzoo cannot finish what he wanted to tell her and runs out of the room. The following morning, Bunzoo feels shy and avoids talking to Osei, but she comes to his room upstairs and asks him: (6-2) → Osei: watashi ga zankoku da tte I nom cruel cop tte taihen okotteirashitta ga, very angry.stat:pol:pst but nani ga zankoku desu no. what nom cruel cop ptcl ‘(You) were very angry (with me) (saying) that I am cruel, but what (about me) is cruel?’ (Ukigumo (1887)) Example (6–2) is different from Example (4), because watashi ga zankoku da ‘I am cruel’, the part preceding tte, is taken as Osei’s summary of what Bunzoo said about her, rather than reporting the utterance made by Bunzoo the night before shown in (6–1). Notice in Bunzoo’s “original” utterance in (6–1), he uses the second person pronoun anata towards Osei. So Osei’s closer-to-authentic reproduction of Bunzoo’s utterance would look like “anata ga zankokuda” tte taihen okotteirashita ((You) were angry (with me) (saying) “you are cruel”). In Osei’s utterance in (6–2), the personal pronoun watashi ‘I’ turns the entire clause (watashi ga zankoku da ‘I am cruel’) into a description. Tte in Example (6–2) thus can be viewed as forming an adverbial clause of manner, describing the way in which Bunzoo was angry. Example (6–2) shows that the function of tte as a quotative expands to cover not only reported speech but also description based on someone’s utterance, thus more indirect in nature. The ‘main verb’ iu ‘say’ is missing but the meaning ‘say’ persists in tte itself, and at the same time, the range of main verbs following the adverbial tte clause expands: It is not limited to verbs of saying, but includes verbs describing mental states, such as okoru ‘get angry/upset’, yorokobu ‘be pleased’, ochikomu ‘get 6. The descriptive part is a summary by me. The conversational parts are from the original novel.

The emergence of quotative constructions in Japanese 

depressed’ and so on. The following is a schematization of (6), as a more general use of tte in an adverbial clause:

(7)

4.2

Original speaker + [summary of utterance + tte] + main clause 0 (you=Bunzoo) watashi ga zankoku da tte okotterashita ‘(You) were angry (with me) (saying) that I am cruel.’

Tte clause as independent clause: Quotative tte in utterance-final positions

Let us now turn to the instances of tte found in utterance-final positions. In the early data of the 1830s and 1880s, there are three types of constructions in which the quotative tte occurs in utterance-final position. 4.2.1 Type A [Utterance + tte]: The 2nd pair part of a question-answer sequence Type A is an instance of an utterance-final tte in a question-answer sequence. Among the total of four tokens in the data from the 1830’s, i.e., the earliest instances of tte in the written conversational data, three appear in response to a question. Observe (8):

(8) Conversation between Yoiyoi, who has palsy, and a young man in front of the public bath. When asked how he is, Yoiyoi replies that his legs are fine. He says he ran to his aunt’s place to help her when there was a fire in the area.

1 Yoiyoi: hataetaa. hataetaa. omotama hataetaa. work:pst work:pst a.lot work:pst obatan ometa, obatan hometaa.7 aunt praise.pst aunt praise.pst ‘(I) worked and worked. Auntie was thankful. Auntie was thankful.’ 2 Young man: nante tte hometa. what say praise ‘What did she say to thank/praise (you)?’ 3→ Yoiyoi:

ee dejobu da tte dejobu da tte. fil brave.man cop tte brave.man cop tte ‘“brave man” “brave man” (she said).

4

sunjaa sankinokonpiyatamee, konpiyatamee then place.name place.name

7. Yoiyoi’s words are repetitive and his pronunciation is childlike. This is the author (Shikitei Sanba)’s way of portraying Yoiyoi’s palsy.

 Ryoko Suzuki

5→ odeme ni ika, ikaa, ikadee tte thank.visit to go tte ‘“(We should) go to Konpira Shrine in Sanuki to thank God” (she said.)’ (Ukiyoburo (1809)) As indicated in the arrowed lines 3 and 5, the aunt’s words are quoted and accompanied by tte. Notice that in line 2, the main verb itte appears as tte with the initial vowel contracted. 8 Therefore, the second pair part, i.e., Yoiyoi’s answer, does not contain the main verb ‘say,’ but it is clear that Yoiyoi’s answer refers to what the aunt said. These are the earliest instances of the utterance-final reportative use of tte. Let us consider another instance from a novel from the late 1800s: (9) Mother: hakama wa itsu dekiagaru to ittai? hakama top when ready quo say:pst ‘Did (the kimono shop) say when the hakama (=formal skirt) will be ready?’ Oyuki: asatte wa zehi the.day.after.tomorrow top definitely → dekiagarimasu tte... ready tte “The day after tomorrow (it) will definitely be ready (the shop said).” (Otomegokoro (1889)) The tte construction in Oyuki’s utterance constitutes the response in this questionanswer sequence. The final tte in Oyuki’s utterance introduces the utterance by the kimono shop personnel. The polite masu ending instead of the plain form dekiagaru contributes to the authenticity (‘verbatim-ness’) of the original utterance. The tte clause is not followed by any main verb because it is already expressed in the preceding question by Oyuki’s mother (i.e., ittai? ‘did (they) say’). It is interesting that the writer uses three dots to convey the impression of Oyuki’s utterance as including an implication left hanging. 4.2.2 Type B: Original Speaker + [Utterance + tte] The next type of utterance-final tte construction is shown in Example (10) below: (10) Elder Sis: kinchan nani o shite iru nda nee name what acc do stat nom:cop ptcl 8. Based on the Chinese character and kana characters alongside it indicating pronunciation used in this particular clause, I take tte here to be a contraction of the gerundive form itte of the verb iu ‘to say.’ I treat this tte as different from the quotative tte in the following lines by Yoiyoi, which is assigned a different Chinese character.

The emergence of quotative constructions in Japanese 

hayaku kikaete oshimai yo fast change.clothes finish ptcl

→ sooshite koto no osarai o ohajime and Japanese.harp gen practice acc pol:start tte okkasan ga ... tte mother nom ‘“Kin-chan, what are you doing? Change quickly and practice the harp,” Mom (says).’ Kinko: hai ima... yes now ‘Yes (I will).’ (Satsuki no koi (1888)) The characteristic of Type B is that we do not find any ‘main’ clause (including a ‘main’ verb) either preceding or following this dialog. However, in Type B, the original speaker is explicitly indicated (e.g., okkasan ‘mother’), or is easily identifiable from the previous discourse. Therefore, in this type, an instance of tte is used in order to frame the preceding portion as a reported utterance.9 4.2.3 Type C [Utterance + tte]: Immediate (partial) repetition of the interlocutor’s previous utterance In Types A and B, what was said plays an integral part in the independent clause ending with tte. As for Type C, not only what was said, but also how it was said by the interlocutor, seems to matter, since this is the most emotionally loaded type. Consider Example (11): (11) Heme: ohamasan wa, mada kimaranai kai. name top yet decide.neg q.ptcl ‘Hasn’t Ohamasan decided?’ → Denkichi: kimaranai tte. decide.neg tte ‘“(lit.) Hasn’t decided”?’/‘What do you mean by “Hasn’t decided”?’ Heme: dokoka e oyome ni ikinasaru n somewhere to bride to go.polite nmz

9. As it happens with many novels published around this period, the reported speech is not always bracketed with the quotation marks in the original. Accordingly, the beginning of the quoted part (i.e., mother’s utterance) is not clear in this excerpt (though the ending is clear because tte is expected right after). The reported utterance here could be the entire utterance preceding tte, or could be koto no osarai o ohajime ‘start the harp practice.’

 Ryoko Suzuki

jaa nai kane. cop:top neg ptcl ‘Isn’t (she) going somewhere to get married?’ (Hemeden (1895))

Beginning in the late 1800s, we find instances in which the speaker immediately repeats back the preceding speaker’s (i.e. the interlocutor’s) utterance. This often happens in contexts in which the speaker did not understand the meaning or import of an utterance (as in the above example), and is taken as a request directed to the prior speaker to elaborate his/her utterance. So in (11), Heme’s second utterance is a paraphrase of what he originally meant by kimaranai ‘hasn’t decided.’ Immediate repetition also occurs when the speaker does not understand the intention of the preceding speaker, or when the speaker immediately reacts with an expression of surprise having either an implication of ‘why-ask-the-obvious’ or of ‘non-acceptance/rejection’ of the preceding utterance (Suzuki R. 2007). In other words, Type C differs from Types A and B in that it is not a genuine, neutral quotative; rather, it seems to have developed into an independent construction. The evidence for this is that (a) there is no indication of the original speaker; (b) often, there is no verb of saying accompanying tte in the surrounding context; and (c) the construction is associated with particular types of speaker’s reactions, e.g., a request for an elaboration or expressions of surprise or rejection, which shape the following utterance, as we saw in (11), for example, where after Denkichi’s utterance with the final-tte, Heme elaborates his question. 4.2.4 Distribution of each type of tte We have seen that even the early data from the 1830s and the 1880s show that there are instances of utterance-final tte. In other words, the “full version” of the complementizer tte accompanied by a main clause (Examples 1 and 4 and Section 4.1) is not the only syntactic pattern that we find in the early data. We have seen that when reporting someone else’s (or the speaker’s own) prior utterance, tte can appear in utterance-final position in three types of syntactic environments. First, tte occurs utterance-finally as an answer to the information question ‘what did X say’ (Type A). Since the main verb of saying is explicit in the question, the response is not accompanied by a main clause. This type can be viewed as “elliptical.” Next, we find an utterance-final tte, accompanied by the originator of the utterance (Type B). This is directly related to the “reportive” final particle discussed in Okamoto (1995: 239). The origin of the information is not always explicitly stated. Lastly, an independent clause ending with tte is found when the speaker immediately repeats what the prior speaker has just uttered (Type C). The identity of the original speaker is clear, so it need not be mentioned. The verb of saying is not typically mentioned either, since the utterance itself is at issue.

The emergence of quotative constructions in Japanese 

Table 1.â•‡ The quotative tte constructions in the 1800s the 1830s

the 1880s

1

15

utterance-medial (dependent tte clause)

tte complement followed by the main clause

utterance-final (independent tte clause)

Type A 2nd pair part of a Q-A sequence (+verb of saying in 1st pair part)

3

â•⁄ 4

Type B introducing utterance (+original speaker indication)

0

11

Type C immediate repetition of the prior speaker’s utterance

0

13

Table 1 shows the token counts of dependent tte clause and independent tte clause. Clearly, for Types A through C (utterance-final tte without any following main clause), the 1880s data show greater variation in the types of utterance-final tte construction than the 1830s data: Given the small number of tokens from the 1800s, it would not be appropriate to make a direct linkage between the utterance-final uses in the early data and the utterance-final pragmatic particles that we see in the 20th century. However, these data clearly illustrate that utterance-final tte without any accompanying main clause, used with several different interactional motivations, is robustly found in the earliest data available. 5. Final remarks Introducing someone’s utterance to the addressee is an interactional act. The evolution of tte and the related constructions over the last 160 years show that tte does not simply mark an utterance as ‘reported speech,’ but rather indicates how the speaker frames, presents, and evaluates an utterance or thought for the addressee. Numerous studies in functional linguistics and conversation analysis show that participants produce utterances collaboratively in natural conversation. Through an investigation of quotative constructions involving tte, we now see that participants make use of multiple constructions using this one morpheme in a

 Ryoko Suzuki

variety of contexts. Hence the hypothesis that the tte clause as a dependent clause historically precedes the independent-clause counterpart is not supported by the data, and needs to be reconsidered. Looking at the relationship between the dependent-clause construction and the independent-clause construction (Types A through C), we could take a more cognitive approach: following Okamoto (1995) on tte, we could say that the main clause is omitted through reanalysis and conventionalization of speech act meanings, presumably through frequent use. From this perspective, the utterance-medial complementizer tte could be claimed to be reanalyzed as an utterance-final modality marker. In this study, I suggest a more interactionally oriented view: the interactional practices with which tte is associated result in tte occurring both in utterancemedial and utterance-final positions, even in the early stages of the emergence of tte. These specific communicative practices, such as offering information, clarification, asking for elaboration etc., are frequently associated with tte, and may feed into the emergence of utterance-final tte as an independent syntactic construction. I hope to have shown that the syntactic development of tte is not simply from a bound clause to an independent clause. Interactional practices, as well as the syntax involved in the practices, need to be carefully examined. The early data on tte is thus messier, richer and more interesting than expected. Acknowledgements My deepest gratitude goes to Ritva Laury for her constant support and encouragement. I am grateful to Pat Clancy and Sandy Thompson for their valuable discussions and comments to the earlier version of this paper, and to the panelists and the audience at the tenth International Pragmatics Conference in 2007 for their stimulating discussions. Thanks are also due to Maggie Camp for proofreading. None of them is responsible for any shortcomings of this paper. Abbreviations ACC COMP COP EMPH FIL GEN

accusative complementizer copula emphatic filler genitive

The emergence of quotative constructions in Japanese 

INTERJ NEG NMZ NOM POL PRG PST PTCL Q QUO STAT TOP

interjection negative nominalizer nominative polite progressive past particle question quotative stative topic

Text references 1809. Ukiyoburo, by Shikitei Sanba. Nihon koten bungaku taikei 63 (1957). Tokyo: Iwanami Shoten. 1887. Ukigumo, by Futabatei Shimei. Gendai nihon bungaku taikei 1 (1971). Tokyo: Chikuma Shoboo. 1888. Satsuki no koi, by Iwaya Sazanami. Meiji bungaku zenshuu 20 (1968). Tokyo: Chikuma Shoboo. 1889. Otomegokoro, by Ishibashi Shian. Meiji bungaku zenshuu 22 (1969). Tokyo: Chikuma Shoboo. 1895. Hemeden, by Hirotsu Ryuuroo. Gendai nihon bungaku taikei 3 (1970). Tokyo: Chikuma Shoboo.

References Barth-Weingarten, Dagmar and Couper-Kuhlen, Elizabeth. 2002. “On the development of final though: A case of grammaticalization?” In New Reflections on Grammaticalization, Ilse Wischer, and Gabriele Diewalds, (eds), 345–361. Amsterdam: John Benjamins. Bybee, Joan. 2003. “Mechanisms of change in grammaticization: The role of frequency.” In Handbook of Historical Linguistics, Richard D. Janda and Brian D. Joseph (eds), 602–623. Oxford: Blackwell. Couper-Kuhlen, Elizabeth and Ono, Tsuyoshi (eds). 2007. Turn Continuation in Cross-linguistic Perspective [Pragmatics 17.4]. Antwerp: International Pragmatics Association. Englebretson, Robert (ed.). 2007. Stancetaking in Discourse: Subjectivity, Evaluation, Interaction. Amsterdam: John Benjamins. Heine, Bernd and Kuteva, Tania. 2002. World Lexicon of Grammaticalization. Cambridge: Cambridge University Press.

 Ryoko Suzuki Higashiizumi, Yuko. 2006. From a Subordinate Clause to an Independent Clause: A History of English Because-clause and Japanese Kara-clause. Tokyo: Hituzi Shobo Publishing. Hopper, Paul J. and Traugott, Elizabeth C. 2003. Grammaticalization (2nd ed.). Cambridge: Cambridge University Press. Mulder, Jean and Thompson, Sandra A. 2006. “The grammaticization of but as a final particle in English conversation.” In Selected Papers from the 2005 Conference of the Australian Linguistic Society, K. Allan (ed.), http://www.als.asn.au/proceedings/als2005/mulderthompson.pdf. Mulder, Jean and Thompson, Sandra A. 2008. “The grammaticization of but as a final particle in English conversation.” In Ritva Laury (ed.), Crosslinguistic Studies of Clause Combining: the Multifunctionality of Conjunctions, 179–204. Amsterdam: John Benjamins. Nakayama, Toshihide and Ichihashi-Nakayama, Kumiko. 1997. “Japanese kedo: Discourse genre and grammaticization.” In Ho-min Song and John Haig (eds), Japanese/Korean Linguistics 6: 607–618. Stanford: CSLI Publications. Nihon Kokugo Daijiten (NKD) [The dictionary of the Japanese language] (2nd ed.). 1972–76. Tokyo: Shogakkan. Ochs, Elinor, Emanuel A. Schegloff, and Sandra A. Thompson (eds). 1996. Interaction and Grammar. Cambridge: Cambridge University Press. Ohori, Toshio. 1995. “Remarks on suspended clauses: A contribution to Japanese phraseology.” Essays in Semantics and Pragmatics: In Honor of Charles J. Fillmore, Masayoshi Shibatani and Sandra A. Thompson (eds), 201–218. Amsterdam: John Benjamins. Ohori, Toshio. 1997. “Framing effects in Japanese non-final clauses: Toward and optimal grammar-pragmatics interface.” Berkeley Linguistics Society 23: 471–480. Okamoto, Shigeko. 1995. “Pragmaticization of meaning in some sentence-final particles in Japanese.” Essays in Semantics and Pragmatics: In Honor of Charles J. Fillmore, Masayoshi Shibatani and Sandra A. Thompson (eds), 219–246. Amsterdam: John Benjamins. Okamoto, Shigeko and Ono, Tsuyoshi. 2008. “Quotative -tte in Japanese: Its multifaceted functions and degrees of ‘subordination.’ In Ritva Laury (ed.), Crosslinguistic Sudies of Clause Combining: the Multifunctionality of Conjunctions, 205–230. Amsterdam: John Benjamins. Selting, Margret and Couper-Kuhlen, Elizabeth (eds). 2001. Studies in Interactional Linguistics, 287–215. John Benjamins. Amsterdam. Suzuki, Ryoko. 1999a. “Mutifuncitonality: The developmental path of the quotative tte in Japanese.” Cognition and Function in Language, Barbara A. Fox, Dan Jurafsky, Laura A. Michaelis (eds). Stanford, CA: CSLI Publications, 50–64. Suzuki, Ryoko. 1999b. Grammaticization in Japanese: A Study of Pragmatic Particle-ization. Ph.D. dissertation, University of California, Santa Barbara. Suzuki, Ryoko. 2007. “(Inter)subjectification in the quotative tte in Japanese conversation: Local change, utterance-ness and verb-ness.” Journal of Historical Pragmatics 8(2): 207–237. Suzuki, Satoko. 1998. “Tte and nante: Markers of psychological distance in Japanese conversation.” Journal of Pragmatics 29: 429–462. Suzuki, Satoko. 2001. “Self-mockery in Japanese.” Linguistics 40(1): 163–189. Thompson, Sandra A. and Anthony Mulac. 1991. “A quantitative perspective on the grammaticization of epistemic parentheticals in English.” In Approaches to Grammaticalization, Elizabeth C. Traugott and Bernd Heine (eds), vol 2: 313–329. Amsterdam: John Benjamins. Traugott, Elizabeth C. and Richard B. Dasher. 2005. Regularity in Semantic Change. Cambridge: Cambridge University Press.

Clines of subordination – constructions with the German ‘complementtaking predicate’ glauben* Wolfgang Imo The German “complement-taking predicate” (Thompson/Mulac 1984, 1991) glauben occurs in a number of different syntactic constructions. For example, it can be realized as part of a matrix clause followed by a subordinate clause with or without a complementizer. As the term “complement-taking predicate” suggests, these matrix constructions should be expected to be the standard ones with glauben. An empirical study of the uses of glauben in spoken German reveals, though, that the majority of all cases involving glauben are hybrid constructions resembling modal particles or adverbs. With those instances of glauben, a new construction is about to be grammaticalized into a kind of epistemic qualifier. The process of this reanalysis is not yet finished, resulting in constructions that cannot be allocated to any conventional category such as matrix clause or modal particle.

1. Introduction The German ‘complement-taking predicate’ (Thompson/Mulac 1984, 1991) glauben (‘to believe’ / ‘to think’) can be used in a wide variety of syntactic constructions. For example, it can be realized within a matrix clause which is followed by a complementizer and a subordinate clause. According to valence-based entries in dictionaries (e.g. the “Valenzwörterbuch deutscher Verben” (Schumacher et al. 2004: 429f)), this type of construction is usually regarded as the standard one for German complement-taking predicates. The following transcript (which is transcribed according to the GAT-conventions developed by Selting et al. (1998)) illustrates precisely such a case where the verb glauben is used within a matrix clause which is then followed by the complementizer dass (‘that’) and a subordinate clause. The conversation between two friends (W and H) is taking place in a car while they are visiting their friends

 Wolfgang Imo

on Christmas Eve (because of the length of the transcripts I will only give a condensed English translation that tries to stay as closely to the original as possible): (1) Example 1 private conversation in a car 2056 W deine ELtern gehen nich in die kIrche? your parents don’t go to church? 2057 H NE:. no:. 2058 (0.5) 2059 W .hh des WUNdert mich irgendwie. I’m surprised to hear that. 2060 (1.0) 2061 H NE:. no:. 2062 machen se NET, they don’t do it, 2063 → wobei ich nich glaub (.) dass ses aus überZEUgung nich machen sondern weils halt dann halt (noch) zu wenig lUscht oder (.) zu fAUl oder sOnschtwas sin; though I don’t think (.) that they don’t do it because they are convinced of not going but because in the end they’ve too little interest or they are too lazy or whatever; 2064 also sie sin schon potentielle .hh WEIHnachtskirchenâ•‰ kandidaten. well in fact they are potential .hh Christmas-church candidates. After W expresses his surprise at H’s statement that his parents do not go to church on Christmas (line 2059), H starts to provide an explanation for his claim. In line 2063 he does not credit his parents with a conscious decision against going to church but claims the reason is just laziness. H introduces his utterance by a matrix clause (wobei ich nich glaub ‘though I don’t think’), which places the following subordinate utterance under the modalizing scope of glauben. Some remarks on the concept of subordination are needed in order to understand German syntactic patterns. Subordination refers to the fact that one clause syntactically relies on another. In German, there are two main ways of marking subordination, which are usually combined: the use of a subordinating conjunction and the placement of the finite verb at the end of the subordinate clause. The latter structure is reserved for subordinate clauses only (except for a rather rare type of optative sentences, which cannot be combined with matrix clauses, such as Wenn es doch morgen nur regnen würde. lit.: ‘If it tomorrow only rain would.’) and is therefore a very strong marker of subordination. Many

Clines of subordination 

subordinating conjunctions in German (such as weil ‘because’ or obwohl ‘although’) can also occur with a main clause structure (finite verb in second position) and are much weaker markers of subordination. Freywald (2008) even showed that it is not uncommon to use the subordinating conjunction dass (‘that’) together with a verbsecond main clause in spoken German, a construction nobody had thought probable before. So there is a choice of the following structures: a. Main clause plus subjunction plus subordinate clause: Ich glaube, dass es heute regnen wird. lit.: ‘I think that it today rain will.’ b. Main clause plus subjunction plus main clause: Ich glaube, dass es wird heute regnen. lit.: ‘I think that it will today rain.’ c. Main clause plus main clause: Ich glaube, es wird heute regnen. lit.: ‚I think it will today rain.’ In example c, the concept of subordination already becomes problematic because syntactically, the main clause is no longer dependent on the matrix clause. Nevertheless, there is some syntactic connection left, which is revealed by a reformulation into the matrix clause-plus-subordinate clause structure, which does not change the meaning of the sentence. Auer (1998) proposes the term abhängiger Hauptsatz (‘dependent main clause’) for these structures to highlight their oscillation between dependence and autonomy. To return to the discussion of Example 1, the subordinating syntactic structure has the effect of marking the following explanation of H merely as a surmise, not as certain knowledge. By using the constructional pattern of placing a subordinate clause under the scope of a matrix clause containing glauben, H hedges his utterance as being based on his personal and possibly biased opinion rather than on certain knowledge. This way of using glauben is quite typical for written German (exempting computer-mediated communication), where the prescriptive effect of grammars has a strong hold. Data from spoken German quickly reveal, though, that matrix clauses with complementizers are not often used in actual spoken interaction. In the data I analyzed (see Section 2) there occurred 169 instances of glauben, of which a mere 22 were realized together with a subordinate clause (for a more detailed analysis of the different constructions glauben is used with, see Imo 2006 and 2007a). The example cited above actually shows the only common way in spoken German to use glauben as part of a matrix clause which is then followed by a subordinate clause introduced by a complementizer. This structure is used whenever the matrix clause containing glauben is negated (see Auer 1998). According to Auer (1998: 293f) this can be explained by the fact that when the matrix clause is negated, it is asserted and highlighted relative to the presupposed subordinate clause. In other words: When a matrix clause is negated, according to Auer (1998), it becomes more salient than the following subordinate clause. If, on the other hand, the matrix clause is not negated,

 Wolfgang Imo

it is usually treated as less salient and important and thus as presupposed relative to the following (subordinate) clause. The combination of matrix clause and subordinate clause therefore always codes relative levels of saliency. In most of the cases where (positive) matrix clauses with glauben are produced in spoken German, they are realized with a following main clause (as Auer’s (1998) and Imo’s (2007a) data show), indicating that there is little need in spoken interaction for a construction that places a high level of salience on a matrix clause. While Auer (1998) uses the term “salience” to refer to the differences between the construction of a matrix clause followed by a subordinate clause (e.g. Ich glaube, dass es heute regnen wird. lit.: ‘I believe that it today rain will’), in comparison to a matrix clause followed by a main clause (Ich glaube, es wird heute regnen. lit.: ‘I believe it will today rain’), I would prefer the concept of profile determinacy (Langacker 1987: 288) to capture the phenomenon. For complex constructions, Langacker assumes that one component usually carries the main information (or meaning). This is called the “profile”, which is “maximally prominent and can be thought of as a kind of focal point” of a predication. In complex constructions, profiles are inherited: For the most part, a composite structure simply inherits the profile of one of its components. The component structure whose profile is inherited will be termed the profile determinant of the construction. (Langacker 1987: 289)

If someone utters a sentence such as Ich glaube, dass es heute regnen wird. (lit.: ‘I think that it today rain will’), the profile determinant should be Ich glaube (‘I think’), while in Ich glaube, es wird heute regnen. (lit.: ‘I think it will today rain’), the profile determinant should be es wird heute regnen (‘it will today rain’). In Langacker’s (1991: 436) terms, a subordinate clause is “describable as one whose profile is overridden by that of a main clause”. The analysis in terms of profile determinacy allows for a less ambiguous and more universal concept than the rather vague notion of salience, as it can draw on Langacker’s (1987) elaborate application of the idea, including a classification of profile types and inheritance relations. Langacker (1987: 292) especially stresses that the identification of a profile determinant is by no means a mechanical process but depends on a lot of factors – contextual information above all:1 As a general matter, it is somewhat inaccurate to regard the composite structure as being constructed out of its components [...]. It is more appropriate to say that the component structures motivate aspects of the composite structure, and that the degree of motivation is variable.

For example, the more formal a speech situation becomes, the more likely it will be that speakers use a subordinate structure instead of a main clause following the 1. For a newer and more detailed analysis with a special focus on subordination in Cognitive Grammar, see Langacker (2009: 327–340).

Clines of subordination 

matrix clause (see Imo 2007a). This context-dependent variability of syntactic constructions ties in with conceptualizations of grammar as an emergent phenomenon, as proposed by Hopper (e.g. 1998). In his view, grammar is not a more or less permanently fixed code but is shaped and re-shaped continually to fit into specific – and always changing – contexts according to the needs of the interactants. The phrases glaub(e) ich2 and ich glaub(e) provide a good illustration of this variability or emergence of grammar. It is almost impossible to provide fixed constructions or patterns that can account for all of the instances of glauben, because in many cases, as will be shown, several constructions seem to overlap and be activated at the same time. From the point of view of a dialogically oriented grammarian, phrases with glauben can therefore be treated as evidence for Hopper’s (2004: 1) assumption of open grammatical constructions: Grammar is an epiphenomenon of frequent combinations of constructions. Because grammar is a result of interactions rather than a prerequisite to them, it is not a fixed code but is caught up in a continual process of local adaption (emergence).

In the following analysis I will try to show that phrases with glauben are indeed an example of a fragmentary grammar in which profile structures as well as constructions are on a flow to create all kinds of instances of amalgamated uses. Matrix clauses, discourse markers, modal particles, modal adverbs and increments all contribute aspects of meaning and function that are needed for the interactants to make sense of actual utterances containing the verb glauben. 2. The data The data that provide the basis for this analysis consist of approximately 30 hours of spoken German including radio phone-in programmes, radio interviews and informal conversations between family members and friends (mostly students). The data cover different German-speaking regions (Brandenburg, North Rhine-Westphalia, Thuringia, Bavaria, Baden-Wuerttemberg, Austria and Switzerland) and are part of the database of spoken German of Susanne Günthner (University of Münster). They were transcribed according to the GAT conventions (Selting et al. 1998). The audio material was searched for all forms of the verb glauben except those with the semantic content of “believing in God”. Although glauben literally has to be translated with to believe, it is often used as an equivalent of to think, in spite of the fact that there is another German verb, denken, which also means to think. Because of this ambiguity I will translate glauben with to believe/think. 2. In spoken German, the verb-final schwa marking the 1st person is usually not pronounced.

 Wolfgang Imo

The aim of my search was to give a full description of all of the constructions glauben is used with in ordinary everyday talk (for a comprehensive analysis see Imo 2007a). The total sum of all phrases with the verb glauben amounted to 169 instances. The following table gives a ranking oriented to aspects of the temporal and personal deixis of glauben. The verb glauben most often occurs in the 1st person singular present tense (156 of 169 cases, i.e. more than 92%). This is the usual distribution of many verba sentiendi et dicendi in spoken German (Imo 2007a), which usually refer to the deictic centre of I-here-now. Furthermore, the 1st person singular present tense phrases ich glaub(e) (‘I think’) and the inverted form glaub(e) ich (lit.: ‘think I’) occur so often and with such special functions that they can be said to have an almost formula-like quality (see also Thompson/Mulac (1984, 1991) for a more general discussion of this quality of complement-taking predicates (CTPs)). While the phrase ich glaub(e) – with the predicate following the subject – is most common, a very substantial number of instances are realized in the inverted form glaub(e) ich (lit.: ‘think I’), with the subject following the predicate. When the order of predicate and subject is inverted in German, such a phrase cannot be placed in front of an utterance but only within or after utterances. There is a division of work between utterance-initial phrases consisting of subject and verb (Ich glaub das wird nichts. ‘I think/believe that won’t work’) and utterance-medial and final phrases consisting of verb and subject (Das wird glaub ich nichts. lit.: ‘That won’t believe I work’ and Das wird nichts glaub ich. lit.: ‘That won’t work believe I’). Whether it is still possible to analyze the phrase with glauben as having a subordinate clause within its scope when it is embedded in the subordinate clause or follows it will be discussed in Section 4. In the following section I will first look at instances of the non-inverted form ich glaub(e) (‘I think’), asking whether they can indeed be classified as matrix clauses or, rather, as some kind of “markers” or “CTP phrases” in the terminology of Thompson/Mulac (1984, 1991; see also Thompson (2002) and Thompson/Hopper (2008)). Table 1.â•‡ Distribution of glauben in terms of deixis 1st person singular present tense (ich glaub(e) ‘I think’) 1st person singular present tense (inversion: glaub(e) ich ‘think I’) glaub (‘think’) without a personal pronoun (i.e. subject) others (2nd and 3rd persons, plural forms, future and past forms)

78 71 â•⁄ 7 13

Clines of subordination 

3. From matrix clause to discourse marker In his study of matrix clauses, Auer (1998) argues that the difference between whether a main clause or a subordinate clause follows the matrix clause can be explained in terms of salience. In Langacker’s (1987) terms, the profile determinant in these cases is changed from the matrix clause towards the dependent clause. The following two examples show typical uses of glauben with a following main clause in spoken German. The changed profile structure of the matrix clause immediately becomes obvious in these two instances. Example 2 is taken from a private family conversation (the daughter T has just discovered ants crawling around, which is a very unexpected event as the conversation is taking place in winter). In Example 3, two student friends are designing an invitation to a barbecue party on their computer, trying to work out the directions for getting to the location of the party. (2) Example 2 family conversation 47 T Ameisen im winter. ants in winter. 48 → ich glaub ich SPINN. I think I’m crazy. 49 (3.0) (3) Example 3 private conversation between students 253 A wie schreibt man eigentlich torMINbrücke; (1.0) how does one spell Tormin-Bridge; 254 B ich kenn den ausdruck erst seitdem du den verwendest. I know this expression only after you’ve started to use it. 255 A → ich glaub ich guck mal aufm STADT[plan. ] I think I will look it up on the map. 256 B [ich hab] hier n STADTplan; I’ve got a map here; In the data I searched I could find very few instances of matrix clauses with glauben followed by an independent main clause where the transformation of the post-positioned main clause into a subordinate clause (involving a subordinate conjunction and the obligatory shift of the finite verb from the second position typical for German main clauses to the final position necessary for subordinate clauses) would be unproblematic. In other words, it is quite uncommon that the difference between the construction [matrix clause + subordinate clause] and [matrix clause + main

 Wolfgang Imo

clause]3 only rests in the higher or lower profile of the matrix clause in relation to the following (subordinate or main) clause. Usually, such a transformation would seem decidedly odd or would even create a wrong or absurd meaning, i.e. involving not just a shift in perceptual profile but also in the semantics of the utterance. Examples 2 and 3 illustrate this point: If one were to transform the utterances in lines 48 and 255 into matrix clauses with following subordinate clauses (Ich glaube, dass ich spinne. ‘I believe/think that I am crazy’; Ich glaube, dass ich auf der Karte nachschaue. ‘I believe/think that I will look it up on the map’), the matrix clause would become much too prominent in German, a consequence of being marked as the profile determinant of the combined clause. Both cases illustrated above are typical for the uses of glauben in spoken interaction insofar as the matrix clause usually does not carry much information on a presuppositional basis. In most instances, speakers do not explicitly state what they think or believe (i.e., they do not place much stress on the matrix clause) but rather use short phrases to modalize one of their utterances and mark it as an exaggeration, a proposal, an offer or a conjecture. In Example 2, the speaker expresses her astonishment at seeing living ants in winter. She is not crazy, nor does she believe she is. The phrase “ich glaub ich SPINN” (‘I think I’m crazy’) in Example 2, line 48, is in fact a conventionalized and routinized formula for expressing disbelief or sometimes disbelief coupled with anger. The fact that such a construction [matrix clause + following main clause]4 has become sedimented into an idiomatic expression is revealing because it shows that matrix constructions are open to reinterpretation and can easily be used as building blocks to construct phrases which are working on an interactional level, expressing – as is the case here – speaker stance, rather than being used with a full propositional content. In Example 3, the combination of the phrase with glauben and the following clause does not constitute a fixed formula. Nevertheless, it, too, cannot be interpreted in such a way that glauben carries a full semantic load. Instead, “ich glaub” (line 255) serves as a kind of pragmatic marker or discourse marker,5 framing the 3. To facilitate the recognition of complex constructional patterns in the text I will place all constructions (ranging from word to sentence level) in square brackets. For a discussion of different types of constructions see Croft (2002: 17). 4. In a broad sense, the first part can still be interpreted as a matrix clause because ich glaub (‘I believe/think’) is not complete on its own and it is still possible to ask after the complement, as in What do you believe?. 5. A number of terms have been proposed for those small words or phrases which are placed in front of an utterance and have pragmatic (i.e. textual or interpersonal) functions: “Pragmatic markers” (Brinton 1998), “discourse markers/Diskursmarker” (Auer/Günthner 2005, Fraser 1990, Gohl/Günthner 1999, Lenk 1998, Schiffrin 1987), “operators” (Barden 1998, Barden/Elstermann/Fiehler 2001), “pragmatic expressions” (Erman 1987), “discourse particles” (Kroon 1995, 1998, Schourup 1985) and “pre-front-field expressions” (Thim-Mabrey 1988).

Clines of subordination 

following clause as a proposal or an offer. Speaker A cannot plausibly use glauben in the context of Example 3 to describe an action she is going to undertake because she knows what she will do next. The phrase rather serves as a “gambit” which acts as a “discourse lubricant” in the terms of Edmondson/House (1981: 69). If one were to transform the main clause into a subordinate clause, the quality of the “gambit” would be lost in favour of the full semantic meaning of glauben (‘to believe’), which would then imply that the speaker utters a belief about her next action – which would clearly sound absurd and would imply that the speaker has no control over her own intentions and actions. The two examples presented here show that while matrix clauses with following subordinate clauses and those with following main clauses are somehow related, it is not possible simply to transform Examples 2 and 3 into matrix clauses with following subordinate clauses. Elmauer (1973: 169f), for example, claims that in German it is always possible to transform a matrix clause with a following main clause into a matrix clause with a following subordinate clause, using dass (‘that’) or ob (‘if ’) as subordinating conjunctions. In much the same vein, Helbig/Buscha (2001: 566; my translation) state: “Subordinate clauses without complementizers are equivalent to those with complementizers and the former can be transformed into the latter by eliminating the conjunction and changing the position of the finite verb”. Auer (1998: 285; my translation) is more cautious about categorizing what he calls dependent main clauses as object complements of verbs, he merely treats them as “object clauses belonging to verbs which are not saturated in terms of their valence”. With this position he avoids problematic claims about the possibilities of interchanging the constructional patterns of [matrix clause + main clause] and [matrix clause + subordinate clause]. Neither Ich glaube, dass ich spinne (‘I think/believe that I am crazy’) nor Ich glaube, dass ich mal auf dem Stadtplan gucke (‘I think/believe that I will look it up on the map’) make much sense and both would yield a strikingly different meaning from the actual utterances of the speakers in Examples 2 and 3. 4. From matrix clause to increment In the previous section, phrases with glauben were analyzed which are placed utterance-initially and follow the internal constructional schema [subject(ich) + verb(glaube)]. Almost as common as this pattern is the constructional schema [verb(glaube) + subject(ich)], where the order of subject and predicate is inverted. Phrases following this pattern can be used either within utterances (where they occur most often) or at the end of an utterance. The following three examples illustrate varieties of utterance-final placement of glaub(e) ich.

 Wolfgang Imo

Example 4 is taken from a radio interview. In the transcript presented here, the radio host interrupts his questioning routine because he thinks that the mobile phone of one of his guests is ringing. Example 5 is taken from the popular German radio phone-in talk show “Domian”. A caller is chatting with the host Domian (D) about his experiences as a porn actor. Example 6 is taken from a private conversation between university students. Speaker S is describing his problems finding both a suitable topic and enough material for an essay he has to write. (4) Example 4 radio interview 18 N eh herr KOLLmanntaler, .hh erm Mr. Kollmanntaler, .hh 19 ehm (0.5) die eh erm (0.5) the erm 20 aber das IS irgendwas; but that is something; 21 n geRÄUSCH, a sound, 22 aus ihrem (.) moBILtelefon; from your mobile phone; 23 → glaub ich. I think. (lit.: think I) (5) Example 5 radio phone-in “Domian” 91 D also ist der begriff GANG bang schon gut getroffen, so the term gang bang fits quite well, 92 ne, doesn’t it, 93 wenn von mir if from me 94 A BITte? sorry? 95 D ich SAGte dann ist der begriff gang bang → SCHON ganz gut getroffen glaub ich. I said then the term gang bang fits quite well I think. (lit.: think I) (6) Example 6 private conversation between students 886 S und da hätt ich aber da hab ich nur zwei textstellen geHABT and there I would have had only two quotable passages 687 und da wärs das wär so SCHWIErig geworden → glaub ich and it would have become so difficult I think (lit.: think I)-

Clines of subordination 

688

da noch mehr zu FINden; to find more of these;

In Example 4, the speaker’s utterance is possibly complete in line 22 after “mobile phone” because there is no syntactic or semantic projection (Auer 2002, 2007b, 2008) left open which might project the glaub ich in line 23. Therefore, glaub ich can be analyzed here as occupying the post-field of the sentence starting in line 18. In terms of prosodic realization, glaub ich is realized in a separate tone contour6 apart from the previous utterance. The marked prosodic realization increases the profile of glaub ich so that it appears as an increment. Following Ford et al. (2002: 16), an increment can be described as a nonmain-clause continuation after a possible point of turn completion. That is, an increment will be defined here as any nonmain-clause continuation of a speaker’s turn after that speaker has come to what could have been a completion point, or a ‘transition-relevance place’, based on prosody, syntax and sequential action.7

The phrase glaub ich in Example 4 cannot be classified as a typical main clause, because it is not clear whether the preceding sentence can be interpreted as its complement clause, i.e. whether there is a relation of super- and subordination at work here at all. Such a classification would be very artificial. Especially in those cases where the preceding utterance ends on a falling tone (as in Example 4, line 22), the expression glaub ich acquires an independent status, i.e. it appears as a unit of its own with little syntactic relation to the preceding utterance. Therefore, it is not a matrix clause but meets Ford et al.’s characterization of a special type of increment. According to Ford et al.’s (2002: 18) typology of subclasses of increments, post-positioned glaub(e) ich can be classified broadly as an extension:8 6. The term tone contour will be used as an equivalent to “Intonationsphrase” in Selting et al. (2009), where it is defined as a coherent unit where the final pitch movement is the most prominent marker of the end of the tone contour, which contains at least one accent and which – optionally – can have additional markers of its initial and final borders (creaky voice, pauses, lengthening, pitch jumps and others). 7. It is not always clear whether all of these conditions have to meet simultaneously. In Examples 5 and 6, there is no prosodic break but the non-finite verb (which forms the right verb bracket in German) provides a very strong syntactic signal of completeness. 8. For a newer and very detailed analysis of increments see Couper-Kuhlen/Ono (2007). While Ford et al. (2002) use the concept of the “transition-relevance place” as the main feature of their classification, Couper-Kuhlen/Ono (2007: 515) differentiate between “non-add-ons”, which are placed after a syntactic closure but without a prosodic break, and “glue-ons”, which fit to the previous TCU syntactically but are set off prosodically by pauses, a new onset or accents. I will use the broader classification suggested by Ford et al. because the syntactic status of postpositioned glaub(e) ich is unclear and the examples show that it can both be prosodically latched onto the previous utterance (Examples 5 and 6) or set off (Example 4). See also Auer (2006; 2007a)

 Wolfgang Imo

An extension is a short syntactic unit with the function of repairing a previous utterance before any misunderstandings can occur. It can repair possible problems of understanding, possible threat of face, and it can mark the utterance as not based on certain knowledge.

If post-positioned glaub(e) ich is indeed to be classified as an increment,9 this would imply that the previous utterance to which it is pragmatically – but not syntactically – attached is complete by itself and the phrase glaub(e) ich then has to be recategorized as a modalizing phrase or post-positioned discourse marker. In Langacker’s (1987) terms, this would mean that there occurs a change in profile determinacy. It is not the phrase with glauben that is the profile determinant, but rather the preceding utterance. However, exactly as in those cases where a prepositioned ich glaub(e) is to be classified as an utterance-initial discourse marker, the change of profile determinacy is by no means complete. Both discourse markers and extensions have a status somewhere outside the canonical sentence structure and operate between syntactic disintegration and pragmatic integration. In the case of the post-positioned glaub ich in Example 4, its function is clearly to mark the previous utterance as a conjecture and to save the face of the interviewee who forgot to turn off his mobile phone. So far, the description of postpositioned glaub(e) ich as a conventionalized increment to tone down utterances seems to be pretty convincing. Example 5 shows, though, that this classification cannot be held up without problems. Here, the phrase with glauben is latched directly onto the previous utterance and is realized within one single tone contour: “ich SAGte dann ist der begriff gang bang SCHON ganz gut getroffen glaub ich” lit.: ‘I said then the term gang bang fits quite well think I’ (Example 5, line 95). The character of an increment is lost by this process of latching because increments require some autonomous prosodic status – at least according to Ford et al. (2002: 16). Furthermore, the phrase with glauben loses even more of its salience and becomes part of the ordinary sentence structure, just like any other part of speech (e.g. an adverb).

for a discussion of the as-yet unclear status both of syntactic and prosodic features that should be used to determine the borders of utterances and, thus, the classes of increments. 9. It is important to note that the definition of increments is a highly contested area. While Couper-Kuhlen/Ono (2007) insist that only when a prosodic break occurs does the category increment apply, Auer (2007a) claims that prosody can be overridden in some types of increments. This is because in German the right verb bracket is such a strong marker of syntactic completeness that few constituents are expected after it. Therefore, a phrase such as glaub ich is marked strongly enough as being out of place by the preceding closure of syntactic structure and does not need prosody to become marked as an increment.

Clines of subordination 

In Example 6 the integration of glaub(e) ich into the preceding utterance is even stronger. Here, glaub ich (line 687) is uttered between the right verb bracket (finite verb) “geworden” (‘have become’) and the post-field (“da noch mehr zu FINden” ‘to find more of these’): “und da wärs das wär so SCHWIErig geworden glaub ich- da noch mehr zu FINden;” (lit.: ‘and there would that would so difficult have become think I- there even more to find’). Not only is the phrase with glauben prosodically latched onto the preceding clause, there is also a strong aspect of syntactic integration of glaub ich into the surrounding clause, which does not stop after glaub ich. The phrase is actually integrated within a clause, not just preceding or following it. This integration is responsible for the perceptory effect that glaub ich appears more like an adverb or modal particle and not like a discourse marker. In order to interpret the heterogeneous phrases with glauben that have been presented so far, all relevant linguistic levels have to be taken into account, ranging from syntactic features via semantic and pragmatic ones to prosody.10 On the level of meaning, there is a slight tendency of glauben to carry a full semantic content, i.e. there is much less “semantic bleaching” than with other pragmaticalized phrases (see Auer/Günthner 2005). Exceptions are provided by Examples 2 and 3 discussed above, where the semantic content is reduced in favour of a new semantic meaning of the whole fixed expression Ich glaub ich spinn. ‘I think/believe I am crazy’ (Example 2) or reduced because of contextual factors making the interpretation of glauben as believing highly improbable (Ich glaub ich guck mal aufm Stadtplan. ‘I think/believe I will look it up on the map’). So, in terms of semantics there is no certain proof as to a change of category of the phrases with glauben. Syntactically, it would be possible to transform all sentences into matrix clauses with following subordinate clauses. Such a transformation, though, would change the profiles of both constructions ([matrix clause followed by a subordinate clause] and [independent clause followed by the phrase glaub(e) ich]) so strongly that meanings would be created that might even be absurd in given contexts. On a prosodic level, the phrase with glauben can either be uttered within an independent tone contour or it can be latched onto or integrated into the previous utterance. In the first case, glaub(e) ich is profiled and has the character of an increment. In the second case it is back-grounded and appears more like an adverb or modal particle, i.e. like any other part of speech embedded into a sentence. All of these factors taken together lead to the assumption of a gliding scale between instances of glaub(e) ich which are placed in an utterance-final position (and can be called 10. A description covering a broad range of linguistic and extra-linguistic features is needed to describe constructions with glauben: “I do not believe that grammar, in all its actual complexity, is coherently describable independently from processing and discourse, or that those are neatly distinct from another. I believe we need an integrated account of all three domains for any of them to be properly understood.” (Langacker 2009: 340)

 Wolfgang Imo

increments) and those which are embedded into the utterance (and behave more like adverbs or modal particles). This gliding scale corresponds to the change in profile (see also Langacker 2009: 337 for “transitional” cases of subordinate structures) – which is determined by prosodic, semantic, syntactic and pragmatic features – from the phrase with glauben towards the surrounding utterance it is pragmatically associated with (from top to bottom): phrase with glauben is profiled

ich glaub(e) + subordinate clause ich glaub(e) = matrix clause

ich glaub(e) + main clause / utterance ich glaub(e) = matrix clause or discourse marker

utterance + glaub(e) ich (prosodically separated) glaub(e) ich = incremental discourse marker

utterance + glaub(e) ich (prosodically integrated, uttered after the main clause) glaub(e) ich = incremental discourse marker, adverb or modal particle

utterance + glaub(e) ich surrounding utterance is profiled

(prosodically and syntactically integrated, i.e. realized within the main clause) glaub(e) ich = adverb or modal particle

Schema 1.â•‡ From subordinating to embedded structure

Clines of subordination 

As the last entry suggests, glaub(e) ich can actually be integrated completely within an utterance. Indeed, it is uttered much more often within the so-called “middle field” of a sentence (i.e. within an utterance, between the brackets formed by the finite and non-finite verb parts) than after an utterance. 5. From matrix clause to modal particle The majority of cases following the constructional schema [verb(glaube) + subject(ich)] are found not at the end of, but more or less seamlessly embedded within an utterance. Example 6, which was discussed above, already showed such a pattern. The following examples illustrate further typical instances of the phrase glaub(e) ich realized completely within a sentence. Example 7 is taken from a radio phone-in show where callers (C) can talk to the host of the programme (H) about their experiences with foreigners in Germany and Example 8 is taken from a radio counselling show with a priest (P) offering help to people with psychological problems (caller C). (7) Example 7 radio phone-in (psychological counselling) 876 H ja entSCHULdigen sie wenn ich sie jetzt unter[BRECH]e; yes excuse me if I interrupt you now; 877 C [ja. ] yes. 878 H → ihr problem ist glaub ich ganz GUT bei unsern hörern und bei unserer expertin angekommen, your problem has I think (lit.: think I) reached our audience and our expert very well, (8) Example 8 radio phone-in (psychological counselling) 321 C und ich möchte auch ein KLEInes bisschen and I also want a little bit of 322 ein KLEInes stückchen DAseinsberechtigung möchte ich SCHON noch haben; I really do want a little bit of the right to exist; 323 P naTÜRlich, (.) of course, (.) 324 C nech; don’t I; 325 P die sie sich SELBST AUCH geben; (.) which you also give yourself; (.)

 Wolfgang Imo

326 diese DAseinsberechtigung; .h this right to exist; .h 327 C → das KANN man sich glaub ich nicht selbst a- geben. one can’t I think (lit.: think I) give that to oneself. In both these examples glaub(e) ich is morphologically and phonologically reduced (the flectional ending -e is omitted) and prosodically completely integrated. Neither pauses nor accents nor stresses mark out the phrases with glauben as not belonging to the sentence like any other part of speech. Would it still be possible to interpret these instances of glaub(e) ich as parenthetically realized matrix clauses? There are several reasons for not following that interpretation: – First, the so-called “typical” indicators of parentheses are lacking. Parentheses, according to Schönherr (1997), have to be uttered faster, in a lower key and less loudly than the rest of the utterance. – Instead, on a prosodic level all tokens of glaub(e) ich which are realized within an utterance are neither marked by stress or accent nor highlighted by (micro)pauses before or after, but are seamlessly integrated into the surrounding utterance. So prosodically, too, they behave more like modal particles or modal adverbs. – Third, even if one takes a purely syntactic view of parentheses and defines them as interruptions of a syntactic projection (see Auer (2002; 2007b; 2008) or Stoltenburg (2003), who, for example, defines parentheses as “interruptions of emergent syntactic projections”) there are no reasons for classifying glaub(e) ich as a parenthesis. The phrase is completely integrated within the utterance and, again, seems to belong to the classes of modal adverbs or modal particles, which are also not projected by a verb but can be added freely into a sentence. – The last argument is a pragmatic one. Although it is syntactically possible to transform all of the instances of integrated glaub(e) ich into prepositioned matrix clauses followed by a subordinate clause, this would profile ich glaub(e) so strongly that the meaning of the sentence would change – sometimes in quite a radical way. When glaub(e) ich is uttered within a clause, it is not asserted in comparison to the rest of the utterance. Rather, it serves as a modalizing element toning down the utterance. The phrase with glauben appears much more like a modal particle (a category that is usually strongly integrated into the clause in German) and one could indeed speak of an inversion of profile determinacy (Langacker 1987) in these cases. It is the utterance in which glaub(e) ich is embedded in, not the phrase with glauben, that provides the profile determinant of the complete utterance.

Clines of subordination 

6. Clines of subordination: A discussion of the results The results of the preceding analysis can be summarized as follows: 1. If ich glaub(e) is uttered sentence-initially and followed by a subordinate clause, the phrase is asserted and usually carries strong syntactic and pragmatic weight. It is interpreted as a matrix clause which has the following subordinate clause within its scope and it provides the profile determinant of the utterance. 2. If ich glaub(e) is uttered sentence-initially and followed by a main clause, it carries much less weight. Often, it can no longer be interpreted as a matrix clause at all because pragmatically and semantically the following main clause cannot be interpreted as filling the complement slot of the matrix clause. Instead, ich glaub(e) is being reanalyzed as a kind of pragmatic particle, pragmatic marker or discourse marker. While the main clause provides the basis for the profile determinant in that case, it is unclear whether the discourse marker should be subsumed under that profile or be treated as an independent unit (and, thus, as providing a special pragmatic profile determinant of its own). 3. If glaub(e) ich is uttered sentence-finally, its relative weight is reduced even further. Sometimes, the phrase appears as an increment. In terms of profile determinacy, the same holds for these cases as for ich glaub(e) as an utteranceinitial discourse marker. The phrase can neither be subsumed under the previous utterance nor be classified as a completely independent unit – it is somewhere “in-between”. Often, though, glaub(e) ich is prosodically latched onto the previous utterance. If that is the case, the semantic and pragmatic weight of glaub(e) ich is reduced even further and it starts to appear as a modal particle or adverb, i.e. like any other part of speech that is integrated completely into the sentence. 4. If glaub(e) ich is produced within an utterance, its relative weight is very small. The phrase is prosodically and syntactically completely integrated into the surrounding clause and has lost any appearance of a matrix clause. What happens, in fact, is that the direction of embedding vs. embedded has been reversed.11 When the phrase is uttered sentence-initially, the following clause 11. With “embedded” I do not refer to the assumptions of transformative grammar but rather use the term in a cognitive way to describe the gestalt phenomenon that parts of speech can be integrated seamlessly into a sentence prosodically and semantically – i.e. no prosodic break or pitch change occurs and no unusual semantic contribution is added. For example, adverbs are often neatly embedded, as in the following sentence: It will definitely rain tomorrow. It is possible, though, to highlight them, causing a break in the sentence: It will (.) DEfinitely (.) rain tomorrow. (The brackets indicate pauses and the capital letters a stressed syllable). For a detailed discussion of such phenomena see the special issue on increment structures of Pragmatics (2007), especially the discussion of prosodically integrated vs. non-integrated increments by Auer (2007a: 634f.).

 Wolfgang Imo

seems to be embedded into the matrix clause. When glaub(e) ich is uttered within a clause, it appears to be embedded into that clause and, thus, the profile determinant lies in the whole clause and not in the phrase with glauben. Auer/Günthner (2005) as well as Stoltenburg (2003) come to the conclusion that glaub(e) ich is being reanalyzed as a modal adverb or modal particle because of its loss of syntactic powers of projection, its morphological and phonological reduction and its prosodic downscaling and integration. This means that ich glaub(e) shows strong signs of a process of recategorization. If it is uttered sentence-initially, it leaves the mainly syntactic category [matrix clause] and moves towards the mainly pragmatic category [discourse marker]. In that process, the complete construction moves from [matrix clause + dependent main clause12] towards [discourse marker + following utterance]. This process has been amply described for a lot of German and English “ex”-matrix clauses. Rehbein (2003: 270), for example, states that matrix constructions show the tendency to congeal into fixed phrases or formulae and Thompson/Mulac (1984, 1991) as well as Bybee (2001) and Thompson (2002) have shown that matrix clauses tend to be separated syntactically from their following main clauses, thus gaining an autonomous status. In the end, they become recategorized (or pragmaticalized) as “epistemic qualifiers” or “CTP-phrases as epistemic/evidential/evaluative frames for declarative and interrogative clauses” (Thompson/Mulac 1984: 315; see also Pekarek Doehler, Okamoto/Laury and Günthner (this volume)). This certainly holds true for matrix clauses which are uttered sentence-initially, such as ich glaub(e). The inverted construction glaub(e) ich has moved even further away from a matrix clause or from any subject-plus-predicate-plus-complement combination. It can no longer be treated as a concatenation of a predicate and a subject but has to be reanalyzed as a fixed, frozen phrase. But in which direction is this phrase moving? It definitely cannot be classified as a discourse marker because that class of expressions is not embedded into sentences in German (Gohl/Günthner 1999). In fact, there are at least three neighbouring constructions which form a taxonomic network13 within which the phrase glaub(e) ich is moving freely. These three schematic constructions (shown below) fulfil the functions of attractors and it depends on the actual context in which the phrase glaub(e) ich is uttered to decide which construction works as main attractor:

12. The term “dependent main clause” is a literal translation of Auer’s (1998) “abhängiger Hauptsatz”, referring to main clause sentences (i.e. sentences with the finite verb in the second position) which follow a matrix clause (e.g. Ich glaube, es wird heute regnen. ‘I think it will rain today.’). 13. On the concept of taxonomic networks within the theory of construction grammar or cognitive grammar see Croft (2002), Croft/Cruse (2004) and Langacker (1987).

Clines of subordination  i. glaub(e) ich as a modal particle: pro

syntactically fixed phrase (sedimentation); modalizing function; cannot carry stress/accent; strong tendency to be placed in the middle field14

contra

the phrase with glauben can theoretically be transformed into a matrix clause (which a modal particle cannot); the phrase can occupy the post-field and (partly) the front field of clauses – both these positions are blocked for modal particles; rather strong semantics with little bleaching in most cases

ii. glaub(e) ich as a modal adverb: pro

syntactically fixed phrase (sedimentation); modalizing function; can be transformed into a matrix clause; double front-field position can be a step towards the ability to occupy the front field on its own15

contra

unlike a modal adverb, the phrase with glauben cannot be placed in the front field on its own; the phrase cannot be stressed, while a modal adverb can carry stress; while modal adverbs can be used as an answer to a yes/no question, glaub(e) ich cannot be used that way

iii. glaub(e) ich as an inverted matrix clause: pro

on a purely formal syntactic level, it can be transformed into a prepositioned matrix clause

contra

glaub(e) ich has become a fixed phrase or formula via sedimentation processes; glaub(e) ich has a low profile (it is much less salient than a matrix clause); on a pragmatic and sentence-semantic level, the transformation into a prepositioned matrix clause is not possible

Schema 2.â•‡ Network of possible constructions 14. According to topological grammar (Drach 1963), the middle field of a main clause is the part between the left and (optional) right verb brackets. The structure of German main clauses is: front field / left verb bracket / middle field / (right verb bracket) / post-field. 15. Though glaub(e) ich can never occur alone in the front field (before the finite verb), it can occur in a kind of double front-field position together with another element: “über WEIHnachten glaub ich isch er SCHON daheim.” (lit.: ‘during Christmas think I is he certainly at home’).

 Wolfgang Imo

The more gestalt features glaub(e) ich shares with the schematic construction [modal particle] (i.e. positioned within the middle field of the sentence, no pauses before and after and no accent/stress), the stronger the construction [modal particle] appears as the superordinate construction which glaub(e) ich is an instance of. If glaub(e) ich is realized in the front field or post field of a sentence, it moves closer to the construction [modal adverb] or [(inverted) matrix clause]. In any case, because of the formal possibility of changing the phrase into a prepositioned matrix clause, the construction [matrix clause] always stays active to at least some small degree. These conflicting results suggest that the findings of Thompson/Mulac (1984, 1991) are also valid for German complement-taking predicates. Actually, glaub(e) ich is a good illustration of what they have noted about certain English verbs, which they call “epistemic verbs”: These epistemic verbs together with their subjects behave very much like unitary epistemic morphemes in other languages, to the point of being ‘transportable’ to positions other than that which they could occupy if they were only functioning to introduce a complement. (Thompson/Mulac 1984: 315)

What this implies is that because of the feature of being transportable to different positions within an utterance, some verbs have to be interpreted as being part of completely different constructions. Thompson/Mulac (1991: 239) name to believe as an example of a verb that oscillates within a network of constructions. In the sentence I believe it’s going to snow., I believe is seen as a complement-taking predicate, while in the sentence It’s going to snow, I believe. the phrase is reanalyzed as an epistemic morpheme and thus becomes an instantiation of such constructions as [discourse marker] or [modal particle]. So the change of position within a clause implies a change of construction, too. Post-positioned, I believe no longer has the power of demanding a complement (loss of valence). 7. Phrases with glauben as instances of fragmentary language Phrases with glauben oscillate between several different constructions. First, there are the syntactic patterns of [matrix clause + subordinate clause], [matrix clause + dependent main clause] and [parenthetically realized matrix clause] that always remain more or less closely connected in the network of constructions that provides the area in which actual instances of ich glaub(e) and glaub(e) ich can be interpreted. For the non-inverted form ich glaub(e), the construction [discourse marker] even seems to be the core construction for some cases. For the inverted form, traditional constructions such as [modal particle] or [modal adverb] have to

Clines of subordination 

be considered – or, maybe, a new construction, such as [incremental discourse particle], ought to be added. While there have been precedents for a process such as the development of ich glaub(e) into a discourse marker,16 which involves a path that runs from [matrix clause + subordinate clause] via [matrix clause + dependent main clause] to [discourse marker + following utterance], the development of inverted glaub(e) ich is much more difficult to sketch. One possible solution for the situation of uncertain target constructions might be found in Hopper/Traugott’s (1993: 17) concept of grammaticalization. During grammaticalization processes it often happens that linguistic elements can no longer be classified in terms of unequivocal syntactic categories.17 “Earlier forms may coexist with later ones [...], earlier meanings may constrain later meanings and/or structural characteristics.” So the phenomenon of overlaying multiple forms and meanings/functions (divergence) might be responsible for the constructional uncertainty. I would argue, though, that at the present stage it is not possible to decide whether glaub(e) ich really can be described as being about to be lexicalized or pragmaticalized. If this were the case, what would be the target construction? Would it be that of a modal particle or a modal adverb? How, then, can one account for the post-positioned forms? Will they vanish? Will there be two, or even three, target constructions (incremental discourse markers, modal particles, modal adverbs)? Or will the result be a completely new construction? What is certain, so far, is the fact that glaub(e) ich can no longer simply be interpreted as a parenthetically realized matrix clause. On top of that, the morpho-syntactic and phonological reduction of the phrase glaub(e) ich to glaub in some of the data18 seems to point to a process of sedimentation into a fixed phrase. Haspelmath (2002: 274) comes to the conclusion that phonological reduction is a natural result of the high-frequency use of a phrase: “Phonological reduction is a natural result of increased frequency, and frequency of elements and patterns automatically leads to an increased mental entrenchment and thus results in fixed grammatical patterns.” (Haspelmath 2002: 16. The phrase ich mein(e) (I mean) is a case of a discourse marker that is already very strongly grammaticalized (or, rather, pragmaticalized), according to Günthner/Imo (2003). See also Traugott (1995: 38): “Yet another example of a verbal construction coming to acquire discourse particle function is I think. This is currently undergoing a change from main-clause status, where the first person is selected from among the set of possible subject arguments of think, to a parenthetical construction with greater freedom of position, to a fixed phrase indicating speakers’ epistemic attitude.” 17. In functional linguistics, the feature of categorical vagueness or openness is assumed to be typical for grammatical categories in general, not just for processes of grammaticalization. 18. As in the following case: “da hab i glaub sogar beDIENT;” (lit.: ‘there have I think even worked as waiter’). The pronoun “i”, a dialectal variety of “ich”, is the subject of the verb “beDIENT”, not of “glaub”.

 Wolfgang Imo

274; My translation). This, in turn, usually leads to a strong mental implementation which ends in the formation of new grammatical patterns (see also Bybee/Scheibman 1999). In the case of glauben, though, this is just a hypothetical development which has by no means proceeded so far as to lead to any fixed construction in which the verb has lost all of its “verbiness” (i.e. its ability to demand a complement) and given up its – albeit vague – association with a matrix clause. The phrase glaub(e) ich is placed within a “no-man’s land” between different constructions, and a fixed category cannot be decided upon. The only “movement” along a “path” that all of the phrases with glauben (those in front of, embedded in and after an utterance) have in common is what Traugott (1995) and Langacker (1990) call a path towards “subjectification”, namely the movement towards marking an “alignment to the speaker’s perspective” (Traugott 1995: 39). This broad pragmatic development as a marker of speaker stance does not correlate closely with the coagulation into fixed forms, i.e. the development of constructions (form-meaning pairings). A syntactic theory that aims to sketch an accurate picture of the present state of language cannot ignore the fact that language is always on the flux (see Hopper’s (1998) concept of “emergent grammar”). Therefore, the fragmentary character of language has somehow to be taken into account: In casual spoken discourse constructions appear not as neatly bounded sentences or clauses but as unstructured fragments that are habitually combined with other fragments to make utterances. [...] They [constructions W.I.] are open because their structure is emergent, that is to say, their structure never reaches a point of closure and completion as a construction. [...] They are intrinsically indeterminate. (Hopper 2004: 19)

This quote seems to be written to account exactly for the phenomena discussed here. Constructional schemata are only prototypes which interactants orient to (cf. Imo 2007b). Depending on interactional needs, actual utterances can be realized in all kinds of different ways and draw on a wide range of related constructions. They can be reanalyzed, amalgamated, aborted or reinterpreted ad hoc. Norén/Linell (2007) refer to this inherent openness of lexical or grammatical constructions as their “meaning potentials”: The meaning potential of a lexical item or a grammatical construction is the set of properties which together with contextual factors, including features of the linguistic co-text as well as various situational conditions, make possible all the usages and interpretations of the word or construction that language users find reasonably correct, or plainly reasonable in the actual situations of use.

The meaning potentials of the phrases involving glauben can draw on a range of different sources. First, the semantic content of the verb – i.e. expressing a belief about something, thinking something – provides a starting point. This basic

Clines of subordination 

semantic content can lead down the path of subjectification, transforming phrases with glauben into mere pragmatic markers indicating speaker stance. Second, the routinized and highly grammaticalized constructions glauben is used in (such as matrix clauses or transitive constructions) provide further sources for the activation of meaning potentials. Here, the full semantic content is usually needed for glauben. Third, constructions not involving glauben but showing similar syntactic patterns of distribution, function or internal structure by analogy endow a phrase with glauben with aspects of the meaning and function of, for example, a modal particle. The concept of meaning potentials can explain both the apparent lack of fixed structure as well as the processes of making sense of an utterance in given actual situations. It may well be that the extreme case of syntactic indecidability illustrated in this analysis could actually be at the heart of the grammar of spoken interaction in general. While this is still an open question, what has become clear is the fact that one has to abandon the idea of constructions which can be determined in a binary yes/no fashion.19 If one takes Hopper’s (1998) idea of “emergent grammar” seriously, all language is fragmentary. The constructs with glauben provide a strong hint that this is indeed the case. Literature Auer, Peter. 1998. “Zwischen Parataxe und Hypotaxe: ‘abhängige Hauptsätze’ im Gesprochenen und Geschriebenen Deutsch.” ZGL 26: 284–307. Auer, Peter. 2002. “On line-Syntax – Oder: was es bedeuten könnte, die Zeitlichkeit der mündlichen Sprache ernst zu nehmen.” Sprache und Literatur 85 (31): 43–56. Auer, Peter. 2006. “Increments and more. Anmerkungen zur augenblicklichen Diskussion über die Erweiterbarkeit von Turnkonstruktionseinheiten.” In Grammatik und Interaktion, Arnulf Deppermann, Reinhard Fiehler and Thomas Spranz-Fogasy (eds), 279–294. Radolfzell: Verlag für Gesprächsforschung. Auer, Peter. 2007a. “Why are increments such elusive objects? An afterthought.” Pragmatics 17 (4): 647–658. Auer, Peter. 2007b. “Syntax als Prozess.” In Gespräch als Prozess, Heiko Hausendorf (ed.), 95–124. Tübingen: Narr. Auer, Peter. 2008. “On-line syntax: Thoughts on the temporality of spoken language.” Language Sciences 31 (1): 1–13.

19. See also Seppänen and Laury (2007: 569), who in their discussion of Finnish et(tä) clauses come to the conclusion that these clauses are “not very good examples of traditional complementation and thus not at all clear examples of subordination”. Et(tä) clauses are used by interactants in such a wide variety of constructions and functions that they, too, could be seen as examples for fragmentary language.

 Wolfgang Imo Auer, Peter and Günthner, Susanne. 2005. “Die Entstehung von Diskursmarkern im Deutschen – ein Fall von Grammatikalisierung?.” In Grammatikalisierung im Deutschen, Torsten Leuschner and Tanja Mortelsmans (eds), 335–362. Berlin: de Gruyter. Barden, Birgit. 1998. “Prosodische Merkmale von Operator-Skopus-Strukturen.” Unpublished manuscript. Halle, 3.3.98. Barden, Birgit, Elstermann, Mechthild, and Fiehler, Reinhard. 2001. “Operator-Skopus-Strukturen in gesprochener Sprache.” In Pragmatische Syntax, Frank Lidtke and Franz Hundsnurscher (eds), 197–232: Tübingen: Niemeyer. Brinton, Laurel J. 1998. “‘The flowers are lovely; only, they have no scent’: The evolution of a pragmatic marker.” In Proceedings of the Conference of the German Association of University Teachers of English, Raimund Borgmeier, Hubert Grabes and Andreas H. Jucker (eds), 9–33. Trier: WVT. Bybee, Joan 2001. “Main clauses are innovative, subordinate clauses are conservative.” In Complex Sentences in Grammar and Discourse, Joan Bybee and Michael Noonan (eds), 1–17. Amsterdam: Benjamins. Bybee, Joan and Scheibman, Joanne. 1999. “The effect of usage on degrees of constituency: the reduction of don’t in English.” Linguistics 37 (4): 575–596. Couper-Kuhlen, Elizabeth and Ono, Tsuyoshi (ed.). 2007. Turn Continuation in Cross-linguistic Perspective. Special Issue of Pragmatics 17 (4). Couper-Kuhlen, Elizabeth and Ono, Tsuyoshi. 2007. “‘Incrementing’ in conversation. A comparison of practices in English, German and Japanese.” Pragmatics 17 (4): 513–552. Croft, William. 2002. Radical Construction Grammar. Oxford: Oxford University Press. Croft, William and Cruse, Alan D. 2004. Cognitive Linguistics. Cambridge: Cambridge University Press. Drach, Erich. 1963. Grundgedanken der deutschen Satzlehre. Darmstadt: Wissenschaftliche Buchgesellschaft. Edmondson, Willis and House, Juliane. 1981. Let’s Talk and Talk about it. A Pedagogic Interactional Grammar of English. München: Urban & Schwarzenberg. Elmauer, Ute. 1973. “Abhängige Hauptsätze in gesprochenen und verschriftlichten Texten.” In Institut für deutsche Sprache. Forschungsberichte Band 7, Ulrich Engel and Irmgard Vogel (eds), 193–217. Tübingen: Niemeyer. Erman, Britt. 1987. Pragmatic Expressions in English. Stockholm: Almquist & Wiksell. Ford, Cecilia E. et al. 2002. “Constituency and the grammar of turn increments.” In The Language of Turn and Sequence, Cecilia E. Ford, Barbara A. Fox and Sandra A. Thompson (eds), 14–38. New York: Oxford University Press. Fraser, Bruce. 1990. “An approach to discourse markers.” Journal of Pragmatics 14: 383–395. Freywald, Ulrike. 2008. “Zur Syntax und Funktion von dass-Sätzen mit Verbzweitstellung.” Deutsche Sprache 36: 246–285. Gohl, Christine and Günthner, Susanne. 1999. “Grammatikalisierung von weil als Diskursmarker in der gesprochenen Sprache.” Zeitschrift für Sprachwissenschaft 18 (1): 39–75. Günthner, Susanne and Imo, Wolfgang. 2003. “Die Reanalyse von Matrixsätzen als Diskursmarker: ich mein-Konstruktionen im gesprochenen Deutsch.” In Jahrbuch der ungarischen Germanistik, Magdolna Orosz and Andreas Herzog (eds), 181–216. Budapest: DAAD. Haspelmath, Martin. 2002. “Grammatikalisierung: von der Performanz zur Kompetenz ohne angeborene Grammatik.” In Gibt es eine Sprache hinter dem Sprechen?, Sybille Krämer and Ekkehard König (eds), 262–286. Frankfurt: Suhrkamp. Helbig, Gerhard and Buscha, Joachim. 2001. Deutsche Grammatik. Berlin: Langenscheidt.

Clines of subordination 

Hopper, Paul J. 1998. “Emergent grammar.” In The New Psychology of Language, Michael Tomasello (ed.), 155–175. Mahwah: Psychology Press. Hopper, Paul J. 2004. “The openness of grammatical constructions.” Chicago Linguistic Society 40: 239–256. Hopper, Paul J. and Traugott, Elizabeth C. 1993. Grammaticalization. Cambridge: Cambridge University Press. Imo, Wolfgang. 2006. “‘Da hat des kleine glaub irgendwas angestellt’ – Ein construct ohne construction?” In Konstruktionen in der Interaktion, Susanne Günthner and Wolfgang Imo (eds), 263–291. Berlin: de Gruyter. Imo, Wolfgang. 2007a. Construction Grammar und Gesprochene-Sprache- Forschung: Konstruktionen mit zehn matrixsatzfähigen Verben im gesprochenen Deutsch. Tübingen: Niemeyer. Imo, Wolfgang. 2007b. “Der Zwang zur Kategorienbildung: Probleme der Anwendung der Construction Grammar bei der Analyse gesprochener Sprache.” Gesprächsforschung – Online Zeitschrift zur verbalen Interaktion 8: 22–45. Kroon, Caroline 1995. Discourse Particles in Latin. Amsterdam: Gieben. Kroon, Caroline. 1998. “A framework for the description of Latin discourse markers.” Journal of Pragmatics 30: 205–223. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, Vol. 1. Stanford: Stanford University Press. Langacker, Ronald W. 1990. “Subjectification.” Cognitive Linguistics 1 (1): 5–38. Langacker, Ronald W. 1991. Foundations of Cognitive Grammar, Vol. 2. Stanford: Stanford University Press. Langacker, Ronald W. 2009. Investigations in Cognitive Grammar. Berlin/New York: de Gruyter. Lenk, Uta. 1998. “Discourse markers and global coherence in conversation.” Journal of Pragmatics 30: 245–257. Norén, Kerstin and Linell, Per. 2007. “Meaning potentials and the interaction between lexis and contexts: An empirical substantiation.” Pragmatics 17 (3): 387–416. Rehbein, Jochen. 2003. “Matrixkonstruktionen in Diskurs und Text.” Zeitschrift für Interkulturellen Fremdsprachenunterricht 8 (2/3): 252–276. Schiffrin, Deborah. 1987. Discourse Markers. Cambridge: Cambridge University Press. Schourup, Lawrence C. 1985. Common Discourse Particles in English Conversation. New York: Garland Publishing. Schönherr, Beatrice. 1997. Syntax – Prosodie – nonverbale Kommunikation. Tübingen: Niemeyer. Schumacher, Helmut, Kubczak, Jacqueline, Schmidt, Renate and de Ruiter, Vera. 2004. VALBU – Valenzwörterbuch deutscher Verben. Tübingen: Narr. Selting, Margret et al. 1998. “Gesprächsanalytisches Transkriptionssystem.” Linguistische Berichte 173: 91–122. Selting, Margaret, et. al. 2009. “Gesprächsforschung-Online-Zeitschriftzur Verbalen Interaktion 10: 353–402. Seppänen, Eeva-Leena and Laury, Ritva. 2007. “Complement clauses as turn continuations: the Finnish et(tä)-clause.” Pragmatics 17 (4): 553–572. Stoltenburg, Benjamin. 2003. “Parenthesen im gesprochenen Deutsch.” InLiSt (Interaction and Linguistic Structure) 34: 1–44. (www.uni- potsdam.de/u/inlist/issues/34/index.htm) Thim-Mabrey, Christiane. 1988.“Satzadverbialia und andere Ausdrücke im Vorvorfeld.” Deutsche Sprache 16: 52–67. Thompson, Sandra A. 2002. “‘Object complements’ and conversation: towards a realistic account.” Studies in Language 26: 125–163.

 Wolfgang Imo Thompson, Sandra A. and Mulac, Anthony. 1984. “A quantitative perspective on the grammaticization of epistemic parenetheticals in English.” In Coherence and Grounding in Discourse, Russel S. Tomlin (ed.), 131–339. Amsterdam: Benjamins. Thompson, Sandra A. and Mulac, Anthony. 1991. “The discourse conditions for the use of the complementizer that in conversational English.” Journal of Pragmatics 15: 137–251. Thompson, Sandra A. and Hopper, Paul J. 2008. “Projectability and clause combining in interaction.” In Crosslinguistic Studies of Clause Combining. The Multifunctionality of Conjunctions, Ritva Laury (ed.), 99–123. Amsterdam: Benjamins. Traugott, Elizabeth C. 1995. “Subjectification in grammaticalization.” In Subjectivity and subjectivisation, Dieter Stein and Susan Wright (eds), 31–54. Cambridge: Cambridge University Press.

Are kara ‘because’-clauses causal subordinate clauses in present-day Japanese? Yuko Higashiizumi

Tokyo Gakugei University This paper reviews subordination in conversation through grammaticalization. Traditionally, the Japanese kara ‘because’-clause has been characterized as a causal subordinate clause. This study describes the various uses of karaclauses in present-day Japanese and the historical background to the variety in clause combination. According to Hopper & Traugott (2003), complex clause constructions are traditionally classified into subordination and coordination. More recent studies, however, have suggested that the dichotomy calls for further investigation. Building on these recent studies, Hopper & Traugott (2003) propose a subordination – hypotaxis – parataxis continuum of complex clause constructions in grammaticalization. I illustrate that a continuum is more relevant than the traditional dichotomy in understanding both the functional diversity of kara-clauses in present-day Japanese and the historical process in kara-clauses.

1. Introduction The aim of this study is to reconsider the concept of subordination in conversation from the viewpoint of grammaticalization. According to Hopper & Traugott (2003: 176), complex clause constructions are traditionally classified into two types, i.e., subordination and coordination. On the other hand, recent work suggests that this traditional dichotomy requires further discussion (e.g., Haiman & Thompson 1984; Lehmann 1988; Matthiessen & Thompson 1988; Cristofaro 2003; Verstraete 2007). Building on more recent studies, Hopper & Traugott (2003: 176–184) propose a subordination – hypotaxis – parataxis continuum of complex clause constructions in grammaticalization. This paper will show that the continuum is more useful than the dichotomy in illuminating a wide variety of kara ‘because’-clauses, so-called causal subordinate clauses, in present-day Japanese, as well as the historical process in clause combination. The kara-clause consists of a clause and the so-called subordinate clause marker kara added as a suffix, referred to as a setuzoku-zyosi ‘conjunctive particle’.

 Yuko Higashiizumi

Canonically, a pair of clauses in which one clause is called the subordinate clause and the other the main clause can be schematized as in (1a). (Hereafter, CL stands for clause. CL1 stands for a clause that has a subordinate clause marker attached as a suffix when it is associated with a main clause CL2.) In the same way, the karaclause and its syntactic main clause can be schematized as in (1b). (1) a. CL1-subordinate clause marker CL2. b. CL1-kara CL2. A number of recent studies, however, have indicated that what are traditionally called subordinate clauses are not always immediately followed by a syntactic main clause in present-day Japanese, as in (2a). In the same way, the kara-clause may stand alone with no preceding or following syntactic main clause in present-day Japanese, as schematized in (2b). (2) a. CL-subordinate clause marker. b. CL-kara. Such subordinate clause markers occupy the position in which a syuu-zyosi ‘final particle’ usually appears and express various discourse-pragmatic functions (Nakayama & Ichihashi-Nakayama 1997; Iguchi 1998; Ohori 1995; Okamoto 1995; Suzuki 1999). This study focuses on the historical backgrounds to the present-day usage of kara-clauses. The organization of this paper is as follows: Section 2 provides notes about the examples while Section 3 gives the background to the present study and the various relationships of kara-clauses to their main clauses in present-day Japanese conversation. Section 4 introduces the diachronic process in kara-clauses. Section 5 discusses what the functional extension of kara-clauses can tell us about subordination in conversation. Finally, Section 6 offers conclusions to the current research. 2. Data The data come from examples collected for Higashiizumi (2006). It consists of a total of 435 kara-clauses collected from the conversational parts of novels and play scripts from 1592 to 1985 and a total of 107 kara-clauses taken from conversation transcripts in Nihongo Journal and in Lee (2000) dated 1995–2000 in the Edo/ Tokyo dialect in Japan. In order to minimize any bias in sampling, no more than ten relevant examples are taken from novels and play scripts by a single author or from the same conversation transcript.1 1. See the appendices for the data list. Jacobs & Jucker (1995) provide a discussion of spoken language in diachronic data. For further details on collecting diachronic conversational data in Japanese, also see Suzuki (1999) and Onodera (2004).

Are kara ‘because’-clauses causal subordinate clauses in present-day Japanese? 

3. Background 3.1

Cline of clause-combining constructions in grammaticalization

In the literature of grammaticalization, it has been pointed out that when two clauses are combined, they tend to develop into a more integrated clause combination over the course of grammaticalization (Hopper & Traugott 2003: 178), as shown in Figure 1. 3.2

Previous studies on the history of kara-clauses

This subsection gives a quick look at the history of kara-clauses based on Ishigaki (1955), Kobayashi (1996), Konoshima (1966, 1983), and Yamaguchi (1996). Kara in the 8th century is said to have been a noun-like word which occurs “either typically or extensively with adnominal modification” (Martin 1975: 664). The meaning of kara at that time is considered to have been ‘following’, ‘with’, ‘along’, ‘by way of ’, etc. It is thought to have developed into an ablative marker following a noun, i.e., Noun-kara ‘from Noun’, and then into a causal marker following a finite clause. It came into wide use as a causal clause marker in the eastern part of Japan beginning around 1600. Yoshii (1977) studied causal clauses that appeared in the conversational parts of novels in the Tokyo area from around 1700 to 1960. He makes several important statements on the development of kara-clauses. I will take up three of these points that are relevant to the present study. First, his data showed that karaclauses began to appear in conversation around 1700 and came into use in approximately 80% of all causal expressions around 1760. Second, he examined whether or not the syntactically associated main clause of kara-clauses expresses a description of facts. To put it differently, the kara-clause is considered to be associated with the verb in the main clause when a simple finite verb appears in the main clause, whereas the kara-clause is considered as being associated with the modal expressed by a certain type of post-verbal expression that appears in the main clause. He observed such expressions as conjecture (e.g, kamosirenai ‘maybe’), opinion (e.g., V-beki ‘must, should’), volition (e.g., V-(y)oo ‘will, let’s’), order (e.g., imperative form), request (e.g., V-tekure/tyoodai ‘please do something’), and question based on Nagano (1979 [1952]). Yoshii found that main clauses without such expressions gradually decreased, whereas main clauses including such expressions increased from around 1850. If we use the set of terms given in Sweetser (1990), we can say that kara-clauses appearing to correspond with a content conjunction gradually decrease in proportion, whereas karaclauses appearing to correspond to an epistemic or speech-act conjunction increase from around 1850 onwards. In other words, the kara-clause construction

 Yuko Higashiizumi parataxis

> hypotaxis

> subordination

– dependent

+ dependent

+ dependent

– embedded

– embedded

+ embedded

Figure 1.â•‡ Cline of clause-combining constructions in grammaticalization (Hopper and Traugott 2003: 178)

is becoming more subjective in the sense of Traugott (2003). Third, he pointed out that the proportion of kara in sentence-final position rapidly increased in the twentieth century, although whether kara in sentence-final position refers to the inversion of a kara-clause and its syntactically associated main clause (i.e., CL2, CL1-kara) and/or an independent kara-clause without an accompanying syntactically associated main clause (i.e., CL-kara in Schimatization (2b) above) is not clarified in his study. See Higashiizumi (2006: 141–146) for further details of Yoshii (1977). 3.3

Previous studies on kara-clauses in present-day Japanese

Kara-clauses are commonly characterized as subordinate clauses with causal meaning in present-day Japanese. More recent researchers, however, have noticed that they are not necessarily considered to be subordinate clauses or as a means of expressing cause or reason (e.g., Alfonso 1980; Masuoka 1997; Minami 1974, 1993; Shirakawa 1995, 2009; Takubo 1987; Uno 2009). Before Sweetser’s (1990) systematic analysis of logical connectives, Takubo (1987) differentiated kara-clauses that correspond to content conjunction in Sweetser’s (1990) sense, as shown in Example (3), from those that correspond to epistemic conjunction, as shown in Example (4), based on what the kara-clause modifies in the main clause. (3)

[kare ga itta kara kanozyo mo itta] no desyoo. he nom went kara she also went nmlz guess ‘(Lit.) (I) guess [because he went she also went].’ ‘I guess [she also went because he did].’ (Takubo 1987: 43)

(4)

kare ga itta kara [kanozyo mo itta] desyoo. he nom went kara she also went guess ‘(Lit.) Because he went, (I) guess [she also went].’ ‘I guess she also went, [because (I know) he did].’

(Takubo 1987: 43)

Example (3) expresses a real-world causality between the events described in the kara-clause and its associated main clause. The kara-clause in Example (3) modifies

Are kara ‘because’-clauses causal subordinate clauses in present-day Japanese? 

the proposition that ‘she went’ in the main clause since it is within the scope of the nominalizer no. In other words, the kara-clause in Example (3) is an adjunct and restrictive clause (Quirk et al. 1985: 1076) and is a hypotactic clause in terms of the cline of clause-combining constructions in grammaticalization. On the other hand, the kara-clause in Example (4) can be interpreted as the speaker’s grounds for drawing the conclusion that ‘she went’ in the main clause. It does not modify the proposition but the modal expressed by desyoo in the main clause. That is to say, it is a disjunct and non-restrictive clause (Quirk et al. 1985: 1076) and is a paratactic clause in terms of the cline of clause-combining constructions in grammaticalization. Takubo noticed that kara-clauses are non-restrictive when they express a motive for the speech-act performed in their main clause as in Example (5). It can be considered to correspond to Sweetser’s (1990) speech-act conjunction. (5) P: sono uuron-tya mottainai kara nom-oo yo the oolong-tea wasteful kara drink-vol ptcl ‘because (it is) wasteful, let’s drink the oolong tea’ Q: un yeah ‘Yeah’

(Z95–15)

The kara-clause in Example (5) expresses the speaker’s motive for inviting the addressee to drink tea in the main clause. It does not modify the proposition that they drink tea but the modal expressed by the suffix -oo in the main clause. It is also a disjunct and non-restrictive clause and is a paratactic clause in terms of the cline of clause-combining constructions in grammaticalization. See Alfonso (1980) and Shirakawa (1995) for further details of kara-clauses of this type. In present-day Japanese, kara-clauses may stand alone with no preceding or following syntactic main clause (Higashiizumi 2006; Honda 2005; Iguchi 1998; Ohori 1995, 1997; Shirakawa 2009; Suzuki 2009; Yabe 1997, 2002; Yamaguchi 1996). An example of an independent kara-clause is given in Example (6). The interlocutors Q and P talk about the security guard whom they have not seen for a long time. (6) P: watasi are kara at-te nai yo I then since see-ger not ptcl ‘I have not seen (him) since then.’ Q: kinoo watasi mo hazimete atta no yo yesterday I also for.the.first.time met mnlz ptcl ‘I also saw (him) for the first time (since I last saw him) yesterday’

 Yuko Higashiizumi

P: a hontoni [rising intonation] int really ‘oh, really?’ Q: un yeah ‘Yeah.’ P:

zikantai ga nee aru kara ne work.hours nom ptcl exist kara ptcl ‘(lit.) because (his) work hours exist.’ ‘(you saw him after a long time) because (you had been missing) (his) work hours.’

Q: tyotto mat-te ne ... little wait-ger ptcl ‘Wait a moment.’ P: un yeah ‘Yeah.’

(Z95–6)

The kara-clause in P’s third turn in Example (6) can be considered as a single independent clause with a final-particle-like pragmatic marker or as a marker of discursive subordination (Lindström and Londén 2007). There is no linguistic main clause in Example (6), although we may interpret her utterance as meaning that she has not seen him for a long time because she does not know his work hours or because she cannot pass by the security office during his work hours. It has a pragmatic meaning that can be translated as ‘you know’ in English (Ohori 1995). The current study will term this type of kara a “pragmatic marker” in the sense of Brinton (1996). Table 1 summarizes the characteristics of kara-clauses in present-day Japanese, using Sweetser’s (1990) terms. So far, we have seen how and where kara-clauses function in present-day Japanese. Next, we will observe the historical process in kara-clauses. In the following section, I will explore how the relative distribution of examples, such as Examples (3)–(6), changes historically. 4. The diachronic process in kara-clauses 4.1. Clause combination Table 2 shows the transition of the distribution of kara-clauses by position from the 18th to the 21st centuries. The position of kara-clauses is classified as follows: (i)

Are kara ‘because’-clauses causal subordinate clauses in present-day Japanese? 

Table 1.â•‡ Characteristics of PDJ kara-clauses Examples

Example (3)

Examples (4) and (5)

Clause combination kara-clause

Schematization (1b) CL1-kara CL2. hypotactic (or subordinate) subordinate clause marker

CL1-kara, CL2. paratactic (or non-subordinate) conjunction-like clause marker

kara

Interpretation using Sweetser’s (1990) terms

content conjunction

Example (6)

Schematization (2b) CL-kara. more paratactic (or independent) final-particle-like discourse-pragmatic marker epistemic conjunction or N/A speech-act conjunction

Table 2.â•‡ Development of the positional distribution of kara-clauses

i. CL1-kara CL2 (canonical) ii. CL2, CL1-kara (inverted) iii. CL-kara (with no main clause) iv. Others TOTAL TOTAL

–1750

–1800

–1850

–1900

–1950

–2000 (wr)

–2000 (sp)

9

84

50

38

67

44

48

(69.2%) 1

(89.4%) 3

(86.2%) 3

(76.0%) 8

(55.8%) 27

(44.0%) 35

(44.9%) 11

(7.7%) 0

(3.2%) 2

(5.2%) 0

(16.0%) 1

(22.5%) 17

(35.0%) 9

(10.3%) 45

(0%) 3 (23.0%) 13

(2.1%) 5 (5.3%) 94

(0%) 5 (8.6%) 58

(2.0%) 3 (6.0%) 50

(14.2%) 9 (7.5%) 120 335

(9.0%) 12 (12.0%) 100

(42.1%) 3 (2.8%) 107 207

the kara-clause precedes a main clause (i.e., traditional, canonical clause order); (ii) it follows a main clause (i.e., newer, inverted clause order); (iii) it accompanies no main clause; and (iv) other types such as an answer to an explicit why-question. What is noticeable from Table 2 is that the proportion of (i) (i.e., traditional, canonical clause order) decreases from the 19th century onwards, whereas that of (iii) (i.e., kara-clauses with no main clause) increases in the 20th century. In other

 Yuko Higashiizumi

words, the trend is for a kara-clause to become more paratactic to its main clause or more independent from its main clause in terms of clause combination.2 Further, the proportion of (ii) (i.e., newer, inverted clause order) gradually increases from the late 19th century. 4.2

Interpretations of kara-clauses

In order to observe the developmental process of kara-clauses in terms of the correlation between their interpretation and their restrictiveness, I examined types of expressions that occur explicitly in both a kara-clause and the main clause when they have an epistemic or speech-act conjunction interpretation. The explicit expressions concerned in the current study are desyoo ‘(I) guess’ as in Example (4), the volitional form of verbs (nom-oo ‘will drink’ or ‘let’s drink’) as in Example (5), and other suffixes or suffix-like expressions such as conjecture, opinion, volition, order, request, and question, as in Yoshii (1977) above. The current study observed the transition in the distribution of evident overt expressions like these within each clause. Table 3 presents the distribution of such expressions occurring in the main clauses in the case of clause order (i) (i.e., canonical). Table 4 shows the case of clause order (ii) (i.e., inverted). What is noteworthy in Table 3 is that expressions of epistemic or speech-act conjunction interpretation appear explicitly in the main clause in more than half of all the examples from around 1800 onwards in the case of canonical clause order. Table 3.â•‡ Development of the distribution of overt expressions in main clauses in (i) ‘CL1-kara CL2’

with overt expressions without overt expressions TOTAL TOTAL

–1750

–1800

–1850

–1900

–1950

–2000 (wr)

–2000 (sp)

0 â•⁄ (0%) 9 (100%) 9

55 (65.7%) 29 (34.3%) 84

25 (50.0%) 25 (50.0%) 50

26 (68.4%) 12 (32.6%) 38

45 (57.1%) 22 (32.9%) 67 248

27 (61.4%) 17 (38.6%) 44

31 (64.6%) 17 (35.4%) 48 92

2. This may be a phenomenon that is observable in conversational language and, thus, genrespecific. My sincere thanks go to Elizabeth Traugott for her comment on this point. Further study of kara-clauses based on a larger corpus in various genres will be one of our future tasks.

Are kara ‘because’-clauses causal subordinate clauses in present-day Japanese? 

Table 4.â•‡ Development of the distribution of overt expressions in main clauses in (ii) ‘CL2, CL1-kara’

with overt expressions without overt expressions TOTAL TOTAL

–1750

–1800

–1850

–1900

–1950

–2000 (wr)

–2000 (sp)

0 (0%) 1 (100%) 1

0 (0%) 3 (100%) 3

0 (0%) 3 (100%) 3

6 (75.0%) 2 (25.0%) 8

18 (66.7%) 9 (33.3%) 27 42

21 (60.0%) 14 (40.0%) 35

5 (45.5%) 6 (54.5%) 11 46

Table 4 shows that such expressions came to appear explicitly in the main clause from around 1900 onwards in the case of inverted clause order. That is to say, karaclauses often give epistemic or speech-act conjunction interpretation rather than content-conjunction interpretation in conversational data in more recent years. Next, let us examine the distribution of evident overt expressions in karaclauses. Table 5 indicates the distribution of the overt expressions that appear in the kara-clause in the case of clause order (i) (i.e., canonical). Table 6 shows the case of clause order (ii) (i.e., inverted). What is noteworthy here is that evident expressions of epistemic or speech-act conjunction occur in kara-clauses from around 1800 onwards in the case of canonical clause order and from around 1900 onwards in the case of inverted clause order. Kara-clauses having these evident expressions in themselves can not be considered as modifying the modal in their main clauses, thus can be considered to be an indication that they were becoming more independent from the main clause gradually from around 1800. Table 5.â•‡ Development of the distribution of overt expressions in kara-clauses in (i) ‘CL1-kara CL2’

with overt expressions without overt expressions TOTAL TOTAL

–1750

–1800

–1850

–1900

–1950

–2000 (wr)

–2000 (sp)

0 (0%) 9 (100%) 9

25 (29.8%) 59 (70.2%) 84

17 (34.0%) 33 (66.0%) 50

18 (47.4%) 20 (52.6%) 38

24 (35.8%) 43 (64.2%) 67 248

15 (34.1%) 29 (65.9%) 44

21 (43.8%) 27 (56.3%) 48 92

 Yuko Higashiizumi

Table 6.â•‡ Development of the distribution of overt expressions in kara-clauses in (ii) ‘CL2, CL1-kara’

with overt expressions without overt expressions TOTAL TOTAL

–1750

–1800

–1850

–1900

–1950

–2000 (wr)

–2000 (sp)

0 (0%) 1 (100%) 1

0 (0%) 3 (100%) 3

0 (0%) 3 (100%) 3

3 (37.5%) 5 (62.5%) 8

10 (37.0%) 17 (63.0%) 27 42

10 (28.6%) 25 (71.4%) 35

4 (36.4%) 7 (63.6%) 11 46

Table 7.â•‡ Development of the distribution of overt expressions in (iii) ‘CL-kara’

with overt expressions without overt expressions TOTAL TOTAL

–1750

–1800

–1850

–1900

–1950

–2000 (wr)

–2000 (sp)

0 (0%) 0 (0%) 0

0 (0%) 2 (100%) 2

0 (0%) 0 (0%) 0

0 (0%) 1 (100%) 1

5 (18.5%) 12 (81.5%) 17 20

4 (44.4%) 5 (55.6%) 9

17 (37.8%) 28 (62.2%) 45 54

Lastly, Table 7 presents the distribution of evident expressions when a kara-clause appears without any accompanying main clause. The expressions increase in proportion from around 1950. To sum up, the evident expressions of epistemic or speech-act conjunction interpretation occur explicitly in both kara-clauses and main clauses. In other words, kara-clauses are used to express epistemic or speech-act conjunction rather than content conjunction in conversational data in more recent years. Roughly speaking, the trend is for epistemic or speech-act conjunction interpretation. 4.3

Summary of the diachronic process of kara-clauses

As noted in Table 1 and at the beginning of Section 3, kara-clauses with epistemic or speech-act conjunction interpretation are more paratactic than those with content conjunction interpretation. Kara-clauses can be considered as not modifying the proposition in the main clause but rather modifying the speech-act being performed in the main clauses.

Are kara ‘because’-clauses causal subordinate clauses in present-day Japanese? 

In terms of clause combination, the relationship between kara-clauses and their main clause has become more paratactic to each other. As Table 2 shows, the proportion of kara-clauses without a main clause is increasing. That is to say, the relationship between kara-clauses and the main clause has extended from hypotactic to paratactic to independent over time. In terms of interpretation, kara-clauses tend towards an epistemic or speech-act conjunction interpretation. As shown in Tables 3–7, the expressions of epistemic or speech-act conjunction interpretation appearing in both kara-clauses and main clauses are on the increase. The diachronic process under investigation can be summarized as shown in Figure 2. 5. Discussion The diachronic process and synchronic diversity of kara-clauses adds new evidence to oppose the sharp dichotomy between coordination and subordination. As I have shown so far, there are examples of kara-clauses that cannot be considered coordinate or subordinate (i.e., hypotactic in Hopper & Traugott’s continuum) in the strict sense. Moreover, these clauses illustrate that the three-way continuum proposed in Hopper & Traugott (2003) is more useful in describing their diachronic process and synchronic diversity in present-day Japanese conversation. As Figure 2 shows, kara-clauses are becoming more paratactic clauses. I propose that the three-way continuum is more relevant to describe the diachronic process and synchronic diversity of kara-clauses than the two-way distinction, although they are extending in a direction away from the cline of clause-combining constructions in grammaticalization postulated by Hopper & Traugott (2003). Günthner (1996) claims that the weil ‘because’-clause in present-day German conversation is developing from hypotactic to paratactic clause. In sum, the three-way continuum is more useful than the sharp two-way distinction for comprehensively understanding the diachronic process and synchronic variation in clause combination as far as kara-clauses conversational data are concerned. hypotactic

→ paratactic

→ more paratactic or independent

CL1-kara CL2. → CL1-kara CL2. → CL-kara 0 CL2, CL1-kara. → 0 CL-kara

Figure 2.â•‡ Diachronic process of kara-clauses

 Yuko Higashiizumi

6. Conclusion The subordination – hypotaxis – parataxis continuum of complex clause constructions is more useful than the traditional dichotomy in analyzing the diachronic process and synchronic diversity of kara-clauses in conversational data in Japanese. Acknowledgements This paper is a revision of the one presented to the panel on “Subordination in conversation: crosslinguistic analyses of form-function matches” at the 10th International Pragmatics Conference in Gothenburg, Sweden, 8–13 July 2007. My special thanks go to Ritva Laury, Ryoko Suzuki, and the panel members and participants for their stimulating discussions and comments. A contrastive study of the English because-clause and the Japanese kara-clause from the perspective of grammaticalization was given at the 4th International Conference “New Reflections on Grammaticalization” in Leuven, Belgium, 16–19 July 2008. I sincerely thank Elizabeth Traugott and the participants for their insightful comments. I am grateful to Jane Boughton and Maggie Camp for correcting my English. All remaining errors and faults are my own. Abbreviations cl ger int nmlz nom ptcl vol

clause gerundive interjection nominalizer nominative particle volitional

References Alfonso, Anthony. 1980. Japanese Language Patterns: A Structural Approach. Tokyo: Sophia University L. L. Center of Applied Linguistics. Brinton, Laurel J. 1996. Pragmatic Markers in English: Grammaticalization and Discourse Functions. Berlin: Mouton de Gruyter. Cristofaro, Sonia. 2003. Subordination. Oxford: Oxford University Press. Günthner, Susanne. 1996. “From subordination to coordination?: Verb-second position in German causal and concessive constructions.” Pragmatics 6 (3): 323–356.

Are kara ‘because’-clauses causal subordinate clauses in present-day Japanese? 

Haiman, John, and Thompson, Sandra A. 1984. “‘Subordination’ in universal grammar.” Berkeley Linguistics Society 10: 510–523. Higashiizumi, Yuko. 2006. From a Subordinate Clause to an Independent Clause: A History of English because-clause and Japanese kara-clause. Tokyo: Hituzi Syobo Publishing. Honda, Akira. 2005. Affordance no Ninti Imiron: Seitai-Sinrigaku kara Mita Bunpoo Gensyoo (An Affordance-theoretical Cognitive Semantics: Grammatical Phenomena Seen from an Ecological-Psychological View). Tokyo: University of Tokyo Press. Hopper, Paul J. and Traugott, Elizabeth C. 2003. Grammaticalization. Cambridge: Cambridge University Press. Iguchi (Higashiizumi), Yuko. 1998. “Functional variety in the Japanese conjunctive particle kara ‘because’.” In Studies in Japanese Grammaticalization: Cognitive and Discourse Perspectives, Toshio Ohori (ed.), 99–128. Tokyo: Kurosio. Ishigaki, Kenji. 1955. Zyosi no Rekisiteki Kenkyuu (Historical Study of Particles). Tokyo: Iwanami Shoten. Jacobs, Andreas and Jucker, Andreas H. 1995. “The historical perspective in pragmatics.” In Historical Pragmatics: Pragmatic Developments in the History of English, Andreas H. Jucker (ed.), 3–33. Amsterdam/Philadelphia: John Benjamins. Kobayashi, Kenji. 1996. Nihongo Zyooken Hyoogen-si no Kenkyuu (A Study of the History of Conditionals in Japanese). Tokyo: Hituzi Syobo. Konoshima, Masatoshi. 1966. Kokugo Zyosi no Kenkyuu: Zyosi-si Sobyoo (A Study of Particles in Japanese: An Outline of the History of Particles). Tokyo: Ofusha. Konoshima, Masatoshi. 1983. Zyodoosi Zyosi Gaisetu (An Outline of Auxiliaries and Particles). Tokyo: Ofusha. Lee, Reien. 2000. Nihongo Bogo Wasya no Zatudan ni Okeru “Monogatari” no Kenkyuu (A study of spoken narratives by native speakers of Japanese). Tokyo: Kurosio. Lehmann, Christian. 1988. “Towards a typology of clause linkage.” In Clause Combining in Grammar and Discourse, John Haiman and Sandra A. Thompson (eds), 181–225. Amsterdam/Philadelphia: John Benjamins. Lindström, Jan and Londén, Anne-Marie. 2007. “From syntactic to discursive subordination: Consecutive, adversative and causal conjunctions in Swedish conversation.” Paper presented to the panel on “Subordination in conversation: Crosslinguistic analyses of form-function matches” at the 10th International Pragmatics Conference in Göteborg, Sweden, 8–13 July 2007. Martin, Samuel. 1975. A Reference Grammar of Japanese. New Haven: Yale University Press. Masuoka, Takashi. 1997. Hukubun (Complex sentences). Tokyo: Kurosio. Matthiessen, Christian, and Thompson, Sandra A. 1988. “The structure of discourse and ‘subordination.’” In Clause Combining in Grammar and Discourse. John Haiman and Sandra A. Thompson (eds), 275–329. Amsterdam/Philadelphia: John Benjamins. Minami, Fujio. 1974. Gendai Nihongo no Koozoo (The Structure of Modern Japanese). Tokyo: Taishukan Shoten. Minami, Fujio. 1993. Gendai Nihongo Bunpoo no Rinkaku (The Outline of Modern Japanese Syntax). Tokyo: Taishukan Shoten. Nagano, Masaru. 1979 [1952]. “Kara to node wa doo tigau ka (How are kara and node different from each other?).” In Nihon no Gengogaku 4, Shiro Hattori, Susumu Ono, Atsuyoshi Sakakura, and Akira Matsumura (eds), 467–488. Tokyo: Taishukan Shoten. Nakayama, Toshihide, and Ichihashi-Nakayama, Kumiko. 1997. “Japanese kedo: discourse genre and grammaticization.” In Japanese/Korean Linguistics 6, Ho-min Sohn and John Haig (eds), 607–618. Stanford, CA.: CLSI.

 Yuko Higashiizumi Ohori, Toshio. 1995. “Remarks on suspended clauses: A contribution to Japanese phraseology.” In Essays in Semantics and Pragmatics, Masayoshi Shibatani and Sandra A. Thompson (eds), 201–218. Amsterdam/Philadelphia: John Benjamins. Ohori, Toshio. 1997. “Framing effects in Japanese non-final clauses: Toward an optimal grammar – pragmatics interface.” Berkeley Linguistics Society 23: 471–480. Okamoto, Shigeko. 1995. “Pragmaticization of meaning in some sentence-final particles in Japanese.” In Essays in Semantics and Pragmatics, Masayoshi Shibatani and Sandra A. Thompson (eds), 219–246. Amsterdam/Philadelphia: John Benjamins. Onodera, Noriko O. 2004. Japanese Discourse Markers: Synchronic and Diachronic Discourse Analysis. Amsterdam/Philadelphia: John Benjamins. Quirk, Randolph, Greenbaum, Sidney, Leech, Geoffrey, and Svartvik, Jan. 1985. A Comprehensive Grammar of the English Language. London: Longman. Shirakawa, Hiroyuki. 1995. “Riyuu o arawasanai kara (Kara that does not express reason).” In Hukubun no Kenkyuu, zyoo (Studies of Complex Sentences, vol. 1), Yoshio Nitta (ed.), 189–219. Tokyo: Kurosio. Shirakawa, Hiroyuki. 2009. ‘Iisasi-bun’ no Kenkyuu (Studies on Suspended Sentences). Tokyo: Kurosio. Suzuki, Ryoko. 1999. Grammaticization in Japanese: A Study of Pragmatic Particle-ization. Ph.D. dissertation, University of California, Santa Barbara. Suzuki, Ryoko. 2009. “Interactional profile of causal kara-clause in Japanese conversation: Because ‘main’ clause is there or not.” Paper presented to the panel on “Clause combining in discourse: the interplay between structure and pragmatics” at the 11th International Pragmatics Conference, Melbourne, Australia, 12–17 July 2009. Sweetser, Eve. 1990. From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. Takubo, Yukinori. 1987. “Toogo koozoo to bunmyaku zyoohoo (Syntactic structure and contextual information).” Nihongogaku 6 (5): 37–48. Traugott, Elizabeth Closs. 2003. “From subjectification to intersubjectification.” In Motives for Language Change, Raymond Hickey (ed.), 124–139. Cambridge: Cambridge University Press. Uno, Ryoko. 2009. Detecting and Sharing Perspectives Using Causals in Japanese. Tokyo: Hituzi Syobo Publishing. Verstraete, Jean-Christophe. 2007. Rethinking the Coordinate – Subordinate Dichotomy: Interpersonal Grammar and the Analysis of Adverbial Clauses in English. Berlin: Mouton de Gruyter. Yabe, Hiroko. 1997. “Nokketyau kara ne-kara soositeorimasu node-made (From nokketyau kara ne to soositeorimasu node).” In Zyosei no Kotoba: Syokuba-hen (Women’s language: In the workplace), Gendai Nihongo Kenkyuukai (ed.), 139–154. Tokyo: Hituzi Syobo Publishing. Yabe, Hiroko. 2002. “Kara-to node-no siyoo ni miru syokuba no dansei-no gengokoodoo (Men’s language use in the workplace seen from the use of kara and node).” In Dansei no Kotoba: Syokuba-hen (Men’s Language: In the workplace), Gendai Nihongo Kenkyuukai (ed.), 133–148. Tokyo: Hituzi Syobo Publishing. Yamaguchi, Gyoji. 1996. Nihongo Setuzokuhoosiron (A Theory of the History of Clause Linkage in Japanese). Osaka: Izumi Shoin. Yoshii, Kazuto. 1977. “Kindai Tokyoogo ingakankei hyoogen-no tuuziteki koosatu: kara to nodeo tyuusin-to-site (A diachronic consideration of the expression of causal relation in the modern Tokyo dialect: With special reference to kara and node).” Kokugogaku 110: 19–36.

Are kara ‘because’-clauses causal subordinate clauses in present-day Japanese? 

Appendix List of Data and URL 1592. Amakusaban Heike Monogatari [NBTF] 1593. Amakusaban Isoho Monogatari [NBTF] 1703. Sonezaki sinzyuu, by Chikamatsu Monzaemon. Chikamatsu Zyoorurisyuu. Tokyo: Yuhodo. 1912–1914. [JTIC]. 1713. Sukeroku. Koten Nihon Bungaku Taikei 98. Tokyo: Iwanami Shoten. 1965. [NKB]. 1721. Sinzyuu yoigoosin, by Chikamatsu Monzaemon. Koten Nihon Bungaku Taikei 49. Tokyo: Iwanami Shoten. 1958. [NKB]. 1736. Sibaraku. Koten Nihon Bungaku Taikei 98. Tokyo: Iwanami Shoten. 1965. [NKB]. 1742. Kenuki. Koten Nihon Bungaku Taikei 98. Tokyo: Iwanami Shoten. 1965. [NKB]. 1748. Kanadehon tyuusingura, by Izumo Takeda, Miyoshi Shoraku, and Namiki Senryu. Tokyo: Iwanami Shoten. 1937. [JTIC]. 1770. Tatumi no sono, by Muchusanjin Negoto Sensei. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB]. 1770. Yuusihoogen, by Inakaroojin Tadano Jijii. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB]. 1772. Kanoko moti, by Kimuro Booun. Koten Nihon Bungaku Taikei 100. Tokyo: Iwanami Shoten. 1966. [NKB]. 1773. Kiki zyoozu, by Komatsuya Hyakki. Koten Nihon Bungaku Taikei 100. Tokyo: Iwanami Shoten. 1966. [NKB]. 1776. Koomanzai angya nikki, by Koikawa Harumachi. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB]. 1778 (?). Dootyuu sugoroku. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB]. 1779. Tai no misozu, by Shokusanjin. Koten Nihon Bungaku Taikei 100. Tokyo: Iwanami Shoten. 1966. [NKB]. 1782. Gozonzi no syoobaimono, by Kitao Masanobu. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB]. 1783. Uzisyuui, by Shumokuan Aruji. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB]. 1785. Daihi no senroppon, by Shiba Zenko. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB]. 1785. Edo umare uwaki no kabayaki, by Santo Kyoden. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB]. 1785. Kiruna no ne kara kane no naru ki, by Tourai Sanna. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB]. 1787. Soomagaki, by Santo Kyoden. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB]. 1788. Bunbu nidoo mangoku toosi, by Houseido Kisanji. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB]. 1789. Koosizima tokini aizome, by Santo Kyoden. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB].

 Yuko Higashiizumi 1790. Keiseikai syizyuuhatte, by Santo Kyoden. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB]. 1790. Shingaku hayasomekusa, by Santo Kyoden. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB]. 1791. Nisiki no ura, by Santo Kyoden. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB]. 1795. Katakiuti gizyo no hanabusa, by Nansensho Somahito. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB]. 1798. Keiseikai hutasuzimiti, by Umebori Kokuga. Koten Nihon Bungaku Taikei 59. Tokyo: Iwanami Shoten. 1958. [NKB]. 1801. Meika no tokuu mimasu no tamagaki, by Sakurada Jisuke. Koten Nihon Bungaku Taikei 54. Tokyo: Iwanami Shoten. 1961. [NKB]. 1802–1822. Tokaidootyuu hizakurige, by Jippensha Ikku. Koten Nihon Bungaku Taikei 62. Tokyo: Iwanami Shoten. 1958. [NKB]. 1809. Ukiyoburo, by Shikitei Samba. Koten Nihon Bungaku Taikei 63. Tokyo: Iwanami Shoten. 1957. [NKB]. 1813. Osome Hisamatu ukina no yomiuri, by Tsuruya Namboku. Koten Nihon Bungaku Taikei 54. Tokyo: Iwanami Shoten. 1961. [NKB]. 1832–1833. Syunsyoku umegoyomi, by Tamenaga Shunsui. Koten Nihon Bungaku Taikei 64. Tokyo: Iwanami Shoten. 1962. [NKB]. 1833–35. Syunsyoku tatumi no sono, by Tamenaga Shunsui. Koten Nihon Bungaku Taikei 64. Tokyo: Iwanami Shoten. 1962. [NKB]. 1842 (?). Kagekiyo. Koten Nihon Bungaku Taikei 98. Tokyo: Iwanami Shoten. 1965. [NKB]. 1843. Narukami. Koten Nihon Bungaku Taikei 98. Tokyo: Iwanami Shoten. 1965. [NKB]. 1859. Kosode soga azami no ironui, by Kawatake Mokuami. Koten Nihon Bungaku Taikei 54. Tokyo: Iwanami Shoten. 1961. [NKB]. 1887. Ukigumo, by Futabatei Shimei. Tokyo: Shun’yodo. 1947. [JTIC]. 1895. Takekurabe, by Higuchi, Ichiyo. Shinchoo-bunko no 100-satsu. Tokyo: Shinchosha. 1896. Wakaremiti, by Higuchi, Ichiyo. Shinchoo-bunko no 100-satsu. Tokyo: Shinchosha. 1897. Konzikiyasya, by Ozaki, Koyo. Shinchoo-bunko no 100-satsu. Tokyo: Shinchosha. 1902. Zyunsa, by Kunikida, Doppo. Shinchoo-bunko no 100-satsu. Tokyo: Shinchosha. 1906. Kusamakura, by Natsume, Soseki. Shinchoo-bunko no 100-satsu. Tokyo: Shinchosha. 1907. Heibon, by Futabatei, Shimei. Shinchoo-bunko no 100-satsu. Tokyo: Shinchosha. 1912–1913. Gan, by Mori Ogai. Tokyo: Iwanami Shoten. 1948. [JTIC]. 1914. Kokoro, by Natsume, Soseki. Shinchoo-bunko no 100-satsu. Tokyo: Shinchosha. 1918. Umareizuru nayami, by Arishima, Takeo. Shinchoo-bunko no 100-satsu. Tokyo: Shinchosha. 1918. Zigokuhen, by Akutagawa Ryunosuke. Tokyo: Hosokawa Shoten. 1947. [JTIC]. 1920. Yuuzyoo, by Mushanokoji, Saneatsu. Shinchoo-bunko no 100-satsu. Tokyo: Shinchosha. 1924. Tizin no ai, by Tanizaki, Jun’ichiro. Shinchoo-bunko no 100-satsu. Tokyo: Shinchosha. 1927. Kappa, by Akutagawa Ryunosuke. Tokyo: Hosokawa Shoten. 1946. [JTIC]. 1935. Aside, by Ishikawa, Jun. Shinchoo-bunko no 100-satsu. Tokyo: Shinchosha. 1936. Kaze tatinu, by Hori, Tatsuo. Shinchoo-bunko no 100-satsu. Tokyo: Shinchosha.

Are kara ‘because’-clauses causal subordinate clauses in present-day Japanese? 

1941. Roboo no isi, by Yamamoto, Yuzo. Shinchoo-bunko no 100-satsu. Tokyo: Shinchosha. 1947. Syozyo kaitai, by Ishikawa, Jun. Shinchoo-bunko no 100-satsu. Tokyo: Shinchosha. 1952. Nobi, by Oka, Shohei. Shincho-bunko no 100-satsu. Tokyo: Shinchosha. 1956. Kinkakuzi, by Mishima, Yukio. Shincho-bunko no 100-satsu. Tokyo: Shinchosha. 1964. Nireke no hitobito, by Kita, Morio. Shincho-bunko no 100-satsu. Tokyo: Shinchosha. 1967. Amerika hiziki, by Nosaka, Akiyuki. Shincho-bunko no 100-satsu. Tokyo: Shinchosha. 1970. Kokoo no hito, by Nitta, Jiro. Shincho-bunko no 100-satsu. Tokyo: Shinchosha. 1978. Shimbashi karasumoriguti seisyun-hen, by Shina, Makoto. Shincho-bunko no 100-satsu. Tokyo: Shinchosha. 1982. Issyun no natu, by Sawaki, Kotaro. Shincho-bunko no 100-satsu. Tokyo: Shinchosha. 1985. Taroo monogatari, kookoo hen, by Sono, Ayako. Shincho-bunko no 100-satsu. Tokyo: Shinchosha. (Z95–6), (Z95–15): 1995. Zatudan siryoo (Transcriptions of chats) in Lee (2000). 1999–2000. Osyaberi namatyuukei (Real-life chats) in Nihongo Journal (Japanese Journal), 1999.4–2000.3. Tokyo: Aruku.

URLs NBTF: Nihon Bungaku-too Text File (www.let.osaka-u.ac.jp/~okajima/bungaku.htm) JTIC: J apanese Text Initiative Collection, University of Virginia (http://etext.virginia.edu/japanese/texts.html) NKB: Nihon Koten Bungaku Sakuhin Database (http://www.nijl.ac.jp)

Teyuuka and I mean as pragmatic parentheticals in Japanese and English Ritva Laury and Shigeko Okamoto

University of Helsinki and UC Santa Cruz The English I mean and the Japanese teyuuka differ syntactically and semantically, but they have similar pragmatized uses. Both verbs, mean and yuu, function as regular verbs in main clauses and also as part of formulaic expressions which indicate a modal meaning with respect to an utterance, or project back to an earlier utterance and index it as inadequate or in need of modification. Both constructions can also frame another expression as a modification of the earlier utterance. They also function metacommunicatively to manage the interaction on a strategic level. The article compares the structure and functions of these two constructions in conversation and shows how structurally different expressions used in certain kinds of discourse and interactional contexts have come to serve similar but not identical pragmatic needs.

1. Introduction. In this study, we consider two constructions: the English I mean and the Japanese teyuuka (and its variant forms, such as tteyuuka, toiimasuka). Each of these expressions has been studied extensively, but separately (e.g. Schiffrin 1987; Imo 2006; Brinton 2007; Maynard 2004; Kitano 2005; McGloin 2007). The verbs mean and yuu ‘say’ both can and do function as complement-taking predicates (CTPs; Thompson 2002), as illustrated in Examples (1) and (2).

(1) By reading I mean that I treat the charts as texts and attempt to tease out the assumptions that go into constructing it (FLOB; Brinton 2007: 44).

(2) Uun, watashi mata tsuretette tte yuu no ne. no me again take prt say prt prt ‘No, please take me (with you) again, she says.’ In its CTP use, as in (1), I mean is followed by the complementizer that, which in turn is followed by the complement clause; that is, it has the structure of ‘I mean

 Ritva Laury and Shigeko Okamoto

that X (complement clause)’. In (2) tte is a quotative particle, indicating that what precedes it, or its complement, is reported speech; it is then followed by yuu ‘say’ without an overt subject; that is, it has the structure of ‘X (complement clause) tte yuu’. The particle ka in teyuuka is a question marker (cf. McGloin 2007). As these descriptions show, the syntactic and semantic composition of I mean and teyuuka are quite different from each other. However, I mean and teyuuka have similar pragmaticized uses in which they do not function as prototypical matrix clauses but instead serve as formulaic expressions or grammatical constructions which indicate a modal meaning with respect to the “main” part of the utterance, or serve to project back to an earlier utterance and index it as inadequate or in need of modification. Both can also frame another expression as a modification of the earlier utterance. That is, they function metalinguistically. They also appear to function metacommunicatively; in other words, they serve to manage the interaction underway on a strategic level (Hopper & Thompson 2008; see also Günthner, this volume; Pekarek Doehler, this volume; Koivisto, Laury & Seppänen, this volume). In this study, we examine the structure and functions of these two constructions in conversation and consider how and why they are similar to and different from each other. Through this comparison, we hope to show how structurally quite different expressions used in certain kinds of discourse and interactional contexts come to “acquire” similar but not identical pragmatic functions. 2. Data The data for I mean come from the Santa Barbara Corpus of Spoken American English (Du Bois et al. 2000–20051); those for teyuuka are drawn from a corpus of twelve Japanese conversations (each from 20 to 30 minutes long) taped and transcribed by the second author, as well as transcripts of twenty two interviews of female high school students by a male writer (Murakami 1998).2

1. In the Du Bois et al. transcription system, line breaks indicate intonation unit boundaries, and line final punctuation reflects intonation contours. Two dots represent short pauses, while three dots represent longer pauses. Brackets are used to indicate overlap between speakers. Numbers within brackets indicate who overlaps with whom. The following symbols are used: @ (laughter quality), % (glottalization), = (lengthening), H (inhalation). Segments of talk marked with X represent uncertain hearings, segments marked with MRC indicate marked voice quality, and segments marked with P are said more quietly than surrounding talk. 2. Ryu Murakami, a well-known writer, interviewed 51 female high school students (2, 3, or 4 students at a time). The transcripts of these interviews are published in a book form; however, they are not detailed transcripts that would include pauses, overlaps, etc.

Teyuu ka and I mean as pragmatic parentheticals in Japanese and English 

3. The structure and function of I mean 3.1

The syntactic environments of I mean

Although it is clear that mean functions in English as a complement-taking predicate (see discussion in Section 1 and Example (1)), in many previous studies, I mean has been classified as a – pragmatic parenthetical3 (e.g. Stenström 1995; Brinton 2007) – discourse marker (Schiffrin 1987, Imo 2005) – discourse particle (Goldberg 1980) The use of these terms shows that in its use as a formulaic expression, I mean has come to resemble other expressions in English and in other languages typically containing speech act verbs or mental verbs such as the English think, guess and know, which, together with a first- or second-person subject, have become through frequent use phonetically reduced and de-syntacticized, or particle-ized, so that they are no longer confined to use as complement-taking constructions, but can instead occur in a number of syntactic positions within an utterance without being followed by a nominal or clausal complement. (See, for example, Östman 1981, Thompson & Mulac 1991, Brinton 1996, 2007, Scheibman 2001, Thompson 2002, Kärkkäinen 2003 on English; Saari 1984, 1986 on Swedish; Keevallik 2003 on Estonian; Hakulinen, Keevallik Eriksson & Lindström 2003 on Finnish, Estonian and Swedish; Hakulinen & Seppänen 1992 on Finnish). Further, as we will show, and as shown by Kärkkäinen (2003, 2007) for I think and I guess, the particle-ization of I mean has advanced to the degree where it is used to link to other speakers’ utterances, and in that use, its meaning has been bleached to the degree where I can no longer be said to index the current speaker. In our data, I mean occurs in a variety of syntactic environments, as noted by other researchers studying this expression (see e.g. Imo 2005; Brinton 2007). Table 1 roughly summarizes the syntactic environments in which I mean occurred in our data. In the table, X is the utterance that I mean indexes as somehow inadequate or in need of modification, and Y is the utterance that does the modification. X and Y can be either phrases or clauses.

3. The term was originally (as far as we know) due to Urmson, who used it to refer to a class of verbs such as know and guess which are used “as signals guiding the hearer to a proper appreciation of the statement in its context, social, logical or evidential (1952: 495).” The term ‘(epistemic) parenthetical’, as well as ‘epistemic phrase’, is also used by Thompson and Mulac (1991). Another term we think would be appropriate for the use of I mean in our data is the term ‘(evaluative) fragment’ used by Thompson (2002) and Hopper & Thompson (2008).

 Ritva Laury and Shigeko Okamoto

Table 1.â•‡ Syntactic environments of I mean Utterance-medially:

X I mean Y

Utterance-initially:

X.4 I mean Y

Utterance-finally

X. Y I mean

(X and Y are uttered by the same speaker or different speakers; another turn may intervene.) (Another turn may intervene.)

The most frequent syntactic environment in which I mean occurs, as also noted in earlier studies (Brinton 2007: 43), is at the beginning of a clause, as in Example (3) below. This can be seen to reflect its original use as a main clause. However, in our spoken data, consistent with its status as a pragmatic parenthetical, I mean is never followed by the complementizer that. In Example (3), I mean, although it is followed by a full clause, is not best analyzed as a complement-taking main clause, because it is not followed by that, and because the clause that follows it is not an indirect question (and therefore not a complement clause) as shown by the word order. (3)

SBC0012.TRN 1 ... Who the hell do you know, 2 goes to a party in a suit. 3 (H) I mean who .. is this g=eek.

In our data, I mean also occurs clause-internally, although this is much more rare than clause-initial use. Example (4) shows a clause-internal use of I mean. (4)

SBC 0010.TRN 1 PHIL: Because in board discussion, 2 ... I mean, 3 BRAD: [Yeah]. 4 PHIL: [o-] on one hand, 5 as a member she has every right.

The syntactic unit which I mean occurs with or initiates does not need to be a clause, as shown in the next example. (5)

SBC 001 1 (H) And like in some of em? 2 ... (TSK) blood will spurt out, 3 you know, 4 and I mean,

4. In the table, the period indicates that the preceding utterance (X) has been brought to a close syntactically, prosodically and pragmatically (in the sense of Ford & Thompson 1996).

Teyuu ka and I mean as pragmatic parentheticals in Japanese and English 

5 oh. 6 (H) It’s @nothing @ – 7 <@SM (H) it’s the grossest thing SM@>. In Example (5), I mean is followed by a particle, oh5. I mean has an especially strong affinity to you know; they often occur in the same turn, as in Example (5), and even immediately adjacent to each other, with either one occurring first. In the next example, the syntactic unit with which I mean occurs is a prepositional phrase. Example (6) also shows that in contrast to claims made in earlier research (e.g. Imo 2005), I mean also occurs in final position in an utterance. This excerpt comes from a multi-party conversation. Miles has just been telling about rather shocking events he has witnessed at a club that several of the participants are familiar with. Jamie then asks a clarification question. (6)

SBC0002.TRN 1 JAMIE: [So] – 2 This is all [2at Bahia2]? 3 MILES: [2strangers2]. 4 JAMIE: All this stuff happens at [3Bahia3]. 5 MILES: [3Yeah3]. 6 JAMIE: Hunh? 7 MILES: ... To %other people I mean. 8 H,_J,_P_LAUGHING_8.0_SEC))

It should be noted that in this example, in contrast to the other examples in our data, I mean is produced with a final intonation contour. Unlike the uses of I mean in our other examples, where it is produced with a non-final intonation contour, it does not project forward to more talk to follow here. The prepositional phrase uttered by Miles in line 7 can be analyzed as a syntactic extension of Jamie’s preceding turn in line 4 or perhaps as a continuation of Miles’ own turn in line 56. Obviously, in this syntactic position, as in clause-internal position (Example 4) and in 5. An interesting question here is how far I mean carries here. It is quite possible that in addition to the particle in line 5, the utterances in lines 6 and 7, which are said with a smile, are also meant as a modification (amplification or explanation) of what is reported in line 2. Oh is not lengthened or prosodically marked here, but it could still be affectively loaded; if I mean carries to lines 6 and 7, then the affect is not shown until the later lines. On the other hand, lines 6 and 7 could amplify the particle. Prosodically, of course, oh is more clearly linked to I mean than lines 6 and 7 are: line 4 ends with a non-final intonation contour, while line 5 ends with a final contour. 6. In a sense, it may not matter to the participants which utterance the prepositional phrase supplements or links to. Jamie’s turn in line 4 is a candidate understanding of Miles’ earlier telling, which Miles confirms in line 5, and then clarifies further in line 7. At this point, the stories have become a shared understanding.

 Ritva Laury and Shigeko Okamoto

a position preceding a particle (Example 5), it cannot be considered a CTP, but rather it functions as a pragmatic parenthetical or particle. In this section, we have seen that I mean can occur in a number of syntactic positions. We have seen that it occurs initially in a clause (Example 3), clause-internally after an initial adverbial (4), before a particle (5), and after a clause-final adverbial, which may be retrospectively added to an already syntactically complete utterance, as in (6). In our data, we also found it preceding phrasal utterances and between clauses. However, I mean is at least rare constituent-internally: we did not find it in such environments in our data. This may be connected to the ‘tightness’ of English syntax, noted by Fox, Hayashi & Jasperson (1996). As can be seen from these examples, I mean occurs in syntactic positions where it cannot be characterized as a complement-taking construction, but rather functions as a particle. 3.2

The pragmatic function of I mean

Formulaic, grammaticalized expressions which originate from a combination of a speech act verb or mental verb with a first- or second-person subject have been characterized by Thompson (2002: 142–146) as “epistemic/evidential/evaluative fragments”, which are used by speakers to frame or to take a stance toward an utterance. The type of stance that I mean has been said to take toward an utterance could perhaps be most closely described as metalinguistically evaluative. In many previous studies the function of I mean has been discussed in terms of modification or ‘editing’ of various sorts, which may also be termed repair, in particular self-repair (Schiffrin 1987: 299). In that sense, it can be considered a metalinguistically evaluative device, since repair may be connected to a speaker’s evaluation of his or her own utterance as somehow inadequate or at least requiring modification (although it is clear that repair is associated with functions much beyond the correcting of mistakes, as already pointed out by Schegloff, Jefferson & Sacks 1977). This evaluation, of course, may be based on the speaker’s own intentions, and/or it may be based on feedback of various sorts from recipients. In addition to repairing a previous utterance or a part of it, I mean, as a metalinguistically evaluative device, can also be used to supplement it in various ways by elaborating, clarifying, expanding on, or explaining what has previously been said (for a useful summary, see Brinton 2007). In this paper, we take the position that the function, or basic meaning, of I mean is actually much more simple than has been previously discussed (see also Fox Tree & Schrock 2002). In our view, I mean indexes a connection between its associated utterance (the utterance it precedes or, in some cases, follows) and preceding talk, indicates that that preceding utterance is not quite adequate, and frames its associated utterance as an adjustment or modification (Schiffrin 1987),

Teyuu ka and I mean as pragmatic parentheticals in Japanese and English 

most often to the speaker’s own talk but also, in a few examples in our data, to talk by others. More specific functions such as repair, elaboration, clarification, exemplification, mitigation, etc. are not specified by I mean but derive from the context and the semantic relationship between the earlier utterance and the utterance accompanied by I mean. Thus, I mean always links backward to the talk it makes relevant or evaluates as inadequate. In contrast to teyuuka (see below), I mean is always accompanied by an utterance intended as an adjustment or modification of the prior utterance. This utterance, as we saw in the preceding section, can either follow or, less commonly, precede I mean. In initial use, when it is pronounced with a non-final intonation contour, it projects more talk to follow, but it can also be used in syntactically and prosodically final position, where it does not project more talk, as shown in Example (6) above. Previous studies have noted that I mean is used in contexts of self-initiated same-turn self-repair (Goldberg 1980). We find it in just those contexts in our data as well. Example (7) contains such a use of I mean. (7)

SBC007.TRN 1 MARY: ... Hand me that ashtray. 2 ... Or your light, 3 I mean. 4 ...

. 5 ... (H) It’s behind the sewing machine. 6 (H)

In line 1, Mary asks the other participant in the conversation to hand her an ashtray. She then repairs her utterance with the phrase or your light in line 2, and appends I mean with a final intonation. The repair is accomplished, on the one hand, with the conjunction or which, in this pragmatic context, cannot be taken to index an alternative, but rather, a repair. On the other hand, it is accomplished by the use of I mean, which indexes that the preceding utterance is not adequate, and that the utterance associated with I mean (or your light) is meant as an adjustment. This is a typical use of I mean in self-repair: it occurs in the transition space between turns (Schegloff, Jefferson & Sacks 1977). Since Mary’s utterance in line 1 is a complete turn-constructional unit (TCU) semantically, syntactically and prosodically, a turn transition (either a verbal turn or other appropriate action) becomes relevant after Mary finishes. However, she produces another utterance, which is marked as a modification of prior talk both in the beginning of the TCU (or) and at the end (I mean). Although I mean is used in the context of self-repair in our data, this is not its most common function. It is most commonly used in contexts where a speaker, often in the middle of a lengthy turn, modifies her prior utterance in some other

 Ritva Laury and Shigeko Okamoto

fashion. We found uses of I mean in all the contexts mentioned in earlier studies: in our data, it was used in contexts of elaboration, clarification, expansion, explication and reformulation, as well as specification, exemplification, and mitigation of various types (for a summary of use types, see Brinton 2007: 41). The next example comes from a two-party conversation and illustrates elaboration of a prior utterance. One of the participants is describing a party she recently attended. (8)

SBC0006 Cuz (795.93–811.11) 1 (H) I walk in, 2 and the first hint that there’s – 3 this is gonna be something a little bit unusual, 4 is there’s all these helium balloons? 5 ... ut- – 6 .. up at the ceiling? 7 .. (H) With X- uh metallic= color=ed streamers coming down, 8 three streamers, 9 each balloon, 10 I mean the whole ceiling is just packed with these helium balloons. 11 It was so much fun.

The speaker is describing entering the party, and seeing the balloons that the room was decorated with. I mean in line 10 initiates a clause which elaborates and upgrades the preceding description – the ceiling was just packed with the balloons. However, I mean can occur in contexts where the relationship of the utterances linked with I mean is just the opposite: I mean can also initiate a modification which mitigates, rather than upgrades, earlier talk, as in the next example. This example comes from the same conversation as Example (3); a lawyer is preparing a witness for trial. (9)

SBC008.TRN Tell the jury 639.92–651.90 1 REBECCA: Was he looking at you at [all during that time]. 2 RICKIE: [Ye=s. 3 The] <MRC who=le ... enti=re ... ti=me MRC>. 4 REBECCA: Okay. 5 RICKIE: I mean, 6 (H) not looking, 7 (H) at <
Teyuu ka and I mean as pragmatic parentheticals in Japanese and English 

10 11 12

... And look at me. ... You know, trying to get my attention PAPER>>.

In lines 2–3, in answer to Rebecca’s question, Rickey answers emphatically that the suspect (he) was looking at her the who = le enti=re ti=me. After Rebecca’s short receipt of Rickey’s answer in line 4, in her I mean -initial turn, Rickie mitigates her answer by downgrading it to not looking, at first he would go like this, even, with the paper. Again, we suggest that the function of mitigation does not arise from the use of I mean but rather from the semantic relationship between the two utterances linked by I mean. Here, as in the previous example, I mean simply serves to index the relevance of what follows it to what Rickie has said earlier. As we have noted, most of the uses of I mean serve to modify the speaker’s own, earlier talk. This tendency is strong enough so that there can be an intervening turn by another participant, but the I mean-initial turn is still understood to modify its speaker’s earlier turn. Consider Example (10) below. It is taken from a three-party conversation. Pete is visiting his friends, Marilyn and Roy. Dinner is under preparation. (10)

SBC0003 Conceptual Pesticides 1414.57–1440.35 1 MARILYN: ... Should I make like tartar sauce, 2 or something really decadent? 3 ROY: ... What’s this in this squirt bottle? 4 MARILYN: ... U = m, 5 ... Windex. 6 ROY: ... I’m gonna use some of that. 7 PETE: Does tartar sauce go with .. with Cajun spices. 8 MARILYN: ... Well we could make – 9 PETE: ... I mean, 10 that doesn’t matter, 11 <X I suppose [it just X>] – 12 MARILYN: [Oh], 13 you know what, 14 we have this neat ... [island] ... man[2go sauce2]. 15 ROY: [<X Fields X>] – 16 PETE: [2Mm2] 17 ROY: [2((SNEEZE))2] 18 MARILYN: ... Chutney [3sauce3]. 19 ROY: [3Chut3]ney. 20 MARILYN: That would be good. 21 PETE: ... Yeah.

 Ritva Laury and Shigeko Okamoto

22 MARILYN: ... I have tons of chutney. 23 PETE: ... [Well that sounds good]. 24 MARILYN: [Chutney]. In lines 1–2, Marilyn suggests that she might make a tartar sauce to go with the fish that is being prepared for dinner. In line 7, after an exchange between Roy and Marilyn about cleaning supplies he needs, Pete asks whether tartar sauce is appropriate with the spices the fish is flavored with. Marilyn responds by starting a turn which appears to begin an alternative suggestion. She cuts off her turn, leaving it syntactically incomplete. At that point, Pete starts an I mean -initial turn in line 9. This turn links back to his earlier turn in line 7; it is interpretable as a modification (mitigation) of his earlier turn and expresses that it does not matter whether tartar sauce goes with Cajun spices or not. It is not interpretable as a mitigation of Marilyn’s just preceding turn, which offers alternatives; that is, Pete is not suggesting that it does not matter that alternatives exist. Marilyn then comes up with an alternative, chutney, and Marilyn and Pete arrive at a consensus on the appropriateness of chutney in lines 20–24. This example shows that speakers can, with the use of I mean, link back to their own, previous utterance across another speaker’s turn. It also shows that utterances marked with I mean can serve to increase agreement, harmony or alignment in conversation. One point of disagreement in previous studies has been precisely whether the utterances marked with I mean are self-directed or other-directed. While it has been argued that I mean is speaker-oriented (Schiffrin 1987: 299) and not interactive (Edmondson 1981), it has also been suggested, by some of the same authors, that I mean is interactive, cooperative, and hearer-oriented (House & Kasper 1981, Schiffrin 1987: 305). Maynard (2006) found that utterances prefaced with I mean occur “with extreme regularity” in complaint sequences, where they accompany utterances that are threatening or delicate in some ways. In Maynard’s data, the I mean-prefaced utterances functioned to defuse the speaker’s own action and to elicit aligning responses from other participants. In other words, Maynard shows that I mean is used to elicit certain types of behavior from coparticipants, and in that sense, it is used in an other-directed way. In the preceding Example (10), Pete’s utterance in line 7 can be seen as threatening or at least delicate: since he is a guest, it may not be appropriate for him to suggest that his host’s choice of a food item is inappropriate, and his I mean -prefaced utterance in lines 9–11 can be seen as defusing his inappropriate comment. In other words, while Pete’s I mean-initial utterance links back to his own utterance, the comment does not deal with the accuracy, comprehensibility or adequacy of his earlier utterance, but rather with its social acceptability. As suggested by Maynard (2006) for I mean-prefaced utterances in his data, this utterance functions

Teyuu ka and I mean as pragmatic parentheticals in Japanese and English 

to defuse the speaker’s utterance and to elicit alignment. In that sense, it can be seen as other-directed. In other examples in our data, speakers also use I mean in contexts where they are modifying their earlier utterances in such a way that they can be seen to be eliciting alignment from, or seeking alignment with, other participants. In the light of the preceding example, I mean can indeed be said to function in ways which are prosocial, interactive and addressee-oriented (House & Kasper 1981, Schiffrin 1987; Brinton 2007: 41). In fact, at the extreme end of the prosocial uses of I mean, we found uses of I mean in our data where the utterance it was associated with did not relate to any earlier utterance by its speaker, but rather was connected with an earlier utterance by another participant, or more broadly to a stance being collaboratively constructed by the participants. As we have seen, in Example (6), it is quite possible to analyze Miles’ turn To other people I mean as continuation of another participant’s turn, or, even more appropriately, more generally of an already shared understanding. The ability of I mean to link to utterances not produced by the speaker of I mean is shown in our next example, which comes from a conversation relating to problems within an arts organization. In the part of the conversation our example comes from, the participants are engaged in a lengthy negotiation about recent events at a meeting and their potential repercussions. The men appear to be in agreement that one of the members had acted inappropriately. (11)

SBC0010 Letter of concerns 1 BRAD: [She shouldn’t <X have done that X>]. 2 PHIL: and I didn’t [2wanna do anything2]. 3 BRAD: [2I know2]. 4 PHIL: But I just saw, 5 [(H) people] kinda going, 6 BRAD: [That was out] – 7 ... Yeah. 8 PHIL: You [know]? 9 BRAD: [Yeah]. 10 PHIL: And, 11 [(H)] 12 BRAD: [That’s true]. 13 She shouldn’t’ve been [2in on that <X at all X>2]. 14 PHIL: [2She shouldn’t’ve been2], 15 % you know, 16 cause that’s not part of her capacity, 17 she’s not [on that com]mittee, 18 BRAD: [Mhm].

 Ritva Laury and Shigeko Okamoto

19 [2Mhm2]. 20 PHIL: [2you know2], 21 and if she wants – 22 (H) you know, 23 ... % basically the thing is, 24 you know, 25 if she wanted to be a boa = rd person. 26 BRAD: Mhm. 27 PHIL: That’s fine. 28 I [mean let’s look] into b- pursuing that. 29 BRAD: [Right]. 30 Right. 31 PHIL: (H) = But see I’m afraid we’re gonna lose 32 Jim because of that. 33 ... he feels there’s a conflict. 34 And uh, 35 it’s definitely gonna [happen]. 36 BRAD: [(H)] I mean he says. 37 ... There’s no question about it, 38 we’ve [talked], 39 PHIL: [Yeah], 40 and maybe it’s right. 41 .. Maybe it s- [it is] gonna be right. 42 BRAD: [Yeah]. 43 PHIL: ... You know? 44 BRAD: (H) .. Well he – 45 ... d%= – 46 I mean he doesn’t seem to – 47 % % you’ve talked to him and I’ve [talked to him and], 48 PHIL: [(H) Well], 49 ...he [2seems to be2] – 50 BRAD: [2I mean he says2] mayb- – 51 you know, 52 after the board retreat. 53 [now]. 54 PHIL: [Yeah].

In this lengthy excerpt, where there is a considerable degree of agreement, there are four uses of I mean. We will concentrate here on the second use, in line 36.

Teyuu ka and I mean as pragmatic parentheticals in Japanese and English 

I mean in initial position in Brad’s turn in line 36 precedes a clause which does not seem to modify any earlier utterance by Brad himself, but instead, it seems to be connected to Phil’s just preceding utterance. In his turn in lines 31–35, Phil proposes that Jim is going to resign as a consequence of the proposed addition to the board discussed just previously. Brad’s I mean -prefaced utterance can be interpreted as confirming or emphasizing the likelihood of the consequence that Phil has just suggested. He provides evidence for its likelihood by an utterance that functions evidentially. He says in Brad’s utterance can be taken as a retrospective supplement, an insertable (Couper-Kuhlen and Ono 2007), to the beginning of Brad’s utterance in line 33, he feels there’s a conflict or to his utterance in line 35, it’s definitely gonna happen, or both. The import of Brad’s turn consists of reporting that Jim has said himself that he will resign. That is, Brad’s turn counts as an agreeing response and moreover, he suggests that Jim’s intent to resign is something Brad has heard from Jim himself. Ultimately, just as in Example (6) above, it is not crucial to the participants whose utterance he says supplements. Its function is to add further strength to the co-constructed stance that the conflict is likely to result in Jim’s resignation. This example makes it seem likely that the interpersonal nature of the contexts in which I mean occurs, such as complaint sequences to elicit alignment, and in other contexts where it is used in prosocial ways, have broadened the function of I mean so that it can also be used in contexts where it marks utterances intended as modifications of other speakers’ utterances. This emphasizes the important point that utterances which express a speaker’s point of view or a stance and are in that sense subjective, are always also public actions, done in a dialogic and interactional context, always have an intersubjective component as well (e.g. Kärkkäinen 2003; Kärkkäinen 2007; Du Bois 2007). In this section, we have shown that I mean occurs regularly in contexts where an utterance is linked to a just preceding utterance and is designed to modify it in a number of ways. We suggest that the function of I mean is to link the utterance it is associated with to an earlier utterance, and to mark it as a modification of that earlier utterance. The exact nature of the modification, whether it be a repair, explanation, elaboration or a mitigation, is determined by the syntactic, semantic and pragmatic relationship of the utterances thus linked. Previous studies have suggested that I mean is used to modify its speaker’s own, earlier utterance, but our data show that like the Japanese teyuuka, which will be discussed next, I mean is also used to modify utterances produced by other speakers. We take it to be its use in contexts where it is used prosocially use to seek alignment from other speakers which has motivated this perhaps relatively novel use.

 Ritva Laury and Shigeko Okamoto

4. Structure and Function of teyuuka 4. 1

The syntactic environments of teyuuka

The syntactic position of teyuuka is quite flexible. It can occur in an utterance initial, -medial, or -final position (Lauwereyns 2001, 2002; Maynard 2004; Kitano 2005; McGloin 2007). It may also occur within a phrase or word (Kitano 2005). Although both I mean and teyuuka are quite flexible in the syntactic position in which they occur, teyuuka occurs in a greater variety of syntactic positions than I mean in that it can occur within a syntactic unit such as a noun phrase or a compound verb, and that it can be used without being followed by any expression intended to remedy or supplement the expression that precedes it. Example (12) below shows an example of utterance-medial use of teyuuka. (12) [SM, talking with her mother SH at home] no, 1 SM: Jooshi tteyuuka7 ue superior higher gen 2 SH: un, uh huh 3 SM: Itoo-san te hito ga konaida, Ka, Kankoku prt person sm the other day Korea 4 SH: un, uh huh 5 SM: Chuugoku itte, China went ‘My superior, or/I mean higher(-ranking), person, (uh huh) Ito went to Korea, (uh huh), China, the other day,’ Table 2.â•‡ Syntactic environments of teyuuka Utterance-medially:

X teyuuka Y

Utterance-initially:

X. Teyuuka Y (X and Y are uttered by the same speaker or different speakers.) X teyuuka. X1 teyuuka X2 (X1 and X2 constitute a syntactic unit that normally occurs as a whole without any interruption.)

Utterance-finally: Within a phrase/word:

7. Forms such as tteyuuka, toyuuka, and toiimasuka are variants of teyuuka. Tteyuuka tends to appear after a vowel (Example 12), teyuuka tends to appear after a consonant (Example 13). Toyuuka and toiimasuka are stylistically more formal teyuuka, with the latter being the most formal.

Teyuu ka and I mean as pragmatic parentheticals in Japanese and English 

Example (13) shows an utterance-initial use. (13) [MR, a writer, interviewing two female high school students, including MK. Source: Murakami (1998: 46)] 1 MR: Hoka ni kyoodai wa? other siblings tm ‘Do you have siblings?’ 2 MK: Shita ni futari. below two ‘I have two younger (siblings).’ 3 MR: Kyoodai naka ii n desho? Siblings relation good right ‘You (siblings) get along well, right?’ 4 MK: E:: demo, watashi no koto wa kiratteru. Uh, but me gn thing tm dislike 5

Sono futari wa naka yokute, those two tm relation good

6 atashi dake zenzen yoritsukanai. me only at all approach neg ‘Uh, they dislike me. Those two get along well, but they don’t approach me at all.’ 7 MR: Ammari hanasanai no? Not so much talk neg prt ‘You don’t talk with them much?’ 8 MK: U::n, teyuuka watashi ga chotto kite tte ittemo konai uh I sm a little come prt say even if come neg 9 Yada toka itte. do not want or something saying ‘No, it’s not (like we don’t talk), but even if I tell them to come, they won’t come, saying “they don’t want to” or something.’ In Example (13), MK uses a token of teyuuka which projects back to the immediately preceding utterance made by her interlocutor MR. We have also found cases in which the utterance-initial teyuuka that projects back to the immediately preceding utterance made by the same speaker. Example (14) illustrates the utterance-final use of teyuuka.

 Ritva Laury and Shigeko Okamoto

(14) [MR, a male writer, interviewing two high school girls. Source: Murakami (1998: 70)] 1 MR: Demo, futaritomo nanka shikkari-shita kanji ga surushi, but two both somehow reliable feeling sm do and 2 dochira ka tte ittara otonappoi tteyuuka. which Q prt say if adult-like ‘But I feel both of you are reliable and, rather (lit. if I say which one), adult-like, I might say. Example (15) illustrates the toyuuka (a variant of teyuuka) that appears within a syntactic unit. (15) [Source: Kitano 2005: 108] 1 De daisetsugen toyuuka and large snow prairie 2 no tokoro de hikooki ga tomarimashite ne. gn place in plane sm stop ah prt ‘And the plane stopped in a place like a large snow prairie/a place that is a large snow prairie, so to speak, right?’ In (15) toyuuka is inserted within a constituent, the genitive postpositional phrase daisetsugen no ‘of a large snow prairie’; that is, it interrupts this syntactic unit in that the genitive marker no normally follows a noun phrase immediately as in daisetsugen no. In this section, we have seen that teyuuka can occur in a number of syntactic environments; it can occur clause-medially, clause-initially, or clause-finally. In our data, teyuuka was used in all of these three positions relatively frequently, in contrast to I mean, which was only rarely used in utterance-final position. Further, teyuuka, unlike I mean, may also occur within a constituent, or a syntactic unit that is normally uninterrupted, although we found only a few cases of this use in our data. 4.2

Pragmatic functions of teyuuka

The functions of teyuuka have been characterized variously by previous studies. The utterance-medial teyuuka is most commonly regarded as a repair-initiator (Lauwereyns 2001, 2002; Wakamatsu and Hosoda 2003; Kitano 2005; McGloin 2007; Rosenthal 2008); it is also recognized as an initiator for rephrasing, specification, or supplementary explanation (Lauwereyns 2001). The utterance-initial teyuuka is characterized as an initiator for a repair or metalinguistic negation (Lauwereyns 2001, 2002; McGloin 2007; Rosenthal 2008), a preface for expressing

Teyuu ka and I mean as pragmatic parentheticals in Japanese and English 

one’s true feelings (Maynard 2004), a preface for a (sub-)topic shift (Lauwereyns 2001, 2002; Maynard 2004; McGloin 2007), or a device for reframing what has been said in the preceding discourse (Ono 2006). The utterance-final teyuuka is treated as a means for mitigation or blurring, or for expressing hesitation (Lauwereyns 2001; Maynard 2004; McGloin 2007). Taking the findings of these previous studies into consideration, we attempt to provide further analyses of the functions of teyuuka. In this study, we characterize the basic function of teyuuka in terms of the speaker’s evaluative stance toward the adequacy of the expression that precedes it. That is, teyuuka gives a metalinguistic comment on the preceding expression. More specifically, teyuuka indicates that the speaker thinks that the expression in question is not fully adequate or appropriate in the context as it evolves, and that another expression could be given to remedy or supplement its inadequacy (cf. Maynard 2004). Other functions, such as initiating an attenuated correction, elaboration, specification, and generalization can be considered ramifications of this basic function of teyuuka that arise in specific contexts. That is, the inadequacy of the utterance that precedes teyuuka may be remedied or supplemented by another utterance, which in turn serves a variety of functions such as a mitigated correction and elaboration. Alternatively, the speaker may choose not to provide another utterance at all (which is not possible with I mean). That is, he/she simply indicates that what has just been said is not adequate, which in turn may serve to mitigate or qualify that utterance. We will discuss these points with examples. In the case of utterance-medial teyuuka (X teyuuka Y), the speaker first indicates the inadequacy of what he/she has just said by teyuuka and then attempts to remedy or supplement this inadequacy in some way by adding another expression, which in turn may be interpreted in a variety of ways, including a mitigated correction, rephrasing, specification, expansion, elaboration, explanation, and generalization. In other words, teyuuka can be considered a prelude to another expression that serves these functions. Let us discuss these functions with examples. As pointed out by previous studies, teyuuka may be used as a (weak) repairinitiator (e.g. Lauwereyns 2001; Wakamatsu and Hosoda 2003). We emphasize that the basic function of teyuuka is not to point out that the utterance that precedes teyuuka is totally incorrect or wrong, but to indicate that it is not fully adequate or appropriate (cf. Maynard 2004). If the speaker wishes to make an overt correction, he/she can use a more direct expression, such as X ja nakute Y ‘it’s not X, but Y’. Teyuuka, on the other hand, can be used to make a correction, repair, or modification more indirectly. Accordingly, it is often used to correct what the interlocutor said in an oblique way, as illustrated in Example (16):

 Ritva Laury and Shigeko Okamoto

(16) [MR, a male writer, is interviewing SK and YK, female high school students. Source: Murakami (1998: 24)] 1 MR: U::n, demo atama no warui otoko no ko tte omoshiroku nai desho? uh but unintelligent boy interesting neg right? ‘Uh, but unintelligent boys are not interesting, are they?’ 2 SK: Ima no kookoosei minna atama warui yo ne. nowadays high school all unintelligent prt prt ‘High school students these days are all unintelligent, aren’t they?’ 3 YK: Atama warui tteyuuka, unintelligent 4

benkyoo toka no hoo ja nakute, study etc. gen side neg

5 chigau hoo de atama ii. different side in intelligent ‘It’s not like they are unintelligent, but rather they are intelligent in areas other than academic areas/study and the like.’ In this example, tteyuuka is used in other-repair. YK first repeats the expression used by SK and MR in their previous turns, and adds tteyuuka to indicate that the expression is inadequate. She then offers another expression (lines 4–5) to remedy the inadequacy of the expression that preceded tteyuuka. This use of teyuuka differs from the use of I mean; in our data, I mean was not used in other-repair.8 However, like I mean, teyuuka may also be used in self-repair, when the speaker wishes to mitigate his/her own utterance, as illustrated in Example (17): (17) [TK and YM, female college students, talking about beauty shops.] 1 TK: Aa, demo biyooin sagasanakya::. uh, but beauty shop find have to inakunatchatta kara. disappear pst because ‘Uh, I have to find a (new) beauty shop. Because (it) disappeared.’ 2 YM: A, soo ka. Inakunatchatta n da ne. Oh, right disappear pst aux prt ‘Oh, that’s right. (It) disappeared, didn’t it?’ 8. Although, as shown in Example (11), it can be used to comment (amplify) an utterance by another speaker.

Teyuu ka and I mean as pragmatic parentheticals in Japanese and English 

3 TK: Sugo::i hetakuso tteyuuka sa::, very unskilled prt 4

umakunai good neg ‘(She) was very bad/unskilled, uh, I mean not good.’

5 YM:

Aa. yeah ‘Yeah.’

6 TK: Soo, konaida itta tokoro ga. Right, the other day go pst place sm ‘Right, the place I went the other day.’ In this example, in line 3, TK first characterizes (the hair-stylist at) the beauty shop as sugo::i hetakuso ‘very bad/unskillful’, and, within the same turn, uses tteyuuka, followed by the particle sa::, a filler that suggests TK’s hesitation about her word choice; she then uses umakunai ‘not good/skillful’, which is milder than sugo::I hetakuso. Like I mean, teyuuka may also be used in self-repair in contexts that do not involve mitigation, when the speaker is not satisfied with an expression and attempts to rephrase it, or give an alternative expression, as in Example (12) above. The use of teyuuka in that example is parallel to the use of I mean in our Example (7), where it is found in same-turn self-repair. The speaker may also attempt to supplement the inadequacy of an expression by adding teyuuka followed by another expression that serves as a specification, expansion, elaboration, or explanation of the first expression, as illustrated in (18) and (19): (18) [HE and NY are female college students; NY, talking about her experience she had with her host family in England.] 1 NY: De:: tamago mo ikkai-shika denakute, and eggs also once only served ‘And eggs were also served only once, and’ 2 HE: I::! ah ‘Ah!’ 3 NY: Dakara atashi sa Nihonshoku tteyuuka, sa, so I prt Japanese food prt 4

ano, o, o, osenbei suggoi takusan motte itta no ne. uh rice cracker very much take prt prt ‘So I took a lot of Japanese food with me, I mean, rice crackers.’

 Ritva Laury and Shigeko Okamoto

In this example, the expression which follows tteyuuka, osenbei ‘rice crackers’, functions as a specification of the type of Japanese food that speaker NY reports taking with her. (19) [HE and NY, female college students, talking about going skiing.] 1 HE: Ne:: sukii ikitai. hey ski go want ‘Hey, I want to go skiing.’ 2 NY: Dokka ii toko, nai? somewhere nice place not exit 3 NY: Ii tooko tteyuuka, nice place 4 NY: yasui, yasuku, yasukute tomareru tte tokoro cheap cheap cheaply stay can prt place nai kana. not exist prt ‘Is there any nice place? Nice place, I mean, a place where we can stay cheaply.’ Note that the expression that follows teyuuka may be interpreted as serving more than one function. Thus, for example, in (19), what follows tteyuuka may be considered a rephrasing of what precedes it as well as an elaboration or explanation. The expression that follows in tteyuuka in (16) can be interpreted as serving to mitigate as well as to explain. In contrast to the previous usage, in which teyuuka serves to initiate a specification or elaboration of the preceding expression, the speaker may use teyuuka to initiate a generalization of what was said before teyuuka, as illustrated in (20): (20) [IT, a graduate student, talking with OY, her former professor] 1 IT: Ano:: toriaezu, Uh- first ‘Uh’ 2 OY: Un. Yeah ‘Yeah’ 3 IT: Ano::, puririkuijitto tteyuuka, soo yuu kurasu wa Uh- prerequisites such classes tm zenbu oetan desu yo. all finish pst AH prt ‘Uh-, I’ve finished (taking) all prerequisites, I mean courses like that.’

Teyuu ka and I mean as pragmatic parentheticals in Japanese and English 

The functions of utterance-initial teyuuka are basically the same as those of utterance-medial use except for a couple of properties discussed below. In the case of teyuuka occurring utterance-initially (X. Teyuuka Y), a particular utterance reaches some kind of completion point (e.g. the end of a sentence, intonation unit, or turn). The same speaker or her/his interlocutor then says teyuuka and supplies another expression to make up for the inadequacy of the preceding utterance in some way. The utterance that follows teyuuka then may be interpreted as serving a variety of functions – the same functions as in the case of the utterance-medial teyuuka discussed above. For example, in the earlier Example (13), what follows teyuuka serves as a rephrasing of the preceding utterance suggesting that the speaker may have thought it was not quite accurate. In Example (21), the expression that follows teyuuka serves as a mitigation of the preceding utterance, suggesting that the speaker may have thought that her preceding utterance was too negative. (21) [MR, a male writer, interviewing two female high school students, including MK. Source: Murakami (1998: 36)] 1 MR: Matsuda-san wa okaasan toka otoosan no koto doo? hon tm mother etc. father gen thing how 2

Suki desu ka? like AH Q ‘Do you like your mother and father (and the like)?’

3 MK: E? kirai. [laugh] what? dislike ‘What? I don’t like them.’ 4 MR: Soo, dooshite? I see why 5

Chitchai koro wa? small time tm

6

Shoogakusei toka sa, primary school pupil etc. prt

7 shoogakkoo no teigakunen toka. primary school gen lower level etc. ‘I see. Why? (How about) when you were small, like when you were a primary school, lower-level primary school pupil and the like?’ 8 Shoogakoo no toki mo sukijanai. primary school gen time also like neg 9 Tteyuuka oboetenai. remember neg

 Ritva Laury and Shigeko Okamoto

10 Ichiban kirai ni natta no ga shoo-yon. most dislike become nom tm grade 4 ‘I didn’t like them either when I was in the primary school. I mean I don’t remember. It was when I was in the 4th grade that I came to dislike them most.’ In this example, speaker MK first indicates in line 8 that she did not like her parents when she was in primary school. After the utterance reaches a completion point, she uses tteyuuka and adds a mitigating utterance, oboetenai ‘ I don’t remember’, which is vaguer than sukijanai ‘do not like’. Similarly, in Example (17), what follows tteyuuka serves to mitigate the force of the preceding expression. One function that seems peculiar to the utterance-initial teyuuka is its role as an initiator for a (sub-)topic shift (Lauwereyns 2001; Maynard 2004), as in (22). (22) [MR, a male writer, interviewing four high school girls, including HY. They are talking about the so-called enjo koosai ‘compensated dating’, in which older men pay high school girls for their companionship, which may include sexual “service.” Source: Murakami (1996: 126)] 1 MR: Demo ijoo wa ijoo da yo ne. but abnormal tm abnormal cop prt prt 2 Konna koto yatteru no Nihon dake da yo. such things do prg gen Japan only cop prt ‘But it’s really abnormal (lit. abnormal is abnormal). It’s only in Japan (people) do such things.’ 3 HY: Baka da ne::. stupid cop prt 4 Tteyuuka nande konnan natta n ka ne. why like this become pst prt Q prt 5 Dare da, ichiban saisho ni yatta no. who cop number 1 first do pst prt ‘They are stupid, aren’t they! I mean why have things become like this? Who started it first?” In Example (13), MR criticizes enjo koosai in lines 1 and 2. HY also expresses her critical view of enjo koosai in line 3. She then says tteyuuka and introduces a new sub-topic in lines 4 and 5 – the reasons why many high school girls have come to engage in enjo kossai and who started it. HY’s utterances in lines 4 and 5 are not corrections, exemplifications, or specifications of her earlier utterance in line 3. In the case of utterance-final position (X teyuuka), no overt expression follows teyuuka; the speaker simply indicates that he/she is not fully satisfied with the

Teyuu ka and I mean as pragmatic parentheticals in Japanese and English 

expression he/she has just used, which in turn may serve to mitigate and/or modify the preceding utterance. As pointed out by previous studies (e.g. Maynard 2004; McGloin 2007), the utterance-final teyuuka is often used for mitigation of the force of the preceding utterance, as illustrated in Example (23): (23) [MR, a writer, is interviewing YM and MK, female high school students. Source: Murakami (1998: 43–44)] 1 MR: Kore yatteru toki tanoshii toka tte yuu koto aru? this do prg when fun or something prt say thing exist ‘Is there anything that you feel fun or something when you’re doing it?’ 2 YM: Ee. Nai. uh. No ‘Uh. No (there isn’t).’s 3 MR: Karaoke shiteru toki tte? do prg time prt ‘When you are doing karaoke, (is it fun)?’ 4 YM: Un, sore mo tanoshii, tanoshii kedo, yeah that also fun fun but 5 YM: sonna sugoku tanoshiku wa nai tteyuuka. that extremely fun tm neg ‘Yeah, that’s also fun, it’s fun, but it’s not that fun, I mean.’ 6 MR: Maa sonna ni ne. well not so prt ‘Well, (it’s not) that much (fun), right?’ 7 YM: Tanoshii koto dewa nai. fun thing cop neg ‘It’s not (such) a fun thing.’ In line 3 in (23), MR asks if YM enjoys doing karaoke. YM first responds positively in line 4, but then in line 5 she says she doesn’t enjoy it that much and then adds tteyuuka and ends her utterance. By adding tteyuuka at the end of her utterance, YM can indicate (at least on the surface) that she is not quite satisfied with what she has just said, which in turn can make her utterance less forceful in giving a negative answer to MR. It seems YM used this strategy because she might have thought MR was expecting a positive answer that can lead to further talk about the topic. Similarly, in the earlier Example (14), the addition of tteyuuka utterance-finally in line 2 can serve to indicate that the expression (otonappoi ‘adult-like’) that the speaker has chosen may not be a fully appropriate description that is acceptable

 Ritva Laury and Shigeko Okamoto

to the interlocutors. In examples like these, teyuuka could be said to have the interactional motivation of seeking alignment or approval from another participant. The speaker may also use teyuuka simply to indicate that his/her word choice may not be perfect, as illustrated in (24), in which NY is not sure if the word abutta is appropriate to describe the food she was talking about: (24) [NY and HE are female college students; NY, talking about the experience she had with her host family in England.] 1 NY: De, koshoo mo shio mo ajitsuke- shitenakute, and black pepper also salt also seasoning do neg 2 roosuto beefu data no ne? roast beef was prt prt ‘And it wasn’t seasoned either with black pepper or salt, right? It was roast beef.’ 3 HE: Un. yeah ‘Yeah.’ 4 NY: Maa, chotto abu, abutta tteyuuka. uh a little bit grill pst ‘Uh, it was grilled a little bit, so to speak.’ As we have seen so far, teyuuka normally precedes or follows a word, phrase or clause. However, as noted earlier, teyuuka may also occur within a phrase or word (Kitano 2005), as in Example (25). Further, teyuuka may occur more than once in the same utterance, as illustrated also in Example (25): (25) [NY and IN, middle-aged women, talking about the trip they made recently with other women.] 1 NY: Ano:: watashi-tachi ga kankoo-annai de mita uh we sm travel guide in see pst basho yori wa motto kitsukute ne? place than tm more tough prt ‘It was tougher (to walk) than what we expected from (lit. saw in) the travel guidebook, right?” 2 IN: Soo. right ‘Right.’ 3 NY: Chotto kokorogamae mo nakatta kara, a little preparedness also have neg so ‘I wasn’t prepared (for it), so,’

Teyuu ka and I mean as pragmatic parentheticals in Japanese and English 

4 IN: ==Un un. yeah yeah ‘Yeah, yeah.’ 5 NY: ==chotto dokidoki tteyuuka, a little anxious ‘I was a little kind of anxious, or’ 6 IN: ==Un yeah ‘Yeah’ 7 NY: bikkuri tteyuuka shita kedo ne. surprise do pst but prt ‘kind of surprised, right?’ In (25), NY first says dokidoki and adds tteyuuka, indicating she is not fully content with this word choice; she then offers bikkuri as an alternative to dokidoki but adds tteyuuka to it, too, indicating she is not fully happy with that word, either. But she does not offer another alternative expression and instead links tteyuuka directly to shita, or the second part of the compound verb bikkuri-shita ‘was surprised’. Thus, Example (25) also illustrates the case in which teyuuka interrupts a particular syntactic unit (in this case, a compound verb) that normally occurs as a whole. Example (15), as we discussed earlier, also shows the use of teyuuka within a phrase (i.e., daisetsugen toyuuka no tokoro ‘a place like a large snow prairie’) in order to indicate that the speaker is not fully happy with the word daisetsugen. In this section, we have shown that the basic function of teyuuka is to indicate that the speaker regards the expression in question as inadequate. The speaker then may make another explicit expression, which can serve various functions, including repair, attenuated correction, specification, elaboration, explanation, exemplification, and generalization. Alternatively, the speaker may end the utterance with teyuuka without providing any utterance that could supplement or remedy the inadequacy of the preceding utterance. This serves to mitigate the force of the utterance that precedes teyuuka. 5. Discussion and conclusion The foregoing discussion of the properties of I mean and teyuuka show that the two expressions are quite different from each other in their syntactic and semantic makeup and yet have similar but not identical pragmatic functions. In the

 Ritva Laury and Shigeko Okamoto

remainder of this chapter, we consider further how and why these two constructions are similar to and different from each other. On the basis of our close analyses of conversational data, we argue that both I mean and teyuuka can be used to index and negotiate a metalinguistically evaluative stance toward an expression that precedes them rather than contributing directly to the main content of the conversational sequence they appear in. More specifically, both expressions serve to look back to an earlier utterance, either by the same speaker or by another speaker, and to mark it as inadequate or in need of modification. Both can also frame another expression as a modification of the earlier utterance, and in this way they may function as an initiator for, or project, more talk which serves as a modification or expansion of the prior expression, such as repair, elaboration, exemplification, generalization, specification, or and mitigation. However, unlike previous studies which assign specific functions (e.g. repair, mitigation) to these constructions, we suggest that the function of I mean and teyuuka is to simply mark the earlier expression as inadequate, and the latter expression as a modification of the former. The exact nature of the modification is not specified by I mean or teyuuka per se, but arises from the semantic, pragmatic and syntactic relationship between the two utterances. We also argue that the pragmatic functions associated with I mean and teyuuka, may be considered conventionalized implicatures – meanings associated with their use in certain types of contexts – contexts which serve to evoke certain metalinguistic as well as interactional needs. The properties of I mean and teyuuka we have analyzed illustrate flexibility and creativity in language use. English and Japanese make use of (syntactically and semantically) quite different resources available to them to satisfy similar, if not the same, pragmatic needs that arise in discourse. At the same time, they also show that while the pragmatic functions of the two constructions are very similar to each other, they are not identical, a finding that seems to be related to the syntactic and semantic differences of the two expressions. In our data, I mean is always accompanied by the expression intended to modify or supplement the earlier expression, but teyuuka can be used without the modification. This can be seen to stem from the syntactic and semantic nature of teyuuka as a clause-final expression, which simply questions the adequacy of its “complement,” that is, what precedes teyuuka, and from the syntactic and semantic nature (and still most frequent use) of I mean as a clause- and phrase-initial expression that projects forward, in a sense expects the “complement” to follow. In addition, I mean cannot do otherrepair in our data (although it can supplement or support talk by another, as in Examples 6 and 11), but teyuuka can. This can be attributed to the fact that I mean is person-marked and thus, in its original use, indexes its speaker’s meaning, while teyuuka is not person-marked. The fact that teyuuka comes after its “complement”

Teyuu ka and I mean as pragmatic parentheticals in Japanese and English 

clause makes it a useful device for other-repair, since something that is said can be made into a complement after the fact (cf. Tanaka 2001). The original, or basic, meaning of I mean has to do with speaker intentions (Brinton 2007), while yuu is a verb of saying. As mentioned above, I mean is marked for person, while teyuuka is not. Both are used to make a metalinguistic comment, that is, they are used to express the speaker’s evaluation of the adequacy of an earlier utterance. Both are also used in contexts where there is negotiation or adjustment of meanings of the expressions used based on interactional concerns (Maynard 2004). Maynard discusses the dialogicity of teyuuka. As we have shown above, I mean also occurs in contexts where the modifications its associated utterance accomplishes are motivated in the needs, actions and utterances of other participants. In these contexts, the indexical features of the utterance are also bleached, so that I no longer indexes the current speaker. It appears that I mean has more of a meta-communicative, intersubjective function than teyuuka, which may be related to the fact that the latter is not marked for person and can function more on the metalinguistic level by interpreting the agent of yuu generically; however, this point requires more study. Both constructions can be said to function as pragmatic parentheticals. However, teyuuka is more flexible syntactically in that it can used within a syntactic unit such as a phrase or word. This may mean that teyuuka is further advanced in its development toward a parenthetical, or it may reflect differences in repair practices in the two languages. If in fact I mean is less advanced in its grammaticalization path, but has more of a meta-communicative, intersubjective function than teyuuka, it provides a puzzle for the theory of grammaticalization, since it has been assumed that as grammaticalization proceeds, items progress from subjective to intersubjective functions (Traugott 2003: but see Brinton 2007). These observations indicate that even if two constructions in two separate languages may share structural and functional similarities, they are still not identical, and one may have functions the other one does not share, so they may not even conveniently fit underneath one crosslinguistically valid, universal category label (Dryer 1997; see also Englebretson 2008). Dryer suggests that rather than reflecting unified crosslinguistic phenomena, similarities between structures in different languages actually reflect a high degree of isomorphism between structure and function, an observation consistent with the findings of the present study. References Brinton, Laurel. 1996. Pragmatic Markers in English: Grammaticalization and Discourse Functions. Berlin/New York: Mouton de Gruyter.

 Ritva Laury and Shigeko Okamoto Brinton, Laurel. 2007. “The Development of I mean: Implications for historical pragmatics.” In Methods in Historical Pragmatics, Susan Fitzmaurice and Irma Taavitsainen (eds), 37–79. Berlin and New York: Mouton de Gruyter, Couper-Kuhlen, Elizabeth and Ono, Tsuyoshi. 2007. “‘Incrementing’ in conversation. A comparison of practices in English, German and Japanese.” Pragmatics 17 (4): 513–522. Dryer, Matthew. 1997. “Are grammatical relations universal?” In Essays on Language Function and Language Type: Dedicated to T. Givón. Joan Bybee, John Haiman and Sandra A. Thompson (eds), 115–143. Amsterdam: John Benjamins. Du Bois, John W. 2007. “The stance triangle.” In Stancetaking in Discourse, Robert Englebretson (ed.), [Pragmatics & Beyond New Series 164]. 183–220. Amsterdam: Benjamins. Du Bois, John W. et al. 2000–2005. Santa Barbara Corpus of Spoken American English. Parts 1–4. Philadelphia: Linguistic Data Consortium. http://www.ldc.unpenn.edu Edmondson, Willis. 1981. Spoken Discourse. A Model for Analysis. London/New York: Longman. Englebretson, Robert. 2008. “From subordinate clause to noun phrase. Yang-constructions in colloquial Indonesian.” In Crosslinguistic Studies of Clause Combining. The multifunctionality of conjunctions, Ritva Laury (ed.), [Typological Studies in Language 80]. 1–34. Amsterdam: John Benjamins. Ford, Cecilia E. and Thompson, Sandra A. 1996. “Interactional units in conversation: syntactic, intonational, and pragmatic resources for the projection of turn completion.” In Interaction and grammar, Elinor Ochs, Emanuel A. Schegloff and Sandra A. Thompson (eds), [Studies in Interactional Sociolinguistics 13.]135–184. Cambridge: Cambridge University Press. Ford, Cecilia E.,. Fox, Barbara A. and Thompson, Sandra A. 2002. “Constituency and the Grammar of Turn Increments.” In The Language of Turn and Sequence, Cecilia Ford, Barbara A. Fox and Sandra A. Thompson (eds), [Oxford Studies in Sociolinguistics].14–38. Oxford: Oxford University Press. Fox, Barbara A. and Jasperson, Robert. 1995. “A syntactic exploration of repair in English conversation.” In Alternative Linguistics. Descriptive and Theoretical Modes, Philip W. Davis (ed.), [CILT 102]. 77–134. Amsterdam: Benjamins. Fox, Barbara A., Hayashi, Makoto and Jasperson, Robert. 1996. “Resources and repair: a crosslinguistic study of syntax and repair.” In Interaction and grammar. Elinor Ochs, Emanuel A. Schegloff and Sandra A. Thompson (eds). [Studies in Interactional Sociolinguistics 13]. Cambridge: Cambridge University Press. Fox Tree, Jean E. and Schrock, Josef C. 2002. “Basic meanings of you know and I mean.” Journal of Pragmatics 34: 727–747. Goldberg, Julia Anna. 1980. Discourse particles: An analysis of the role of y’know, I mean, well, and actually in conversation. Ph.D. dissertation. University of Cambridge. Hakulinen, Auli, Keevallik, Leelo and Lindström, Jan. 2003. “Kuule, kule, hördu – projicerande praktiker i finska, estniska och svenska samtal.” [Kuule, kule, hördu – projecting practices in Finnish, Estonian and Swedish conversations.] In Grammatik och Samtal: Studier till minne av Mats Eriksson, Bengt Nordbert, Leelo Keevallik Erikson, Kerstin Thelander and Mats Thelander (eds), [Studier utgivna av institutionen för nordiska språk vid Uppsala universitet 653]. 199–218. Uppsala. Hakulinen, Auli and Seppänen, Eeva-Leena. 1992. Finnish kato. From verb to particle. Journal of Pragmatics 18(6): 527–549. Hopper, Paul J. and Thompson, Sandra A.. 2008. “Projectability and clause combining in interaction.” In Crosslinguistic Studies of Clause Combining. The multifunctionality of conjunctions, Ritva Laury (ed.), [Typological Studies in Language 80]. 1–34. Amsterdam: John Benjamins.

Teyuu ka and I mean as pragmatic parentheticals in Japanese and English  House, Juliane and Kasper, Gabriele. 1981. “Politeness markers in English and German.” In Conversational Routine: Explorations in Standardized Communication Situations and Prepatterned Speech, Florian Coulmas (ed.), 157–185. The Hague: Mouton. Imo, Wolfgang. (2005): “A Construction Grammar approach to the phrase “I mean” in spoken English.” In InLiSt (Interaction and Linguistic Structure) 42. Kärkkäinen, Elise. 2003. Epistemic Stance in English Conversation. A description of its interactional functions, with a focus on I think. [Pragmatics and Beyond 115]. Amsterdam: Benjamins. Kärkkäinen, Elise. 2007. “The role of I guess in conversational stancetaking.” In Stancetaking in Discourse, Robert Englebretson (ed.), [Pragmatics & Beyond New Series 164]. 183–220. Amstedam: Benjamins. Keevallik, Leelo. 2003. From Interaction to Grammar. Estonian Finite Verb Forms in Conversation. [Studia Uralica Upsaliensia 34]. Uppsala: Acta Universitatis Upsaliensis. Kitano, Hiroaki. 2005. “Shizendawa ni mirareru itsudatsutekina bun no koochiku: Shikootekiteiji no teme no keeshiki “... toyuuka” “... desu ka” nado.” [Deviant sentence construction in naturally occurring discourse: The forms “... teyuuka,” “... desu ka,” etc. for trial presentations]. In Katsudoo toshite no Bun to Hatsuwa [Sentences and Utterances as Activities], 91–121. Tokyo: Hituzi-shobo. Lauwereyns, Shizuka. 2001. “Ripea-maakaa “teyuuka” no kinoo” [The functions of teyuuka as a repair marker]. Paper presented at the 7th Sociolinguistic Science Conference. Lauwereyns, Shizuka. 2002. “Hedges in Japanese conversation.” Language Variation and Change 14: 239–269. Maynard, Douglas. 2006. “Defensiveness in social interaction: On the use of I mean -prefaced utterances in complaint and other conversational sequences.” Presentation given at the Department of Sociology, University of Helsinki, November 23rd. Maynard, Senko K. 2004. Danwa-gengogaku: Nihongo no Disukoosu o Soozoo-suru Koosei, Retorikku, Sutorategii no Kenkyuu [Discourse Linguistics: A Study of Structures, Rhetoric, and Strategies for Creating Japanese Discourse]. Tokyo: Kurosio. McGloin, Naomi Hanaoka. 2007. “Utterance-initial teyuuka and metalinguistic negation.” In Aspects of Linguistics, Susumu Kuno, Seiichi Makino and Susan Strauss (eds), 155–176. Tokyo: Kurosio. Murakami, Ryu. 1998. Yumemiru Koro o Sugireba: Murakami Ryu vs. Joshi-koosei 51-nin [After the Dreaming Period: Murakami Ryu vs. 51 Female High School Students]. Tokyo: Rikuruuto Davinchi. Ono, Tsuyoshi. 2006. “Postpredicate elements in Japanese conversation: Nonmodularity and Panchrony.” In Japanese/Korean Linguistics 14, Timothy Vance and Kimberly Jones (eds), 381–391. Stanford: CSLI. Östman, Jan-Ola. 1981. You know: A Discourse-Functional Approach. Amsterdam: Benjamins. Rosenthal, Benjamin M. 2008. “A resource for repair in Japanese talk-in-interaction: The phrase tteyuuka.” Research on Language and Social Interaction 4: 227–240. Saari, Mirja. 1986. “Jag menar” [I mean]. In Xenia Huldeniana. En vänskrift tillägnad Lars Huldén på hans 60-årsdag 5.2.1986. Mirja Saari, Anne-Marie Londen and Kim Nilsson (eds), Publications of the Department of Nordic languages and literatures, Department of Helsinki. Series B, Nr. 10. 329–337. Saari, Mirja. 1984. “Några pragmatiska partiklar i svenskt talspråk” [Some pragmatic particles in spoken Swedish]. In Studier i nordisk filologi. Lars Huldén & Carl-Eric Thors (eds), 213–221. Helsinki: Svenska litteratursällskapet i Finland.

 Ritva Laury and Shigeko Okamoto Schegloff, Emanuel, Gail Jefferson and Harvey Sacks. 1977. “The preference for self-correction in the organization of repair in conversation.” Language 53: 361–382. Scheibman, Joanne. 2001. Point of View and Grammar. [Studies in Discourse and Grammar 11]. Amsterdam: Benjamins Schiffrin, Deborah. 1987. Discourse Markers. Cambridge: Cambridge University Press. Stenström, Anna-Brita. 1995. “Some remarks on comment clauses.” In The Verb in Contemporary English, Bas Aarts and Charles F. Meyer (eds), 290–302. Cambridge: Cambridge University Press. Tanaka, Hiroko. 2001. “The implementation of possible cognitive shifts in Japanese conversation: Complementisers as pivotal devices.” In Studies in Interactional Linguistics. Margaret Selting and Elizabeth Couper-Kuhlen (eds), 81–109. Amsterdam: John Benjamins Thompson, Sandra A. and Anthony Mulac. 1991. “A quantitative perspective on the grammaticization of epistemic parentheticals in English.” In Approaches to grammaticalization (II), Elizabeth Closs Traugott and Bernd Heine (eds), 313–329. Amsterdam: John Benjamins. Thompson, Sandra A. 2002. ““Object complements” and conversation: towards a realistic account.” Studies in Language 60(1): 125–164. Traugott, Elizabeth C. 2003. “Constructions in grammaticalization.” The Handbook of Historical Linguistics. Brian D. Joseph and Richard D. Janda (eds), 624–647. Oxford: Blackwell. Urmson, James O. 1952. “Parenthetical verbs.” Mind 61: 480–496. Wakamatsu, Mikiko and Yuri Hosoda. 2003. “Soogo-kooi, bunpoo, yosokukanoosei: «teyuuka» no bunnseki” [Interaction, grammar and projectability: Using an analysis of teyuuka as an example]. Goyooron Kenkyuu [Discourse Studies] 5: 31–43.

Name index A Aijmer, Karinâ•‡ 14, 18, 24, 27, 29, 30 Alfonso, Anthonyâ•‡ 194, 195, 202 Altmann, Hansâ•‡ 19, 23, 30 Auer, Peterâ•‡ 2–4, 6–8, 12, 15, 18, 19, 22, 23, 25, 28, 30, 31, 41, 60, 66, 73, 89, 97, 98, 103, 104, 106, 109, 111, 113, 117, 127, 141, 142, 145, 147, 167, 168, 171–173, 175–177, 180–182, 187, 188 B Bakhtin, M. M.â•‡ 71, 98 Barden, Brigitâ•‡ 172, 188 Barth-Weingarten, Dagmarâ•‡ 153, 163 Berrendonner, Alainâ•‡ 104, 120, 145 Birkner, Karinâ•‡ 12, 31 Blanche-Benveniste, Claireâ•‡ 105, 120, 124, 145 Brinton, Laurel J.â•‡ 49, 66, 172, 188, 196, 202, 209, 211, 212, 214, 216, 219, 235, 236 Bücker, Jörgâ•‡ 12, 31, 32 Buscha, Joachimâ•‡ 173, 188 Bybee, Joanâ•‡ 1, 8, 37, 66, 149, 163, 182, 186, 188, 236 C Couper-Kuhlen, Elizabethâ•‡ 1, 2, 12, 15, 19, 48, 59, 94, 103, 109, 110, 127, 139, 141, 149, 153, 175, 176, 221 Cristofaro, Soniaâ•‡ 2, 8, 14, 20, 31, 69, 72, 98, 191, 202 Croft, Williamâ•‡ 12, 19, 31, 172, 182, 188 Cruse, Alan D.â•‡ 182, 188 Curl, Traciâ•‡ 57, 66 D Dasher, Richardâ•‡ 149, 164 Deulofeu, Henri-Joséâ•‡ 104, 145

Deppermann, Arnulfâ•‡ 12, 31, 103, 147, 187 De Stefani, Elwysâ•‡ 121, 147 Diessel, Holgerâ•‡ 23, 31 Dixon, R.M.W.â•‡ 2 Drach, Erichâ•‡ 183, 188 Drew, Paulâ•‡ 57, 66, 94, 98, 129 Dryer, Matthewâ•‡ 235 Du Bois, John W.â•‡ 210, 221 Duvallon, Outiâ•‡ 137, 145 E Edmondson, Willisâ•‡ 23, 31, 173, 188, 218, 236 Elmauer, Uteâ•‡ 173, 188 Elstermann, Mechthildâ•‡ 12, 31, 172, 188 Englebretson, Robertâ•‡ 2, 8, 38, 66, 149, 163, 235, 236 Erelt, Matiâ•‡ 38, 57, 66 Erman, Brittâ•‡ 172, 188 Etelämäki, Marjaâ•‡ 2 F Fiehler, Reinhardâ•‡ 147, 172, 187, 188 Ford, Ceciliaâ•‡ 89, 98, 100, 103, 123, 143, 145, 175, 176, 188, 212, 236 Fox, Barbaraâ•‡ 89, 98, 164, 188, 214, 236 Fox Tree, Jean E.â•‡ 214 Fraser, Bruceâ•‡ 172, 188 Freywald, Ulrikeâ•‡ 167, 188 Fried, Miriamâ•‡ 12, 31 G Givón, Talmyâ•‡ 2 Goffman, Ervingâ•‡ 7, 8, 70, 71, 76, 82, 85, 98 Gohl, Christineâ•‡ 23, 31, 172, 182, 188 Goldberg, Adele E.â•‡ 12, 31, 107, 145 Goldberg, Julia Annaâ•‡ 211, 215, 236

Goodwin, Charlesâ•‡ 70, 85, 104, 135, 137 Goodwin, Marjorie H.â•‡ 70 Günthner, Susanneâ•‡ vii, 2–8, 11–14, 18, 19, 22, 23, 27, 28, 30–32, 49, 66, 74, 75, 104–106, 109, 113, 117, 123–125, 127, 134, 139, 141, 145, 146, 169, 172, 177, 182, 185, 188, 189, 201, 202, 210 H Haiman, Johnâ•‡ 1, 8, 9, 30, 32, 33, 67, 69, 98, 99, 146, 191, 203, 236 Hakulinen, Auliâ•‡ 43, 66, 70, 86, 91, 98, 145, 211, 236 Halonen, Miaâ•‡ 94, 98 Haspelmath, Martinâ•‡ 185, 188 Hayashi, Makotoâ•‡ 103, 106, 110, 135, 137, 146, 214, 236 Heine, Berndâ•‡ 1, 9, 10, 34, 67, 147, 149, 163, 164, 238 Helbig, Gerhardâ•‡ 173, 188 Higashiizumi (Iguchi), Yukoâ•‡ vii, 5–7, 153, 164, 191, 192, 194, 195, 203 Honda, Akiraâ•‡ 195, 203 Hopper, Paulâ•‡ 1, 2, 6, 7, 9, 12, 14, 18, 19, 28–30, 32, 33, 37, 49, 66, 73–76, 80, 98, 99, 105, 106, 108, 109, 123–127, 133–135, 138, 139, 141–143, 146, 149, 151–154, 164, 169, 170, 185–187, 189–191, 193, 194, 201, 203, 210, 211, 236 Horlacher, Anne-Sylvieâ•‡ 121, 147 Hosoda, Yuriâ•‡ 224, 225 House, Julianeâ•‡ 173, 218, 219 Hwang, Shin Jaâ•‡ 2, 10 I Ichihashi-Nakayama, Kumikoâ•‡ 153, 164, 192, 203 Imo, Wolfgangâ•‡ vii, 5, 7, 12, 19, 22, 23, 28, 30–33, 74, 109, 113, 117, 127, 141, 146, 165, 167–170, 185, 186, 188, 189, 209, 211, 213, 237

 Subordination in Conversation Ishigaki, Kenjiâ•‡ 193, 203 J Jacobs, Andreasâ•‡ 192, 203 Jacques, Francisâ•‡ 112, 146 Jasperson, Robertâ•‡ 214 Jefferson, Gailâ•‡ 37, 67, 88, 89, 98, 100, 103, 130, 147, 214, 215, 238 Jucker, Andreas H.â•‡ 188, 192, 203 Jullien, Stéphaneâ•‡ 108, 120, 146, 147 K Kärkkäinen, Eliseâ•‡ 73, 74, 99, 100, 211, 221, 237 Kasper, Gabrieleâ•‡ 218, 219 Keevallik, Leeloâ•‡ vii, 3–5, 7, 18, 23, 37, 42, 43, 54, 56, 57, 60, 62, 66, 67, 74, 95, 99, 211, 236, 237 Keevallik Erikson, Leeloâ•‡ 43, 211 Kelly, Johnâ•‡ 87, 99 Kitano, Hiroakiâ•‡ 209, 222, 224, 232 Kitzinger, Celiaâ•‡ 2, 4, 9, 77, 80, 99 Kobayashi, Kenjiâ•‡ 193, 203 Koivisto, Ainoâ•‡ vii, 3, 4, 7, 69, 70, 86, 89, 92, 98, 99, 142, 210 Konoshima, Masatoshiâ•‡ 193, 203 König, Ekkehardâ•‡ 2 Kroon, Carolineâ•‡ 172, 189 Kuteva, Taniaâ•‡ 149, 163 L Laitinen, Leaâ•‡ 75, 99 Lambrecht, Knudâ•‡ 12, 13, 33, 108, 120, 146 Langacker, Ronald W.â•‡ 2, 9, 13, 14, 18, 21, 33, 38, 67, 69, 72, 73, 80, 96, 99, 108, 146, 168, 171, 176–178, 180, 182, 186, 189 Laury, Ritvaâ•‡ vii, 1–7, 9, 11, 18, 20, 23, 33, 38, 66, 67, 69–71, 74, 76, 81, 86, 95, 98–100, 103, 113, 142, 146, 162, 164, 182, 187, 189, 190, 202, 209, 210, 236 Lauwereyns, Shizukaâ•‡ 222, 224, 225, 230 Lee, Reienâ•‡ 192, 203, 207 Lehmann, Christianâ•‡ 191, 203 Leino, Jaakkoâ•‡ 81, 99 Lenk, Utaâ•‡ 172, 189 Lerner, Gene H.â•‡ 2, 4, 9, 77, 80, 99, 103, 111, 146 Lindén, Eevaâ•‡ 81, 99 Lindholm, Camillaâ•‡ 43, 46, 54, 67

Lindström, Janâ•‡ 43, 46, 48, 54, 57, 66, 67, 196, 203, 211, 236 Linell, Perâ•‡ 186, 189 Local, Johnâ•‡ 87, 99, 164 Londén, Anne-Marieâ•‡ 237 Longacre, Robert E. â•‡ 2, 10 M Martin, Samuelâ•‡ 193 Masuoka, Takashiâ•‡ 194, 203 Matthiessen, Christianâ•‡ 2, 9, 20, 30, 33, 67, 69, 99, 104, 146, 191, 203 Maynard, Douglasâ•‡ 218 Maynard, Senkoâ•‡ 209, 222, 225, 230, 231, 235 McCarthy, Michael J.â•‡ 2 McGloin, Naomi H.â•‡ 209, 210, 222, 224, 225, 231 Miller, Jimâ•‡ 16, 31, 33 Minami, Fujioâ•‡ 194, 203 Mulac, Anthonyâ•‡ 7, 9, 23, 30, 34, 38, 57, 67, 109, 113, 117, 147, 153, 164, 165, 170, 182, 184, 190, 211, 238 Mulder, Jeanâ•‡ 86, 89, 99, 153, 164 Müller,Gabrieleâ•‡ 108, 123, 132, 135, 143, 147 Munro, Pamelaâ•‡ 81, 99 N Nagano, Masaruâ•‡ 193, 203 Nakayama, Toshihideâ•‡ 153, 164, 192, 203 Norén, Kerstinâ•‡ 186, 189 O Ochs, Elinorâ•‡ 1, 9, 33, 67, 103, 125, 145, 147, 149, 164, 236 Ogden, Richardâ•‡ 87, 88, 99, 100 Ohori, Toshioâ•‡ 153, 164, 192, 195, 196, 203, 204 Okamoto, Shigekoâ•‡ vii, 5–7, 18, 23, 86, 100, 113, 150, 151, 153, 154, 160, 162, 164, 182, 192, 204, 209 Ono, Tsuyoshiâ•‡ 12, 15, 31, 33, 86, 100, 149, 150, 163, 164, 175, 176, 188, 203, 221, 225, 236, 237 Onodera, Noriko O.â•‡ 192, 204 Östman, Jan-Olaâ•‡ 12, 31, 73, 100, 211, 237

P Payne, Thomas E.â•‡ 37, 67 Pekarek Doehler, Simonaâ•‡ vii, 3–5, 7, 18, 74, 75, 103, 106, 107, 111, 121, 143, 147, 182, 210 Pomerantz, Anita M.â•‡ 29, 33, 112, 147 Q Quirk, Randolphâ•‡ 195, 204 R Raussi, Anneâ•‡ 70, 71, 100 Raymond, Geoffreyâ•‡ 92, 95, 100, 204 Rehbein, Jochenâ•‡ 23, 30, 33, 182, 189 Routarinne, Saraâ•‡ 82, 88, 100, 137, 145 S Saari, Mirjaâ•‡ 211, 237 Sacks, Harveyâ•‡ 27, 33, 37, 67, 85, 88, 100, 103, 147, 214, 215, 238 Schegloff, Emanuel A.â•‡ 1, 2, 9, 12, 18, 33, 37, 41, 46, 67, 85, 87, 88, 100, 103, 104, 125, 126, 141, 145, 147, 149, 164, 214, 215, 236, 238 Scheibman, Joanneâ•‡ 74, 100, 186, 188, 211, 238 Scheutz, Hannesâ•‡ 19, 23, 34 Schiffrin, Deborahâ•‡ 94, 100, 172, 189, 209, 211, 214, 218, 219, 238 Schmid, Hans-Jörgâ•‡ 13, 16, 27, 34 Schmidt, Renateâ•‡ 34, 189 Schönherr, Beatrice â•‡ 180, 189 Schourup, Lawrence C.â•‡ 172, 189 Schrock, Josef C.â•‡ 214 Schumacher, Helmutâ•‡ 165, 189 Schwitalla, Johannesâ•‡ 23, 34 Selting, Margaretâ•‡ 1, 8, 9, 19, 23, 34, 98, 103, 104, 125, 145, 147, 149, 164, 165, 169, 175, 189, 238 Seppänen, Eeva-Leenaâ•‡ viii, 3, 4, 7, 69–71, 74, 76, 81, 85, 86, 95, 99, 100, 142, 187, 189, 210, 211, 236 Shirakawa, Hiroyukiâ•‡ 194, 195, 204 Shore, Susannaâ•‡ 77, 100 Stenström, Anna-Britaâ•‡ 211 Stoltenburg, Benjaminâ•‡ 180, 182, 189

Name index  Suzuki, Ryokoâ•‡ viii, 3, 5, 7, 149, 150, 153, 154, 160, 192, 195, 202 Suzuki, Satokoâ•‡ 150 Sweetser, Eveâ•‡ 193–197, 204 T Takubo, Yukinoriâ•‡ 194, 195, 204 Talmy, Leonardâ•‡ 2, 8, 9 Tanaka, Hirokoâ•‡ 235 Tannen, Deborahâ•‡ 75, 100 Tao, Hongyinâ•‡ 2, 9 Thim-Mabrey, Christianeâ•‡ 172, 189 Thompson, Sandra A.â•‡ 1–4, 7–10, 12, 18–20, 23, 29–34, 38, 45, 49, 57, 59, 66, 67, 69, 70, 72–76, 80, 86, 89, 94, 96, 98–100, 103–106, 108–110, 113, 117, 123–125, 127, 133, 135, 139, 141, 142, 145–147, 149, 153, 162, 164, 165, 170, 182, 184, 188–191, 203, 204, 209–212, 214, 236, 238

Tomasello, Michaelâ•‡ 23, 31, 66, 189 Traugott, Elizabeth C.â•‡ 1, 6, 9, 10, 34, 67, 126, 141, 146, 147, 149, 151–154, 164, 185, 186, 189–191, 193, 194, 198, 201–204, 235, 238

Weinert, Reginaâ•‡ 16, 33

U Uno, Ryokoâ•‡ 194, 204 Urmson, James O.â•‡ 211

Z Zifonun, Giselaâ•‡ 19, 23, 34

V Valli, Andréâ•‡ 124, 132, 147 van der Auwera, Johanâ•‡ 2, 9 Verhagen, Arieâ•‡ 29, 34 Verstraete, Jean-Christopheâ•‡ 191, 204 W Walker, Garethâ•‡ 87, 99 Waltereit, Richardâ•‡ 45, 67 Wakamatsu, Makikoâ•‡ 224, 225

Y Yabe, Hirokoâ•‡ 195, 204 Yamaguchi, Gyojiâ•‡ 193, 195, 204 Yoshii, Kazutoâ•‡ 193, 194, 198, 204

Subject index A actionâ•‡ 2–3, 17, 18, 62, 70, 72–73, 76–77, 79–82, 84–86, 91, 95–97, 109 mainâ•‡ 55, 65, 72, 76, 81, 96–97, 116 142 projectionâ•‡ 41–54, 64–65, 70, 85, 111, 113, 116–117, 126, 127–142 sequenceâ•‡ 38, 44, 46, 48, 61 65, 103–104, 110–125 typeâ•‡ 42, 45, 65 see also subordination activity mainâ•‡ 2, 70, 76, 77, 82, 95, 97 assessmentâ•‡ 58–63, 134 B biclausal sentence/constructionâ•‡ 29–30, 104–105, 117 C clause dependentâ•‡ 15, 20 38, 42 57 152–155, 162, 171; see also subordinate clause dependent mainâ•‡ 167, 173, 182, 184–185 complementâ•‡ 13–18, 20–24, 27–30, 37–39, 41–49, 51–53, 59–60, 70, 72 87–88, 91, 107–108, 113, 142, 151–154, 209–210, 212 independentâ•‡ 20–21, 23, 38, 42, 47, 64, 113, 120, 150–155, 157–160, 162, 177, 196 mainâ•‡ 2–3, 13–24, 26–30, 38–39 45, 47–49, 51–52, 55, 60, 70, 72, 73ff, 75–76 82, 96, 104, 113, 117, 142, 150–151, 153–156, 160–162, 166–187, 192–201, 210, 212 matrix clause, see main clause

subordinateâ•‡ 2–3, 13–19, 20, 23, 27–30, 37–39, 47–48, 51–52, 63, 64–65, 70, 75–76, 81–82, 84–85, 104, 106, 117, 133, 142, 152–154, 156, 165–168, 170, 171–173, 177, 180, 181, 184, 185, 191–202, clause combiningâ•‡ 104, 151–152, 196Â�–198 clause structureâ•‡ 152 cleft presentational cleftâ•‡ 105, 106, 108, 117–123, 142, 143â•‡ pseudocleftâ•‡ 103, 105–108, 123–125, 135, 138 wh-cleftâ•‡ 107Â�–108, 109–110 123–125 127 135–138, 139–141 clinesâ•‡ 181–184, 193–195, 201 communicative practicesâ•‡ 162 complementationâ•‡ 37–38 complement clause, see clause complementizerâ•‡ 14, 20, 27, 28, 60, 70–71, 73–82 88–89, 95, 96, 150–151, 154, 160, 162, 165, 167 complementâ•‡ 13, 24, 30, 37–39, 45, 47, 49, 51, 53, 55, 61, 81, 86–91, 96, 105–110, 113, 117, 161, 181–184, 234–235 subject complementâ•‡ 13, 39, 61 object complementsâ•‡ 39 53, 173 complement-taking constructionâ•‡ 23, 86, 211–214 complement-taking predicatesâ•‡ 37–58, 64, 70–82, 86, 88, 96–97, 165, 170, 182–184, 209 conjunctionâ•‡ 38, 70ff, 89, 166–167, 173 speech-actâ•‡ 193, 195, 197–201 subordinatingâ•‡ 17, 38 166, 167

Construction Grammarâ•‡ 11–13 constructionâ•‡ 11–14, 16–19 27–30, 74, 107–110 143–144, 149, 150–154, 169 182–187, 238 conversational action, see action coordinationâ•‡ 29–30, 104, 191 CTP, see complement-taking predicate CTP-phrase, see complementtaking predicate; see also complement-taking construction CTV, see complement-taking construction D declarativeâ•‡ 22, 59–60 discourse markerâ•‡ 113, 117, 141, 171–173, 176–177, 181–182, 184–185, 211 displacement markerâ•‡ 44 E ellipticalâ•‡ 39, 58, 60, 151, 160 Emergent Grammarâ•‡ 1, 9, 142–143, 186–187 epistemicâ•‡ 38, 39, 42, 49, 53, 55–58, 62–64, 72–76, 82 112–113, 117, 153, 182, 184, 185ff, 198–201, 211ff conjunctionâ•‡ 193, 194, 197 particleâ•‡ 38 verbâ•‡ 184 F footingâ•‡ 74, 76 formulaicâ•‡ 29, 42, 74, 76, 81, 97, 117, 123, 124, 209–211, 214 G grammaticalizationâ•‡ 1, 28–29, 51, 54, 60, 108–109, 113, 126, 138, 141, 151–154, 185–186, 193–195, 201 grammaticization see grammaticalization

 Subordination in Conversation H hedgeâ•‡ 61 I imperativeâ•‡ 38, 40–42, 45, 46, 48–49, 50, 53, 54ff, 57 incrementâ•‡ 175–178, 181 Interactional Linguisticsâ•‡ 1, 11, 104, 106, 110 interrogative clauseâ•‡ 38, 40, 43, 45, 47–59, 61–64 intonation unitâ•‡ 48, 58, 64, 210ff, 229 inversionâ•‡ 42, 47, 48, 51, 57, 64, 170, 194 M meaning potentialsâ•‡ 186, 187 metacommunicativeâ•‡ 71, 76, 82, 85 96, 97, 209, 210 multi-unitâ•‡ 46, 105, 106, 121–123, 125, 130, 133–139, 143 O on-lineâ•‡ 104, 106, 142, 143 P parataxisâ•‡ 29, 104, 121, 194, 195, 197–198, 201 participantframeworkâ•‡ 69–71, 73, 80–81, 96–97 particle conjunctiveâ•‡ 191 finalâ•‡ 70, 71 86–96, 150–151, 160, 192

pragmaticâ•‡ 42ff, 47 65, 150, 161, 181 questionâ•‡ 38 quotativeâ•‡ 210 turn-finalâ•‡ 57–58, 63, 64, 86–95, 97 utterance-finalâ•‡ 154 utterance-initialâ•‡ 65, 82–86 politenessâ•‡ 56, 58 pragmatic markerâ•‡ 141, 172, 181, 187, 196, 197 prefaceâ•‡ 42ff, 45–48, 56–57, 224, 225 profileâ•‡ 2, 18, 38, 45, 57, 64–65, 72–73 80, 96–97, 168–169, 175–179, 180–181, 183 determinacyâ•‡ 18, 168, 176, 180, 181 interactionally profiledâ•‡ 45 projectionâ•‡ 2–3, 14, 15–19, 20–30, 41–54, 64–65, 70, 71, 74–75, 76 84, 85, 87–97, 103–144, 175, 180, 182, 210, 213, 215, 223, 234 routinizedâ•‡ 143 see also action projection prosodyâ•‡ 17, 20, 22, 23, 38, 46, 48, 52, 58, 63, 64, 74, 82, 83, 86–87, 116, 117, 123, 124, 126, 132–133, 139, 141, 175, 177, 180, 213 Q questionâ•‡ 37–65, 79–81, 87–91, 94, 133, 160, 193, 197 210, 213 frameâ•‡ 43, 53–54, 63 indirectâ•‡ 36, 79, 212

see also interrogative complements quotativeâ•‡ 149–162, 210 R referentâ•‡ 46, 108 routinizationâ•‡ 29, 110, 124, 126, 141, 143–144, 172, 187 S sequencing of actions see action stanceâ•‡ 60 strategic management of interactionâ•‡ 73, 76, 80, 82, 97, 209, 210 subjectificationâ•‡ 186, 187 subjectiveâ•‡ 38, 193–194, 221, 235 subordination,â•‡ 1–3, 13, 23–24, 27–30 45, 57–58, 64–65, 69–72, 84–86, 91–97, 98–100, 104–106, 108–110, 111, 116–117, 129, 138, 139, 141–142, 152–153, 164–168, 175, 177–178, 181–184, 187, 191, 192, 196, 201–203 actionalâ•‡ 2–3, 82, 86, 91–96, 116, 129, 138 discursiveâ•‡ 196 subordinate clause, see clauseâ•‡ subordinating conjunction, see conjunction T topicâ•‡ 42–44, 49, 53, 62ff, 63–64, 65, 73, 87–89, 230