Fresh Perspectives on Major Issues in Pragmatics 2020035270, 9780367856366, 9781003017462

Citation preview

i

Fresh Perspectives on Major Issues in Pragmatics

This book showcases new and innovative developments and approaches in pragmatics, spotlighting perspectives from an international range of emerging scholars undertaking cutting-​edge research pushing the field in new directions. The volume begins by taking stock of the most up-​to-​date developments in pragmatics research, as embodied by the work of a new generation of pragmaticists. Chapters are organized around key areas of development within pragmatics, including intercultural and cross-​cultural pragmatics, cognitive pragmatics, and new perspectives on referencing, implicating, and inferring, shedding further light on the ways in which pragmatics increasingly interfaces with other linguistic disciplines and on innovative methodologies. The book also places the focus on pragmatics approaches in languages other than English, further expanding the borders of research. This book will be of particular interest to scholars in pragmatics interested in staying on top of the latest developments and future directions for the field. Monika Kirner-​Ludwig is Assistant Professor of Linguistics in the Department of English at Leopold-​Franzens-​University Innsbruck, Austria, and is affiliated with the University at Albany (SUNY), USA, and Tomsk State University, Russia.

ii

Routledge Research on New Waves in Pragmatics

Fresh Perspectives on Major Issues in Pragmatics Edited by Monika Kirner-​Ludwig For more information about this series, please visit: www.routledge.com/​ Routledge-​Research-​on-​New-​Waves-​in-​Pragmatics/​book-​series/​RRNWP

iii

Fresh Perspectives on Major Issues in Pragmatics Edited by Monika Kirner-​Ludwig

iv

First published 2021 by Routledge 52 Vanderbilt Avenue, New York, NY 10017 and by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN Routledge is an imprint of the Taylor & Francis Group, an informa business © 2021 Taylor & Francis The right of Monika Kirner-​Ludwig to be identified as the author of the editorial material, and of the authors for their individual chapters, has been asserted in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. Library of Congress Cataloging-​in-​Publication Data Names: Kirner-​Ludwig, Monika, 1981–​editor. Title: Fresh perspectives on major issues in pragmatics /​ edited by Monika Kirner-​Ludwig. Description: London; New York: Routledge, 2020. | Series: Routledge research on new waves in pragmatics; 1 | Includes bibliographical references and index. Identifiers: LCCN 2020035270 | ISBN 9780367856366 (hardback) |  ISBN 9781003017462 (ebook) Subjects: LCSH: Pragmatics. Classification: LCC P99.4.P72 F74 2020 | DDC 401/​.45–​dc23 LC record available at https://​lccn.loc.gov/​2020035270 ISBN: 978-0-367-85636-6 (hbk) ISBN: 978-1-003-01746-2 (ebk) Typeset in Sabon by Newgen Publishing UK

v

Contents

List of Figures  List of Tables  List of Contributors  Acknowledgements  List of Abbreviations  1 Introduction: New Waves in Pragmatics 

vii ix xi xiii xv 1

I S TVA N K E C S K E S AN D MO N IKA KIRN E R- ​L UDWIG

2 Can Cats Pick Up Chainsaws? The Interaction of Context and Plausibility 

29

S TA N L E Y A L E X AN DE R DO N AH O O

3 Scalar Implicature through the Lens of Distributional Linguistics 

42

M A X I M E C O D È R E CO RB E IL

4 “We Have a Big Crowd”: The Different Referents of the First-​Person Plural in U.S. Presidential Candidates’ Talk on Entertainment-​Political Interviews 

61

E E A N G R I M S H AW A N D ME N N O H . RE IJVE N

5 Whatever Can Be Meant Can Be Echoed: A Comparison of Since When Questions and Echo Declarative Questions 

82

ANGELIKA KISS

6 The Pragmatics of Japanese Quotative Constructions: A Comparative Study between Japanese and Japanese Heritage Language Families  K I YO N O F U J I N AGA -​G O RDO N

101

vi

vi Contents

7 Managing Epistemic Asymmetry through Dialogic Resonance in Therapy Interactions 

120

RO N G   L E I

8 At-​Issue or Not-​At-​Issue Discourse Contribution by Puisque (F ‘Since’)? Information Structure and Discourse Structure 

141

H A S M I K J I VA N YAN

9 A Discourse Model for “Undirected Speculation” 

163

E R I K A B E L L I N GH AM, H AN N O B E CK, AN D RICHAR D HAT C HER

10 Pragmatic Aspects of Translation: A Contrastive Analysis of Translation Processes Illustrated by Inductive-​ Empirical Eye-​Tracking, Writing Process Analysis, and a Questionnaire 

181

K Y E O N G H WA   L E E

Index 

203

vi

Figures

1.1 Pragmatics –​many streams to a river  3 1.2 Screenshot of record citations extracted from WSCC  8 1.3 Topics most frequently collocating with Pragmatic(s)  9 1.4 Distribution of topical linguistic spheres  13 1.5 Range of pragmatic topics (arranged clockwise)  16 1.6 Personal names cited in ACAC  17 1.7 Speech communities mentioned in ACAC  17 1.8 Varieties, styles, registers referred to in ACAC  18 1.9 Humor-​related terms in ACAC  18 1.10 In- and out-​grouping labels in ACAC  18 1.11 Doctor/​therapist/​caregiver-client/patient frame in ACAC  19 1.12 TV and CMC-​related terms in ACAC  20 1.13 Politics and media in ACAC  20 1.14 Forensic terms in ACAC  20 1.15 Terms pertaining to study objects, methods, and tools in ACAC  21 1.16 Miscellaneous terms in ACAC  21 2.1 Sequence of frames in a G-​maze (for every frame, only one word is a grammatical continuation of the sentence)  33 3.1 Example of a word-​vector for the word candies  44 3.2 Vector representations for tea and coffee  45 6.1 Summary of to-​marked quotative constructions uttered by caregivers  111 6.2 Summary of tte-​marked quotative constructions uttered by caregivers  112 6.3 Summary of to-​marked quotative constructions uttered by children  113 6.4 Summary of tte-​marked quotative constructions uttered by children  114 10.1 Screenshot of a recorded video by an eye-​tracker gaze point  192 10.2 Source text translated by the test person shown in Figure 10.1  193

vi

viii  List of figures 0.3 1 10.4 10.5 10.6 10.7 10.8

Korean subject 2’s fixation duration on Text 1  Korean subject 2’s fixation count on Text 1  Fixation duration on Text 1  Fixation duration on Text 2  Fixation count on Text 1  Fixation count on Text 3 

xi

Tables

1.1 Pragmatics series since 1980  1.2 Pragmatics journals since 1977  1.3 Pragmatics handbooks since 2008  1.4a Distribution of collocates with Pragmatic(s) in WSCC  1.4b Topics most frequently collocating with Pragmatic(s)  1.5 Pragmatics as occurring in journals cited by WSCC  1.6a Salient streams in pragmatics (according to Liedtke & Tuchen 2018)  1.6b Salient analytical frames in pragmatics (according to Liedtke & Tuchen 2018)  1.6c Salient topical foci in pragmatics (according to Liedtke & Tuchen 2018)  2.1 Mean decision times (ms) and percent error rates (in parentheses) for bias-​implausible and neutral-​plausible conditions  3.1 Cosine similarity measures between John ate candies and John ate some candies  4.1 Number and types of we per presidential candidate  4.2 Number and types of general we per presidential candidate  4.3 Number and types of delimited social we per presidential candidate  5.1 EDQs and SWQs conveying different propositional attitudes towards p = ‘John’s not coming with us’  6.1 Summary of participant pool  6.2 Total number of occurrences of to/​tte-​marked quotative constructions  8.1 Prototypical statuses of the clauses related by PSQ  8.2 Statuses of the clauses in non-​prototypical PSQ-​usages ((22)–​(24), newspaper examples)  9.1 A asserts p, and B accepts A’s assertion  9.2 A asks {r,¬r}, B responds by asserting r, A accepts B’s answer 

4 5 6 9 9 10 14 15 15 36 54 65 69 75 88 108 110 145 153 173 173

x

x  List of tables 9.3 A makes a USU p containing the inquisitive component {r,¬r}, B takes up the inquisitive component and asserts r, A accepts B’s assertion  9.4 A makes a USU p containing the inquisitive component {r,¬r}, B does not take up the inquisitive component (and does not challenge A’s assertion)  10.1 Selected terms and compound words marked in the experimental texts (German and English)  10.2 Selected terms and compound words marked in the experimental texts (German and Korean)  10.3 Explanations and comparison between the Korean and the English subject 

175 176 189 189 194

xi

Contributors

Hanno Beck is a doctoral candidate at the University at Buffalo (State University of New York, USA). His research focuses on formal models of discourse, as well as the syntax, semantics, and pragmatics of infinitival complement constructions.  Erika Bellingham is a doctoral candidate at the University at Buffalo (State University of New  York, USA). Her research focuses on the syntax, semantics, and pragmatics of event representation, as well as formal models of dialogue. Maxime Codère Corbeil is a doctoral candidate in Linguistics with a graduate minor in Cognitive Sciences at the University of Quebec in Montreal, Canada. His research interest lies at the boundary of pragmatics, semantics, cognitive sciences, and computational linguistics, specifically focusing on computational approaches in pragmatics and semantics and on the modelling of pragmatic processing. His dissertation focuses on linguistic prediction. Stanley A. Donahoo received his Linguistics MA with Merit from University College London and is currently a Linguistics doctoral candidate with a minor in Cognitive Sciences at the University of Arizona, USA. His dissertation, supported by the Bilinski Foundation, focuses on swearing. His research has received numerous awards and honors, including a grant from the German Federal Government (DAAD), a selective Fulbright Research Grant, a fellowship from the State Parliament of Berlin, Germany, and a Social and Behavioral Sciences Research Institute grant from the University of Arizona. Kiyono Fujinaga-​Gordon is a doctoral candidate in the Department of Linguistics at the University at Buffalo (State University of New York, USA). She specializes in aspects of second language acquisition and Sociolinguistics. Eean Grimshaw is a doctoral candidate in the Department of Communication at the University of Massachusetts, Amherst, USA. Using and developing ethnographic studies of communication, including cultural discourse analysis, he specializes in studies of communication and culture. Specifically, he

xi

xii  List of contributors is interested in studies of cultural logic, including political discourse in the U.S. context, and intercultural encounters. Richard Hatcher is a doctoral candidate at the University at Buffalo (State University of New  York, USA). His research involves the investigation of Cayuga and Seneca, Northern Iroquoian languages, focusing on the influence of pragmatics and information structure on word and sentence prosody. Hasmik Jivanyan is currently a PhD-​ researcher and teaching assistant at the University of Geneva, Switzerland. Her research focuses on causal connectives and causal relations in French and other languages,  from the perspective of the discourse-​grammar interface, involving the investigation of adverbial clause-​linking at syntactic, semantic, and discourse-​ pragmatic dimensions. Her primary research interests lie in such aspects as subordination-​coordination, hypotaxis-​parataxis, information structure at sentence and discourse levels, the taxonomy of coherence (causal) relations and their expression by (causal) connectives, various syntactico-​semantic uses of connectives and their profiling at the level of discourse-​pragmatics. Angelika Kiss is a doctoral candidate in the Department of Linguistics at the University of Toronto, Canada. Her doctoral research focuses on dynamic pragmatic models of non-​canonical questions, incorporating results from production and perception experiments on declarative and rhetorical questions in various languages. KyeongHwa Lee is currently a doctoral candidate in the Department of Applied Linguistics and Translation Studies at Leipzig University, Germany. She completed her Master’s degree in Translation Studies at Hankuk University of Foreign Studies, Seoul, Korea, and also in Linguistics at Heidelberg University, Germany. Her main research interests include neurocognitive linguistics and cognitive pragmatics. Rong Lei is a doctoral candidate at the National Key Research Center for Linguistics and Applied Linguistics, Guangdong University of Foreign Studies, which is the only key national research center of linguistics and applied linguistics in China. Her research interests include cognitive pragmatics and interpersonal pragmatics. Menno H.  Reijven is currently a doctoral candidate in the Department of Communication at the University of Massachusetts, Amherst, USA, and a lecturer at the University of Amsterdam, the Netherlands. Using discourse analytic methods, he studies the rhetoric and communication of politicians during political campaigns as well as the commentary produced by ordinary citizens.

xi

Acknowledgements

The present volume appears as the first of hopefully many in the newly established series Routledge Research on New Waves in Pragmatics. This title is to be understood in two ways: not only does it refer to the innovative approaches and ‘hot’ topics in pragmatics that the chapters in this and upcoming volumes and monographs address, but also does ‘New Waves’ refer to the fact that this format exclusively gives the floor to excellent early-​ career scholars in the discipline. Each one of the following selected chapters has not only been supervised and mentored by well-​established scholars of pragmatics, but underwent a rigorous double-​blind peer review before being accepted. The idea for this new series is one of several that Istvan Kecskes, Distinguished Professor at the State University of New York, USA, has been kind enough to share and discuss with me over the past five years. In fact, not only did he share the idea with me but also the credit for it by entrusting me with the task and responsibility of editing this very volume. I am tremendously grateful to him for his unceasing mentorship. I would also like to pay tribute to the American Pragmatics Association, especially President Istvan Kecskes, Executive Secretary Anne Bezuidenhout, and the American Pragmatics Association (AMPRA) Executive Committee for paving the way for innovative and open-​minded projects such as this one. Clearly, a volume like the present one could neither be composed in an efficient and diligent manner nor take anywhere near a decent shape, if it weren’t for those many perceptive minds and helping hands I was fortunate enough to have the support of. My gratitude goes to two anonymous reviewers for their benevolent and constructive input provided on the chapters and my book proposal as a whole. I would like to particularly thank Elysse Preposi, Helena Parkinson, Kelly Winter, and Sue Cope at Routledge for their patient and professional guidance and transparency throughout the (pre-​)publishing and copy-​editing process. For their meticulous work on the formatting and indexing of this volume, I am immensely grateful to my assistants Julia Heiss and Lena Langegger. I am very much indebted to the Office of the Vice Rector of Research and the Faculty of Language, Literature and Culture as well as my

vxi

xiv Acknowledgements home department at the University of Innsbruck, Austria, all of whom have supported me and my research unwaveringly throughout the past three years. Last but clearly not least, I am extremely grateful to the contributors to this volume. Without their enthusiasm for and commitment to this undertaking, this book, purely and simply, could not have been produced.

v x

Abbreviations

ACAC Corpus of AMPRA-​4 conference abstracts ACC accusative AF absolute frequency AI at-​issue AMPRA American Pragmatics Association AU audience BG background CG Common Ground CG* Projected set CMC computer-​mediated communication COCA Corpus of Contemporary American English COND conditional COP copula DA distributional approach DC Discourse Commitment DC* Projected Discourse Commitment DT Donald Trump EDQ echo declarative question EEG electroencephalography EN English EPI entertainment-​political interview F French F-​domain Focus  domain GE German G-​maze task grammaticality maze task HC Hillary Clinton HOPS Handbooks of Pragmatics, de Gruyter (2011) IPrA International Pragmatics Association IS information structure JF Jimmy Fallon JK Jimmy Kimmel KO Korean LOC locative ms millisecond(s)

xvi

newgenprepdf

xvi Abbreviations MT machine translation NAI not-​at-​issue NOM nominative NOMNL nominalizer PCQ parce que (F, ‘because’) POL polite form POSS possessive PSQ puisque (F ‘since’) Q question marker QUD question under discussion QUO quotative RF relative frequency RSVP task Rapid Serial Visual Presentation task SBC Santa Barbara Corpus SI scalar implicature S-​O-​V Subject-​Object-​Verb S-​V-​O Subject-​Verb-​Object SWQ since when question Table* Projected Table TOP topic TPR translation process research T-Vdistinction tu-vos distinction USU Undirected Speculation Utterance V2 verb-​second (order) WSCC Web of Science Core Collection * indicates ungrammatical construction or non-​attested occurrence

1

1  Introduction New Waves in Pragmatics Istvan Kecskes and Monika Kirner-​Ludwig

1  A Brief Introduction to this Introduction The field of pragmatics is a continuously advancing, dynamic conglomerate of frameworks and approaches. Today’s pragmaticists surely have the luxury of being able to resort to a number of detailed and comprehensive accounts that diligently dismantle the complex interrelations and philosophical streams that have been opening up and forming pragmatics as a heterogenous yet game-​ changing new field of linguistics (e.g. Koyama 2011; Liedtke & Tuchen 2018). Thus, we shall only dedicate a brief section to sketching the tradition of main influences into and in pragmatics (Section 1). The present introductory chapter is primarily concerned with the facets of pragmatics in today’s research agendas and with what pragmatics holds in store with regard to ongoing innovation and interdisciplinary approaches. Section 1 will thus essentially lay the foundation for dissecting the main currents that have continuously been shaping the field, including research on direct and indirect speech acts (Austin 1962[1955]; Searle 1975, 1982) and performativity (Bach & Harnish 1979; Fraser 1975; Recanati 1987; Searle 1989), interactional linguistics (Hopper 1988; Ochs et al. 1996; Couper-​Kuhlen & Selting 2001), conversation analysis (Schegloff & Sacks 1973; Sacks et al. 1974; Sidnell & Stivers 2013) and the Cooperative Principle (Grice 1975). As an additional layer of context, our groundwork account of the history of pragmatics uses up-​to-​date figures from pragmatics-​related publications in leading journals, volumes and handbooks in order to trace its development since the 1970s. Section 3 zooms in on the major research spheres and objects in pragmatics that have taken shape in the last two decades. After all, it is only if we seek answers to and insights into where it is that pragmatics currently stands with regard to the most frequently and saliently addressed topics that we can infer what further developing tendencies there may be. Thus, in order to establish a status quo of the main current research interests in pragmatics, Sections 3.1 and 3.2 resort to corpus linguistic and scientometric resources and tools. Tackling these issues will inevitably and aptly frame and contextualize the innovative contributions assembled in the present volume –​the new waves, so to speak (cf. Section 4).

2

2  Kecskes and Kirner-Ludwig

2  Pragmatics in Context and Context in Pragmatics, or: Many Streams to a River The field of pragmatics has come a long and windy way since the 1960s, essentially witnessing a shift from contextualism (in the Firthian sense) to context, or, more simplistically, from sentence to utterance. In other words, structuralism’s context-​isolated conceptions gave way to context-​dependent approaches, and the traditionally prescribed scholarly interest in fictional and historical text shifted to an acknowledgement of real discourse as an equally worthwhile object of investigation (Searle 1982). As Koyama puts it, context came to be a notion broadened from pertaining only to “tokens of linguistic types” to “communicative-​event participants and their pragmatic acts” (2011: 140; cf. Mey [1993]2001: 206–​235). From the perspective of pragmatics, disciplines like philosophy of language, anthropology, ethnomethodology, European sociology, psychiatry, psychology and ethology were confluent streams, generating a think tank of pragmatic ideas and research desiderates. Consequently, fresh views on text and both intra-​and extralinguistic context –​which arose from the seeds ultimately planted by scholars of semiotics (cf. e.g. Parret 1983; Deledalle 1989)  –​sprouted some of the most important pragmatic foundations that still serve as major benchmarks for today’s pragmatics (cf. Figure  1.1). These publications include the groundbreaking works on speech acts and performativity by John L. Austin ([1955]1962), John R. Searle (1969, 1975; also cf. Wunderlich 1972; Maas & Wunderlich 1972, 1976; Fraser 1975; Bach & Harnish 1979), Harvey Sacks et al.’s work on turn-​taking (1974), Herbert H. Clark & Edward F. Schaefer’s notions of Grounding and Common Ground (1989; also cf. Clark & Brennan 1991), and H. Paul Grice’s introduction of the Cooperative Principle and the notion of implicature (1975). It was only due to such fresh deliberations –​and initially strictly theoretical approaches to language use in context –​that the way was paved for such revolutionary frameworks as Politeness (Brown & Levinson 1987) and Relevance Theory (Sperber & Wilson 1995, 2004). As pragmatics was emancipating itself, its dissociation from other linguistic disciplines –​foremost Lexical Semantics (cf. e.g. McCawley 1978) –​was most effectively accomplished by channeling the rapidly growing body of pragmatic research towards new publication organs and international conferences. Thus, the 1970s and 1980s witnessed the launch of the Journal of Pragmatics (founding eds. Haberland & Mey, 1977–​; current eds. Haugh & Terkourafi, 168 volumes as of October 2020) and the Benjamins journal Pragmatics (founding ed. Gruber, 1986–​). The latter was started in lockstep with the first International Pragmatics Conference held in Italy in 1985 (and every two or three years since then) and the founding of the International Pragmatics Association (IPrA) with John Gumperz, the figurehead of Interactional Sociolinguistics (cf. 1978, 1982), as its first president (1986–​1990).1 In the 1980s Benjamins also launched the Pragmatics and Beyond book series: Pragmatics and Beyond (eds. Parret & Verschueren, 1980–​1986, 56 volumes), Pragmatics & Beyond

3

Introduction: New Waves in Pragmatics 3 Peirce (1932); Morris (1938, 1946); Bar-Hillel (1946); Carnap (1942) Structuralist methodologies + semiotic views 1950

Philosophy of language (Strawson 1950; Wittgenstein 1921; Carnap 1955; Morris 1965) Anthropology (e.g. Malinowski 1923, 1935)

Language and social context (Goffman 1955, 1959, 1963, 1967; Labov 1963, 1966)

Text linguistics (Dressler 1972)

Ethnomethodology (Garfinkel 1984)

European sociology (Habermas 1971, 1976; Brown and Levinson 1987) Psychiatry, Psychology (Goffman 1955, 1967) 1980

Sociolinguistics (Hymes 1974)

Conversation analysis (Sacks et al. 1974; Gumperz 1978)

Contextualism (Firth, e.g. 1957) Functionalism (Bühler 1934; Jakobson 1960; Hymes 1972; Halliday & Hasan 1976; Leech 1983)

PRAGMATICS Context Deixis Implicature Presupposition Perlocutionary effect Felicity conditions Communicative functions Performative utterances Fictional vs. real discourse

Silverstein (1976) “ indexicality” “metapragmatics”

Ethology, Behaviorism

Figure 1.1 Pragmatics –​many streams to a river

Companion Series (eds. Parret & Mey, 1985–​1987, 6 volumes) and Pragmatics & Beyond New Series, ed. Fetzer 1988–​). The latter has published 318 volumes to date (cf. Table 1.1).2 In the 1990s, pragmatics was productive enough for Benjamins to launch yet another handbook series, i.e. The Handbook of Pragmatics (eds. Jef Verschueren, Jan-​Ola Östman, Jan Blommaert & Chris Bulcaenthat, 1995–​), which, since 2003, has also been available as a continuously expanded online resource. Thus, the discipline of pragmatics had become so well established at that point that –​rather than being patronized and informed by other core linguistic disciplines –​the tables were beginning to turn: Blutner established the subfield of Lexical Pragmatics and postulated that it be needed to provide a “systematic and explanatory account of pragmatic phenomena that are connected with the semantic underspecification of lexical items” (1998: 115). By 2000, a strong pragmatic camp had worked its way even into Historical Linguistics. Thus, what Labov once coined the problem of “bad data” (1994: 11) is what papers published in the Journal of Historical Pragmatics (founding eds. Jucker & Taavitsainen 2000, current ed. Archer) have been making the very best of since its launch (i.e. in 21 volumes to date).

4

4  Kecskes and Kirner-Ludwig Table 1.1 Pragmatics series since 1980 Title of series, editors

Publisher

Active

Pragmatics and Beyond, eds. Herman Parret & Jef Verschueren Pragmatics & Beyond Companion Series, eds. Herman Parret & Jacob L. Mey Pragmatics & Beyond New Series, ed. Anita Fetzer The Handbook of Pragmatics, eds. Jef Verschueren, Jan-​Ola Östman, Jan Blommaert & Chris Bulcaen Palgrave Studies in Pragmatics, Language and Cognition, eds. Richard Breheny & Uli Sauerland Studies in Pragmatics, eds. Maj-​Britt Mosegaard Hansen, Kerstin Fischer & Anne Barron Handbook of Pragmatics Highlights, eds. Jef Verschueren & Jan-​Ola Östman Handbooks of Pragmatics [HOPS], eds. Wolfram Bublitz, Andreas H. Jucker & Klaus P. Schneider Key Topics in Semantics and Pragmatics [no series editor] Perspectives in Pragmatics, Philosophy & Psychology, ed. Alessandro Capone Yearbook of Corpus Linguistics and Pragmatics, ed. Jesús Romero-​Trillo

Benjamins

1980–​1986

Benjamins

1985–​1987

Benjamins

1988–​

Benjamins

1995–​

Palgrave [since 2015: Springer]

2004–​

Brill

2006–​

Benjamins

2009–​

De Gruyter

2010–​

Cambridge UP

2012–​

Springer

2013–​

Springer

2013–​

At the same time, completely new streams of pragmatics emerged and established themselves, amongst these Cognitive Pragmatics (cf. e.g. HOPS 4, ed. Schmid 2012), Formal Pragmatics (Potts 2010; Asher & Lascarides 2003; Kadmon 2001), Corpus Pragmatics (Romero-​Trillo 2008, 2018; Kopaczyk 2015), Interlanguage Pragmatics (Kasper & Blum-​Kulka 1993; Bardovi-​Harlig & Hartford 2005), Intercultural Pragmatics (Kecskés 2011, 2013, 2015; Kecskes & Assimakopoulos 2017) and Experimental Pragmatics (Noveck & Sperber 2004; Noveck 2018; Meibauer 2012).3 Intercultural Pragmatics and Historical Pragmatics have been represented by high-​profile journals for over 15 years (cf. Table 1.2). The majority of handbooks on pragmatics published since 1995 are displayed in Table  1.3,4 not including handbooks dedicated to individual foci within the field, for example, politeness and impoliteness (cf. Culpeper et al. 2017).

newgenrtpdf

5

Table 1.2 Pragmatics journals since 1977 Editors

Publisher

Active since

Number of volumesa

Impact factor

Journal of Pragmatics

Michael Haugh & Marina Terkourafi Helmut Gruber

Elsevier

1977–​

155

1.350 (5-​year: 1.580)

Benjamins

1986–​

29

0.605 (5-​year: 0.903)

Dawn Archer

Benjamins

2000–​

20

Istvan Kecskes Paul Chilton & Monika Kopytowska Jacob L. Mey Jesús Romero-​Trillo

De Gruyter De Gruyter

2004–​ 2005–​

16 14

0.941  (5-​year: 0.714) 0.290 (5-​year: 0.407) 1.158 (5-​year: 1.489) —​

Benjamins Springer

2010–​ 2017–​

10 4

0.429 (5-​year: 0.542) —​

Pragmatics [Quarterly publication of the International Pragmatics Association (IPrA)] Journal of Historical Pragmatics Intercultural Pragmatics Lodz Papers in Pragmatics Pragmatics and Society Corpus Pragmatics [International Journal of Corpus Linguistics and Pragmatics] a As of October 2020.

Introduction: New Waves in Pragmatics 5

Title of journal, editors

6

6  Kecskes and Kirner-Ludwig Table 1.3 Pragmatics handbooks since 2008 Title of handbook

Publ. year

Publisher

Series

Output

The Handbook of Pragmatics, eds. Laurence R. Horn & Gregory Ward The Cambridge Handbook of Pragmatics, eds. Keith Allan & Kasia M. Jaszczolt, The Routledge Handbook of Pragmatics, eds. Anne Barron, Yueguo Gu & Gerard Steen The Oxford Handbook of Pragmatics, ed. Yan Huang Handbuch Pragmatik, eds. Frank Liedtke & Astrid Tuchen

2008

Blackwell

Blackwell Handbooks in Linguistics

1 out of 60 books in that series

2012

Cambridge University Press

Cambridge Handbooks in Language and Linguistics

1 out of 36 books in that series

2017

Routledge/​ Taylor & Francis

Routledge Handbooks in Applied Linguistics

1 out of 38 books in that series

2017

Oxford

Oxford Handbooks in Linguistics

1 out of 57 books in this series

2018

Metzler/​ Springer

—​

—​

3  A Sea of Possibilities: Pragmatics as a “Functional Perspective on Language rather than a […] Linguistic Theory”5 It has been pointed out that, since the 1990s, social sciences have increasingly been forced to challenge “disciplinary thinking” and “traditional disciplinary boundaries” (Holden 1993: 1796). Gradually, such delimitations and categorizations have been blurred and bent almost out of recognition to accommodate torrents of new knowledge, to respond to the demand for socially relevant research by funding agencies, and to reflect the fact that the problems of greatest moment today have to be tackled by multiple approaches. (Holden 1993: 1796) Although precisely this notion is being postulated and implemented by the majority of today’s linguistic scholars, pragmatics has been claiming its very own ground next to –​rather than as part of or merely the opposite pole to –​ semantics. Yet, the issue of the semantic-​pragmatic interface and the debate

7

Introduction: New Waves in Pragmatics 7 “about how much actual situational context the linguistic signs need in order for them to be meaningful in the communicative process” (Kecskes 2019: 489) is still a current one. In 2012, Verschueren, advised that we should stop trying to assign to pragmatics its own set of linguistic features in contradistinction with phonology, morphology, syntax and semantics. If, for the sake of argument, pragmatics is to be defined as the study of meaning in context, it should study whatever meaning emerges as a result of the contextual use of any linguistic feature (including phonological, morphological, or syntactic ones), whether this feature has a ‘semantics’ of its own or not; hence semantics should not be the primary point of comparison, though it usually is treated that way in attempts at defining pragmatics. Given the research spheres and objects that pragmaticists have developed in the last two decades based on such significant groundwork as was provided by Austin ([1955]1962), Searle (1969, 1975), Fraser (1975), Grice (1957, 1975), Brown & Levinson (1987) and Sperber & Wilson (1995), where do we stand now with regard to the most frequently discussed topics in pragmatics? In order to provide an answer to this question and use it as a springboard for framing and contextualizing the contributions assembled in the present volume (cf. Section 3), we shall here confine ourselves to corpus linguistic tools applied to two specific corpora. First, we make use of the extensive data made available by the Web of Science Core Collection (WSCC) –​a corpus holding data from 1900 to the present  –​in order to track the evolution of certain Pragmatic research foci. Secondly, we collected a homogenous corpus of conference abstracts (including titles and keywords, when given) submitted for consideration of acceptance to AMPRA-​4, held at Albany, NY, in November 2018. By carving out collocates and nodes, we expected to find that those topics and pragmatic disciplines are currently be in alignment with what the most recent handbooks suggest, but also to discover which topics and focal areas have been emerging recently –​possibly off the grid. 3.1  Scientometric Insights from the Web of Science Corpus Web of Science (WSCC) is a publisher-​independent, multidisciplinary global citation database, channeling over 1.7 billion cited references from over 159  million records. Relevant to our case study is the pool of over 23,000 international and peer-​ reviewed scholarly journals across sciences in the WSCC, which is updated weekly to ensure the continuously up-​to-​date availability of browsable cited sources. For our case study, we narrowed down our search to only such journals in WSCC that pertain to linguistics and language studies. We set the relevant timeframe for our search at 2015–​2019. The dataset we extracted includes 99,437 words, made up only of full citations of records (i.e. not full texts), that is, records that contain the search string in their

8

8  Kecskes and Kirner-Ludwig

Figure 1.2 Screenshot of record citations extracted from WSCC

title, abstract or keywords. The screenshot in Figure 1.2 shows how these raw results are rendered in txt-​format. Tables 1.4a and 1.4b list the queries, search strings and results in descending order, with Table 1.4a displaying the count of journal records containing the topic and Table 1.4b, together with Figure 1.3, presenting the number and distribution of collocates with Pragmatic(s) in WSCC. Table 1.4b shows the search strings [Pragmatics + collocating topic] with the topics or foci most frequently collocating with Pragmatic(s) being context (35%), corpus (23%), culture (12%), politeness (9%) and speech act (7%), implicature (6%), awareness (5%) and humor (3%). This suggests two things: For one, only 15.1% of the records extracted explicitly commit themselves to a particular subfield of pragmatics in their title, abstract or keywords. In other words, 84.9% of 1499 records published on pragmatic phenomena do not restrict their scope to any of the established and browsed-​for labels. However, when it is ascertained which journals these records have appeared in, it is not very surprising: as Table 1.5 shows,6 only 3 journals out of the 46 listed are journals exclusively concerned with pragmatics, i.e. the Journal of Historical Pragmatics, the International Review of Pragmatics and the Journal of Pragmatics. It is obvious why in-​group peers within pragmatics would opt for detailed pragmatic foci rather than a more general understanding and application of the notion. Thus, the majority, i.e. 43 records are either much broader in scope (cf. 1., 3., 5., 6., 7., 11., 16., etc.), or particularly narrow. In the case of the latter, in most cases these deal with specific methodologies (35.), particular varieties (32.), certain phenomena in communication (20., 36.) or neighboring disciplines, such as translation (39.), language teaching and learning (17., 21., 37.), multilingualism (10., 31., 34.), language acquisition (18., 29.), cognition (24., 25.) and psychology (14., 41.). When grouping the 46 records from Table  1.5 together into topical linguistic spheres, the distribution is as shown in Figure 1.4: The three pragmatics journals (Historical Linguistics, International Review of Pragmatics, Journal of Pragmatics) cover only 10% of the data extracted. Journals dedicated to English Linguistics on the one hand and linguistics and language studies without any specification of object language restrictions on the other together cover 32% of the whole, which means that pragmatic issues are addressed

9

Introduction: New Waves in Pragmatics 9 Table 1.4a Distribution of collocates with Pragmatic(s) in WSCC Topic [double quotation marks ensure exact wording searched for]

AF of records referring to topic

RF

“Pragmatics” “Experimental Pragmatics” “Intercultural Pragmatics” “Historical Pragmatics” “Corpus Pragmatics” “Cognitive Pragmatics” “Gricean Pragmatics” “Formal Pragmatics” “Lexical Pragmatics” “Neo-​Gricean Pragmatics” “Prosodic Pragmatics” Total

1499 49 40 32 26 21 19 16 13 7 1

100% 3.3% 2.7% 2.1% 1.7% 1.4% 1.3% 1.1% 0.9% 0.5% 0.1% 15.1%

Table 1.4b Topics most frequently collocating with Pragmatic(s) Topic

+ collocating topic

Record counts

Pragmatics

+ context + corpus + culture + politeness + speech act + implicature + awareness + humor/​humour + deixis + intentionality + situational context + referentiality + performativity

938 611 327 251 179 156 143 75 32 11 11 9 6

3% 5% 6% 7%

35%

9% 12%

+ Context

+ Corpus

+ Politeness

+ Speech act

+ Awareness

+ Humor

+ Culture

+ Implicature

23%

Figure 1.3 Topics most frequently collocating with Pragmatic(s)

01

10  Kecskes and Kirner-Ludwig Table 1.5 Pragmatics as occurring in journals cited by WSCC Rank

Journals cited by WSCC

Publisher/​ Affiliate

No. of citations

1. 2.

Journal of English Linguistics Journal of Historical Pragmatics

54 53

3.

International Journal of English Linguistics

4. 5.

Advances in Social Science, Education and Humanities Research Lingvistika/​Russian Journal of Linguistics

6.

Glossa: A Journal of General Linguistics

7.

Arab World English Journal

8.

Journal of Politeness Research: Language, Behaviour, Culture International Review of Pragmatics

Sage, UK Benjamins, the Netherlands Canadian Center of Science and Education, Canada Atlantis Press, France RUDN University, Moscow, Russia Ubiquity Press, UK Malaysia /​USA, sponsored by Arab Society of English Language Studies, www.asels.org  De Gruyter, Germany Brill, the Netherlands De Gruyter, Germany Benjamins, the Netherlands Universidad Complutense, Madrid, Spain Elsevier, the Netherlands Sage, UK State University of Tomsk, Russia National University of Malaysia KARE Publishing Company, Turkey Cambridge University Press, UK

9. 10. 11. 12.

Multilingua –​Journal of Cross-​Cultural and Interlanguage Communication International Journal of Language and Culture Circulo de Linguistica Aplicada a la Comunicacion

13.

Procedia –​Social and Behavioral Sciences

14. 15.

Journal of Language and Social Psychology Universiteta Filologiya –​Tomsk State University Journal of Philology

16.

The Southeast Asian Journal of English Language Studies

17.

Modern Journal of Language Teaching Methods

18.

Studies in Second Language Acquisition

51

49 48

48 47

40 37 33 30 30 27 26 24 20 22 21

1

Introduction: New Waves in Pragmatics 11 Table 1.5 Cont. Rank

Journals cited by WSCC

Publisher/​ Affiliate

No. of citations

19.

Annual Review of Applied Linguistics

21

20.

Journal of Language Aggression and Conflict

21.

Studies in Second Language Learning and Teaching

22.

Rivista Italiana di Filosofia del Linguaggio

23.

Rilce: Revista de Filologia Hispanica

24.

Journal of Memory and Language

25.

Language, Cognition and Neuroscience

26. 27.

Estudios de Linguistica-​Universidad de Alicante-​ELUA Revista de Estudos da Linguagem

28.

Journal of Psycholinguistic Research

29.

Journal of Child Language

30.

Zeitschrift für romanische Philologie

31.

The Journal of English as a Lingua Franca

32.

Stellenbosch Papers in Linguistics Plus

33.

Natural Language and Linguistic Theory

34.

Linguistic Approaches to Bilingualism

35.

International Journal of Corpus Linguistics

36.

Journal of Communication Disorders

Cambridge University Press, UK Benjamins, the Netherlands Adam Mickiewicz University, Poland Italian Society of Philosophy of Language, Italy University of Narvarra, Spain Elsevier, the Netherlands Taylor & Francis, UK University of Alicante, Spain Federal University of Minas Gerais, Brazil Springer, Switzerland Cambridge Core, Cambridge University Press, UK De Gruyter, Germany De Gruyter, Germany Stellenbosch University, South Africa Springer, Switzerland Benjamins, the Netherlands Benjamins, the Netherlands Elsevier, the Netherlands

21 21

20 18 18 18 17 16

15 14

13 13 12 12 12 12 12

(continued )

21

12  Kecskes and Kirner-Ludwig Rank

Journals cited by WSCC

Publisher/​ Affiliate

37.

Innovation in Language Learning and Teaching Journal of Pragmatics

Taylor & Francis, UK Elsevier, the Netherlands Benjamins, the Netherlands Oxford Academic, Oxford University Press,UK Sage, USA Iranian Institute for Encyclopedia Research, Iran Oxford Academic, Oxford University Press, UK Taylor & Francis, UK National University of Malaysia Future Academy, UK

38. 39. 40.

Babel-​Revue internationale de la traduction-​ International Journal of Translation Digital Scholarship in the Humanities

41. 42.

Journal of Language and Social Psychology International Journal of Language Studies A Quarterly Journal of Applied Linguistics

43.

Journal of Applied Linguistics

44.

Inquiry: An Interdisciplinary Journal of Philosophy Gema Online Journal of Language Studies

45. 46. Total

The European Journal of Social & Behavioural Sciences

No. of citations 12 11 10 10

10 10

10

10 10 10 1048

more than three times more often in journals with much wider linguistic scopes than in the journals specifically concerned with pragmatics. What is particularly surprising, however, is the fact that journals from neighboring disciplines like psychology, philosophy, social and behavioral sciences, translation studies and education have produced almost 50% of the output on pragmatics. In other words, pragmatics as a discipline clearly spread extensively beyond its own in-​group during the period 2015–​2019. These numbers confirm the following: While pragmaticists have spent years establishing the boundaries that would justify their discipline’s raison d’être next to other linguistic core disciplines, the period 2015–​ 2019 presents a highly dispersed and interdisciplinary picture overall. Pragmatics has emerged as being a cross-​disciplinary enriching and informing benchmark for scholars of various disciplines, while itself mastering the skill of both aimed expansion and zooming: On the one hand, pragmaticists have been continuing to

31

Introduction: New Waves in Pragmatics 13

22% 10% 10%

9% 4% 45%

Pragmatics

Linguistics, language studies

English linguistics

Language teaching and learning

Multilingualism

Cross-disciplinary

Figure 1.4 Distribution of topical linguistic spheres

advance into the traditionally salient and characteristic key foci in their field, such as further exploring the intricacies of relevance (e.g. Padilla Cruz 2016) and speech acts (cf. e.g. Kissine 2009, 2013), or seeking to pin down the “precise relation between linguistic expression and our mental representation” (de Ponte & Korta 2017: 2), or challenging what has conventionally been referred to as the semantics-​pragmatics interface (e.g. Kecskes 2019). On the other hand, academic papers document the great potential for the interdisciplinary expansion of scope and the enhancement of analytical and methodological tools (cf. e.g. Jucker et al. 2018; Schneider et al. 2019). 3.2  A Corpus Compiled of American Pragmatics Association Conference Abstracts Our self-​compiled corpus of AMPRA conference abstracts (ACAC) provides assuring and more in-​depth evidence of the salience of certain topics in current pragmatic research. It is made up of 332 individual files, each one containing an anonymized abstract, title and keywords. ACAC was compiled on Sketchengine, totaling a solely English corpus of 138,665 words. The validity of this corpus case study is ensured by the fact that all the abstracts included responded to a call for papers that were inviting submissions on any “theoretical and applied research in pragmatics and […] different subfields of pragmatics (philosophical, linguistic, cognitive, social, intercultural, interlanguage, etc.)”. The four main topical strands mentioned in the call were all wide-​reaching in that they invited

41

14  Kecskes and Kirner-Ludwig any contributions concerned with: “(i) Pragmatics theories; (ii) Experimental pragmatics; (iii) Intercultural, cross-​cultural and societal aspects of pragmatics; (iv) Applications”. All abstracts submitted to the conference organizers –​i.e. abstracts for both long and short presentations, posters and invited key notes – have been included in the corpus under investigation. 3.2.1  Comparison of Streams, Topics and Frames with Liedtke and Tuchen’s List (2018) One of the latest pragmatics handbooks, edited by Liedtke and Tuchen (2018), lists the following as the most salient streams (Table 1.6a), analytical frames (Table 1.6b) and topical foci (Table 1.6c) in today’s pragmatic research. We juxtapose these with the absolute frequencies in ACAC. Tables  1.6a–​1.6c show that overall 80% of the research streams, topics and analytical frames listed by Liedtke and Tuchen (2018) are represented, in various degrees, in ACAC; 20% are not. Thus, the level of overlap is very high overall. As Table 1.6a shows, speech act theory is by far the highest-​ranking and rather isolated stream with 123 mentions across ACAC, whereas Interactional Sociolinguistics, Lexical Pragmatics and Critical Discourse Analysis do not feature at all in the data  –​explicitly, that is. Analytical frames (cf. Table  1.6b) that are particularly prominent in our corpus are politics, legal scenarios and language education (cf. education, classroom), which together cover 66% of this dimension. The range of topics as displayed in Table 1.6c is fairly spread out and corresponds in the main to Liedtke and Tuchen’s list (2018): However, studies on politeness and implicature on their own cover 55% of the data. Mentions of presupposition, humor, pragmatic competence and translation together cover a further 28% (cf. Figure 1.5 for the distribution of topics referred to in ACAC). Table 1.6a Salient streams in pragmatics (according to Liedtke & Tuchen 2018) Pragmatic streams

ACAC token count

Speech act Intercultural Pragmatics Neo-​Gricean [cf. also Grice 40; Gricean 14] Historical Pragmatics Relevance Theory [BUT: relevance 63] Interactional Linguistics Evolutionary Anthropology Semiotics Text Linguistics Interactional Sociolinguistics [BUT: Sociolinguistics 6] Lexical Pragmatics Critical Discourse Analysis [BUT: Discourse Analysis 13]

123 30 9 7 7 6 1 1 1 0 0 0

51

Introduction: New Waves in Pragmatics 15 Table 1.6b Salient analytical frames in pragmatics (according to Liedtke & Tuchen  2018) Salient analytical frames

ACAC token count

participation, participant, participate politics, political, politically, politician law, legal, lawyer education classroom science, sciences formula, formulaic (language) ritual(s) computer (mediated communication) religion, religious (verbal) aggression aphasia

262 155 97 76 58 56 30 28 19 9 6 0

Table 1.6c Salient topical foci in pragmatics (according to Liedtke & Tuchen 2018) Salient topical foci

ACAC token count

politeness, polite implicature presupposition humor [humour] pragmatic competence translation narrative, narration lying deixis quotatives salience expressive dimension/​component fiction gestures [BUT also: non-​verbal 7] pragmatic awareness [BUT also: pragmalinguistic awareness 2] ellipses anaphora conventional implicatures [BUT cf. implicature] defaultness interjections mediality

252 98 48 45 44 41 34 27 21 26 17 10 5 4 2 2 1 0 0 0 0

3.2.2  Saliencies in ACAC As already established, the ACAC data confirms to a large extent the topics and research directions that leading handbooks and journals promote. At the same time, we have not yet paid enough attention to those aspects that are

61

16  Kecskes and Kirner-Ludwig 2% 3% 2% 3% 3%

Politeness, polite Implicature Presupposition

4% 40%

6%

Narrative Expressive dimension/component Salience

7%

Lying 7%

Deixis 8%

15%

Quotatives Translation Humor Pragmatic competence

Figure 1.5 Range of pragmatic topics (arranged clockwise)

unique in ACAC in comparison to the other datasets and orientation points we have used. The first aspect we focus on  –​by using word lists and collocations  –​is the personal names that are cited most frequently in the abstracts. The word cloud in Figure  1.6 visualizes the relationship between individuals referred to, with a cut-​off line at 15 < citations. It is obvious that the scholars in our corpus cited Kecskes significantly more often than other researchers in the field, which correlates with the relatively strong emphasis on Intercultural Pragmatics covered by the abstracts submitted (cf. Table 1.6a). What is particularly curious in this collection of personal references is the fact that Donald Trump and the Clintons are amongst the most frequently cited individuals, with the former even ranking third overall in our list, thus even preceding Grice, Wilson and Levinson with regard to frequencies.7 This is an expressive reflection not only of political events accumulating and ongoing in 2017 and 2018, but mostly of their relevance to pragmatic researchers and, vice versa, the high level of applicability of pragmatic approaches to respective data from, for example, news outlets and social networking sites. What is more, the salience of Trump and Clinton correlates with what we have shown in Table  1.6b with regard to political discourse ranking amongst the most cited topics of research. Another peculiarity of the data compiled in ACAC is the wide range of languages, varieties and speech communities addressed (cf. Figure  1.7). English and Chinese (Mandarin) almost lie level with each other in terms of frequency in the corpus, followed by American, Japanese and Spanish, the latter two of which are also head to head. Twenty-​four more variety specifications occur with a frequency of ten and higher, which paints a highly diversified picture of languages and varieties tackled by pragmaticists in

71

Introduction: New Waves in Pragmatics 17

Figure 1.6 Personal names cited in ACAC

Figure 1.7 Speech communities mentioned in ACAC

ACAC within frames of, for example, intercultural, interlanguage, variational and social pragmatics. In order to identify further prominent topics and subtopics present in ACAC, we manually grouped about 400 key words into topical frames, displayed as Figures  1.8–​1.15. The word clouds have been created on https://​wordart. com with the images reflecting the hierarchies between terms in relation to absolute frequencies in the corpus. Word clouds are a handy tool to visualize the relationship between the major research objects chosen by the authors of the abstracts compiled in ACAC and thus create implications as to what the intra-​and interdisciplinary trends in pragmatics have been and may continue to be: Sociopragmatics and multilingualism (Figure 1.8), pragmatics and humor (Figure 1.9), (in-​and) out-​grouping labels in ACAC (Figure 1.10), and telecinematic and computer-​mediated communication (CMC) (Figure  1.12). Legal and forensic pragmatics (Figure 1.13) as well as data from doctor-​patient/​ therapist-​patient/​caregiver-client conversations (Figure  1.11) are clearly two of the most recent topics of major interest in pragmatics. Figures  1.15 and 1.16 display, for one, terms referring to study objects methods and tools in pragmatics research, and, for the other, miscellaneous terms not falling clearly into any of the other groups just listed. Personal names were not included in Figures 1.8–​1.16.

81

18  Kecskes and Kirner-Ludwig

Figure 1.8 Varieties, styles, registers referred to in ACAC

Figure 1.9 Humor-​related terms in ACAC

Figure 1.10 In- and out-​grouping labels in ACAC

91

Introduction: New Waves in Pragmatics 19

Figure 1.11 Doctor/​therapist/​caregiver-client/patient frame in  ACAC

4  Making (New) Waves: The Contributions to the Present Volume Although from its early days onwards, pragmatics has inherently acknowledged the central role of interlocutors using language in context, it is certainly safe to say that pragmatics research has become much more interaction-​or discourse-​ centric than it used to be in the 1970s, when the new discipline was centered much more on intention and the individual. The range of chapters and topical foci assembled in the present volume not only strongly affirms this shift overall, but also aligns significantly with the observations made and implications given in Section 3. Roughly, one half of the chapters included in the present volume respond exactly to the trending streams we have detailed above: For instance, the saliently pragmatic topics of political, telecinematic and medical discourse (cf. Figures  1.11, 1.12 and 1.13) are addressed by Grimshaw and Reijven in Chapter  4 as well as by Rong in Chapter  7. (Scalar) implicature, which ranks amongst the most prominent topics in today’s pragmatic research (cf. Figure  1.16), is addressed by Maxime Codère Corbeil in Chapter  3. Experimental points of departure are taken by Stanley Alexander Donahoo in Chapter 2, on the one hand, who tackles issues of plausibility and context (cf. Table 1.4b), and, on the other hand, by KyeongHwa Lee in Chapter 10, who employs eye-​tracking technology in order to visualize cognitive processes in pragmatic translation (cf. Table 1.6c, Figure 1.5). The other half of the chapters contributed to this volume may represent more isolated, yet no less innovative approaches to pragmatic phenomena situated at various intriguing interfaces: The chapter contributed by Angelika Kiss (Chapter 5) sheds new light on echoing utterances and argues in favor of a to-​be-​established distinction between metalinguistic and meta-​conversational echoes. In Chapter 9 Erika Bellingham, Hanno Beck and Richard Hatcher propose a new utterance type not previously recognized, which they refer to as ‘Undirected Speculation’. Hasmik Jivanyan studies the discourse-​contribution

02

20  Kecskes and Kirner-Ludwig

Figure 1.12 TV and CMC-​related terms in ACAC

Figure 1.13 Politics and media in ACAC

Figure 1.14 Forensic terms in ACAC

12

Introduction: New Waves in Pragmatics 21

Figure 1.15 Terms pertaining to study objects, methods, and tools in ACAC

Figure 1.16 Miscellaneous terms in ACAC

of puisque (‘since’) connective clauses in French in Chapter 8 by applying a formal discourse-​pragmatic model rather than a more conventionally employed discourse-​ analytical framework. Kiyono Fujinaga-​ Gordon in her chapter (Chapter  6) focuses on language use in Japanese heritage families in the US as a sofar much overlooked socio-​linguistic layer of investigation, focusing on their use of quotative constructions. As these glimpses should already show, each one of the chapters contributed to this volume combines interdisciplinary and innovative points of departure in various layers of pragmatic research, topical foci, methodological approaches and various cultural and linguistic spheres. We have therefore decided to arrange the chapters in a rather random manner within this volume. Detailed accounts

2

22  Kecskes and Kirner-Ludwig on each chapter’s accomplishments, however, will be given by presenting the authors with their chapters in alphabetical sequence in the following. The contribution by Erika Bellingham, Hanno Beck and Richard Hatcher introduces an as yet uncovered new utterance type not previously recognized and presents an account of how Undirected Speculation Utterances interact with other elements in a formal discourse model. Their research study demonstrates a graceful extension of prior work within a dialogue gameboard approach and takes an innovative step towards a treatment for tentativity in human language, establishing a flexible model that accurately reflects the effects of tentative discourse moves. The study presented by Maxime Codère Corbeil is one of the few strongly computational-​ theoretical contributions to this volume, in which he seeks to bring together implicated meaning in pragmatics, on the one hand, and distributional approaches, on the other hand, in order to compare the ‘complete meaning’ of two sentences. Demonstrating that different derivation procedures for scalar implicatures will lead to different sentence-​vectors when using distributional approaches to represent the meaning of a sentence, he stresses the mutual benefits for pragmatics and computational approaches in linguistics in recognizing and making use of each other’s set of tools and approaches. Stanley Alexander Donahoo employs the G-​maze task, a recently developed (2009) experimental paradigm, in order to tackle language use in context. By extending its use beyond sentence-​processing research and applying it to the domain of language research in context and by tying together two distinct bodies of literature, the author explores how novel situations will be interpreted based on interlocutors’ background knowledge about how the world typically functions. Kiyono Fujinaga-​Gordon focuses on language material collected from Japanese heritage families, which has  –​compared to, for example, Spanish and Chinese speech communities in the US –​been highly under-​researched so far. While she claims that research on heritage language acquisition tends to focus on the macro-​social aspects of language use such as language attitudes or ideology rather than on grammar acquisition, her study focuses on the acquisition of grammar indicating one’s pragmatic competence. Unlike previous studies that tend to look at caregivers’ language use, the author asserts and demonstrates how children are active participants in socialization –​focusing specifically on Japanese quotatives in her data. The chapter by Eean Grimshaw and Menno Reijven approaches the issue of political communication and campaigning by conjoining discourse analysis and pragmatics. The authors specifically concentrate on the uses of we in entertainment-​political interviews. To that end, they advance Dori-​Hacohen’s taxonomy (2014) by identifying different uses of the open general we (egalitarian and hierarchical) and two subtypes of the delimited social we (association and campaign). Hasmik Jivanyan assumes a fresh take on the discourse contribution of clauses that are linked by the discourse connective puisque (‘since’)

32

Introduction: New Waves in Pragmatics 23 in French, thereby shedding light on an aspect of the usage of discourse connectives that has not previously received much attention. Contrary to well-​established discourse-​analytical (mainly, coherence-​based) frameworks, where connectives have been studied mainly from the perspective of their import to the interpretation and type of coherence relation, Jivanyan’s chapter situates her study of puisque within a formal discourse-​pragmatic model based on the notion of Question Under Discussion. She does so by approaching the information structure of respective utterances not on the sentence-​level, as it has been traditionally done in terms of new/​given (i.e. theme/​rheme), but at discourse level, in terms of at-​issue vs. not-​at-​issue content. The chapter argues that although parataxis is a link between two independent assertions, each having its own illocutionary force, it cannot be accounted for by default as linking two at-​issue contents, as is the case with coordination. In her chapter, Angelika Kiss discusses two question types that can serve as echoing utterances, comparing the well-​established type of echo declarative questions to the much less salient type of since when questions. The author emphasizes that the range of aspects to be targeted by echoic utterances is wider than has previously been suggested: Besides the informative content, the form and the illocutionary act conveyed by an utterance, a speaker may felicitously echo rhetorical relations and the previous speaker’s stance, using either question type. Kiss’s chapter brings to our attention the necessity of a distinction between metalinguistic and meta-​conversational echoes, the latter being an echo of a discourse move itself. KyeongHwa Lee’s contribution investigates subjects’ cognitive writing and translating processes, when confronted with two language-​pairs that have significantly different syntactic structures, i.e. German into Korean and German into English. The author employs eye-​tracking technology  –​still rarely used in Pragmatics  –​recording and visualizing the participants’ decision-​making processes. Rong Lei draws on a sample of authentic data from therapy sessions to explore the use of dialogic resonance in managing epistemic asymmetry. Issues pertaining to epistemic status have previously mostly been addressed by conversational analysts (e.g. Ekberg & LeCouteur 2015), which makes Lei’s study one of the very few that have been approaching it from an applied pragmatic point of departure. She does so by considering relevance and contextual factors of client-​centeredness in therapy talk, exploring how self-​resonance and other-​ resonance are initiated to cope with asymmetries. Lei’s findings suggest that by resonating the therapist can not only intervene in the brain’s system of processing relevant information, but also may strategically emphasize such information. It is in this way that epistemic asymmetry is signaled, while epistemic symmetry is being affiliated. Through the lens of relevance theory and within the frame of applied linguistics, Lei sheds light on the highly practical issue of how professionals in health settings may make use of specific linguistic resources. Her research findings thus have significant implications for more effective counseling practices and training.

42

24  Kecskes and Kirner-Ludwig The early-​career contributors to this volume represent the enriching inter-​ and cross-​culturality of the field with viewpoints from Anglo-​American as well as Asian and European pragmatic schools and camps. The authors are currently affiliated with research institutions in the US, China, Switzerland, Germany, Austria, the Netherlands and the French as well as the English-​speaking parts of Canada. What is more, our contributors represent a wide range of multilingual backgrounds (e.g. Armenian, English, French, German, Japanese, Korean, Mandarin, Spanish) and their foci on language use pertain to a number of object languages (i.e. English, French, Japanese, Mandarin, Korean and German). The chapters assembled in this volume thus represent an impressive range and representative cross-​section of international pragmaticists’ perspectives inevitably relevant to both up-​and-​coming as well as established scholars across the field and in neighboring disciplines.

Notes 1 The association’s current president is Stephen Levinson (2018–​2023). For more details, cf. https://​pragmatics.international/​page/​Organization. 2 As of October 2020. 3 Also cf. Clinical Pragmatics, e.g. Cummings (2009); Schneider & Ifantidou (2020). 4 As of October 2020. 5 Verschueren (2012). 6 We only extracted those journal citations from Web of Science that were listed 10 times or more. We refer to them by rank no. (based on frequency) here. 7 Note that Obama occurs six times in the corpus.

References Asher, Nicholas and Alex Lascarides. 2003. Logics of Conversation. Cambridge: Cambridge University Press. Austin, John L. 1962[1955]. How to Do Things with Words. The William James Lectures Delivered at Harvard University 1955. Oxford: Oxford University Press. Bach, Kent and Robert M. Harnish. 1979. Linguistic Communication and Speech Acts. Cambridge, MA/​London: MIT Press. Bardovi-​Harlig, Kathleen and Beverly S. Hartford. 2005. Interlanguage Pragmatics: Exploring Institutional Talk. Mahwah, NJ: Erlbaum. Bar-​Hillel, Yehoshua. 1946. Analysis of “correct” language. Mind, 55(219): 328–​340. Blutner, Reinhard. 1998. Lexical pragmatics. Journal of Semantics, 15(2): 115–​162. Brown, Penelope and Stephen C. Levinson. 1987. Politeness. Some Universals in Language Usage. Cambridge: Cambridge University Press. Bühler, Karl. 1934. Sprachtheorie. Die Darstellungsfunktion der Sprache. Jena: Gustav Fischer. Carnap, Rudolf. 1942. Introduction to Semantics. Cambridge, MA: Harvard University Press. Carnap, Rudolf. 1955. On some concepts of pragmatics. Philosophical Studies: An International Journal for Philosophy in the Analytic Tradition, 6(6):  89–​91.

52

Introduction: New Waves in Pragmatics 25 Clark, Herbert H. and Edward F. Schaefer. 1989. Contributing to discourse. Cognitive Science, 13(2): 259–​294. Clark, Herbert H. and Susan E. Brennan. 1991. Grounding in communication excerpt. In Lauren B. Resnick, John M. Levine and Stephanie D. Teasley (eds.), Perspectives on Socially Shared Cognition. Washington: American Psychological Association. 127–​149. Couper-​Kuhlen, Elizabeth and Margret Selting (eds.). 2001. Studies in Interactional Linguistics. Amsterdam: Benjamins. Culpeper, Jonathan, Michael Haugh and Dániel Z. Kádár (eds.). 2017. The Palgrave Handbook of Linguistic (Im)politeness. London: Palgrave Macmillan. Cummings, Louise. 2009. Clinical Pragmatics. Cambridge: Cambridge University Press. De Ponte, María and Kepa Korta (eds.). 2017. Reference and Representation in Thought and Language. Oxford: Oxford University Press. Dori-​ Hacohen, Gonen. 2014. Establishing social groups in Hebrew: ‘We’ in political radio phone-​in programs. In: Pavlidou, Theodossia-​Soula (ed.), Constructing  Collectivity: ‘We’ across Languages and Contexts. Amsterdam/​Philadelphia, PA: Benjamins.  187–​206. Ekberg, Katie  and Amanda  LeCouteur. 2015. Clients’ resistance to therapists’ proposals: Managing epistemic and deontic status. Journal of Pragmatics, 90:  12–​25. Félix-​Brasdefer, J. César. 2012. Interlanguage Pragmatics. Oxford: Blackwell Publishing. Firth, John R. 1957. A synopsis of linguistic theory 1930–​1955. In: Firth, John R. (ed.), Studies in Linguistic Analysis. (Special volume of the Philological Society). Oxford: Blackwell Publishing. 1–​32. Fraser, Bruce. 1975. Hedged performatives. In: Cole, Peter and Jerry L. Morgan (eds.), Syntax and Semantics, Volume 3: Speech Acts. New York: Academic Press. 187–​210. Garfinkel, Harold. 1984. Studies in Ethnomethodology. Malden, MA: Polity Press/​ Blackwell Publishing. Goffman, Erving. 1955. On face-​work: An analysis of ritual elements in social interaction. Psychiatry, 18(3): 213–​231. Goffman, Erving. 1959. The Presentation of Self in Everyday Life. New  York: Doubleday & Company. Goffman, Erving. 1963. Stigma: Notes on the Management of Spoiled Identity. Englewood Cliffs, NJ: Prentice-​Hall. Goffman, Erving. 1967. Interaction Ritual. Essays on Face-​to-​Face Behavior. New York: Pantheon Books. Grice, H. Paul. 1957. Meaning. The Philosophical Review, 66(3): 377–​388. Grice, H. Paul. 1975. Logic and conversation. In: Cole, Peter and Jerry L. Morgan (eds.), Syntax and Semantics, Volume 3: Speech Acts. New York: Academic Press. 41–​58. Gumperz, John J. 1978. The conversational analysis of interethnic communication. In E. Lamar Ross (ed.), Interethnic Communication. Athens: University of Georgia Press,  13–​31. Habermas, Jürgen. 1976. Was heißt Universalpragmatik? In: Apel, Karl-​ Otto (ed.), Sprachpragmatik und Philosophie. Frankfurt am Main: Suhrkamp. 174–​272. Halliday, Michael A.  K. and Ruqaiya Hasan. 1976. Cohesion in English. London: Longman. Holden, Constance. 1993. New life ahead for social sciences. Science 261(5129): 1796–​ 1798. doi: 10.1126/​science.8378781

62

26  Kecskes and Kirner-Ludwig Hopper, Paul J. 1988. Emergent grammar and the a priori grammar postulate. In: Tannen, Deborah (ed.), Linguistics in Context. Connecting Observation and Understanding. Norwood, NJ: Ablex Publishing Corporation. 117–​134. Hymes, Dell H. 1972. Models of the interaction of language and social life. In: Gumperz, John J., and Dell H. Hymes (eds.), Directions in Sociolinguistics: The Ethnography of Communication. New York: Holt, Rhinehart and Winston. 35–​71. Hymes, Dell H. 1974.  Foundations in Sociolinguistics: An Ethnographic Approach. Philadelphia: University of Pennsylvania Press. Jakobson, Roman. 1960. Linguistics and poetics. In: Sebeok, Thomas (ed.), Style in Language. Cambridge, MA: MIT Press. 350–​377. Jucker, Andreas H., Klaus P. Schneider and Wolfram Bublitz (eds.). 2018. Methods in Pragmatics (Handbooks of Pragmatics 10). Berlin/​Boston: De Gruyter Mouton. Kadmon, Nirit. 2001. Formal Pragmatics. Semantics, Pragmatics, Presupposition, and Focus. Oxford: Blackwell Publishing. Kasper, Gabriele and Shoshana Blum-​ Kulka. 1993. Interlanguage Pragmatics. New York: Oxford University Press. Kecskes, Istvan. 2011. Interculturality and intercultural pragmatics. In: Jackson, Jane (ed.), The Routledge Handbook of Intercultural Communication. London: Routledge.  67–​84. Kecskes, Istvan. 2013. Intercultural Pragmatics. Oxford: Oxford University Press. Kecskes, Istvan. 2015. Intracultural communication and intercultural communication: Are they different? International Review of Pragmatics 7(2): 171–​194. Kecskes, Istvan and Stavros Assimakopoulos (eds.). 2017. Current Issues in Intercultural Pragmatics. Amsterdam/​Philadelphia: Benjamins. Kecskes, Istvan. 2019. Impoverished pragmatics? The semantics-​pragmatics interface from an intercultural perspective. Intercultural Pragmatics, 16(5): 489–​515. Kissine, Mikhail. 2009. Illocutionary forces and what is said. Mind and Language, 24(1): 122–​138. Kissine, Mikhail. 2013. From Utterances to Speech Acts. Cambridge: Cambridge University Press. Kopaczyk, Joanna. 2015. Diachronic Corpus Pragmatics: Irma Taavitsainen, Andreas H.  Jucker and Jukka Tuominen (eds.), John Benjamins, Amsterdam, 2014 [book review]. Journal of Pragmatics, 84(4):  33–​36. Koyama, Wataru. 2011. The rise of pragmatics: A historiographic overview. In: Bublitz, Wolfram and Neal R. Norrick (eds.), Foundations of Pragmatics (Handbooks of Pragmatics 1). Berlin/​Boston: De Gruyter Mouton. 139–​166. Labov, William. 1963. The social motivation of language change. WORD, 19(3): 273–​309. Labov, William. 1966.  The Social Stratification of English in New York City. Washington D.C.: Center for Applied Linguistics. Labov, William. 1994. Principles of Linguistic Change. Vol 1: Internal Factors. Oxford: Blackwell. Leech, Geoffrey N. 1983. Principles of Pragmatics. London: Longman. Maas, Utz and Dieter Wunderlich. 1972. Pragmatik und sprachliches Handeln. Mit einer Kritik am Funkkolleg, Sprache. Frankfurt am Main: Athenäum. Malinowski, Bronislaw. 1972[1923]. The problem of meaning in primitive languages. Supplement in: Charles K. Ogden and Ivor A. Richards, The Meaning of Meaning. A Study of the Influence of Language upon Thought and of the Science of Symbolism:

72

Introduction: New Waves in Pragmatics 27 With Supplementary Essays by B.  Malinowski and F. G. Crookshank. London: Routledge and Kegan Paul. 296–​336. Malinowski, Bronislaw. 1935. Coral Gardens and Their Magic. A Study of the Methods of Tilling the Soil and of Agricultural Rites in Trobriand Islands. London: George Allen & Unwin. McCawley, James D. 1978. Conversational implicature and the lexicon. In: Cole, Peter (ed.), Syntax and Semantics, Volume 9: Pragmatics. New  York: Academic Press. 245–​259. Meibauer, Jörg. 2012. Pragmatic evidence, context, and story design: An essay on recent developments in experimental pragmatics. Language Sciences, 34(6): 768–​776. Mey, Jacob L. 2001[1993]. Pragmatics. An Introduction. Oxford/​ Malden, MA: Blackwell Publishing. Morris, Charles W. 1946. Signs, Language, and Behavior. New York: Prentice-​Hall. Morris, Charles W. 1965. On the unity of the Pragmatic Movement. Rice University Studies 51(4): 109–​119. Noveck, Ira A. 2018. Experimental Pragmatics. The Making of a Cognitive Science. Cambridge: Cambridge University Press. Noveck, Ira A. and Dan Sperber (eds.). 2004. Experimental Pragmatics. New  York: Palgrave Macmillan. Ochs, Elinor, Emanuel A. Schegloff and Sandra A. Thompson (eds.). 1996. Interaction and Grammar (Studies in Interactional Sociolinguistics 13). Cambridge: Cambridge University Press. Padilla Cruz, Manuel (ed.). 2016. Relevance Theory. Recent Developments, Current Challenges and Future Directions. Amsterdam/​Philadelphia, PA: Benjamins. Peirce, Charles S. [1897]1932. Division of signs. In: Hartshorne, Charles, Paul Weiss and Arthur W. Burks (eds.), Collected Papers of Charles Sanders Peirce. Cambridge, MA: Belknap Press of Harvard University Press. 227–​234. Potts, Christopher. 2010. Formal Pragmatics. In: Cummings, Louise (ed.), The Routledge Pragmatics Encyclopedia. Abingdon/​New York: Routledge. 167–​170. Recanati, François. 1987. Meaning and Force: The Pragmatics of Performative Utterances. Cambridge: Cambridge University Press. Romero-​Trillo, Jesús (ed.). 2008. Pragmatics and Corpus Linguistics: A Mutualistic Entente. Berlin/​Boston: De Gruyter Mouton. Romero-​Trillo, Jesús. 2018. Corpus pragmatics and second language pragmatics: A mutualistic entente in theory and practice. Corpus Pragmatics, 2(2): 113–​127. Sacks, Harvey, Emanuel A. Schegloff, and Gail Jefferson. 1974. A simplest systematics for the organization of turn-​taking in conversation. Language, 50(4): 696–​735. Schegloff, Emanuel A. and Harvey Sacks. 1973. Opening up closings. Semiotica, 8(4): 289–​327. Schmid, Hans-​Jörg. 2012. Cognitive Pragmatics (Handbooks of Pragmatics 4). Berlin/​ Boston: de Gruyter. Schneider, Cosima, Carolin Schonard, Michael Franke, Gerhard Jäger and Markus Janczyk. 2019. Pragmatic processing: An investigation of the (anti-​) presuppositions of determiners using mouse-​tracking. Cognition, 193: [no pagination]. https://​doi. org/​10.1016/​j.cognition.2019.104024. Schneider, Klaus P. and Elly Ifantidou. 2020. Developmental and Clinical Pragmatics (Handbooks of Pragmatics 13). Berlin/​Boston: de Gruyter. Searle, John R. 1969. Speech Acts. An Essay in the Philosophy of Language. London: Cambridge University Press.

82

28  Kecskes and Kirner-Ludwig Searle, John R. 1975. Indirect speech acts. In: Cole, Peter and Jerry L. Morgan (eds.), Syntax and Semantics, Volume 3: Speech Acts. New York: Academic Press. 59–​82. Searle, John R. 1982. Indirekte Sprechakte. In: Searle, John R. (ed.), Ausdruck und Bedeutung. Untersuchungen zur Sprechakttheorie. Frankfurt am Main: Suhrkamp.  51–​79. Searle, John R. 1989. How performatives work. Linguistics and Philosophy, 12: 535–​558. Sidnell, Jack and Tanya Stivers (eds.). 2013. The Handbook of Conversation Analysis. Chichester, West Sussex, UK: Wiley-​Blackwell. Silverstein, Michael. 1976. Shifters, Linguistic Categories and Cultural Description. In: Basso, Keith H. and Henry A. Selby (eds.), Meaning in Anthropology. Albuquerque: University of New Mexico Press. 11–​55. Sperber, Dan and Deirdre Wilson. 1995. Relevance: Communication and Cognition. Oxford: Blackwell Publishing. Strawson, Peter F. 1950. On referring. Mind, 59(235): 320–​344. Verschueren, Jef. 2012. The pragmatic perspective. In: Östman, Jan-​ Ola and Jef Verschueren (eds.), The Handbook of Pragmatics [16th Installment]. Amsterdam: Benjamins.  1–​36. Wilson, Deirdre and Dan Sperber. 2004. Relevance theory. In: Horn, Laurence R. and Gregory Ward (eds.), The Handbook of Pragmatics. Oxford: Blackwell Publishing. 607–​632. Wittgenstein, Ludwig.  1921. Logisch-​ philosophische Abhandlung. In: Ostwald, Wilhelm (ed.), Annalen der Naturphilosophie [vol. 14]. Leipzig: Veit. 185–​262. Wunderlich, Dieter (ed.). 1972. Linguistische Pragmatik. Frankfurt am Main: Athenäum Verlag.

92

2  Can Cats Pick Up Chainsaws? The Interaction of Context and Plausibility Stanley Alexander Donahoo

1  Research Background and Objectives1 A growing body of literature provides support for the view that participants are faster to respond to words when they are preceded by meaningful sentence contexts (cf. e.g. Duffy et  al. 1989; Sharkey & Sharkey 1992). While long narratives preceding target items have provided facilitatory effects, contexts as small as three words have also been found to be advantageous. Jordan and Thomas (2002) demonstrated that participants were faster to respond to a target preceded by the context the cellar was than when presented with a transposed version (cellar the was). Conversely, Murray (1982) found no difference in reaction time between scrambled plausible and scrambled implausible sentences, concluding that the effect must be due to sentence meaning rather than intra-​lexical priming. Since the work of Grice (e.g. 1991) and Austin (e.g. 1975), it has become clear that sentences are not merely the sum of the compositional parts of the literal meaning of the individual words featured, but that various contextual elements of conversation will essentially influence the interpretation of an utterance. Grice postulated that much of what is implicated meaning is derived from the Cooperative Principle. Inspired by Grice, current pragmatic theories essentially argue that context, linguistic or otherwise, is necessary for listeners to arrive at the intended meaning of a speaker (e.g. Fekete et al. 2014; Sauerland 2004). In the work presented here, context is taken to mean the parts of discourse that surround a particular word, or series of words, and that can serve to provide more information about it. Regarding context effects and lexical access, various mechanisms have been proposed in an attempt to explain how context affects word decisions. However, a unified theory has yet to reach consensus. Broadly though, two views  –​one interactive, the other modular  –​have emerged. Modular theories of lexical access (e.g. Fodor 1983; Cairns 1984) concluded that context effects do not facilitate access, as the process of word determination is self-​directed and therefore not influenced by prior context. Forster and Bednall (1976) proposed that lexical access is autonomous; although context determines which one of the possible meanings is selected in the case of lexical ambiguity, all available meanings are argued to be accessed initially.

03

30 Donahoo Forster and Bednall labeled this process exhaustive access. Thus, context effects are the result of knowledge-​based representations that do not have an influence on lexical access in word perception. In a defining study, Swinney (1979) concluded that sentence contexts do not influence lexical access directly, but rather are used to select between alternative senses at a later stage of processing. That is, Swinney’s results support a view in which lexical access is autonomous; a context only shows effects after access is fully realised. Swinney’s claim is supported by two lexical decision experiments he conducted, where an auditorily presented context preceding a target word provided either no disambiguating context (unbiased condition), or a strongly predictive, disambiguating one (biased condition). For example, in a given context where the word bug was presented to bias the insect interpretation, participants were required to make a lexical decision for either ANT, SPY, or an unrelated control word. The first experiment found that, when the target was presented immediately after the ambiguous word in the context sentence, lexical decisions for both the contextually appropriate and inappropriate meanings of the ambiguous word were facilitated. However, the second experiment showed that participants were faster to respond to targets that related only to the biasing context when the target was presented several syllables after the ambiguous word. Swinney took this to suggest that even a strongly biasing context does not guide access, but that any advantage is post-​lexical, at a later stage of language processing. Modular theories such as these account for contextual effects in the following manner: Preceding sentences facilitate lexical access by regulating non-​perceptual decisions about a word’s identity as determined by the output from the perceptual module. That is, a context does not affect processes involved in the autonomous lexical access process, and any observed facilitatory effects likely arise from knowledge-​based representations (e.g. Fodor 1983; Foss & Speer 1991). A more recent study by Witzel and Forster (2014) implicates that any facilitation a word receives is not because of some aspect of the information in the context, but perhaps is simply the result of basic co-​ occurrence counts. Numerous other theories and models of a modular sort have been proposed to accommodate contextual roles (e.g. Hess et al. 1995). Specifically regarding plausibility effects, Murray argued that, from a Fodorian perspective, plausibility effects cannot be the result of lexical semantics, as this would require “the influence of real-​world knowledge into linguistic processing” (2006: 81). On the other hand, interactive theories claim that lexical access is directly facilitated via spreading activation by the sentence content and structure (e.g. Collins & Loftus 1975; Marslen-​Wilson & Tyler 1980). The spreading activation process pre-​activates likely candidates that best complete a scenario with the implication that visual perception itself is altered to help facilitate selection towards the contextually appropriate lexical candidate, as any level of content (e.g. lexical, sentential, or message-​level) may influence any other level above or below it (e.g. Duffy et al. 1989; McClelland & O’Regan 1981; Morris 1994). Clearly, while modular and interactive theories make distinct

13

Can Cats Pick Up Chainsaws? 31 predictions as to how and whether context influences lexical access, there is consensus amongst both camps that there are multiple levels of representation. When describing context and its role in the lexical access process, it is also important to consider that not all context is facilitatory in the same manner. Here is where the concept of plausibility should be considered. While it is straightforward to distinguish plausible and implausible sentences from a semantic perspective, it is a challenge to provide an objective definition of plausibility itself. Ratcliff (1987) provides examples of what is facile to intuit: The dog chewed the bone is much more plausible than The octopus ate the refrigerator. While both sentences are grammatically well-​formed, only the former conjures a possible daily-​life scenario. Still, these examples do not provide an objective sense, or show clear-​cut criteria, of plausibility; we only arrive at what is plausible in contrast to what is not based on the perceptions of the experimenter. One method that has been used to circumvent experimenter-​ defined cases of plausibility is to ask participants to provide plausibility rankings on some numerical scale, where more plausible sentences are rated higher than those that are less plausible (e.g. Thornton & MacDonald 2003). While this may add a degree of reliability in establishing what is plausible or less so, rating tasks do not provide any explicit definitions, either. The present study, outlined below, specifically addresses the issue of what is and is not plausible experimentally, and how this degree of plausibility established via context is sometimes in direct competition with our world knowledge. Regardless of the lack of an operational definition, plausibility has nonetheless been documented as having facilitatory ramifications on reaction time in a variety of tasks, known as the plausibility effect. It has been found to affect accuracy in the Rapid Serial Visual Presentation (RSVP) task (Forster & Ryder 1971), where participants are asked to recall sentences presented at a rate as fast as 16 words per second. Plausibility effects have also been demonstrated in reaction time tasks where the processing of meaning is likely to be irrelevant to the task, as in sentence-​matching paradigms, where participants are presented with two sentences simultaneously and merely have to state whether they are literally the same or different (e.g. Murray 1982; Ratcliff 1987). Similarly, through grammaticality tasks as tested by, for example, Forster and Olbrei (1973) it was found that participants tend to identify plausible sentences as grammatical more quickly than implausible sentences. In Section 2.2, the methodology adhered to in the present study, the maze task, is described, which differs from the tasks outlined so far in that it avoids any sort of explicit question-​answering. That is, when a task requires a participant to respond to a question, it becomes unclear as to whether or not responses about plausibility or context effects, for example, are due to task effects. Murray (2006) points out that, when readers are presented with implausible phrases, but know that they have to answer questions about these phrases, readers would presumably pause to re-​check the meaning through self-​reflection, ensuring accuracy of their seemingly bizarre interpretation. Is this re-​checking mechanism what accounts for the plausibility effect? When we encounter bizarre strings like The train fell on the umbrella, we tend to go

23

32 Donahoo back and check the semantic content to be sure that we have parsed the string correctly. Now, this ‘boggle’ factor of making sense of an implausible phrase could then be the source of the delay which affords plausible phrases their processing advantage over implausible constructions. But, in a delayed-​response, sentence-​matching task where participants were asked to delay their responses to implausible sentences, so that they were certain they had worked out the unlikely, though ‘correct’ meaning, participants were still faster to respond to plausible strings (Murray 1982). It could be argued, then, that plausibility is lexically driven; something about the words themselves in plausible sentences primes identification of other words more quickly. In a study investigating this idea, Forster and Olbrei (1973) presented participants with phrases like The doctor cured the patient and compared them to scrambled sentences like The patient cured the doctor. Here, where all of the lexical elements are the same in both sentences, the former is the more plausible one. Participants were indeed faster to respond to the plausible version, indicating that the plausibility effect is a sentential, and not merely a lexical, component of language processing. While plausibility and contextual effects have been studied separately, research is still lacking for how the two might interact during processing. As a result, the research question motivating the current study was to ask whether context has an effect on plausibility. It was predicted that a ‘biasing context’ will allow for implausible meaning accommodation. That is, when a biasing context is stipulated, participants will not have difficulty in accepting scenarios in which the world, or characters in the world, behave differently from conventional or normal expectation. As pointed out by McCawley (1971), in certain contexts like dreams, lexical selectional restrictions are violated, and are not really restrictions at all. In adult populations using cartoons, rather than dream scenarios, is this also what we find experimentally? Here, in the first study to examine plausibility with a maze task, I investigate whether, given a specific type of context, the plausibility effect that is associated with our world knowledge can be overridden.

2  Methodology 2.1  Participants A total of nine students from the University of Arizona participated in the experiment. All were native or near-​ native speakers of English and freely volunteered in response to an email recruitment circulated through various campus departments. They did not receive any course credit for their participation, but were made aware via an oral consent process that their participation was voluntary, that the data collected remained anonymous, and that they could stop at any time. No participants were excluded from the analysis. 2.2  Task, Materials, and Design A total of 105 items were created for a G-​maze task. The G-​maze task is a behavioural paradigm in which participants see two words simultaneously

3

Can Cats Pick Up Chainsaws? 33 displayed on a computer screen in front of them (Forster 2010). Depending on their decision as to which one of the two words is a better fit in complementing a sentence, participants press one button to choose the word displayed on the left, or another to choose the word on the right. Employing their working memory, participants decide which word in the set is the best fit to continue the sentence they are building incrementally. To get them started in the maze, the first word in the sentence is paired with a letter string (x-​x-​x), randomly occurring on the left or right. One example of a target maze sentence (with correct choices in bold for ease of reading) is: The x-​x-​x /​ cat before /​retire walked /​ past headcount /​drags the /​ of chainsaw /​and lecture /​fetch ran /​ swim towards /​ fireproof the /​unlucky placed /​mouse. thwart. The slashes indicate the contents of each frame participants see on their screen. Please see Figure 2.1 for further illustration. During the task, reaction times are recorded for how quickly the participant indicates their decision between the two words, and also how quickly or slowly they make progress through the sentence. These reaction times grant us a window into how language parsing unfolds in the mind of the participant, and how this parsing is influenced (or not) by contextual content, as well as plausibility.

Figure 2.1 Sequence of frames in a G-​maze (for every frame, only one word is a grammatical continuation of the sentence)

43

34 Donahoo The maze task affords several advantages as a methodology: As participants must make a decision for every word presented, there is no concern that participants might adopt a relatively constant rate of button-​pressing, which could be observed in self-​paced reading (cf. Witzel et al. 2012). Additionally, since decisions are made for every word displayed, there are no spillover effects, where the result of reading a difficult grammatical construction at one point in the sentence (e.g. a morphologically complex verb) is apparent only in the words further downstream in the sentence (e.g. at the following direct object). Spillover effects can be seen in both self-​paced reading and eye-​tracking (cf. Mitchell 1984; Rayner 1998). However, with the maze task, participants demonstrate forced, incremental processing, which is necessary for insight into the plausibility effect (Witzel et al. 2012). Each item consists of both a context sentence (biasing or neutral) and the target maze sentence over which the data is collected. In the target G-​maze sentence, for each word except for the first, an alternative word is presented simultaneously, and decision times are recorded for the participant’s choice between these two. Whether or not the correct word for continuing the sentence was on the left or the right is randomly determined within the experiment script. The experiment manipulates both the preceding context (a neutral, everyday occurrence, or biased situation) and the plausibility of the maze target sentence (plausible or implausible) in a one-​factor design. Biasing contexts feature television and book characters intended to be salient to all participants (e.g. Snoopy and the Peanuts gang, Teenage Mutant Ninja Turtles, Mickey and Minnie Mouse, Tom & Jerry). Stimuli were adapted from those provided by Filik and Leuthold (2008). To illustrate, sets (1)  and (2)  are actual stimuli, which shows how a slightly bizarre context establishes a reality that is rather different from the actual world, or how a relatively banal context does not provide strong reason to change how reality is generally perceived: (1) Context: Tom chased Jerry into the garage and it looked as though Jerry was cornered. Target: The cat picked up the chainsaw and ran towards the unlucky mouse. (2) Context: The garage was not a suitable place for the pets to play. Target: The cat walked past the chainsaw and ran towards the unlucky mouse. Further, the pairing of the correct alternative with the incorrect alternative was exactly the same in both the plausible and implausible conditions. Two lists of items were prepared, so that a sentence presented in a neutral context + plausible target (pets playing the garage + cat walked past the chainsaw) in the first list was presented in a bias context + implausible target (Tom and Jerry fighting + cat picked up the chainsaw) in the second list. Concerning plausibility, the target sentence is equally plausible at the point of variance (i.e. cats can walk past or pick up many items), but is different at the critical item of analysis (i.e. chainsaw).

53

Can Cats Pick Up Chainsaws? 35 2.3  Procedure The stimulus presentation was controlled by a Pentium PC laptop, using the Windows-​based DMDX software, developed by J. C. Forster at the University of Arizona (Forster & Forster 2003). Items were presented as black letters on a white background. After presenting instructions for the task, and a few practice items, participants were presented with the experimental items while seated comfortably in a sound-​dampened booth. Participants read the context sentence, which was displayed for 3500 ms, after which immediately appeared the word READY in the centre of the screen for 1050 ms, so as to ensure that participants fixate on the correct area of the screen. Immediately after this fixation word had left the screen, the first word in an item, along with its alternative, appeared. As per the task instructions, participants decided which alternative was a grammatical continuation of the sentence and then pressed one of two keys indicating their decision. If the correct alternative was on the left, then the left SHIFT key was to be pressed. If it was on the right, then the  right SHIFT key was to be pressed. If the correct alternative was chosen, the display immediately advanced to the next pair of alternatives in the sentence. If the wrong alternative was chosen, an error message was produced, and the display automatically advanced to the next context (participants could not complete the sentence if they had chosen the wrong word). Participants had 4000 ms to make a decision before the display advanced automatically. In running the experiment, only a negligible number of instances occurred where participants were timed out (93 out of 8815 total observations, or 0.01% of trials), likely because of a failure to attend to the task, and not because the alternative choice was too difficult. Reaction times were recorded for all pairs of two alternatives, but were only analysed at the critical item (cf. chainsaw in (1) and (2) above).

3  Results Prior to any data analysis, it was decided that the data from any participant whose error rate exceeded 21% on the target words would be excluded (cf. e.g. Davis et al. 2008). This did not apply to any participant in our experiment. The overall error rate –​that is, the error rate for the entire task, not just the critical items –​was higher than expected (33% for both conditions). However, given that this was a pilot study, with a small number of participants, and that the task was quite long (105 items), these figures do not indicate that the results are unreliable, but merely that the participant’s needs, like more experiment breaks, must be taken into account in future design. Reaction times on trials where an error occurred were discarded (222 out of 8815 total observations, or 0.02% of trials) as were any reaction times that were less than 300 ms or exceeded 2000 ms, which is a standard practice for the field (e.g. Forster et al. 2009; Witzel & Forster 2014). The results were analysed using linear-​mixed effects modelling in R (Baayen 2008). This method allows for two crossed random effects (participants and items), and assesses the raw reaction times

63

36 Donahoo Table 2.1 Mean decision times (ms) and percent error rates (in parentheses) for bias-​ implausible and neutral-​plausible conditions Bias-​Implausible (Tom & Jerry)

Neutral-​Plausible (Pets play in garage)

673 (33.1)

706 (32.9)

for each trial without aggregating over items or participants. This type of modelling does not result in a traditional p value as usually reported in statistical analyses. Following standard practice, any t value greater than 2.0 was deemed significant (Bell, Forster & Drake 2015). Before attempting to fit a mixed-​effects model to the data, the reaction times were transformed using a reciprocal transformation in order to minimise the effects of positive skew. Back-​transformed reaction times are reported here. Comparing bias-​implausible with neutral-​ plausible, there was no significant difference on percent error. There was a significant reaction time difference of 33 ms (fixed effects t-​value of -​3.372); see Table 2.1. The results overall show that, given a specific kind of context, the plausibility effect could in fact be reversed. That is, local pragmatic anomalies of cats picking up chainsaws can be reconciled by providing a context that counteracts an individual’s world knowledge. To verify that this effect was indeed driven by the context provided, and not some artefact of the design, the same experiment was run again, only this time removing any preceding context. Thirteen students who had not previously participated in the first iteration of the experiment, were recruited. In this case, the participants only completed the maze portion of the task; there was no context sentence provided before any of the items. In this version of the experiment, where the context had been removed, there was only an insignificant 1 ms difference for implausible and plausible maze strings (1004 ms vs 1003 ms for average decision time over the entire maze string). This confirms that participants were indeed reading the context sentence in the first experiment, and that the observed difference is being driven by the context itself.

4  Discussion In this first study examining plausibility with a maze task, we found that, given a specific type of context, the plausibility effect can be reversed; adult participants were faster for implausible sentences when a cartoon-​like scenario was presented immediately beforehand, as opposed to a neutral context (673 ms vs 706 ms for the critical item). While the number of participants is relatively small, the number of items and the robustness of the effect means that a reasonable amount of confidence can be attributed to the observed differences. The experiment results demonstrate that, given a specific type of context, one which is short but strongly biasing, the plausibility effect can be reversed. But what is it about these contexts that produces such an intriguing result? Perhaps the biasing contexts have some

73

Can Cats Pick Up Chainsaws? 37 unique features. The contexts feature well-​known characters, but with rather unusual traits, especially when contrasted with the arbitrary cases of the neutral contexts. Theoretically speaking, a mental model framework, for example, as suggested by Johnson-​Laird (1995), could serve to account for the results, where the default context is built up over time. This default is used to check on the computations required to carry out the experimental task, and perhaps the usage of ‘special’ context is taken into account after the fact, which results in the reaction time difference. It seems that, in order to wipe out the plausibility effect, a strong context with characters that participants have presumably known since childhood is required. It is also worth considering the role that the fictional aspects of the biasing context plays. Filik and Leuthold (2013) have demonstrated N400 effects when fictional characters act in ways that are against their nature (The mouse picked up the dynamite, where the mouse is Mickey Mouse, not Jerry), but would nonetheless be normal actions for ‘ordinary’ humans/​ characters. Would it be odd for Superman to walk through a city, rather than fly? Could the same results be obtained by using well-​known, actual historical figures? In order to make sense of exactly what mechanisms are responsible for this plausibility shift, more work is necessary. Long-​term memory structure has also been implicated in studies addressing context, and may also play a role in the biasing contexts resonating strongly with participants. Federmeier and Kutas (1999) had participants read pairs of sentences for comprehension while electroencephalography (EEG) measurements were recorded. The first sentence established a bias for the second sentence, in which participants expected an exemplar of a semantic category. The second sentence (the one for which the EEG was recorded at the end) ended either with the expected exemplar, an unexpected exemplar (but from the same, expected category), or an unexpected item from an unexpected category. For example, one of the stimulus items was: “Checkmate,” Rosaline announced with glee. She was getting to be really good at chess/​monopoly/​football. Both the unexpected exemplar and unexpected item elicited the N400,2 but the N400 was smaller for the unexpected exemplar condition, despite the poorer fit to the context, as they were lower in plausibility. Thus, category structure has an influence on language comprehension, where context effects might constrain access, though not necessarily online. The authors also report that the reduction in N400 was not due to plausibility alone. The findings align with recent findings as documented in the neuropragmatic literature. Hunt et  al. (2013) report findings which suggest that pragmatic meaning is computed online and integrated into the sentence model rapidly enough to influence comprehension of later words. Yet, our results do conflict with other reports. Warren et  al. (2008) report that providing a context for readers had no effect on facilitating the processing of impossible events, events arguably more likely than even our cartoon scenarios (see the appendix in their work for comparison). More work is needed to reconcile these differences. Concerning the materials, it is debatable as to whether or not the targets described as plausible are truly plausible, or at least more plausible than the

83

38 Donahoo implausible targets. This is admittedly a problem in the design of the experiment. Is it really more plausible for the participant to expect that a cat would walk past a chainsaw, even when the participant knows that the pets were playing in the garage? This is not a trivial point and might possibly need to be addressed in future studies of the same vein if underlying mechanisms of mental architecture are to be truly assessed. While this experiment demonstrates an interesting finding, it does not directly contribute to the modular vs. interactive debate as outlined above, and perhaps raises more questions than it answers. If pragmatic processing is truly optional (e.g. Fodor, Garrett et al. 1980), our research will not inform this debate. On the other hand, if plausibility is a pragmatic effect, as our findings seem to suggest (cf. also Murray 1982), then new proposals will be required. Still, reaction time differences are likely a reflection of post-​perceptual processes, after lexical access has taken place; thus, any differences are likely determined by the executive processing system, not by the lexical access module itself. As Jordan and Thomas aptly point out, “Target word performance may differ between different sentence contexts even when the actual perception of target words remains unchanged across conditions” (2002: 34).

5  Confounds and Future Directions Given the finding that adult participants were faster for implausible sentences when a cartoon-​like scenario was presented immediately beforehand, it would be worthwhile to see how children might respond to the same task. Would children even need a biasing context to allow for cats to pick up chainsaws? What is plausible for children, who engage in play more often than adults, and have less world knowledge to guide their pragmatic inference-​making process, might mean that a gradient account of plausibility be required. Jakobson (2002) points out that children are particularly willing to associate specific colours with particular sounds, so cats picking up chainsaws should hardly seem odd. The results of this experiment show that, given the ‘right’ kind of context, the plausibility effect can be reversed, though some design issues need to be addressed; specifically, a 2x2 design (neutral context + plausible target, neutral context + implausible target, bias context + plausible target, bias context + implausible target) would allow for a stronger analysis. Clearly, much work in this area, both from an experimental and theoretical perspective, remains to be done, but this study suggests there are fruitful gains to be made in pursuing the interaction of world knowledge, pragmatic understanding, and the influence on lexical access.

Notes 1 I am grateful to Ken Forster for his steadfast guidance at all stages of this project, Tom Bever, and members of the AMPRA community for their invaluable feedback. Special thanks to Monika Kirner-​Ludwig for many helpful comments and for facilitating this volume.

93

Can Cats Pick Up Chainsaws? 39 2 The N400 is a negatively deflected voltage change in the evoked response potential, peaking around 400 ms after onset. The N400 is larger for words that are not in consonance with whatever the current semantic context is than for words that match the context. For a review, see Kutas and Federmeier (2011).

References Austin, John. 1975. How to Do Things with Words (2nd edn.). Oxford: Clarendon Press. Baayen, R.  Harold. 2008. Analyzing Linguistic Data: A Practical Introduction to Statistics using R. Cambridge, UK  and New York: Cambridge University Press. Bell, Dane, Kenneth Forster and Shiloh Drake. 2015. Early semantic activation in a semantic categorization task with masked primes: Cascaded or not? Journal of Memory and Language, 85: 1–​14. https://​doi.org/​10.1016/​j.jml.2015.06.007 Cairns, Helen. 1984. Current issues in research in language comprehension. In Naremore, Rita C., Language Sciences: Recent Advances. San Diego, CA: College-​ Hill Press. 221–​237. Collins, Allan and Elizabeth Loftus. 1975. A spreading-​activation theory of semantic processing. Psychological Review, 82(6): 407. Davis, Chris, Kim Jeesun, and Kenneth, Forster. 2008. Being forward not backward: Lexical limits to masked priming. Cognition, 107(2): 673–​684. https://​doi.org/​ 10.1016/​j.cognition.2007.07.016 Duffy, Susan, John Henderson and Robin Morris. 1989. Semantic facilitation of lexical access during sentence processing. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15(5): 791–​801. Federmeier, Kara and Marta Kutas. 1999. A rose by any other name: Long-​term memory structure and sentence processing. Journal of Memory and Language, 41: 469–​495. Fekete, István, Mátyás Gerőcs, Anna Babarczy and Balázs Surányi. 2014. Logical and pragmatic meaning in the interpretation of connectives: Scalar implicatures and “shallow” processing. In: Emonds, Joseph and Markéta Janebová. Language Use and Linguistic Structure. Proceedings of the Olomouc Linguistics Colloquium 2013. http://​real.mtak.hu/​20300/​1/​ORpaper_​Olinco_​2014.pdf Filik, Ruth and Hartmut Leuthold. 2008. Processing local pragmatic anomalies in fictional contexts: Evidence from the N400. Psychophysiology, 45(4): 554–​558. https://​ doi.org/​10.1111/​j.1469-​8986.2008.00656.x Filik, Ruth and Hartmut Leuthold. 2013. The role of character-​based knowledge in online narrative comprehension: Evidence from eye movements and ERPs. Brain Research, 1506: 94–​104. https://​doi.org/​10.1016/​j.brainres.2013.02.017 Fodor, Jerry. 1983. The Modularity of Mind: An Essay on Faculty Psychology. Cambridge, MA: MIT Press. Fodor, Jerry, Merrill Garrett, Edward Walker and Charles H. Parkes. 1980. Against definitions. Cognition, 8(3): 263–​367. https://​doi.org/​10.1016/​0010-​0277(80)900 08-​6 Forster, Kenneth. 2010. Using a maze task to track lexical and sentence processing. The Mental Lexicon, 5(3): 347–​357. https://​doi.org/​10.1075/​ml.5.3.05for Forster, Kenneth and Elizabeth Bednall. 1976. Terminating and exhaustive search in lexical access. Memory & Cognition, 4(1): 53–​61. https://​doi.org/​10.3758/​BF03213255 Forster, Kenneth and Jonathan, Forster. 2003. DMDX: A Windows display program with millisecond accuracy. Behavior Research Methods, Instruments, & Computers, 35(1): 116–​124. https://​doi.org/​10.3758/​BF03195503

04

40 Donahoo Forster, Kenneth, Christine Guerrera and Lisa Elliot. 2009. The maze task: Measuring forced incremental sentence processing time. Behavior Research Methods, 41(1): 163–​171. https://​doi.org/​10.3758/​BRM.41.1.163 Forster, Kenneth and Ilmar Olbrei. 1973. Semantic heuristics and syntactic analysis. Cognition, 2(3): 319–​347. Forster, Kenneth and Leonie Ryder. 1971. Perceiving the structure and meaning of sentences. Journal of Verbal Learning and Verbal Behavior, 10(3): 285–​296. https://​ doi.org/​10.1016/​S0022-​5371(71)80056–​7 Foss, Donald and Shari Speer. 1991. Global and local context effects in sentence processing. In: Hoffman, Robert R. and David S. Palermo (eds.), Cognition and the Symbolic Processes: Applied and Ecological Perspectives. Hillsdale, NJ: Lawrence Erlbaum Associates. 115–​139. Grice, Paul. 1991. Studies in the way of words. Cambridge, MA: Harvard University  Press. Hess, David, Donald Foss and Patrick Carroll. 1995. Effects of global and local context on lexical processing during language comprehension. Journal of Experimental Psychology: 124(1):  62–​82. Hunt, Lamar, Stephen Politzer-​Ahles, Linzi Gibson, Utako Minai and Robert Fiorentino. 2013. Pragmatic inferences modulate N400 during sentence comprehension: Evidence from picture–​sentence verification. Neuroscience Letters, 534: 246–​251. https://​doi.org/​10.1016/​j.neulet.2012.11.044 Jakobson, Roman. 2002 Kindersprache, Aphasie und allgemeine Lautgesetze. In: Jakobson, Roman (ed.), Selected Writings: Phonological Studies (3rd edn). Berlin: Mouton De Gruyter. 328–​401. Johnson-​ Laird, Philip. 1995 Mental Models: Towards a Cognitive Science of Language, Inference, and Consciousness (6th edn). Cambridge, MA: Harvard University Press. Jordan, Timothy and Sharon Thomas. 2002. In search of perceptual influences of sentence context on word recognition. Journal of Experimental Psychology: Learning, Memory, and Cognition, 28(1): 34–​45. https://​doi.org/​10.1037//​0278-​7393.28.1.34 Kutas, Marta and Kara Federmeier. 2011. Thirty years and counting: Finding meaning in the N400 component of the event-​related brain potential (ERP). Annual Review of Psychology, 62: 621–​647. Marslen-​Wilson, William and Lorraine Tyler. 1980. The temporal structure of spoken language understanding. Cognition, 8(1):  1–​71. Mitchell, Donald. 1984. An evaluation of subject-​ paced reading tasks and other methods for investigating immediate processes in reading. In Kieras, David E. and Marcel A. Just (eds.), New Methods in Reading Comprehension Research. Hillsdale, NJ: Erlbaum. 69–​89. McCawley, James. 1971. Where do noun phrases come from? In Steinberg,  Danny D.  and Leon A.  Jakobovits  (eds.), Semantics. Cambridge: Cambridge University Press. 217–​231. McClelland, James and J. Kevin O’Regan. 1981. Expectations increase the benefit derived from parafoveal visual information in reading words aloud. Journal of Experimental Psychology: Human Perception and Performance, 7(3): 634–​644. https://​doi.org/​10.1037/​0096-​1523.7.3.634 Morris, Robin. 1994. Lexical and message-​level sentence context effects on fixation times in reading. Journal of Experimental Psychology: Learning, Memory, and Cognition, 20(1): 92–​103. Murray, Wayne. 1982. Sentence matching: The influence of meaning and structure (Doctoral dissertation). Monash University, Melbourne, Australia.

14

Can Cats Pick Up Chainsaws? 41 Murray, Wayne. 2006. The nature and time course of pragmatic plausibility effects. Journal of Psycholinguistic Research, 35(1): 79–​99. https://​doi.org/​10.1007/​ s10936-​005-​9005-​4 Parret, Hemann, 1983. L’énonciation en tant que déictisation et modalisation. Langages, 18:  83–​97. Ratcliff, Janet. 1987. The plausibility effect: Lexical priming or sentential processing? Memory & Cognition, 15(6): 482–​496. Rayner, Keith. 1998. Eye movements in reading and information processing: 20 years of research. Psychological Bulletin, 124(3): 372–​422. Sauerland, Uli. 2004. Scalar implicatures in complex sentences. Linguistics and Philosophy, 27(3): 367–​391. Sharkey, Amanda and Noel Sharkey. 1992. Weak contextual constraints in text and word priming. Journal of Memory and Language, 31(4): 543–​572. https://​doi.org/​ 10.1016/​0749-​596X(92)90028-​V Swinney, David. 1979. Lexical access during sentence comprehension: (Re)consideration of context effects. Journal of Verbal Learning and Verbal Behavior, 18(6): 645–​659. https://​doi.org/​10.1016/​S0022-​5371(79)90355-​4 Thornton, Robert and Maryellen MacDonald. 2003. Plausibility and grammatical agreement. Journal of Memory and Language, 48(4): 740–​759. Warren, Tessa, Kerry McConnell and Keith Rayner. 2008. Effects of context on eye movements when reading about possible and impossible events. Journal of Experimental Psychology: Learning, Memory, and Cognition, 34(4): 1001–​1010. https://​doi.org/​10.1037/​0278-​7393.34.4.1001 Witzel, Jeffrey and Kenneth Forster. 2014. Lexical co-​occurrence and ambiguity resolution. Language, Cognition and Neuroscience, 29(2): 158–​185. https://​doi.org/​ 10.1080/​01690965.2012.748925 Witzel, Naoko, Jeffrey Witzel and Kenneth Forster. 2012. Comparisons of online reading paradigms: Eye tracking, moving-​window, and maze. Journal of Psycholinguistic Research, 41(2): 105–​128. https://​doi.org/​10.1007/​s10936-​011-​9179-​x

24

3  Scalar Implicature through the Lens of Distributional Linguistics Maxime Codère Corbeil

1 Introduction1 Distributional approaches (DAs) to modelling meaning in use have their origins in the idea that the meaning of a word comes from its use (Firth 1957). These approaches are advantageous because they take some contextual clues into account while simultaneously lending themselves to computational implementations (Clark 2015). In such approaches, the meaning of a word is represented by a vector consisting of the relative occurrence of this word with respect to other surrounding words. Since vectors are well-​defined mathematical objects, one advantage of this way of representation is the straightforward assessment of the similarity of meaning between words. If two words have similar meanings, their vectors will also be similar, which is to say that these words occur at similar frequencies and around similar words in a given corpus. Distributional models were first developed to deal with words in isolation, but they rapidly expanded so they could represent not only words but whole sentences (Lenci 2018). However, measures of word similarity will not transfer easily to measures of sentence similarity, mainly because standard compositional DAs do not take into account implicated meaning. Throughout this chapter, I use the term sentence to stay true to the terminology used in DAs even though in pragmatics the correct way to refer to such linguistic expressions would be utterance. It should be noted that implicated meaning may play an important role in determining whether a sentence is more similar to another, which is demonstrated by the following examples where each sentence could be interpreted differently depending on the context: (1)

[A professor has a meeting with John and is waiting for him] [Professor:] John is punctual.

(2)

[A professor has to write a recommendation letter for John, who is applying for a faculty position in Linguistics] [Professor:] John is punctual.

In (1), the sentence John is punctual means that the professor expects him to be on time, but in (2), the same utterance may also convey that John is not good at linguistics. The two sentences have the same intensional meaning in that

34

Scalar Implicature 43 they both state that John is punctual, but in (2) there is an additional aspect being conveyed implicitly, namely that the professor may think that John is not suitable for the faculty position. In this latter case, this implicated aspect of meaning may even be considered to be the main point of this utterance. Following the Gricean framework, the meaning that I assume to be explicitly expressed in (1) and (2) corresponds to the What is said component of meaning while the meaning implicated in (2)  corresponds to the What is implicated component of meaning (Borg 2009; Carston 2002).2 If we have to compare these two utterances in terms of similarity, i.e. how similar (1)  and (2)  are, then we would probably have to differentiate the contribution from the two components of meaning because (1) and (2) are very similar with respect to their explicit meaning, but dissimilar in terms of their implicated meaning. The main objective of this chapter is to bridge the gap between the treatment of implicated meaning in pragmatics and the similarity measures computed by DAs, which have actually been unable to take into account implicated levels of meaning. I propose the novel idea that it is possible to integrate implicated meaning into similarity measures using a vector representation. I present the main features of DAs to meaning in Section 2 and introduce the notion of sentence-​similarity in Section 3.  After discussing how to take into account implicated meaning in Section 4, I  then look more closely at the two most prominent views about the derivation of scalar implicature (SI; cf. Chierchia 2012; Geurts 2010) and show how to integrate the SIs within DAs in Section 6. Finally, in Section 7, I discuss how DAs could be used in conjunction with pragmatic theories to improve our understanding of both fields.

2  Distributional Approaches to Meaning DAs follow from Harris’s Distributional Hypothesis (1954), which states that words occurring in similar settings will express similar meanings. I  use the notion of distributional approaches here to describe any kind of method concerned with finding and representing the meaning of words from their distributions within a given corpus. DAs offer “both a model to represent meaning and computational methods to learn such representations from language data” (Lenci 2018: 152). The main advantage of such approaches is to be able to represent a meaning in terms of a well-​defined mathematical entity called a vector, which means that different meaning-​vectors could, in turn, be combined or compared easily. 2.1  Word-​vectors To retrieve the meaning of a given word, one has to consider the surrounding words within a given contextual window. The size of this window can be small (considering the preceding and succeeding words only) or it can be huge (taking into account the whole text around the target word), depending on the task one wishes to accomplish (Kiela & Clark 2014).3 The particular meaning of a word is defined via the creation of a co-​occurrence matrix representing

4

44  Codère Corbeil

candies

Sugar

Children

Sweet

Dessert

2

3

3

1

Figure 3.1 Example of a word-​vector for the word candies words as rows and contextual elements as columns (Li et al. 2004). This matrix is derived by counting how many times a target word occurs in a particular (con)text. Co-​occurrence in DAs is related to the notion of collocation which also dates back to Firth (1957) and recent classifications of collocations have been based on DAs (Garcia et al. 2019; Wanner et al. 2017). For example, if we pick the word candies as the target, we can build a row matrix that represents the number of times another word is co-​occurring with it in all contexts like what is illustrated in Figure 3.1. This row-​matrix or vector thus represents the meaning of the word candies in this corpus. 2.2  Sentence-​vectors Once we have represented the meaning of a word as a vector, the next logical step is to combine different word-​ vectors together to build a representation of the meaning of the whole sentence (Clark 2015). The Principle of Compositionality tells us that the meaning of a sentence is a function of its parts and the way they are combined (Pelletier 1994), and with this in mind we can write the representation of a sentence as a function of different word-​ vectors. Various approaches to combining word-​vectors into sentence-​vectors have been proposed (Baroni et al. 2014; Clark et al. 2008; Clarke 2012), but I choose here to adhere to the multiplicative model from Mitchell and Lapata (2010). In their paper, they compared a number of compositional models, arriving at the conclusion that “simple multiplication” is better suited to sentence similarity tasks than the other alternatives they tested. Their model uses simple multiplication as the compositional function, which means that a sentence-​vector will be derived from the multiplication of all the word-​vectors composing a sentence. One limitation of this model is that it does not take into account word ordering as multiplication is commutative, i.e. the composed vector for grey dog will be the same as the one for dog grey. Although other more sophisticated models that can integrate syntactic constraints have been proposed (Coecke et al. 2010), I employ Mitchell and Lapata’s (2010) simple multiplicative model for the present study, because of its simplicity.4 2.3  Measuring Similarity with Vectors Once word-​vectors (or sentence-​vectors) have been obtained, we can compare them to each other and measure their levels of similarity. I follow Landauer and Dumais (1997) as defining word-​meaning similarity as the distance between the vectors representing those words. The method generally used to measure

54

Scalar Implicature 45

tea

coffee 40°

Figure 3.2 Vector representations for tea and coffee

the distance between two vectors is the cosine similarity measure which is equivalent to the angle between the two vectors (Baroni et al. 2014). The more similar two vectors are, the smaller the angle between them will be. The cosine measure yields a real number from -​1 to 1, with 1 indicating perfect similarity, which is visualized by a 0° angle between the vectors. The possible range for the cosine measure will actually depend on the normalization used for the word-​vectors. In other words, if the word-​vectors are represented only within the first quadrant, i.e. if they all have positive components, then the angle between two vectors will not be greater than 90°, which means that the cosine measure will be between 0 and 1. For example, Figure 3.2 displays hypothetical word-​vectors for the words tea and coffee represented in a 2D-​plane. Using the cosine measure, we can compute the similarity between those two word-​vectors. In this fictional case, we obtain that the two word-​vectors have a similarity measure equal to the cosine of 40°, i.e. 0.77, which is quite high. The cosine measure was originally used to compute similarity between words, but it can also be applied to compute similarity between sentences. However, as we will see in the next section, this transition from words to sentences is not that straightforward in practice.

3  From Word-​Vectors to Sentence-​Vectors When measuring the similarity of word-​ vectors, we are comparing their respective vector representations, which are derived from their distribution of use in a corpus. At the word-​level, the correlation between distribution and meaning is supported by Harris’s Distribution Hypothesis (1954): the word-​vector representing the meaning of car will correspond to the meaning expressed by this word in different contexts of use. However, this correspondence does not transpose easily to sentences because unlike single words, the meaning conveyed by a sentence has two different components: the explicit meaning and the implicated meaning. Following the Gricean Framework (Grice 1957), these two components should be written down as the ‘What is said’ part and the ‘What is implicated’ part.

64

46  Codère Corbeil The explicit meaning is the meaning derived at the syntactico-​semantic level by the composition of all the words present in the sentence and it corresponds to the semantic meaning expressed by elements of a language and their combinations (De Swart 1998). On the other hand, the implicated meaning of a sentence is not explicitly expressed by the sentence, but it could still be derived from it at the pragmatic level. While this duality is not present at the word level we certainly have to take this interlinkage into account at the sentence level in order to be able to measure the full similarity between sentences, i.e. both the similarity at the explicit and the similarity at the implicated level.5 However, sentence-​vectors are still most commonly derived by composing word-​vectors together, which allows them to be able to represent only the explicit component of the meaning because words themselves are bearers of explicit meaning only. In turn, this means that the sentence similarity measures currently at our disposal are not exhaustive. A  lack of consideration for implicated meaning does not stem from the compositional function itself, but from the nature of the DA, i.e. no matter how one combines words in order to form sentences, one will be limited to capturing only the explicit component of meaning, unless one works with corpus material specifically tagged for implicated meanings. Not having a representation for the implicated meaning of a sentence may not be an issue if the objective of this similarity measure is to compare only the explicit meaning of two sentences, or if said sentences do not present an implicated meaning in the given context, but it could have great consequences when considering two sentences with similar explicit meanings, but different implicated meanings. Even if the integration of implicated meaning within the sentence-​vector has never been discussed in DAs, other areas have emphasized the difference between explicit and implicated meaning when it comes to comparing sentences. For example, what is described as similarity of sentences in DAs has been described for years in translation studies as “equivalence of sentences” (Koller 1989, 1995). So to better understand this concept of equivalence, Nida and Taber (1982) proposed two types of equivalences: formal equivalence and dynamic equivalence. Formal equivalence pertains to the translation of form and content from one language into another, whereas dynamic equivalence pertains to finding the “closest natural equivalent” (Nida & Taber 1982: 12) of the source message in the target language, which essentially means that the translation should trigger the same response from the reader as the source message. Thus, formal equivalence is about the content expressed by a sentence, i.e. its explicit meaning, while the dynamic equivalence focuses on the effect of a sentence, i.e. its implicated meaning. To illustrate this difference, I  borrow an example from Ervas (2014) concerned with the translation of Friday the 13th from English into Spanish. (3)

John stays home on Friday the 13th.

(4)

John se queda en casa el viernes 13. John stays home on Friday the 13th.

74

Scalar Implicature 47 (5)

John se queda en casa el martes 13. John stays home on Tuesday the 13th.

Here, Sentence (4) is the word-​for-​word translation of (3): both the lexical items and the syntactical make-​up of the sentences are the same. If we consider only the explicit meaning of the two sentences, we can give them a similarity value of 1, i.e. perfect similarity. However, when taking into account the implicated meaning, we arrive at a different result. From (3), we might conventionally infer that the reason John stays home on this date is due to superstition based on his culturally shared knowledge that this date is supposed to be unlucky. However, we could not derive this information from (4), as in most Spanish-​speaking countries the usually unlucky day will be Tuesday the 13th (as in (5)). On the other hand, comparing (3) and (5) looking only at the implicated meaning, we could claim that they have a similarity of 1 because they both give rise to the same implicated meaning that John is superstitious, while (4) does not. This example really shows that for a similarity measure to be really complete, it should encompass both the explicit meaning and the implicated meaning of sentences. Sentence similarity measures are now used in a number of applications like short answer tasks (Koleva et al. 2014), automatic summarization (Erkan & Radev 2004) and machine translation (Liu & Zong 2004), and it is therefore very important to be able to take into account implicated meaning as this may lead to very different results.

4  Pushing Forward: Representing Implicated Meaning Currently, traditional sentence-​ vectors in DAs consider only the explicit meaning, but for sentence similarity measures to be truly comprehensive we need to be able to compare ‘full’ sentence-​vectors; ‘full’ in the sense that the sentence-​vectors represent both the explicit and the implicated meaning. In order to resolve this issue, I propose that we first need to find a way to represent implicated meaning in the form of a sentence-​vector. The most straightforward way to represent the implicated meaning of a sentence while still continuing to use the same mechanics as in traditional DAs would be to translate the implicated meaning into explicit wording. Writing down the implicated meaning explicitly essentially boils down to attributing a sentence-​vector to another sentence and labelling it as the representation of the implicated meaning of this other sentence. According to Sperber and Wilson (1995), implicated meaning is propositional in nature, which is why there is arguably no difference between representing explicit meaning or implicated meaning using a vector. This means that in both cases the sentence-​vector would represent the meaning of a proposition and that they could both be derived by combining word-​vectors. This innovative way of integrating the implicated meaning within a vector representation has thus the advantage of using the same compositional principles of standard DAs while being able to build two different sentence-​vectors, one for the explicit meaning and one for

84

48  Codère Corbeil the implicated meaning. To illustrate this, let me reuse the example of John being punctual: (6)  [A professor has to write a recommendation letter for John, who is applying for a faculty position in Linguistics] [Professor:] John is punctual. a) Explicit meaning: John is punctual. b) Implicated meaning: John is not good at Linguistics. In (6a), the sentence-​vector would be derived from the denotative composition of the word-​vectors for John, is and punctual, and this sentence-​vector would represent the explicit meaning of the sentence John is punctual. In (6b), on the other hand, the sentence-​vector would be derived as the composition of the word-​vectors for John, is, not, good, at and Linguistics, and this sentence-​vector would represent the implicated meaning of the sentence John is punctual. So here because we have two kinds of meanings attached to the sentence John is punctual we need to also have two corresponding sentence-​ vectors: one relates to its explicit meaning and the other to its implicated meaning. If we combine these sentence-​vectors, we can get the sentence-​vector representing the full meaning of John is punctual. I  call this sentence-​vector the full meaning sentence-​vector because it contains both the explicit and the implicated meaning of the sentence John is punctual. To illustrate this idea of full meaning, I use a plus sign in (7a) and (7b) to show the need to combine both components in order to obtain the full meaning of John is punctual. In other words, the plus sign represents the composition function needed to form the full sentence-​vector from the explicit and the implicated sentence-​vectors. (7)

[A professor has to write a recommendation letter for John, who is applying for a faculty position in Linguistics] Full meaning of John is punctual: a) Explicit meaning + Implicated meaning b) John is punctual + John is not good at Linguistics

Constructing the full meaning sentence-​vector like this, i.e. by combining the explicit sentence-​vector with the implicated sentence-​vector that were both derived from word-​vectors, has the advantage of using the same mechanism to derive the full meaning as was done with only the explicit meaning. Since both are formed from word-​vectors, they can be directly compared to each other using the cosine similarity measure. In other words, the word-​vector for John would be the same for both sentence-​vectors even if one were to be part of the explicit meaning and the other one of the implicated meaning components. From a vectorial point of view, the distinction between explicit and implicated meaning does not make any difference as the meaning of a full sentence will still be derived from the meaning of its individual parts. Being able to derive two different sentence-​vectors and to combine them is a step forward towards our objective of being able to measure sentence

94

Scalar Implicature 49 similarity for the implicated meaning of a sentence. However, despite this being, we still need to define what this plus sign is in (7a,b), i.e. what kind of composition function should we use to combine the explicit and the implicated meaning. In the next section, I describe how to apply this proposed method to cases involving scalar implicature so that we are able to measure similarity for full sentence-​vectors.

5  Measuring Full Sentence Similarity for Scalar Implicatures As I have shown, it is possible to represent both the explicit and the implicated meaning using sentence-​vectors, but we also need to find a way to combine them to derive a full sentence-​vector. In this chapter, I am proposing a way out of carrying out this integration within a full sentence-​vector for a well-​known kind of implicated meaning, namely the scalar implicature. 5.1  Deriving a Scalar Implicature While scalar implicature (SI) has been the focus of numerous studies in recent years (e.g. Chierchia 2004; Fox 2007; Geurts 2010), it has so far not been discussed from the perspective of DAs, which is the frame of reference of the present study. SI represents a special kind of quantity implicature, involving a series of alternatives that can be generated by substituting lexical items that are linked on a conventionalized scale (Geurts 2010). Horn defined conventionalized scales as containing different lexical items organized by informativity (Horn 1972). For example, (10) is derived from (8), using the scale in (9), where all is a stronger alternative than some: (8)

John ate some candies.

(9)

{some, all}

(10) John ate some and not all candies. In this study, I focus solely on the opposition between what is called the ‘globalist’ (or post-​compositional) and the ‘localist’ (or compositional) view (Geurts 2009; Geurts & van Tiel 2013). At this point, it is important to note I only consider cases where the scalar implicatures are in fact derived and will not discuss cases where they are not. In other words, I am only interested in finding a way to represent SIs using DAs, and I will not discuss questions of optionality, i.e. whether SIs are derived in a particular context or not. For a discussion about the optionality of scalar implicatures see Fox (2007) or Geurts (2010). 5.1.1  Deriving an SI: The Globalist View The globalist view states that SIs are derived using pragmatic mechanisms (Geurts 2009). To understand the standard derivation procedure as suggested

05

50  Codère Corbeil by Geurts (2010), suppose that Mike utters (11) and not (12), then the SI would take the form shown in (13). (11) John ate some candies. (12) John ate all the candies. (13) John ate some and not all candies. From these, we can symbolically transcribe the derivation procedure like this: (i) (ii) (iii) (iv)

Mike utters (11) instead of (12) He must not believe that (12) is true: ¬BM(12) (read: M does not believe (12)) Mike is likely to know whether (12) is true: BM (12) ∨ BM ¬(12) By combining (ii) and (iii) we get that Mike must believe that (12) is not true: (¬BM(12) ∧ (BM (12) ∨ BM ¬(12))) → BM ¬(12) which gives rise to (13).

As shown, the hearer first derives the weaker implicature (¬BM(12)) based on the fact that the speaker chose not to utter the stronger alternative in (12). The transition from the weaker implicature to the stronger, i.e. from (¬BM(12) to BM ¬(12) is made possible through the Competence Assumption wherein a competent speaker either believes p or ¬p (Geurts 2010). It is thus the combination of the Competence Assumption and the personal choice made by Mike to utter the weaker alternative that leads to the derivation of an SI. Following this view, the SIs are derived using global processes like contextual information and the Competence Assumption rather than local grammatical processes. This view also supports the idea that SIs are never derived by default; rather, they are derived in particular contexts by post-​compositional global processes, which means that the derivation of SIs should not play a role in the compositional process.6 5.1.2  Deriving an SI: The Localist View In contrast to the globalist view, the localist view considers that SIs are derived from specific lexical items. Additionally, and most importantly, SIs are derived compositionally, which means they are derived semantically and not pragmatically (Chierchia et al. 2011). The basic assumption here is that the implicature arises from the presence of an exhaustivity operator exh (Fox & Spector 2018). This operator, exh, acts to negate the stronger alternatives of a sentence S and combine this negation with the sentence itself. It may be expressed in some cases as a silent covert only (see (15) or simply be written as O (Dieuleveut et al. 2019). Although this operator is inspired by Gricean reasoning, it must be integrated within the sentence in order for it to be able to play a role

15

Scalar Implicature 51 locally when in an embedded position (Chierchia 2012; Chierchia et al. 2011; Fox 2007).7 Going back to the example, according to the localist view the SI –​ going from (14) to (16) –​is derived grammatically from the presence of the operator exh (Fox & Spector 2018): (14) John ate some candies. (15) John ate exh some candies. (John ate only some candies) (16) John ate some and not all candies. (i) (ii)

The set of alternatives of (14) is based on the Horn scale ALT((14)) = {some, all} The logical form of (14) is then added to the negation of all stronger alternatives of (14): exhALT (14) = some ∧ ¬all

This operator exh acts locally, and because it can be inserted at different positions within a sentence, it means it would be possible to retrieve different implicatures depending on where the operator is processed (Chierchia et  al. 2011). While this variability actually makes this approach not strictly local because we could still modify the position in which we apply exh, I still consider it to be local in the sense that the operator acts locally, whereas pragmatic processes always act globally. 5.2  Representing the Full Meaning of a Sentence With the two views concerning the issue of SIs in mind, we can now return to our main objective, which is to derive the full sentence-​vector for sentences involving SIs. As discussed in Section 4, we can write down the implicated meaning explicitly using word-​vectors and then combine them to derive the sentence-​vector for the implicated meaning. Once the explicit meaning and the implicated meaning sentence-​vectors are derived, the next step is to combine them to compose the full sentence-​vector. It is important to note that the way this composition occurs will depend on the derivation procedure of the implicated meaning, i.e. a different derivation procedure could lead to a different composition function between the explicit and the implicated sentence-​vectors. Specifically, in the case of SIs, the globalist and the localist views use different mechanisms to derive the SIs, and this has consequences for the way the composition of the full sentence-​vector is performed. I  will describe the relevant differences in the following sections. 5.2.1  Composing the Full Sentence-​Vector in the Globalist View In the globalist view, the SI is derived using global processes that are taking place above the grammatical level. This means that the sentence John ate some candies is composed and interpreted separately from the implicated meaning

25

52  Codère Corbeil John ate not all the candies. The globalist view is post-​compositional because the composition of the explicit-​meaning sentence does not influence the composition of the implicated-​meaning sentence, i.e. the implicated meaning is not part of the composition of the explicit meaning. Once we have acknowledged that the implicated meaning is derived separately from the explicit meaning, we can write it down explicitly as a second sentence-​vector: (17) John ate some candies   (18) Explicit meaning =  John ate some candies   (19) Implicated meaning =  John ate not all candies Next, we have to combine the two vectors to form the full sentence-​vector. The fact that the two sentence-​vectors are compositionally independent allows for a simple composition of the two sentence-​vectors much like when two word-​ vectors are combined: the explicit meaning sentence-​ vector and the implicated meaning sentence-​vector are first composed separately from their respective word-​vectors. As shown in (20a), the explicit meaning sentence-​ vector is derived from the composition of the word-​vectors for John, ate, some and candies and the implicated meaning sentence-​vector is derived from the composition of the word-​vectors for John, ate, not, all and candies. Once the two sentence-​vectors have been composed, the full meaning sentence-​vector is derived by composing the two together using the same composition function as the one that was used to form both sentence-​vectors in the first place as in (22b). In other words, the word-​vectors are composed together using a particular composition function to form the two sentence-​vectors, and then these two sentence-​vectors are in turn combined with the same composition function. As discussed before, if we use Mitchell and Lapata’s (2010) simple multiplication function as the composition function, the full sentence-​vector is equal to the multiplication of the two sentence-​vectors. (20)  (a) Full meaning

=

(b)

=

  Explicit meaning ⋅ Implicated meaning   John ate some candies ⋅ John ate not all candies

5.2.2  Composing the Full Sentence-​Vector in the Localist View In the localist view, the procedure to derive the full sentence-​vector is slightly different. Since the derivation is done compositionally, the SI is derived during the composition of the sentence. This means the full sentence-​vector is now directly composed from the word-​vectors for John, ate, some, and, not, all and candies.   (21) Full meaning = John ate some and not all candies

35

Scalar Implicature 53 In the localist view, the contribution from the implicated meaning is linked to the presence of and not all, which is grammatically integrated within the full sentence-​vector. The fact that the SI is derived compositionally makes it impossible to detach (or extract) the implicated meaning from the full sentence-​ vector (as was the case with the globalist view). This is not a problem per se for measuring the full sentence similarity, but it certainly poses a limitation, if one seeks to measure specifically the contribution of the implicated meaning within a full sentence-​vector. I should also mention that many authors write down the explicit form of the SI using but not all instead of and not all (Fox & Spector 2018; Geurts & van Tiel 2013), but since I am specifically interested in comparative measures of full sentence similarity, a differentiation between but and and will not have much impact on this relative similarity value. 5.2.3  Comparing the Two Views about the Derivation of SIs The principal reason for the difference between full sentence-​vectors in both views is that the globalist view is post-​compositional whereas the localist view is compositional. The compositional derivation process of the localist view makes the contribution from the SI directly integrated within the full sentence-​ vector, whereas the post-​compositional process of the globalist view allows for the two separate sentence-​vectors to be subsumed. This shows that the derivation procedure of the implicated meaning does have an influence on its integration within DAs. In the long run, if our objective is to take into account different kinds of implicated meaning, while measuring sentence similarity, we need not only to work out how to derive the implicated meaning sentence-​ vector and integrate it within the full sentence-​vector, but also we must be aware that the structure of approaches used to account for implicated meaning will influence how this integration is performed. In the case of SIs, the two approaches I  presented here result in different full sentence-​vectors even for a sentence as seemingly simple as John ate some candies. This is surprising considering the fact that the point of contention between the two approaches is embedded cases like Mike believes that John ate some candies, and it was thought that sentences like John ate some candies should be treated the same way by the two views. I do not wish to reopen the whole debate between the localist and globalist views, but it must be mentioned that when we take the perspective of DAs, the differences in the treatment of SI between the localist and the globalist views have a significant influence on how full sentence-​vectors are derived, and this is true even for non-​embedded cases. 5.3  Measuring Sentence Similarity for Full Meaning Involving SIs Putting all of this together, we are finally able to derive the full sentence-​vector for John ate some candies. The only thing left to do is to substitute the word-​ vectors and to compute the sentence-​vector using the multiplication compositional function of Mitchell and Lapata (2010). For this calculation, I  used pre-​trained word-​vectors from a corpus based on Wikipedia English entries

45

54  Codère Corbeil containing more than 9  billion words (Mikolov et  al. 2017). The method proposed in Section 5.2 now allows us to take into account the full meaning, when measuring sentence similarity. For example, we could compare a sentence that involves an SI, e.g. (23), with one that does not, e.g. (22). (22) John ate candies. (23) John ate some candies. If we use the full sentence-​vectors given in Sections 5.2.1 and 5.2.2 for the localist and the globalist views, we can measure the full sentence similarity and compare it to the traditional sentence similarity measures involving only the explicit meaning (Corbeil 2018). The values for the similarity measures between (22) and (23) presented in Table 3.1 clearly show that there is a significant difference between adhering to the full meaning versus exclusively relying on the explicit meaning as both full meaning similarities are lower than the explicit only meaning. When using just the latter, the similarity measure is very high (i.e. at 0.9961) but this should not be surprising: after all, it is but one word that causes a difference between (22) and (23). For both full meanings, the lower values for the similarity measures are due to the fact that (22) does not have an implicated meaning, whereas (23) does; it is this very difference that caused the similarity measure between the two to decrease. Put differently, the more the implicated meaning contributes to the full meaning, the lower the similarity between the explicit meaning and the full meaning sentence-​ vectors will be because adding an implicated component will automatically lower the contribution from the explicit component of meaning to the full meaning sentence-​vector. This relationship still holds if we take sentences of the form John ate some of the candies instead. Adding of the or any other modification in phrasing, might change the relative difference between similarity values slightly, but it would not modify the procedure we used to form full sentence-​vectors. In Table  3.1, we also notice that the localist view has a higher similarity value (0.9343) than the globalist view (0.7759). This relative difference is due to the different compositions for the full sentence-​vectors. In the globalist view, the explicit meaning is combined with the implicated meaning to form a longer (in terms of number of words) full sentence-​vector compared with the one

Table 3.1 Cosine similarity measures between John ate candies and John ate some candies Explicit meaning only

0.9961

Complete meaning Localist view

Globalist view

0.9343

0.7759

5

Scalar Implicature 55 from the localist view, i.e. respectively John ate some candies + John are not all candies versus John ate some and not all candies.

6  Discussion In this chapter I have focused on SI cases involving some but we could easily have extended our scope to include other cases of implicated meaning. For example, the procedure I described for the derivation of the full sentence-​vector would also apply to cases like John ate three candies, where the implicated meaning would be that John ate no more than three candies, or even for cases involving a free choice inference (Fox 2007; Geurts 2010), i.e. cases where John or Mary ate candies leads to the derivation of John or Mary and not both ate candies. The possibility of extending our procedure to these new cases comes from the fact that the treatment of the SI of the form no more than and the free-​ choice inference are accounted for by both the localist and the globalist views (Corbeil 2018). Incidentally, the derivation of the full sentence-​vector for these cases is very similar to the one we have just done (cf. Section 5 above). In fact, we could extend this derivation to other cases as long as we have a theory to support these cases, i.e. a theory able to account for the derivation of the implicated meaning for these particular cases. For instance, we would not be able to use the localist view for our example about the translation of Friday the 13th from English to Spanish because the derivation of the implicature that John is superstitious is not accounted for by the localist view. But the good news is we may still compose the explicit and the implicated meaning together using the globalist view even if it was not intended to deal with implicated meaning like this in the first place. In fact, the structure of the globalist view allows for a smooth integration of any kind of implicated meaning that is derived post-​ compositionally. Additionally, in the globalist view, the full sentence-​vector is defeasible which is an interesting and important feature to have in DAs as per Asher et al.(2016). In other words, we could still use the multiplication function to combine the explicit and the implicated sentence-​vectors: (24)    Full meaning = John stays home on Friday the 13th ⋅ John is superstitious From there, if we compute the full sentence similarity between the two Spanish translations and the original sentence in English we would find that the best overall translation is the one with martes 13, i.e. Tuesday the 13th, which has a similarity score of 0.9997 rather than the one with viernes 13, i.e. Friday the 13th, which has a similarity score of 0.8843. The sentence similarity measures are obtained by comparing the full sentence-​vectors for the two translations to the one from the original sentence in English. It was computed using the English counterpart of all the words and their respective word-​ vectors. This last computation shows that using our derivation for the full

65

56  Codère Corbeil sentence-​vector we would be able to tackle many different kinds of implicated meaning which would be a great improvement for DAs.

7  Distributional Approaches and Pragmatics In this chapter the derivation of sentence-​vectors was improved by integrating two views about the treatment of SIs. In turn, this integration within DAs allowed me to reveal differences between these implicature models. Using DAs in conjunction with theoretical or formal approaches to meaning may provide some new insights into the latter approaches. With this in mind, DAs could be used as a tool in pragmatics (and in other fields interested in the study of meaning). Another advantage of DAs is to provide computational capabilities to the study of sentence-​meaning by making it possible to perform a numerical analysis of sentence-​similarity or to compute how two word-​vectors could be combined. For example, Asher et al. (2016) have already discussed a possible integration of DAs within their logical-​type model. Throughout this chapter, I  deliberately omitted discussing optionality, i.e. the issue of when implicated meaning is derived and why, because this issue is not the focus of DAs the main purpose of which is to model meaning in from words and their distributions within the corpus (Lenci 2018). To resolve the optionality issue, we would have to find a way to identify implicated meaning in the corpus directly. The more recent DA models have changed the way the word-​vectors are extracted in the first place by using full sentence representations instead of word-​vectors (Devlin et al. 2018), and it might not be very long before they are able to recognize implicated meaning, even if that were not explicitly tagged or otherwise indicated in a corpus. In summary, using pragmatic models in conjunction with DAs can be beneficial in two ways: data gathered from DAs could help improve theoretical approaches and these could in turn provide theoretical insights into how we should improve DA to tackle more efficiently issues relevant in pragmatics. In other words, DAs would profit as much from pragmatics as pragmatics would from DAs. In this chapter, I have emphasized the importance of the implicated meaning for full sentence similarity measures and I  have presented a way of taking into account the implicated meaning in DAs. This new improvement makes a significant difference for cases where the implicated meaning of a sentence is relevant in a given context. It was a first step towards bridging the gap between DAs and the theoretical approaches of implicated meaning. I also looked at two views regarding how the derivation of SIs takes place, the localist and the globalist, and I showed that those two views give rise to different sentence-​vectors even for non-​embedded cases. I then used the representational capacity of these DAs to derive full sentence-​vectors and compute sentence similarity measures for the two views. Finally, I  discussed how my results may be extended to include other kinds of implicated meaning and I stressed the importance of the interdependency and the mutual benefits pragmatics would have in continuing

75

Scalar Implicature 57 to work in conjunction with DAs. This really is a new direction and we have yet to discover the full advantages of combining these two approaches.

Notes 1 I gratefully acknowledge the support from the Social Sciences and Humanities Research Council of Canada (SSHRC), and I  would like to thank Elizabeth Allyn Smith for her support during the development of this project. 2 This separation between explicit and implicated meaning is not related to the distinction present in DAs between explicit and dense representations where explicit representations are described as high-​dimensional sparse vectors (Levy & Goldberg 2014). 3 In DAs the context refers to the linguistic context surrounding a given word (Lenci 2018)  while in pragmatics the context may also be extralinguistic as it usually consists of a sum of assumptions or information that could be used to retrieve the meaning of an utterance (Carston 2002). 4 See Corbeil (2018) for a similar approach using Coecke et al. (2010) compositional model. 5 I consider here that a word in itself is not a bearer of implicated meaning, and that the implicated meaning arises when different words are composed together or if a given word is inscribed in a specific context. 6 Geurts (2010, 2013) acknowledges the fact that some kind of local pragmatics must be invoked when considering particular examples such as embedded implicatures, but such structures are outside the scope of this chapter. 7 This view originates from the treatment of embedded implicatures, but also extends to all scalar implicatures (Fox 2007). For more information about the motivations behind the localist view see Chierchia (2004), Chierchia et al. (2011) and Fox (2007).

References Asher, Nicholas, Tim Van de Cruys, Antoine Bride and Marta Abrusán. 2016. Integrating type theory and distributional semantics: A case study on adjective-​noun compositions. Computational Linguistics, 42(4): 703–​725. Baroni, Marco, Raffaela Bernardi and Roberto Zamparelli. 2014. Frege in space: A program of compositional distributional semantics. Linguistic Issues in Language Technology, 9(6): 242–​346. Borg, Emma. 2009. On three theories of implicature: Default Theory, Relevance Theory and Minimalism. International Review of Pragmatics, 1(1):  63–​83. Carston, Robyn. 2002. Thoughts and utterances. In: Carston, Robyn (ed.), Thoughts and Utterances: The Pragmatics of Explicit Communication. Malden, MA: Blackwell Publishing. Chierchia, Gennaro. 2004. Scalar implicatures, polarity phenomena, and the syntax/​ pragmatics interface. In: Belleti, A. (ed.), Structures and Beyond: The Cartography of Syntactic Structures, Vol. 3. Oxford: Oxford University Press. 39–​103. Chierchia, Gennaro. 2012. Scalar implicature as a grammatical phenomenon. In: von Heusinger, Klaus, Claudia Maienborn and Paul Portner (eds.), Semantics. Berlin/​ Boston: de Gruyter. 2297–​2331. Chierchia, Gennaro, Danny Fox and Benjamin Spector. 2011. The grammatical view of scalar implicatures and the relationship between semantics and pragmatics. In:

85

58  Codère Corbeil Portner, Paul, Claudia Maienborn and Klaus von Heusinger (eds.), Handbook of Semantics, Berlin: Mouton de Gruyter. 1–​43. Clark, Stephen. 2015. Vector space models of lexical meaning. In: Lappin, Shalom and Chris Fox (eds.), The Handbook of Contemporary Semantic Theory. Chichester: John Wiley & Sons, Ltd. 493–​522. Clark, Stephen, Coecke, Bob and Sadrzadeh, Mehrnoosh. 2008. A compositional distributional model of meaning. Proceedings of the Second Quantum Interaction Symposium (QI-​2008). College Publications, UK), 133–​140. Clarke, Daoud. 2012. Challenges for distributional compositional semantics. ArXiv:1207.2265. Coecke, Bob, Mehrnoosh Sadrzadeh and Stephen Clark. 2010. Mathematical foundations for a compositional distributional model of meaning. Linguistic Analysis, 36(345): 345–​384. Corbeil, Maxime. 2018. Representing Scalar Implicatures in Distributional Semantics. In: Sikos, Jennifer (ed.), Proceedings of the ESSLLI 2018 Student Session: 30th European Summer School in Logic, Language & Information. Sofia, Bulgaria. 209–​220. De Swart, Henriette. 1998. Introduction to Natural Language Semantics. Center for the Study of Language and Information. Chicago: University of Chicago Press. Devlin, Jacob, Ming-​Wei Chang, Kenton Lee and Kristina Toutanova. 2018. BERT: Pre-​ training of deep bidirectional transformers for language understanding. ArXiv:1810.04805. Dieuleveut, Anouk, Emmanuel Chemla and Benjamin Spector. 2019. Distinctions between primary and secondary scalar implicatures. Journal of Memory and Language, 106: 150–​171. Erkan, Güneş and Dragomir R. Radev. 2004. LexRank: Graph-​based lexical centrality as salience in text summarization. Journal of Artificial Intelligence Research, 22: 457–​479. Ervas, Francesca. 2014. On semantic and pragmatic equivalence in translation. In: Pasa, Barbara and Lucía Morra (eds.), Translating the DCFR and Drafting the CESL: A Pragmatic Perspective. Munich: Sellier European Law Publishers. 87–​101. Firth, John R. 1957. A synopsis of linguistic theory, 1930–​1955. In: Studies in Linguistic Analysis. 1–​32. Reprinted in: Palmer, Frank R. (ed.). 1968. Selected Papers of J. R. Firth 1952–​59. London: Longmans. 168–​205. Fox, Danny. 2007. Free choice and the theory of scalar implicatures. In: Sauerland, Uli and Penka Stateva (eds.), Presupposition and Implicature in Compositional Semantics. London: Palgrave Macmillan. 71–​120. Fox, Danny and Benjamin Spector. 2018. Economy and embedded exhaustification. Natural Language Semantics, 26(1):  1–​50. Garcia, Marcos, Marcos García-​Salido and Margarita Alonso-​Ramos. 2019. Towards the automatic construction of a multilingual dictionary of collocations using distributional semantics. Proceedings of Electronic Lexicography in the 21st Century Conference, 2019. 747–​762. Geurts, Bart. 2009. Scalar implicature and local pragmatics. Mind and Language, 24(1):  51–​79. Geurts, Bart. 2010. Quantity Implicatures. Cambridge: Cambridge University Press. Geurts, Bart and Bob van Tiel. 2013. Embedded scalars. Semantics and Pragmatics, 6(9):  1–​37.

95

Scalar Implicature 59 Grice, Paul H. 1957. Meaning. The Philosophical Review, 66(3): 377–​388. Harris, Zellig. 1954. Distributional structure. Word, 10(23): 146–​162. Horn, Laurence. 1972. On the semantic properties of logical operators in English. (Doctoral Dissertation, UCLA, Los Angeles). https://​linguistics.ucla.edu/​images/​ stories/​Horn.1972.pdf Horn, Laurence. 1995. Presupposition and implicature. In: Lappin, Shalom (ed.), The Handbook of Contemporary Semantic Theory. Oxford: Blackwell. 299–​319. Kiela, Douwe and Stephen Clark. 2014. A systematic study of semantic vector space model parameters. Proceedings of the 2nd Workshop on Continuous Vector Space Models and Their Compositionality, 353. Gottenburg, Sweden: Association for Computational Linguistics. 21–​30. Koleva, Nikolina, Andrea Horbach, Alexis Palmer and Simon Ostermann. 2014. Paraphrase detection for short answer scoring. NEALT Proceedings Series, 22. Uppsala, Sweden: LiU Electronic Press. 59–​73. Koller, Werner. 1989. Equivalence in translation theory. In: Chesterman, Andrew (ed.), Readings in Translation Theory. Helsinki: Finn Lectura. 99–​104. Koller, Werner. 1995. The concept of equivalence and the object of translation studies. Target, 7(2): 191–​222. Landauer, Thomas K. and Susan T. Dumais. 1997. A solution to Plato’s problem: The Latent Semantic Analysis Theory of Acquisition, Induction, and Representation of Knowledge. Psychological Review, 104(2): 211–​240. Landauer, Thomas K., Peter W. Folt and Darrell Laham. 1998. An introduction to latent semantic analysis. Discourse Processes, 25(2): 259–​284. Lenci, Alessandro. 2008. Distributional approaches in linguistic and cognitive research. Italian Journal of Linguistics, 20(May):  1–​31. Lenci, Alessandro. 2018. Distributional models of word meaning. Annual Review of Linguistics, 4(1): 151–​171. Levy, Omer and Yoav Goldberg. 2014. Linguistic regularities in sparse and explicit word representations. In: Proceedings of the Eighteenth Conference on Computational Natural Language Learning. Ann Arbor, MI: Pub: Association for Computational Linguistics.171–​180. Li, Yuhua, Zuhair Bandar, David McLean and James O’Shea. 2004. A method for measuring sentence similarity and its application to conversational agents. In: Proceedings of the Seventeenth International Florida Artificial Intelligence Research Society Conference, Miami Beach, Florida, USA. American Association for Artificial Intelligence (AAAI) Press. 820–​825. Liu, Ying, and Chengqing Zong. 2004. Example-​ Based Chinese-​ English MT. In: IEEE International Conference on Systems, Man, and Cybernetics. 2004 IEEE International Conference on Systems, Man and Cybernetics (IEEE Cat. No.04CH37583): 6093–​6096. Mikolov, Tomas, Edouard Grave, Piotr Bojanowski, Christian Puhrsch and Armand Joulin. 2017. Advances in pre-​training distributed word representations. ArXiv: 1712.09405. Mitchell, Jeff, and Mirella Lapata. 2010. Composition in distributional models of semantics. Cognitive Science, 34(8): 1388–​1429. Nida, Eugene A. and Charles R. Taber. 1982. The Theory and Practice of Translation. Leiden: Brill.

06

60  Codère Corbeil Pelletier, Francis Jeffry. 1994. The principle of semantic compositionality. Topoi, 13:  11–​24. Sperber, Dan and Deirdre Wilson. 1995. Relevance: Communication and Cognition. Behavioral and Brain Sciences (Vol. 10). Oxford: Blackwell Publishers. Wanner, Leo, Gabriela Ferraro and Pol Moreno. 2017. Towards distributional semantics-​ based classification of collocations for collocation dictionaries. International Journal of Lexicography, 30(2): 167–​186.

16

4  “We Have a Big Crowd” The Different Referents of the First-​Person Plural in U.S. Presidential Candidates’ Talk on Entertainment-​Political Interviews Eean Grimshaw and Menno H. Reijven 1 Introduction1 In 2000, both major candidates for the U.S. presidency visited David Letterman’s late-​night talk show the Late Show. Back then, Letterman claimed that “the election will be decided here [on his show]” (quoted in Niven et al. 2003: 118). Over the last two decades, late-​night talk show interviews like those given by Letterman have become a central feature in U.S. presidential elections (Parkin 2014). During the 2016 elections, almost all presidential hopefuls visited a late-​night talk show to be interviewed by the host. Such interviews are unique in that they are conducted between a late-​night talk show host and a politician during an election campaign. Thus, the inherently entertaining nature of the program is blended with politics, as candidates are pressed on furthering their campaign (Parkin 2014). We refer to such interviews as entertainment-​political interviews (EPIs), i.e. a type of interview defined by co-​occurring entertainment and political discourse. This genre has been studied extensively by media scholars (e.g. see Baumgartner & Becker 2018), showing how the use of entertainment affects politics. In contrast, there is a lack of studies that pay close attention to the talk itself (however, see Loeb 2017). Specifically, we need to better understand the linguistic resources politicians use to connect with their audience. One salient linguistic resource used by politicians in the EPI is the first-​person plural (we, our, us –​ henceforth we), which is the focus of this chapter. However, the reference of we often remains vague (Borthen 2010). Despite we being a term including and centering on the speaker, the extralinguistic referent of we across uses can be different groups of people (Dori-​Hacohen 2014; Petersoo 2007). Thus, as we can imply different identities through invoked group membership, politicians can exploit this pronoun, enabling various pragmatic inferences beneficial to their cause (Petersoo 2007; Bull & Fetzer 2006; Bazzanella 2002; Wilson 1990). To improve our understanding of how we is used by politicians in an EPI, we characterize the different kinds of referents as introduced by U.S. presidential candidates’ use of we in EPI discourse.

26

62  Grimshaw and Reijven

2  Literature Review on In-​Grouping in Political Discourse Pragmatic studies of political discourse attempt to understand the production and reception of communication (Fetzer 2013). Studies have been concerned with three types of pragmatic inquiry (see Fetzer 2013). While some focus on how meaning is negotiated among different recipients, and others try to uncover the inference strategies the audience employs to uncover the speaker’s intentions, a third group focuses on clarifying the context-​dependent meanings of a particular genre. We engage in this last form of research. In this regard, we start from the fact that, through talk, the speaker positions themself and their addressee(s) in a presentation of the world, providing a context for producing and interpreting utterances (Chilton 2004). At any point, this discursive world is built by articulating differences and similarities between oneself and others using deixis (Chilton 2004). One tool politicians can exploit in this respect is person-​referencing  –​through person-​referencing, identities are negotiated in discourse (Bull & Fetzer 2006). To improve the understanding of political language in the EPI, we take a closer look at the genre-​specific referents of the first-​person plural (we, us, our  –​ henceforth we), as its referent often remains vague (Borthen 2010). By knowing the genre-​specific referents of we, we improve our understanding of the discursive worlds which are projected to effectively address the audience. Often, first-​person plurals imply membership with a larger group, helping to bridge a supposed or perceived gap between the individual speaking and others (Helmbrecht 2002; Íñigo-​Mora 2004). Sometimes, we is used by individuals to speak on behalf of the group invoked (Mühlhäusler & Harré 1990; Pavlidou 2014). Generally, first-​person plural forms include both the speaker and hearer(s) as referent. Speakers thereby project commonality and in-​group status with their recipients. When excluding the hearer as referent (i.e. by invoking a group the hearer is not a part of, see Dori-​Hacohen 2014; Íñigo-​Mora 2004), the speaker tactically distances themself from their audience to achieve their ends (Íñigo-​Mora 2004). We can also have various types of extralinguistic referents, including organizations like the Republican Party, groups based on relationships like one’s family, and gatherings (Lerner & Kitzinger 2007). In sum, the first-​person plural enables speakers to refer to a plethora of different social groupings and through this they can articulate various relationships they have (or not) with their audience. Since we can be used to invoke many different types of group memberships and relationships with the audience, we is often used by politicians in a strategic manner. Specifically, politicians can use we strategically to accept or deny their individual responsibilities for political actions (to be) committed (Pyykkö 2002; Bull & Fetzer 2006). Sometimes, we can be used to both encourage feelings of solidarity, and to identify and designate both supporters and enemies (Fairclough 2002; Íñigo-​Mora 2004; Skarzynska 2002; Wilson 1990). Through their pronominal choices, politicians can position themselves with respect to their colleagues, sometimes based on their similarities, but sometimes based on their differences (Íñigo-​Mora 2004; Maitland & Wilson 1987). Moreover, we

36

“We Have a Big Crowd” 63 can also be used to introduce political ideologies –​like nationality (Petersoo 2007)  –​into the conversation to form solidarity and separation. Hence, the first-​person plural is a potentially powerful device for politicians (e.g. Bull & Fetzer 2006; Pyykkö 2002) to strategically craft a discursive world wherein they position themselves, their interlocutor(s), and others. By presenting various group memberships and connecting with the audience through various collectives, politicians can realize their political goals (Pavlidou 2014; Petersoo 2007; Wilson 1990). Despite being a central linguistic force in political discourse (Pyykkö 2002), a determination of the exact meanings of we in context –​that is, the precise groups referred to  –​is not a straightforward task (Borthen 2010). In general, pronouns are conventionally classified as function words –​i.e. carrying only grammatical, not semantic meaning (cf. e.g. Hudson 2000; Quirk et al. 1985). However, in use, pronouns take on discursive meaning given the noun (phrases) they substitute (Fontaine 2006; Pyykkö 2002; Hudson 2000; Quirk et al. 1985). Thus, void of explicit reference, we, being a grammatically cohesive device, derives its situated meaning from the specific context it is used in (Borthen 2010). Referents of the first-​person plural can thus only be pragmatically inferred through considering the co-​text. Regarding we, scholars have developed taxonomies clarifying potential referents: a common distinction is between inclusive we, referencing both speaker and hearer (we instead of you and I), exclusive we, referencing speaker and a group not including the hearer (we meaning we without you), and a royal we referencing only the speaker, as traditionally used by a sovereign (Fontaine 2006; Íñigo-​Mora 2004). Some scholars have developed more extensive taxonomies. For example, Quirk et al. (1985: 350–​351) identified eight subtypes of we. However, their taxonomy arguably lacks detail regarding clarifying the potential referents for we. As we are interested in the uses of the first-​person plural by presidential candidates to categorize themselves, having only three types of we where they are the deictic center of a collective is too limited. To understand the variety of referents potentially invoked through we by presidential candidates to position themselves in a discursive world, we need more than a generic (i.e. all human beings), rhetorical (i.e. a specific group of which the speaker is a member), and an inclusive authorial we (i.e. just the speaker and hearer). The other forms of we identified by Quirk et al. (1985) either refer only to the speaker, only to the addressee, or to a third party. Dori-​Hacohen (2014) offers a more refined taxonomy than Quirk et  al.’s (1985) given his recognition of additional types of potential referents for we which center on the speaker as part of a collective. Through examining which groups callers refer to in Israeli radio phone-​ in programs, Dori-​ Hacohen (2014) identifies seven types of groups potentially referenced by we. As we focus on the types of we that can be used by the presidential candidate visiting the late-​night talk show to categorize themselves, two types as proposed by Dori-​Hacohen (2014) are not relevant to our inquiry: the vocal we, which is used when quoting a non-​present party, and the program we, which can only be used by the host to refer to their own show. The other five types of

46

64  Grimshaw and Reijven we (Dori-​Hacohen 2014) can be used by the caller to categorize themself as a member of a group. Four of these types are inclusive, and thus refer to both the speaker and the hearer. First, the humanity we (cf. Quirk et  al.’s 1985: “generic we”) used to refer to all people on Earth. Next, there is the general we, which is used to refer to a collective which includes both the speaker and hearer (cf. Quirk et al.’s 1985: “rhetorical we”). For example, American society includes both the presidential candidate, the host of the late-​night talk show, and the audience, but would not encompass all of humanity. This type of we has two subtypes. It can be open general if only the collective is invoked (e.g. ‘we Americans have opportunities’), but also opposing general if the group invoked is positioned in contrast to another group (e.g. ‘we Americans will beat the Dutch’). Fourth, the conversation we (cf. Quirk et al.’s 1985: “inclusive authorial we”) is used by interlocutors to manage the conversation at hand, referring to just speaker and hearer and no one else. The fifth type is the delimited social we, which is an exclusive form of we. It is used by the speaker to reference a group that the hearer is not a part of (e.g. ‘we as Americans have opportunities you do not have’). Although we has been widely studied in political discourse (e.g. Dori-​ Hacohen 2014; Fontaine 2006; Íñigo-​Mora 2004; Maitland & Wilson 1987; Petersoo 2007), such research has not yet considered the EPI. Given the rising prominence of the EPI in American political discourse (Parkin 2014), studying this genre is vital to understanding political communication in the United States. To better understand how politicians use this genre to connect with the electorate, the first-​person plural is one particularly salient, powerful resource. Therefore, we pose the following research questions: (1) What are the different referents of we in EPI discourse? (2) How may these findings have implications for tightening or expanding the current taxonomies of we?

3  Data and Methodology In order to study the various possible referents of we as used by presidential candidates on EPIs, we collected recordings of four episodes of broadcast late-​night talk shows: two appearances by Donald Trump and two by Hillary Clinton during the 2016 U.S.  presidential election cycle available online.2 Collected episodes were transcribed following Jefferson’s (2004) transcription system, but simplified for this chapter to improve readability. The dataset we filtered from these materials consists of 95 instances of the first-​person plural (i.e. we, our, ours, us), used by the presidential candidate with them at the deictic center (i.e. we excluded all such pronouns and determiners if they were part of direct quotes). Our analysis follows procedures used by Dori-​Hacohen (2014) and Íñigo-​ Mora (2004) in categorizing each one of these 95 instances based on the group referenced in concert with the speaker. We first determined whether the we used includes or excludes an interlocutor. Second, we focused on the exclusivity of membership of the group referred to, i.e. determining who else is part of this group. These steps enabled us to categorize the types of we based on

56

“We Have a Big Crowd” 65 Table 4.1 Number and types of we per presidential candidate

Open general we Delimited social we Conversation we

Trump

Clinton

21 43  1 65

13 16  1 30

34 59  2 95

Dori-​Hacohen’s (2014) taxonomy (see Table  4.1). Note that our upcoming discussion excludes the humanity we as no instance was present in our data (which are also generally rare, see Dori-​Hacohen 2014). We also do not discuss the conversation we as this we always refers to the presidential candidate and the talk show host. In the EPI, candidates use this we to engage with ongoing interaction, including participation in activities with the host. Categorizing all instances of we based on Dori-​Hacohen’s (2014) taxonomy still left us needing further specifications of the referent of we as used in context. As the referent of we is always informed by the local context, our categorization had to include determining the specific referents within a situated interaction (like Petersoo 2007). Therefore, in a third step, we focused on the surrounding discourse to determine the social context of the referenced group, further specifying the precise group referenced by we. Following Fontaine (2006) we also considered the mood and voice of verbs used in concert with we to better understand the referent. This enabled the specification of locally relevant referents of we for EPI discourse. Based on these types, we were also able to refine Dori-​Hacohen’s (2014) taxonomy of possible reference groups for we. We did not conduct a quantitative analysis as this study is decidedly qualitative and aims at elucidating the various possible referents of we within the context of EPI discourse. In the next section, we present our analysis. First, we discuss the general we and show that its genre specific referent is American society. Second, we present the possible referents for the social delimited we. For both, we also show what kind of subtypes should be recognized. Lastly, we discuss the functions these types of we have in EPI discourse.

4  Analysis of Types of We in EPI 4.1  The General We Being a type of inclusive we, the general we refers to both the speaker and the hearer (Dori-​Hacohen 2014). Additionally, the general we refers to a group of people, but a more restrictive group than all of humanity (i.e. the humanity we, see Dori-​Hacohen 2014). This we can be expressed in two ways. Sometimes, this we can be invoked by explicitly introducing an opposing group (i.e. the opposing general we, see Dori-​Hacohen 2014). However, this general group can also be referred to without resorting to making it explicit who is excluded

6

66  Grimshaw and Reijven from the group (i.e. the open general we, see Dori-​Hacohen 2014). In this section, we elucidate which groups are specifically referred to by U.S. presidential candidates (i.e. the genre specific referent) in EPI discourse from our data through the general we. Our first excerpt is taken from Jimmy Kimmel Live! In the course of the interview, Kimmel asks Clinton whether she would want to deport Trump. When Clinton reacts that she would only deport violent people, Kimmel jokes that Trump’s “second amendment comment” (not shown here), which refers to the suggestion made by Trump that people could stop Clinton from nominating judges based on their second amendment rights, would fulfill this criterion. Trump’s comment is then classified by Clinton as part of Trump’s “alternative universe.” Next, while critiquing Trump’s creation of an “alternative universe” to frame the campaign, Clinton responds as shown in Excerpt (1)). (1) JK-​HC:08/​22/​16 1. HC Well, you know, that’s another part of the 2. alternative universe. Look. This is such a serious 3. time in our country. 4. W[e ((have)) so many opportunities 5. JK   [Not really (0.3) not really ((laughing))/​(1.1) 6. AU [((laughing))/​(1.3) 7. HC [and there’s so much we can do and 8. JK Yeah. Clinton notices that it is “such a serious time in our country” (1:2–​3), thus referring to the United States as the country that she affiliates with as a citizen. As it would be possible to substitute our country with the country of both of us (Kimmel also lives in the United States), the first-​person plural is inclusive. Moreover, as Clinton’s comment was raised in response and as a complaint to Trump’s creating an “alternative universe” (1:2), she is implying that Trump is not taking “our country” seriously and is not addressing real problems. The fact that it is “a serious time in our country” is not only relevant to the two interlocutors, but to the U.S. society as a whole. Consequently, this first-​person plural possessive determiner does not just refer to Clinton, Kimmel, and the studio audience, but also to most of their television audience. Thus, this is a general we-​reference which refers to U.S. society. This general we-​reference  –​including Clinton, Kimmel, the present and remotely watching audience, as well as U.S.  citizens not actually watching the show –​is used again later (1:4; 1:7). Clinton follows her “serious time” comment by noticing that “we [have] so many opportunities” (1:4) and that “there’s so much we can do” (1:7). Across these three instances, the group referenced through “we” is not positioned as an active agent. For Clinton, these “opportunities” (1:4) and things “we can do” (1:7) are states of affairs, referencing the possibilities people have during “this serious time in our country” (1:2–​3). As above, it was “a serious time” for U.S. citizens, the U.S. citizenry is also subject of these “opportunities” and “things we can do.”

76

“We Have a Big Crowd” 67 In sharing her impression about the general state of their nation with her fellow Americans, Clinton does not make an explicit call for action by the citizenry. While this may encourage people to get involved themselves, Clinton does not directly order or request them to do so as she simply talks about the situation “in our country” (1:3) and the potentials in regard to positive change (i.e. “opportunities” and “can do”). Thus, Clinton uses these instances of we-​ reference in Excerpt (1) in a manner void of action (see Helmbrecht 2002: uses of we referring to a nation are not usually used to talk about collective action). Through these tokens of we, Clinton also self-​categorizes as a member of the larger American society, positioning herself as part of the same collective as the electorate. Moreover, Clinton does not order others in the collective to act, thus she portrays an egalitarian relationship between herself and the American citizenry through her use of we. Moving on to Excerpt (2)  from an EPI between Donald Trump and talk show host Jimmy Fallon, we notice how Trump also uses the general we. On the one hand, like Clinton, Trump’s uses of we do not ascribe action to the referent. On the other, Trump instead creates a hierarchical relationship between himself and the rest of American society. Excerpt (2) comes following a remark by Fallon where he raises that people perceive Trump and Vladimir Putin, President of the Russian Federation, as being rather close. Such a remark from Fallon could raise doubts as to whether Trump would have the best interests of Americans in mind when elected president. In turn, Trump rebuts this challenge as follows (see Excerpt(2)). (2) JF-​DT:09/​15/​16 1. DT I just think if we get along with Russia that’s not a 2. bad thing, and you know getting along with other 3. countries, [3 lines omitted] 7. I’m going to make great deals for our country (0.2) I’m interested 8. in our country I’m interested in (0.2) the success 9. of our country and right now I mean you you see 10. (0.2) what’s happening you see what’s happening just 11. generally speaking and (0.3) we have a long way to go Excerpt (2)  is part of Trump’s answer to the question about his relationship to Putin. After emphasizing that he does not “know him [Putin]”), Trump explains that it is “not a bad thing” “if we get along with Russia” (2:1–​2), or “with other countries” for that matter (2:2–​3 [emphasis added]). The juxtaposition of we with “Russia” and “other countries” suggests that this we refers to America. More specifically, like Dori-​Hacohen (2014), we believe this means that the references of we are not to the politicians of these respective countries but their societies. After all, any citizen could say that ‘it is not a bad thing if we get along with Russia.’ In this instance, how the American government acts (e.g. “getting along”) is also how the American people act.

86

68  Grimshaw and Reijven Next, Trump produces three instances of we-​reference through mentioning “our country” (2:8–​10). These uses of we are to be classified, like Clinton’s above, as general we-​references with U.S.  society as referent at large. First, Trump wants to “make great deals for our country” (2:8) because when a head of state makes a deal for their country, all its citizens will benefit from it in some way. Second, Trump states that he is “interested in our country” (2:9) and “in the success of our country” (2:9–​10). Similarly to the “great deals” (2:7) to be made on behalf of the country, the “success of our country” (2:8–​9) would also be experienced by all U.S. citizens. Lastly, Trump claims that “we have a long way to go” (2:12) regarding achieving the “success of our country.” Again, here it is all American citizens who have a long way to go until this prosperity is experienced by them. One the one hand, Trump’s use of the general we in Excerpt (2) is similar to Clinton’s uses in Excerpt (1), as both are void of action (e.g. Helmbrecht 2002): “Getting along with Russia” (2:1) is a desirable and potential future state of affairs, while “having a long way to go” (2:12) is the status quo for Americans. In the other instances (2:7; 2:8; 2:9), Americans, should Trump be elected, can experience the effects of Trump making “great deals,” being “interested in our country,” and “interested in the success of our country.” In each of these instances, where the referent of the we is all of American society, Trump refrains from mentioning collective action. On the other, by juxtaposing “I” and “our” (2:7–​10) within his utterances (e.g. “I’m going to make great deals for our country;” “I’m interested in our country”), Trump implies that he is going to be acting for or on behalf of “our country”, i.e. serving it. In contrast to Clinton, Trump’s general we not only references the general state of affairs in the country as experienced by its citizens, but places him in a hierarchical position with regard to the rest of the group being referred to. Thus, the general we in EPI can be used in two ways: to create an egalitarian relationship between the politician and their society or a hierarchical one. In our data, instances of such hierarchical we-​referencing only appear in Trump’s discourse, not once in Clinton’s. To conclude this section, the general we refers to “U.S. society” in EPI discourse. The referred to group is inclusive by referencing the presidential candidate, the host, and the U.S. citizenry. This type of we is not ascribing action to this collective (following Helmbrecht’s (2002) observation that any nation-​ we is too big to act collectively). Instead, the world is described discursively in terms of its current state and possible potentials. Moreover, presidential candidates can use this type of we in two distinct ways: egalitarian and hierarchical. For an overview of the distribution of egalitarian and hierarchical uses of the general we in EPI, see Table 4.2. 4.2  The Delimited Social We In contrast to the general we, which in EPI refers to the entire U.S.  society, the delimited social we is characterized by the speaker referring to a group of which they are a member, but their interlocutor is not. Using the delimited

96

“We Have a Big Crowd” 69 Table 4.2 Number and types of general we per presidential candidate

Egalitarian Hierarchical Total

Trump

Clinton

We 12 Our 2 We 0 Our 7 21

We 12 Our 1 We 0 Our 0 13

27  7 34

social we, the speaker talks as “representative” of the group they invoke (Dori-​ Hacohen 2014). In our data, we have identified two subtypes of this type of we-​reference. On the one hand, such uses of we can refer to groups whereof membership is appealing to the audience, i.e. in our specific case of EPIs, the voter, like family. We call this the association we (see Section 4.2.1 below). On the other hand, this type of reference may also refer to the presidential candidate’s political team (thus: political team-​we, see Section 4.2.2). Both groups not only exclude the host but also (the vast majority of) the audience witnessing the conversation. 4.2.1 The Association We We call the first subtype of the delimited social we the association we, as it presents the candidate as a member of a particular (social) group which they supposedly want to be associated with. In such cases, invoking this membership is expected to contribute to their being perceived positively by the electorate. A group often referenced through this type of we is family, as shown in Excerpt (3). (3) JF-​DT:01/​11/​16 1. JF Are you a grandpa? 2. DT I am. (.) in fact my daughter [6 lines omitted] 9. DT she [Ivanka] is she is going to have her third baby, in 10. two months so we are very happy about that. And we 11. are proud. Responding to Fallon’s question of whether Trump is “a grandpa” (3:1), and some overlapping talk (omitted lines), Trump shares that his daughter Ivanka “is going to have her third baby” (3:9) and that “we are very happy about that and […] proud” (3:10–​11). Thus, Trump positions himself as a member of a small social group, i.e. his family, which excludes Fallon as well as the audience. By starting his response turn with “in fact” (3:2), Trump suggests that the public might not be aware of this new development in his family, as they are not a part of Trump’s family. Their exclusion from Trump’s family explains their lack of knowing.

07

70  Grimshaw and Reijven This we, although excluding the audience, may still appeal to many people, as, conventionally, the vast majority will self-​identify as members of a similar group-​category. Thus, particularly in the United States, a country where family values rank high in importance, it reflects positively on a speaker to affiliate with such a social group membership as everyone can empathize with it. By sharing that he “cares,” is “happy” and “proud,” Trump invokes group-​specific family-​tied behaviors (Sacks 1972), demonstrating his belonging to that social group. By sharing this piece of news with the audience, and specifically by showing his happiness (if not pride) over his daughter expecting her third child, Trump enables the electorate to empathize with him as they would experience the same emotions. The importance of associating with one’s family becomes clear in Excerpt (4), where Clinton repairs an instance of “his” (Kimmel’s referring to her husband Bill Clinton) into we. Aware that it is publicly known her husband was accused of marital infidelity during his presidency, Hillary Clinton stresses her family as a unit –​specifically herself and her husband as a married couple and as grandparents –​the constant expected from any potential president of the United States. Immediately prior to Excerpt (4), Kimmel starts to talk about the Democratic National Convention. He starts this topic with talking about Bill Clinton’s look as if “he ha[d]‌n’t seen balloons in his life” (not shown) when balloons were falling down upon the podium. Next, Kimmel shows a picture of Clinton and her husband’s hotel room. (4) JK-​HC:08/​22/​16 1. JK and this is e this is his hotel room after the 2. [it’s a 3. HC [Yeah that’s our hotel room. He actually a (0.6) 4. don’t tell anybody he took a ballo[on for our= 5. JK                [((Laughs))/​ (1.0) 6. HC =granddaughter. She she could you know she is too 7. little she wasn’t there the night before ar ar our 8. nephew and our niece were so (0.8) Bill took one of 9. the balloons and the next morning (0.6) we played 10. with Charlotte and the balloon. So the balloon 11. fantasy and excitement continued. When Kimmel talks about “his [Bill Clinton’s] hotel room” (4:10) during the Democratic National Convention, Clinton, in her responding turn, references “our hotel room” instead (4:3). This suggests that Clinton is concerned with showing her relationship with him. This bond is reiterated in subsequent lines, as Clinton produces the same we: “our granddaughter” (4:4–​6), “our nephew and our niece” (4:7–​8) and “we played with Charlotte and the balloon” (4:9–​ 10). In each of these cases, Clinton shares insights into her role as a wife, grandmother, and aunt, cultivating her family relationships and engaging in salient group-​specific family activities to strengthen her perceived membership and identification with family values. As with Trump in Excerpt (3), Clinton’s

17

“We Have a Big Crowd” 71 use of we excludes the host and audience from belonging to her family group, yet is used by Clinton to connect with the electorate. Association with social groups through using the delimited social we does not always have to be with family. In Excerpt (5), Clinton’s we references both herself and her previous opponent in the Democratic primary, Bernie Sanders, who was highly popular amongst many Democratic voters whom Clinton would have to convince to vote for her in the general election. (5) JK-​HC:08/​22/​16 1. JK When’s the last time you spoke to Bernie Sanders? 2. (0.2) 3. HC Oh? at the convention. 4. JK At the convent[ion that was the last time he does[n’t 5. call you and go 6. HC         [right              [right 7. Oh no well I mean we are really working hard together 8. he is doing (0.3) a lot for the campaign. I’m very grateful Following their discussion about which major politicians will vote for her, Kimmel asks Clinton when she last spoke to Sanders (5:1). Following Clinton’s “Oh” (5:3), which indicates her surprise at this question (Heritage 1984), Kimmel questions Clinton’s relationship with Sanders by asking “he doesn’t call you?” (5:4–​5). Next, through a delimited social we (5:7), Clinton references herself and Sanders as a collective: “I mean we are really working hard together” (5:7). Namely, both the audience and host are again excluded from this we-​reference (as neither are “working hard together” with Sanders). Moreover, in this fragment, Kimmel has referred to Clinton through a singular “you” and also invoked Sanders. When using we, the topic of talk continues to be about the interactions between Clinton and Sanders, and thus it is only those two who are “working hard together,” subsequently clarified as Clinton explains that Sanders is doing a lot for her campaign. This instance represents an association we, as given Sanders’s popularity amongst Democrats throughout the primaries, Clinton would clearly want to gain his endorsement. Hence, Clinton wants to associate with him. Yet, Clinton also expresses hesitation when associating herself with Sanders through using we (i.e. her pause (5:2), and discourse markers such as “Oh” (5:3); and “Oh no well I mean” (5:7)). We suggest this is likely because too eagerly affiliating with Sanders, who identified as a ‘democratic socialist,’ might alienate more centrist voters whose support Clinton also needs to win the election. In sum, the delimited social we can imply one’s membership with various social groups. In EPIs, membership with such groups may reflect positively on potential voters. To establish such in-​group affiliations, speakers invoke certain ‘normal’ category-​bound activities (Sacks 1972): it is normal for a member of a family to be “happy” and “proud” about a “daughter” having “a baby”; it is normal to share a “hotel room” with one’s husband and normal to “play”

27

72  Grimshaw and Reijven with one’s granddaughter. Similarly, as with Sanders and Clinton, “working together” is the required activity regarding a political endorsement. This subtype of delimited social we, the association we as used by presidential candidates invokes exclusive social groups that voters are expected to appreciate. 4.2.2  The Political Team We The second subtype of the delimited social we in EPI discourse is the political team we, which introduces one of the presidential candidate’s political teams into the interaction (e.g. their campaign; their future administration; their party). While this kind of we-​reference thus excludes the host and audience, in contrast to the association we, the political team we is not used to relate to the audience. Instead, the political team we introduces an explicit separation between the candidate and one of their political teams, on the one hand, and the out-​group, on the other. This exclusion of the audience from the presidential candidate’s political team is shown in Excerpt (6). Here, Clinton talks about the policies she wants to pursue if elected president. Following their discussion about rumors Trump has spread about Clinton’s health to undermine her bid for the presidency, Kimmel and Clinton engage in a short activity where Clinton’s health is tested through her opening a jar of pickles. In the subsequent conversation Clinton uses the delimited social we. (6) JK-​HC:08/​22/​16 1. HC I do feel sometimes like this campaign has 2. entered into (0.2) an alternate universe (0.2) and 3. so I’m out here talking about hey (0.2) here’s how we 4. can create a lot more good jobs here’s how we can 5. help young people pay off their student deb[ts= 6. AU                    [((applause))/​ (2.6) 7. HC =here’s how we can make college more affordable 8. here’s what we do about healthcare lots of issues 9. prescription drugs mental health addiction (0.3) 10. I’m out here talking about all of this (0.3) and then 11. I have to sort of step into the alternative reality The prior discussion of Clinton’s health including the joking activity of opening a jar of pickles is qualified by Clinton as part of Trump’s “alternate universe” (6:2). In contrast to this “alternate universe,” Clinton uses a delimited social we to talk about her political aims. She outlines what her campaign platform seeks to change in the United States to help Americans. Clinton uses we to talk about her future policies, including “creating a lot more good jobs” (6:4), “helping young people pay off their student debts” (6:5), “making college more affordable” (6:7), and what to “do about healthcare” (6:8). The exclusive nature of this we is confirmed through saying “we do” (6:8): as only Clinton’s

37

“We Have a Big Crowd” 73 political team can do this, and not the larger American society. Yet, the we is also not referring to Clinton’s campaign team, as most of the volunteers canvassing for her will not be part of Clinton’s political team that can accomplish these things, should she be elected. Thus, Clinton’s proposals can only be enacted by her and her potential future administration, not the general public or Kimmel. As her future administration does not (yet) exist, this is not a possible referent. Instead, were Clinton to win the election, her cabinet officials would be taken from the Democratic Party. Thus, her social delimited we refers to politicians of the Democratic Party and Clinton outlines what they, with her, can accomplish for the American people. Trump produces a similar referent for we in Excerpt (7). When Fallon asks Trump why he pursues the presidency, he responds as shown in Excerpt (7). (7) JF-​DT:09/​15/​16 1. DT if you wanna help people there’s no better position to 2. do it (0.3) from than the presidency so we can do a 3. lot here Trump claims that in order “to help people” (7:1), “the presidency” (7:2) is the best position to hold: from there, “we can do a lot” (7:2–​3). Only Trump’s political team, not all Americans, will gain access to work immediately with Trump, if he is elected president, and, thus, “can do a lot” (7:2–​3). Similar to Clinton’s use in Excerpt (6), Trump’s future administration does not yet exist, the group of people who can do something if Trump wins the election are those cabinet officials taken from the Republican Party. Hence, this we explicitly introduces a separation between Republican politicians who can execute Trump’s proposals and the American public at large. In fact, both Trump and Clinton use the delimited social we in advocating for and promising future actions, if the public elects them president. In these cases, we only refers to a limited social group that can help the candidate to implement their envisioned policies, which are, in the long run, going to benefit the people they are addressing. However, this type of we-​reference is not only used when candidates talk about their future policies, and thus to refer to their party platform who can execute their plans, if elected, but also when presidential candidates address current activities they are undertaking which are being organized and supported by their campaign team. After Fallon talked about Clinton getting sick on the campaign trail, he inquires whether Trump has been close to getting sick. First, Trump evades the question by talking about the material Fallon’s desk is made of. Next, he affirms that he has not been sick at all, and continues as shown in Excerpt (8). Here, Trump explains that “right after this [appearance on Fallon’s show]” (8:1–​2), he is about going to give a speech to a large crowd of his supporters. Through saying that we are “making a speech” (8:2–​3), he positions himself and his campaign team separate from the crowd who will attend his speech.

47

74  Grimshaw and Reijven (8) JF-​DT:09/​15/​16 1. DT I I’ve really been uh we’re going in fact right after 2. this I’m going up to New Hampshire we’re making a 3. speech up in New Hampshire we have a big crowd and I 4. uh just got back from Ohio and we were in 5. Pennsylvania, Florida we’re all over Talking about the campaign activity he will engage in after this interview with Fallon, Trump shares that “we’re going” to “New Hampshire” to “make a speech” (8:1–​3). As Fallon will not travel with him, this is an exclusive use of we. By using we, the activity Trump engages in, i.e. “making a speech” in New Hampshire, is presented as a group activity that he is participating in. For this speech, they will “have a big crowd” (8:3). As this “speech” activity is not done by this “big crowd”, but for them, this “big crowd” is not included as referent for the we either. Like the “big crowd”, the audience watching the EPI is also not part of this we, as while the team staging the speech travels “all over” (8:5), the ordinary citizen does not do this to support Trump. Hence, the we invoked by Trump is exclusive, referring only to his political team and himself. In our data, both presidential candidates use a subtype of the delimited social we, the political team we to refer to the various political teams they employ to become a successful president. Sometimes, they talk about their plans and what their future administration can do for the American people. Then, the delimited social we refers to their larger party platform, as politicians from their party will become a part of their administration, and their party’s members of Congress will vote supportively of their plans. Thus, it is their party which can change Americans’ lives. At other points, presidential candidates talk about their current political team, which works to get them elected. In such instances, presidential candidates speak as the spokesperson for their volunteers and staff, and other politicians who spend time to convince voters to cast their vote for a specific presidential candidate. In these cases, presidential candidates talk about the current activities they are engaged in to reach the voter. In contrast to the association we which positions the presidential candidate as having desirable social relationships, the political team we refers to the people who are working to help the presidential candidate become successful at benefiting the American citizenry. 4.2.3  Summary: Types of the Delimited Social We in EPI In EPI discourse, the delimited social we has two types of referents: socially valued groups and political teams. Regarding the former, associations can be made with, for example, one’s family or popular political movements. Membership invoked through the association we is reinforced with the doing of stereotypical activities. In contrast, the political team we concerns either the current campaign activities, or the striving towards future policies. The activities conducted by these groups are/​can be done for the U.S. citizenry. For an overview of our corpus, see Table 4.3.

57

“We Have a Big Crowd” 75 Table 4.3 Number and types of delimited social we per presidential candidate

Association we Family Politics Campaign we Campaign trail Future administration

Trump

Clinton

5 We 2 Our 0 We 1 (+ GOP) Us 2 (+ Clinton) 38 We 34 Us 1 We 3 43

8 We 3 Our 4 We 1 (+ Sanders) Us 0 8 We 3 Us 0 We 5 16

13 9 4 46 38 8 59

4.3  Relations between the Different Types of We In Sections 4.1 and 4.2, we identified three types of referents for we which are relevant to EPI discourse. In the EPI, the general we refers to U.S. society. The delimited social we has two subtypes of referents: valued social groups and political teams. The general we is used by candidates void of action. Specifically, while the general we refers to the U.S. electorate at large, presidential candidates do not presume this collective as active agents (see Helmbrecht 2002). For the former subtype of the delimited social we, instances of association we are tied to category-​bound activities. In the latter subtype, the political team we is tied to actions done or to be done for the electorate. In this section, we focus on the placement of different kinds of we in the larger discourse. As noted above, the general we is used in EPI discourse to talk about the current U.S. social environment. In Excerpt (1), the general we was produced after Clinton’s criticized Trump’s “alternate universe” invoking the political team we (Excerpt (6)). In Excerpt (9)  we have reproduced the two previous excerpts in full ((1) and (6)), including the discourse in between, in order to shed full light on the logic underlying the use of we-​references. (9) JK-​HC:08/​22/​16 1. HC that’s so funny (0.4) no I mean I 2. I I do feel sometimes like this campaign has 3. entered into (0.2) an alternate universe (0.2) and so 4. I’m out here talking about hey (0.2) here’s how we can 5. create a lot more good jobs here’s how we can help 6. young people pay off their student 7. deb[ts= 8. AU    [((applause))/​(2.6) 9. HC =here’s how we can make college more affordable here’s 10. what we do about healthcare lots of issues 11. prescription drugs mental health addiction (0.3) 12. I’m out here talking about all of this (0.3) and then

67

76  Grimshaw and Reijven 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42.

JK HC AU JK HC JK AU HC JK AU JK HC JK HC JK AU HC

I have to sort of step into the alternative reality mhm and (0.3) you know (0.5) answer questions about am I al[ive (0.2) and how much longer will I be alive and the like   [((laughing))/​(1.0)   [I think it would be (1.9) if you were elected presid[ent=    [yes =how great would it be if your first act (0.5) was to deport Donald Trump to Mexi[co (1.7) is that (0.2) can the president do that               [((cheering and appla[use))/​(7.2)     [((laughing))/​(1.9) No (0.3) ha well I (0.1) I am not into deporting (0.2) anybody except violent people Okay al[right well that second amendment comment may have=      [((applause))/​ (1.5) =quali[fied ya know     [Well you know that’s another part of the alternative uni[verse I I (0.4) eh you look this is such a serious time in our country=         [((laughing))/​ (0.5) =W[e so many opportunities    [Not really (0.3) not really ((laugh[ing))/​(1.1)                  [((laughing))/​(1.3)                  [an there’s so much we can do and= (0.3)

Following Kimmel’s continuous invocation of Trump’s “alternative universe” (not shown) including suggesting Clinton might drop dead at any moment, Clinton lists those policies she wants to talk about in her campaign (9:4–​14), rather than addressing rumors about her health, part of Trump’s “alternative reality” (9:13–​14). As discussed above following Excerpt (1), these four uses of the we are of the delimited social type. Specifically, the political team we, as Clinton refers to her team which can make such policies (9:4–​11) a reality, should she win the presidency. In turn, Clinton contrasts talking about policy with her having to “step into the alternative reality” (9:13–​14) to deal with such rumors as concern her health. Then, Kimmel asks Clinton whether she wants to “deport Donald Trump to Mexico” (9:23–​24). Kimmel does not take Clinton’s call to step out of Trump’s “alternative universe” seriously. Despite the audiences “cheering and applause” (9:26–​27) Clinton responds that she is only into deporting “violent people” (9:30). Kimmel protests that Trump’s “second amendment comment” (9:31–​32) showed he is “violent” (9:30–​32).

7

“We Have a Big Crowd” 77 Thus, he attempts to still avoid Clinton’s call for political talk. However, after Clinton qualifies such comments as again part of Trump’s “alternative universe” (9:36), she remarks that “this is such a serious time for our country” (9:37) and that “we [have] so many opportunities” (9:39). Clinton’s invocation of the general we reiterates the importance and relevance of talking about real issues. Thus, the general we, void of collective action, can be used by presidential candidates to provide relevance for their political answers. In Excerpt (9) Clinton’s use of the general we to reference the current U.S. social environment helps justify why she should not talk about the “alternate universe”. Through her use of the general we, Clinton provides a common ground which is a foundation for her political talk. As another example of providing support for the acceptability of their answer through a general we, consider Excerpt (10), where Clinton supports her preference to run against someone other than Donald Trump. (10) JK-​HC :08/​22/​16 1. HC =I would be absolutely I mean I think that you 2. know all of the presidents that a a I have seen 3. during my lifetime that I’ve read about in history 4. we had great ones and not so great ones but I don’t 5. think we’ve ever been confronted with somebody who 6. (1.0) we see right now in the midst of this election 7. (0.6) is unqualified and temperamentally unfit Immediately prior to this excerpt, Kimmel asks Clinton whether she “would prefer to be running against somebody boring” (not shown). Clinton responds that someone “qualified” and “temperamentally fit” would be her preference. Next, she provides grounds for her preference (introduced by “I mean”, “I think that” and “you know” (10:1–​2)). According to Clinton, “all of the presidents” she has “seen” (10:2) and “read about” (10:3) have met certain standards Trump falls short of: “we had great ones and not so great ones” (10:4), but never have “we” “been confronted” (10:5), like “we see right now” (10:6). All these instances of we are the general we type –​as Clinton and all Americans “had great and not so great ones” (10:4) and “[are] confronted with somebody [Trump] (…) right now”  –​in Clinton’s justification for why she would “prefer” to be running against someone else. Thus, Clinton uses the general we to explain what is confronting U.S. society (Trump), as a means of supporting the acceptability of her earlier political response  –​that she “absolutely” (10:1) wants to run against someone qualified. While the general we is used as support for the relevance of political answers, instances of the social delimited we are primarily invoked to first respond to the questions on EPI. Given that presidential candidates are themselves usually the topic of discussion on EPIs (Eriksson 2010), candidates use the delimited social we to talk about themselves: via the association we (as a member of various social groups) or via the political team we to express ideas for the country (as the spokesperson for their campaign).

87

78  Grimshaw and Reijven

5  Conclusion As we have shown, on EPI, there are two types of we that have genre-​specific referents: the general we (with the U.S. citizenry as a referent) and the delimited social we, which we subtype into the association we and political team we. The general we is commonly used void of action, which softens the call-​out by presidential candidates for the U.S. public to take action. Candidates use this type to share information regarding the state of the social environment in terms of the state of affairs (i.e. facts) and opportunities (i.e. potentials). Structurally, the use of the general we is to provide relevance for their political responses. Additionally, the presidential candidate may use the general we whilst portraying themself in a hierarchical or egalitarian relationship to the electorate. The social delimited we is used by the candidates in our data in two ways, i.e. to associate with desirable social groups, and to introduce their political team. Regarding desirable social groups, candidates invoke group-​ specific activities to strengthen their perceived membership. Even though the audience is generally excluded from these groups, they too belong to such groups (e.g. family). Such associations are thus likely to invoke appreciation with people who value membership to such groups. When invoking their political team, candidates talk about what they can do if elected (with their party) and what they are currently doing to become elected (with their campaign team). In these cases, politicians separate themselves from their audience in a rather explicit manner, by emphasizing that they, as spokesperson for their political team, would do these things to benefit others –​their voters. While the association we is particularly important on the EPI for showing the candidate’s personal side, the political team we is needed to show their political identity. The general we is often used to make relevant the use of the political team we. Studying the referents of we in EPI is important as it provides us with a window through which we can begin to understand the discursive world created by politicians through this venue. The lack of the humanity we is not unexpected as the U.S. president will never, and is not supposed to, represent everyone on Earth. Instead, presidential candidates use a general we with U.S. society as referent to refer to a group including themselves as well as their audience. Thus, candidates position themselves as a member of the group they will have to lead. While sometimes, the politician invokes an egalitarian membership stressing common understanding, at other times they show a hierarchical relationship where they serve their group. Additionally, part of presidential candidates discursive presentation of the world is their various associations, including through invoking family and popular politicians. However, at other times, candidates need to show that they act for their voters, and invoke their political team. We recognize that our study has its limitations: we have showed qualitatively what possible referents are invoked in EPI discourse, rather than investigating their distribution quantitatively. A future quantitative study utilizing

97

“We Have a Big Crowd” 79 a larger dataset would be highly beneficial to better understand the types of we-​references employed in the EPI-​genre. Other politicians should also be included in this endeavor, as the types of referents for we used may depend on the idiosyncratic preferences of a specific presidential candidate. Accordingly, we suggest that in additional research of we-​references on EPIs, other primary candidates from 2016 might be included. Given that in the 2020 U.S. presidential election period a large number of Democratic candidates have appeared on late-​night talk shows  –​there are opportunities at hand to conduct such additional research of we-​references on EPIs. Moreover, the types of we identified in this study only indicate possible referents in EPIs and would need first to be proven relevant to other political discourses (e.g. journalistic interviews, debates, speeches). In summation, this survey of first-​person plural pronouns and determiners as used by Hillary Clinton and Donald Trump as two U.S.  presidential candidates in EPIs contributes not only to the understanding of the contextual of use and relationships among first-​person plurals but also to a venue for political communication that has increasing importance in the United States, the entertainment-​ political interview. As presidential candidates’ visits to late-​night shows have become commonplace in U.S.  political discourse, it is desirable to better understand presidential candidates’ use of the discursive resources of this genre. The first-​person plural is a rich linguistic device which offers politicians various ways to negotiate their social identity and create a discursive world. In addition to an improved understanding of this genre, our examination advances Dori-​Hacohen’s (2014) taxonomy through subdividing two additional uses of the general we (egalitarian and hierarchical) and two subtypes of the delimited social we (association and political team).

Notes 1 The authors would like to thank Professor Gonen Dori-​Hacohen for his comments, critiques, recommendations, and overall support of earlier versions of this chapter. We also thank him for his transcriptions of part of the corpus. Additionally, the authors would like to acknowledge fellow Ph.D. candidate Anton Dinerstein for his assistance in the transcription of part of our corpus. 2 Trump’s interview on The Tonight Show Starring Jimmy Fallon on 01/​11/​16 and 09/​ 15/​16; Clinton’s interview on The Tonight Show Starring Jimmy Fallon on 09/​16/​15 and on Jimmy Kimmel Live! on 08/​22/​16.

References Baumgartner, Jody C. and Amy B Becker (eds.). 2018. Political Humor in a Changing Media Landscape: A New Generation of Research. Lanham, MD: Lexington Books. Bazzanella, Carla. 2002. The significance of context in comprehension: The we case. Foundations in Science, 7(3): 239–​254. https://​doi.org/​10.1023/​A:1019657025835. Borthen, Kaja. 2010. On how we interpret plural pronouns. Journal of Pragmatics, 42: 1799–​1815. https://​doi.org/​10.1016/​j.pragma.2009.02.008.

08

80  Grimshaw and Reijven Bull, Peter and Anita Fetzer. 2006. Who are we and who are you? The strategic use of forms of address in political interviews. Text & Talk, 26(1): 3–​37. https://​doi.org/​ 10.1515/​TEXT.2006.002. Chilton, Paul. 2004. Analysing Political Discourse: Theory and Practice. London: Routledge. https://​doi.org/​10.4324/​9780203561218. Dori-​ Hacohen, Gonen. 2014. Establishing social groups in Hebrew: ‘We’ in political radio phone-​in programs. In: Pavlidou, Theodossia-​Soula (ed.), Constructing Collectivity: ‘We’ across Languages and Contexts. Amsterdam/​Philadelphia, PA: Benjamins. 187–​206. https://​doi.org/​10.1075/​pbns.239.13dor. Eriksson, Göran. 2010. Politicians in celebrity talk show interviews: The narrativization of personal experiences. Text & Talk, 30(5): 529–​551. https://​doi.org/​10.1515/​ text.2010.026. Fairclough, Norman. 2002. New Labour, New Language. London: Routledge. https://​ doi.org/​10.4324/​9780203131657 Fetzer, Anita. 2013. Anmerican society. In: Fetzer, Anita (ed.), The Pragmatics of Political Discourse: Explorations Across Cultures. Amsterdam/​Philadelphia: Benjamins. https://​doi.org/​10.1075/​pbns.228 Fontaine, Lise. 2006. Where do ‘we’ fit in? Linguistic inclusion and exclusion in a virtual community. In: Bührig, Kristin and Jan D. ten Thije (eds.), Beyond Misunderstanding. The Linguistic Reconstruction of Intercultural Communication. Amsterdam/​ Philadelphia, PA: Benjamins. 319–​356. https://​doi.org/​10.1075/​pbns.144.12fon7 Helmbrecht, Johannes. 2002. Grammar and function of we. In: Duszak, Anna (ed.), Us and Others. Social Identities Across Languages, Discourses and Cultures. Amsterdam: Benjamins. 31–​49. https://​doi.org/​10.1075/​pbns.98.03hel Heritage, John. 1984. A change-​of-​state token and aspects of its sequential placement. In: Atkinson, J. Maxwell and John Heritage (eds.), Structures of Social Action: Studies in Conversation Analysis. Cambridge: Cambridge University Press. 299–​345. Hudson, Richard. 2000. Grammar without functional categories. In: Borsley, Robert D. (ed.), The Nature and Function of Syntactic Categories. Bingley, UK: Emerald. 7–​36. https://​doi.org/​10.1163/​9781849500098_​003 Íñigo-​Mora, Isabel. 2004. On the use of the personal pronoun we in communities. Journal of Language and Politics 3(1): 27–​52. https://​doi.org/​10.1075/​jlp.3.1.05ini Jefferson, Gail. 2004. Glossary of transcript symbols with an introduction. In: Lerner, Gene H. (ed.), Conversation Analysis: Studies from the First Generation. Amsterdam/​ Philadelphia, PA: Benjamins. 13–​31. https://​doi.org/​10.1075/​pbns.125.02jef Lerner, Gene H. and Celia Kitzinger. 2007. Extraction and aggregation in the repair of individual and collective self-​reference. Discourse Studies, 9(4): 526–​557. https://​ doi.org/​10.1177/​1461445607079165 Loeb, Laura. 2017. Politicians on celebrity talk shows. Discourse, Context & Media, 20: 146–​156. https://​doi.org/​10.1016/​j.dcm.2017.08.006 Maitland, Karen and John Wilson. 1987. Ideological conflict and pronominal resolution. Journal of Pragmatics, 11(4): 495–​512. Mühlhäusler, Peter and Rom Harré. 1990. Pronouns and People. Cambridge, MA: Basil Blackwell. Niven, David, Samuel Robert Lichter and Daniel Amundson. 2003. The political content of late night comedy. Harvard International Journal of Press/​Politics, 8(3): 118–​133. Parkin, Michael. 2014. Talk Show Campaigns: Presidential Candidates on Daytime and Late Night Television. New York, NY: Routledge.

18

“We Have a Big Crowd” 81 Pavlidou, Theodossia-​Soula. 2014. Constructing collectivity with ‘we’. In: Pavlidou, Theodossia-​Soula (ed.), Constructing Collectivity. ‘We’ Across Languages and Contexts. Amsterdam/​Philadelphia, PA: Benjamins. 1–​22. https://​doi.org/​10.1075/​ pbns.239.03pav Petersoo, Pille. 2007. What does ‘we’ mean? National deixis in the media. Journal of Language and Politics, 6(3): 419–​436. https://​doi.org/​10.1075/​jlp.6.3.08pet Pyykkö, Riitta. 2002. Who is ‘us’ in Russian political discourse? In: Duszak, Anna (ed.), Us and Others: Social Identities Across Languages, Discourses and Cultures. Amsterdam/​ Philadelphia, PA: Benjamins. 233–​ 248. https://​doi.org/​10.1075/​pbns. 98.14pyy Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik. 1985. A Comprehensive Grammar of the English Language. London: Longman. Sacks, Harvey. 1972. On the analyzability of stories by children. In: Gumperz, John J. and Dell Hymes (eds.), Directions in Sociolinguistics: The Ethnography of Communication. New York: Rinehart & Winston. 325–​345. Skarzynska, Krystyna. 2002. We and they in Polish political discourse: A psychological approach. In: Duszak, Anna (ed.), Us and Others: Social Identities Across Languages, Discourses and Cultures. Amsterdam/​Philadelphia, PA: Benjamins. 249–​264. https://​ doi.org/​10.1016/​j.pragma.2011.06.010 Wilson, John. 1990. Politically Speaking: The Pragmatic Analysis of Political Language. Oxford: Basil Blackwell.

28

5  Whatever Can Be Meant Can Be Echoed A Comparison of Since When Questions and Echo Declarative Questions Angelika Kiss 1 Introduction1 Echoing is commonly considered linguistic metarepresentation (cf. Noh 1998, 2000). When we echo a previous utterance, we represent some aspect of it, attribute it to the author of the previous utterance, and express our own attitude toward the represented aspect (Noh 1998, 2000; Iwata 2003), in a similar fashion to quoting (Bublitz 2015). This makes echoing comparable to holding a mirror to our interlocutor and drawing their attention to some aspect of the reflection while expressing our own opinion about it. Consider Example (1). (1)

A: We have to raise the taxes. B: We have to raise the taxes?

(Adapted from Poschmann 2008: 253)

As an echo of (1A), (1B) expresses the speaker’s surprise, incredulity, or disapproval. The speaker can also point out an issue relating to the form of the preceding utterance as in Example (2). (2)

A: I’ll call the pólice. B: You’ll call the pólice? You’ll call the políce!

(Adapted from Horn 1989: 371)

(2B) notices a metalinguistic issue in the discourse, namely that police is pronounced incorrectly by A as pólice, with stress on the first instead of the second syllable. (2B) seeks to draw A’s attention to this issue by repeating A’s utterance in full, and complementing it by providing the correction. B’s utterance metarepresents A’s utterance by pointing out a metalinguistic issue (cf. Blakemore 1994; Noh 1998; Iwata 2003, Malamud & Stephenson 2015). This chapter investigates utterances like (1B) and (2B), called echo declarative questions (EDQs), and compares them to utterances like (3B), called since when questions (SWQs). (3)

A: We have to raise the taxes. B: Since when do we have to raise the taxes?

38

Whatever Can Be Meant Can Be Echoed 83 I argue here that SWQs perform an echo function as well. When comparing EDQs and SWQs, the following questions arise: i) What aspect of the previous utterance can they target? ii) What attitude can they convey? iii) Can they be used interchangeably? A comprehensive answer to (i)  is crucial for dynamic pragmatic discourse models (cf.  Malamud & Stephenson 2015; Portner 2018) as it can shed light on what gets grounded upon hearing an utterance. Most discourse models focus on the propositional content that is added to the common ground, removed from it, or corrected as the discourse develops (cf. Farkas & Bruce 2010; Farkas & Roelofsen 2017). It has been noted that information other than the truth-​conditional content of an utterance can also be targeted (cf. Horn 1985, 1989; Malamud & Stephenson 2015; Schlöder et al. 2016; Kiss 2017; Geurts 2019) and interlocutors do keep track of such information, to some extent at least (cf. Clark 1996). However, the question whether non-​truth-​conditional meaning should also be traced in a dynamic model is normally not raised (exceptions are Malamud & Stephenson 2015; Geurts 2019). Answering (ii) leads to a better understanding of what function echoing utterances may perform in discourse. EDQs and SWQs can both be used to echo, but they differ in the range of attitudes they can convey: SWQs are more readily accepted by native speakers of English in contexts where there is a conflict between the interlocutors’ attitudes toward the echoed aspect. This property of SWQs makes them similar to metalinguistic negation which is also felicitous if there is a conflict between interlocutors’ commitments (Horn 1985, 1989). The answer to question (iii) follows from the answers given to questions (i) and (ii): since SWQs, but not EDQs, are restricted to contexts of disagreement on a certain issue, the two cannot always be used interchangeably. Thus, EDQs and SWQs can both echo but they differ in what kind of speaker attitude they can express. The present chapter is organized as follows: Section 2 introduces EDQs and SWQs. Section 3 discusses cases of echoing that have been mentioned in the literature (cf. Blakemore 1994; Noh 1998, 2000; Iwata 2003; Gunlogson 2003; Poschmann 2008) and shows that SWQs can target the same aspects of the previous utterance that EDQs can, although they exhibit a contrast in terms of the licensing contexts. Section 4 shows that rhetorical relations and stancetaking can also be echoed. In Section 5, the novel distinction between metalinguistic and meta-conversational echoes is introduced with the aim of revealing the distributional differences between EDQs and SWQs. Section 6 concludes the chapter. Throughout, the examples have either been borrowed from the literature or elicited from native speakers; the judgments have been provided by native speakers of North American English. Although the target language of this study

48

84 Kiss is English, many if not all of the claims made here are applicable to EDQs and SWQs in other languages that use such constructions.

2  Echo Declarative Questions and Since When Questions 2.1  Echo Declarative Questions Rising declaratives in English have been so called because they have declarative word order, and, at the same time, a rising sentence-​final tune (Pierrehumbert & Hirschberg 1990). Gunlogson’s (2003) insight was that such utterances express a commitment shift: the utterer of the rising declarative does not commit himself/​herself to the truth of the proposition conveyed by the sentence radical. The final rising tune commits the addressee to that proposition, not the speaker. While Gunlogson’s observation inspired much fruitful research (cf. e.g. Farkas & Bruce 2010; Poschmann 2008; Malamud & Stephenson 2015; Farkas & Roelofsen 2017; Jeong 2018; Heim 2019), it has to be noted that her term ‘rising declaratives’ is misleading. This is to say that there are languages that express the same discourse function differently (e.g. Hungarian, where rising declaratives are pronounced with a rise-​fall contour on each accentual phrase and are therefore called rise-​fall declaratives, cf. Gyuris 2019). Even within English, Bartels (1999: 157) reports that, when declarative questions request confirmation from the addressee, they do not necessarily have a final rise, while, if they are used as echoes, they obligatorily do. Poschmann (2008) therefore calls rising declaratives declarative questions. Declarative questions can serve the purpose of requesting clarification or conveying speaker attitude. Poschmann calls the latter echo declarative questions (EDQs) –​this is the declarative question type focused on in this chapter. EDQs are echo questions, thus they can be verbatim echoes of the previous utterance.2 But, as Poschmann argues (2008: 264), they can equally echo the information content of the proposition conveyed by the previous utterance or its formal properties such as word order, phonetics, or register (cf. also Noh 1998, 2000). Horn (1985, 1989) first noticed that the same aspects can be targeted by metalinguistic negation: one can object to a previous utterance “on any grounds whatever”, without rejecting the truth of the proposition it conveys. EDQs can serve as metalinguistic negation, although this is not necessarily the case. Examples (4)–​(6) show that EDQs can echo the information content of a previous utterance, which, according to van der Sandt (1991) is the proposition conveyed by the previous utterance (4B1), a presupposition it comes with (4B2), an entailment (4B3), a conversational implicature (5), or a conventional one (6). Echoing, however, can also target the utterance’s form, as is shown in (7). (4)

[Context: A and B are headed to a steakhouse where they plan to meet John.] A: John is not coming with us to the restaurant. He’s stopped eating meat.

58

Whatever Can Be Meant Can Be Echoed 85 B1: John’s not coming? B2: John used to be a meat eater? B3: John is a vegetarian?

proposition presupposition entailment

(5)

A: Some Cretans are liars. B: Not all Cretans are liars?

conversational implicature

(6)

A: She is poor but honest. B: Poor people are not honest?

conventional implicature

(7)

a.  A: I’ll call the pólice. pronunciation   B: You’ll call the pólice? You’ll call (after Horn 1989: 371) the políce! b.  A: Granpa is feeling lousy. style    B: Granpa is feeling lousy? (after Horn 1989: 371)

In addition, Noh (1998) notes that one can also echo previous utterances with different clause types, like interrogative (8a.), imperative (8b.), or exclamative clauses (8c.), which convey different speech acts. (8a) a.

A: Have you read “Great Expectations”? B: Have I read “Great Expectations”? b. A: Talk to a fortune-​teller. B: Talk to a fortune-​teller? B1: You want me to talk to a fortune-​teller? B2: You think I should talk to a fortune-​teller? B3: You’re telling me to talk to a fortune-​teller? c. A: What a great pleasure this is! B: What a great pleasure this is? (Noh 1998: (3)–​(5))

There are various ways to echo a speech act, as shown by B’s possible reactions in (8b.). Noh calls these reactions “reformulations” that make explicit “an aspect of the intended interpretation that would otherwise have to be inferred” (1998: 609); thus, she classifies reformulations as echoes. What is more, a preceding utterance is not even a necessary condition for a felicitous echo (Noh 1998, 2000; Iwata 2003): a salient object or event in the non-​verbal context can trigger an echoing utterance as shown in Example 9. (9)

Looking at a London map, a freshman at University College says: ‘The British Museum is near University College?’ (Noh 1998: (9))

The common trait in the above examples is that when there is a preceding utterance, an EDQ always targets a commitment of its speaker. Gunlogson makes the point that it is far from the case that commitments can only arise by uttering declarative clauses. She emphasizes that utterances realized through

68

86 Kiss other clause types can also give rise to commitments. She claims that “[i]‌n a broader sense, anything an agent publicly treats as true for the purposes of the discourse will qualify as a commitment” (Gunlogson 2008: 10). Thus I  assume that speakers commit to everything that follows from their other commitments (cf. Geurts 2019: 4f.), and to the appropriateness of every aspect of the speech act that conveys p. If a speaker commits to a proposition p, she not only commits to p’s information content, but to the appropriateness of the form of the utterance and the illocutionary act as well. The notion of commitment used here thus differs substantially from the one used in the most influential commitment-​based models, such as that of Farkas & Bruce (2010). Commitments can be negotiated (cf. Farkas & Bruce 2010), and declarative questions are a conventionalized way of doing so: with their interrogative-​like form, EDQs express a call on the addressee to justify the commitment echoed by the speaker. That is, in (8a.), B questions A’s commitment to asking (8a.A); in (8b.), B targets speaker A’s commitment to uttering the directive in (8b.A); and in (8c.), B questions A’s commitment to making an expressive speech act (see also Section 3.3). EDQs can serve various goals. These can be made explicit by adding some follow-​up material (continuation) in such a way that it expresses the speaker’s affective or epistemic stance. The utterance in (4B1) has three different followups as shown in (10), expressing different speaker attitudes. (10) B1: John’s not coming? a. …Great! At least we can discuss his surprise birthday party. b. …I didn’t know that. c. …That’s just not true! In (10a.), B1 expresses surprise and happiness about what A said, and this utterance indicates that the speaker has grounded the proposition p = ‘John’s not coming’. In (10b.), B1 signals that p was new to him; and in (10c.), he does not believe that p. Therefore, (4B1) cannot be associated in itself with a certain speaker attitude, as the attitude it serves to convey is highly context-​dependent. In sum, EDQs target a commitment (in a broader sense) that a previous utterance expresses, or any contextually salient phenomena, and in addition, they allow the speaker to convey a range of speaker attitudes. 2.2  Since When Questions According to Cheung (2009), the negative wh-​construction is a biased wh-​ question expressing propositional negation. Cheung has shown that it is available in a number of languages, and different languages use different subsets of wh-​phrases. Kiss (2017) argued that not all such convey propositional negation. In languages where both since when and where can be used for this purpose (e.g. Korean, Italian, or Russian), the latter is used for propositional negation exclusively, while since when can express the speaker’s disapproval of other aspects of the utterance as exemplified in Example (11).

78

Whatever Can Be Meant Can Be Echoed 87 (11) A: [utterance in the tu-​form] B: Ma da quando mi dai del tu? (Italian) ‘Since when are you using the tu-​form with me?’

(Kiss 2017: (4))

(11B) is a negative wh-​construction that does not express propositional negation: the speaker merely disapproves of the register used by A to address her. The relevance of SWQs to echoing is the following: just like EDQs, SWQs need not target the truth-​conditional content of the preceding utterance, which is a claim I make countering what Cheung has argued (2009). Both EDQs and SWQs may express disagreement with the informational content of the preceding utterance, but do not necessarily do so: they may express disapproval of the form of the utterance (cf. (7a.) and (11B)), which makes them similar to metalinguistic negation (Horn 1985, 1989). I claim that SWQs are echoic utterances, like EDQs, but the two are not alike in all respects. The following section presents both contexts where they are pragmatically equivalent and contexts where they diverge.

3  Echoing Three Kinds of Acts Searle proposed that a speech act embodies a propositional act, an utterance act, and an illocutionary act (1969: 24). Assuming that the information content results from the propositional act and the form of an utterance from the utterance act, this section demonstrates that both EDQs and SWQs can target all three kinds of acts. 3.1  Propositional Act As mentioned before, EDQs can echo the information content of the previous utterance (Noh 1998; Poschmann 2008; cf. (4)–​(6) above). Examples (12)–​(14) demonstrate this for SWQs. (12) [Context: A and B are headed to a steakhouse where they plan to meet John.] A: John is not coming with us to the restaurant. He’s stopped eating meat. B1: Since when’s John not coming? B2: Since when was John a meat eater? B3: Since when is John a vegetarian?

proposition presupposition entailment

(13) A: Some Cretans are liars. B: Since when aren’t all Cretans liars?

conversational implicature

(14) A: She is poor but honest. B: Since when are poor people not honest?

conventional implicature

As these examples show, SWQs can target the proposition conveyed by the previous utterance (12B1), a presupposition it comes with (12B2), an

8

88 Kiss entailment following from the proposition (12B3), a conversational implicature (13B), or a conventional implicature (14B). SWQs thus seem to be able to target anything that EDQs can. Yet, they allow a different range of speaker attitudes and they also differ in their degree of politeness. Examining the ways one can continue the discourse after a SWQ reveals that such utterances are more felicitous in contexts where there is a conflict between the interlocutors’ commitments. The only case unanimously accepted as a possible continuation of (15B1) is (15c.), which is a follow-​up expressing the speaker’s disbelief in the proposition conveyed by the previous utterance (i.e. John is coming with us). (15a.) may be classified as pragmatically odd, and (15b.) may be accepted by some speakers, but not all. (15) B1: Since when is John not coming with us? a. …#This is great! At least we can discuss his surprise birthday party. b. …%I didn’t know that. c. …That’s just not true! There is thus a contrast in the kinds of propositional attitudes EDQs and SWQs are compatible with. This is revealed by adding a follow-​up to the EDQ or the SWQ that makes the speaker’s stance explicit. Consider Table 5.1, where Examples (10) and (15) are juxtaposed. Table  5.1 shows that, while EDQs are felicitous in discourse contexts where p is grounded or rejected, this flexibility does not seem to be available for SWQs. A property that follows from the above asymmetry is that SWQs are less polite than EDQs, a property that further prevents them from being used interchangeably in the same contexts. The difference in propositional attitudes and politeness these utterance types can convey may also stem from the form of these utterances. Both polar and declarative questions denote two alternatives, p and ¬p (Farkas & Bruce 2010; Farkas & Roelofsen 2017); on the other hand, Table 5.1 EDQs and SWQs conveying different propositional attitudes towards p = ‘John’s not coming with us’  

Echo declarative question

Since when question

Propositional attitude towards p

 

(15) Since when is John not coming with us? # This is great! At least we can discuss his surprise birthday party. % I didn’t know that.

 

b.

(10) John is not coming with us? This is great! At least we can discuss his surprise birthday party. I didn’t know that.

c.

That’s just not true!

That’s just not true!

a.

p is easily grounded p is grounded, but was new to the speaker p is not common ground

98

Whatever Can Be Meant Can Be Echoed 89 SWQs are wh-​questions, where since when denotes a set of previous times with an infinitely high number of members. Thus, polar/​declarative questions and wh-​questions differ in the weight of the request towards the addressee: by asking a polar question, the addressee is required to choose between two alternatives, but a wh-​question makes the addressee choose from a larger set of alternatives, which requires more cognitive effort, resulting in a larger imposition on the addressee (Brown & Levinson 1987). For these reasons, Levinson (2012) concludes that asking a polar question is less face-​threatening than asking a wh-​question. The relevance of the contrast in speaker attitude lies in the fact that, while both EDQs and SWQs can echo, the speaker’s propositional attitude for SWQs is limited to objection: it only serves to pinpoint mismatches between the interlocutors’ versions of the common ground, or to signal that the grounding did not come about in a straightforward way. This property is related to the contrast in (im)politeness between EDQs and SWQs, which arguably stems from the form of these questions. 3.2  Utterance Act In Searle’s (1969) view, speech acts are utterance acts displaying form-​related properties such as pronunciation, word order, dialect, style, or register. Properties relating to the speech event, such as the time and place of the utterance or the participants of the discourse all characterize the utterance act. A  speaker can express her attitude toward these properties (more precisely, to the previous speaker’s commitment to the appropriateness of the form of her utterance) by echoing them with an EDQ or a SWQ, as shown in (16). (16) A: I’ll call the pólice. B1: You’ll call the pólice? You’ll call the políce! B2: Since when will you call the pólice? You’ll call the políce! In (16), both B1 and B2 echo the linguistic form of A’s utterance. An example of targeting the register has been given in (11), which is expressed in Italian by the T-​V distinction (cf. Brown & Levinson 1987). The mini-​dialogue in (17) exemplifies a case in which biased questions target the non-​linguistic, event-​ related properties of the utterance. (17) [Context: A and B are at a family gathering.] A: I think we need to have separate bank accounts. B1: We’re talking about money in front of your aunt? B2: Since when are we talking about money in front of your aunt? In (17), B1 and B2 metarepresent an event-​related property of A’s utterance, namely that it occurs in the presence of A’s aunt. It is possible to imagine that both B1 and B2 dissociate themselves from the commitment that it is

09

90 Kiss acceptable to talk about their income in front of A’s aunt. However, as in the case of propositional acts, there is an asymmetry here in that while B1 can also be pronounced in a surprised but non-​disapproving tone, B2 does not seem to have this option available, it favors the disapproval-​reading. 3.3  Illocutionary Act Both EDQs and SWQs can echo the illocutionary act as well as the propositional act and the utterance act. Searle and Vanderveken (1985) classified speech acts according to their illocutionary points, which is the essence or basic purpose of the speech act. The five types they identified are the assertive, directive, commissive, expressive, and declarative illocutionary points. I have shown examples of assertions and directives being echoed in (4B1) and (8b. B1) respectively; Examples (18)–​(20) show expressives, commissives, and declarations echoed by EDQs and SWQs. (18) Expressives A: I’m sorry. B1: You’re apologizing to me? B2: Since when are you apologizing to me? (19) Commissives A: I’ll call the police. B1: You’re threatening me? B2: Since when are you threatening me? (20) Declarations A: I declare you husband and wife. B1: You’re marrying us? B2: Since when are you marrying us? While B2’s utterances in (18)–​ (20) are felicitous, not all EDQs can be replaced by an SWQ in a straightforward manner (cf. B2 in (21a.)): (21) a. b.

A: Have you read “Great Expectations”? B1: Have I read “Great Expectations”? B2: #Since when have you/​I read “Great Expectations”? A: Talk to a fortune-​teller. B: Talk to a fortune-​teller? B1: You want me to talk to a fortune-​teller? B1’: Since when do you want me to talk to a fortune-​teller? B2: You think I should talk to a fortune-​teller? B2’: Since when do you think I should talk to a fortune-​teller? B3: You’re telling me to talk to a fortune-​teller? B3’: Since when are you telling me to talk to a fortune-​teller?

19

Whatever Can Be Meant Can Be Echoed 91 c.

A: What a great pleasure this is! B1: What a great pleasure this is? B2: Since when is this a great pleasure?

(21a.B1) is a verbatim echo. In English, all it takes to turn utterances into their echoes is to modify their intonation by adding a final rise.3 SWQs are wh-​ questions, however, and they cannot be used as literal echoes, given that a polar interrogative can never occur inside a wh-​interrogative clause. If A’s utterance in (21b.) is echoed by a SWQ, one has to make explicit the illocutionary point of the speech act with a performative verb so that it can embed the clause talk to a fortune-​teller. The speaker is compelled to spell out an implicit aspect of the interpretation that “would otherwise have to be inferred” (cf. the discussion of Noh’s reformulations in Section 2.1). If A’s utterance in (21c.) is echoed as in (21c.B2), it does not echo the illocutionary point of the expressive, just the proposition it contributes. This difference between EDQs and SWQs, which does not just apply to the case of echoing illocutionary acts, is discussed in more detail in Section 5. The contrast in speaker attitudes described in Section 3.1 also holds for illocutionary acts: B1’s utterance in (22) are compatible with a context in which B1 has immediately accepted A’s utterance, and the EDQ merely signals surprise. B2’s utterance is not as readily uttered in a context in which B2 accommodates the fact that A is marrying them, which is signaled by the oddness of the follow-​ up that implies acceptance. (22) A: You are now husband and wife. B1: You’re marrying us? Then let me check if I have the rings. B2: Since when are you marrying us? #Then let me check if I have the rings. In sum, echoing speech acts is possible both for EDQs and SWQs, but SWQs cannot reproduce all the forms of echoing that EDQs can. SWQs cannot serve as verbatim quotes, a limitation due to their being wh-​questions. The above facts about SWQs contradict Cheung’s (2009) claim that negative wh-constructions express propositional negation only. While it is possible for SWQs to express negation, they are better analyzed as devices of metalinguistic negation (Horn 1985, 1989), which allows for expressing denial without propositional negation.

4  Echoing Rhetorical Relations and Stance It was shown in the previous section that one can echo different aspects of the preceding utterance not only with EDQs but also with SWQs. This section shows that the range of these aspects can be broadened, as there are further aspects of an utterance that one can commit to and metarepresent, a fact which has received little attention in the literature on echoing. These are rhetorical relations, which provide coherence between the utterances of a discourse, and

29

92 Kiss stancetaking, expressing the speaker’s emotional or epistemic attitude towards the utterance or interlocutors. 4.1  Echoing Rhetorical Relations What Asher and Lascarides (2003) call rhetorical relations can be regarded as belonging to the information content of an utterance: they have truth-​ conditional effects on the interpretation and hence they affect the common ground of the interlocutors. Rhetorical relations provide the implicit glue that makes a discourse coherent; interlocutors rely on them when interpreting utterances, and if they fail to infer these implicitly added rhetorical effects, it may lead to a misunderstanding. The relations tying together the content of two clauses may be made explicit by connectives (because, but, etc.), but it is not crucial that they be added in order to ensure the correct interpretation. Consider the following example: (23) Max had a lovely evening. He had a great meal. He ate salmon. He devoured cheese. He won a dancing competition. (Asher & Lascarides 2003: 9) Example (23) is a coherent piece of discourse, although cohesion is not overtly marked by designated linguistic means (i.e. connectives). Asher and Lascarides  (2003) spelled out the hidden relations that ensure cohesion by the labels Elaboration and Narration. The utterances He ate salmon and He devoured cheese are related by Narration, and they both elaborate on the immediately preceding utterance He had a great meal, to which they relate by Elaboration. The entire sequence starting from He had a great meal functions as an elaboration of the first utterance Max had a lovely evening. In (24), A’s two sentences are tied together by the rhetorical relation called Explanation. Once echoed, the implicit rhetorical relation comes to the surface as because. (24) A: John is not coming with us to the restaurant. He’s stopped eating meat. B1: John’s not coming because he stopped eating meat? B2: Since when’s John not coming because he stopped eating meat? He just wants to avoid Mary! B’s reactions in (24) do not necessarily target the truth-​values of the two propositions contributed by the two clauses in A’s utterance; what they target is the relation, i.e. Explanation. As for (24B2), the speaker can accept that John is not coming, as well as the claim that John became a vegetarian; but he does not believe that the reason why John is not coming is that he became a vegetarian. Rhetorical relations, too, are something speakers commit to, hence they can be echoed, just like other speaker commitments. I consider this idea parallel to Iwata’s (2003) claims, according to whom assumptions and “unspoken thoughts” can also be metarepresented and, hence, echoed.

39

Whatever Can Be Meant Can Be Echoed 93 The asymmetry in speaker attitudes between EDQs and SWQs holds here as well: While (24B1) can possibly convey the full range of speaker attitudes, (24B2) is felicitous only in contexts with conflicting commitments or in cases where the speaker signals that the proposition was new to him/​ her. 4.2  Echoing Stance Stance is defined as the attitude a speaker takes toward the content or form of their message or toward their interlocutors (e.g. Du Bois 2007; Jaffe 2009; Kiesling 2009). Stance can be seen as having two main aspects: one that relates to emotions (affective stance) and one that relates to what the speaker knows (epistemic stance), as proposed by Ochs and Schieffelin (1989). It is so pervasive in discourse that every utterance is a potential act of stancetaking. As stancetaking acts, they commit their speakers; and as commitments, they become potential objects of negotiation in the following discourse: “Stances taken in interaction play a contextualizing role, creating a point of reference for subsequent utterances” (Jaffe 2009: 12). The following examples illustrate that a stance taken by the speaker of a preceding utterance can be echoed by both EDQs and SWQs. (25)

[Affective stance] A: If that fool doesn’t stop playing the trombone, I’ll call the police! a.  B1: John’s a fool? B2: Since when is John a fool? b. B1: You don’t like John? B2: Since when don’t you like John?

(26)

[Epistemic stance] A: Montague was wrong. a.  B1: Montague was wrong?   B2: Since when was Montague wrong? b. B1: You disagree with Montague?     B2: Since when do you disagree with Montague?

In (25), A conveys a negative affective stance toward a situation and/​or its participants, that is, John and him playing the trombone, whereas in (26), A merely makes a statement. B’s echoing utterances in both examples target A’s stance. In (25), B’s utterances signal surprise and/​or disapproval toward A’s negative attitude toward John; and in (26), B targets A’s epistemic stance toward Montague’s ideas. B’s reactions in both examples show that stance can be metarepresented and echoed, exhibiting the asymmetry in speaker attitude observed in other uses of echoing utterances. While EDQs are compatible with contexts in which the speaker accepts whatever she echoed, SWQs are not, B2’s utterances in the above examples require a context in which A and B disagree on whether John is a fool or whether Montague was wrong.

49

94 Kiss (25) and (26) are examples of echoing a conventional implicature and the proposition, respectively. But stance can be expressed in other forms, for example as a conversational implicature, which is not encoded in the utterance but is inferred; consider Example (27). (27) [Epistemic stance] A: Semantics is not yet in a position to solve this problem. B1: You know semantics this well? B2: Since when do you know semantics this well? The stance attributed to A by B is a higher position in the epistemic hierarchy on semantics, the current topic of discourse (Kiesling 2009). This implicature is not linguistically encoded, but it is present by virtue of the conventional assumptions that whoever makes a claim like (27A) is in a position to know whether it is true; and that knowing whether it is true implies that the speaker is knowledgeable (Gunlogson 2008). To conclude this section, EDQs can felicitously target both rhetorical relations and stance, and so can SWQs. While we find examples in the literature of the former, this function of SWQs has received no attention yet. The fact that EDQs and SWQs can echo these two aspects is compatible with previous accounts of echoing that assume that reformulations, which spell out implicit aspects of the previous utterance, are also cases of echoing (cf. Noh 1998: 609; Iwata 2003).

5  Metalinguistic and Meta-​Conversational Echo Echoing according to the standard view involves propositions that directly involve the content or form of what has been said in the conversation, that is, the linguistic object. However, I argue here that echoing can also target propositions at a more abstract level. Such echoes are not directly associated with the linguistic object, but are about the move made in the discourse context. This underlying idea is endorsed by, for example, Heim (2019): speakers can engage their addressees to resolve issues “at the propositional level or at the SA [speech act] level”. As mentioned in Section 2.1, I follow Gunlogson (2008) and Geurts (2019) in arguing that the notion of commitment does not merely involve commitment to the propositional content of an utterance. Rather, with every utterance they make, speakers implicitly commit to the discursive appropriateness of the move that their utterance brings about, as well as to the appropriateness of the form of the utterance. As a result, speakers can be held accountable for these implicit commitments, as they are for the propositional content they commit to. Consider (28a.), which is an example of what I call a metalinguistic echo, metarepresenting the semantic content of the previous utterance. B’s utterances have a direct relation to the linguistic object, that is, to A’s utterance. (28b.), on

59

Whatever Can Be Meant Can Be Echoed 95 the other hand, does not relate directly to the linguistic object but rather to the function that utterance has in the discourse, which is to express a threat. For that reason, I call utterances like (28b.B2) meta-​conversational. (28) A: I’ll call the police. a. B1: You’ll call the police? B2: Since when will you call the police? b. B1: You’re threatening me? B2: Since when are you threatening me? A discourse consists of a sequence of moves: each utterance brings about changes in the interlocutors’ commitments (cf. Stalnaker 1978; Gunlogson 2003, 2008; Farkas & Bruce 2010; Geurts 2019; Heim 2019). The essence of a discourse move can be implicit as in (28a.), but according to Searle’s principle of expressibility, any implicit move can also be said explicitly (“whatever can be meant can be said”, Searle 1969: 19), as shown in (28b.). Meta-​conversational echoes metarepresent moves made in the discourse, whether they are ‘said’ or just ‘meant’, and they are not directly tied in with the content or form of the echoed utterance itself. As another example, one may consider assertions. A  canonical assertion commits the speaker to the truth of the proposition (speaker commitment) and invites the other interlocutors to commit themselves to it, too (invitation to grounding; cf. Stalnaker 1978; Clark 1996). An assertion places a proposition on the ‘Table’ where it can be grounded, rejected, or discussed further if necessary (cf. Farkas & Bruce 2010; Heim 2019). These two operations are meta-​ conversational, because they relate to types of changes made to the common ground. Consequently, these two operations can be felicitously echoed by both EDQs and SWQs, as shown in (29). (29) A: John has stopped eating meat. a. echoing speaker commitment B1: You believe that? B2: Since when do you believe that? b. echoing invitation to commitment B1: You expect me to believe that? B2: Since when do you expect me to believe that? B1’s reaction in (29a.) metarepresents A’s speaker commitment; in (29b.) it metarepresents the invitation to grounding, that is, A’s call on B to believe that John has stopped eating meat. Note that (29b.B2) could felicitously echo any assertion, as a purely meta-​conversational echo is not sensitive to the semantic content of the echoed utterance. Example (30) shows that the utterance act can be targeted by both metalinguistic and meta-​conversational echoes.

69

96 Kiss (30) A: I’ll call the pólice. a. metalinguistic echo B1: You’ll call the pólice? You’ll call the políce! B2: Since when will you call the pólice? You’ll call the políce! b. meta-​conversational echo B1: You stress the first syllable? Stress the second one. B2: Since when do you stress the first syllable? Stress the second one. Metalinguistic echoes metarepresent the form-​ related feature: the meta­ representation contains a mention or pure quotation of the echoed aspect. In (30aB1), the metalinguistic echo has the form of a pure mention. The meta-​ conversational echoes in (30b.) are reformulations that may or may not involve pure mentions. Note that Noh’s examples of what she calls “reformulations” are both metalinguistic and meta-​conversational. In (31), the matrix clauses in B’s utterances metarepresent the move expressed by the command or advice in A (you want me to P, you think I should P, you’re telling me to P). At the same time, linguistic material from the previous utterance is metarepresented as well, which in this case is the property talk to a fortune-​teller.4 (31) A: Talk to a fortune-​teller. B1: You want me to talk to a fortune-​teller? B2: You think I should talk to a fortune-​teller? B3: You’re telling me to talk to a fortune-​teller?

(Noh 1998: (4))

This distinction is relevant when one seeks to visualize the fact that echoing the illocutionary act differs crucially from echoing the propositional act or the utterance act. SWQs can be metalinguistic or meta-​conversational echoes of propositional or utterance acts, but illocutionary acts seem to require a meta-​conversational component, as in (31). One can argue that a verbatim echo of the previous utterance could count as a metalinguistic echo and a metarepresentation of the illocutionary act at the same time. In this view, (32B1) would count both as a metalinguistic and a meta-​conversational echo, metarepresenting not just the form and content of the preceding utterance but also its illocutionary act. (32) A: Sorry. B1: Sorry? B2: Since when *(are you) sorry? This is possible, as illocutionary acts are sometimes determined by their linguistic form (clause type and intonation), although it is well known that there is not a one-​to-​one relation between form and illocutionary act (cf. Heim 2019).

79

Whatever Can Be Meant Can Be Echoed 97 However, (32B2) seems to target not the illocutionary act but the proposition that ‘A is sorry’. This is due to the obligatory reformulation of the preceding utterance into a declarative clause in order to transform it into a SWQ. Although (32B1) is a literal echo, the SWQ is formally a wh-​question with a verbal predicate and a tensed subject, in the order required by wh-​questions in English. This is also the intuition that underlies the contrast in interpretation between B2’s reactions in (33a.) and (33b.): unless the illocutionary act is made explicit (e.g. by a performative verb) in the SWQ, it is not easily understood as targeting the speech act itself. (33) A: You are now husband and wife. a. metalinguistic echo B1: We’re husband and wife? B2: Since when are we husband and wife? b. meta-​conversational echo B1: You’re marrying us? B2: Since when are you marrying us? What B2 in (33a.) conveys is that ‘we’re not husband and wife’, which is a dissociation from a state of affairs brought about by A’s declaration. B2 in (33b), on the other hand, is a dissociation from A’s illocutionary act, disapproving of it for example on the grounds of not having checked the preparatory conditions (Searle 1969; Searle & Vanderveken 1985). In sum, distinguishing metalinguistic and meta-​conversational echoes lets us see a difference in the echoing potentials of EDQs and SWQs. While EDQs can arguably be metalinguistic or meta-​conversational echoes of all three kinds of acts embodied by the previous speech act, SWQs cannot: they can always serve as meta-​conversational echoes, but not always as metalinguistic ones (cf. illocutionary acts).

6  Summary and Conclusion The present study showed that any aspect of a preceding utterance (or a salient phenomenon in the immediate context) can be targeted by echoing utterances. In addition to utterance types widely discussed in the literature, such as echo declarative questions (EDQs), I have demonstrated that since when questions may function the same way. Both can echo all three kinds of acts Searle (1969) identified in a speech act: the propositional act, the utterance act, and the illocutionary act. However, echoing is not restricted to these three types: discourse pragmatic aspects such as rhetorical relations and stancetaking are also suitable targets for echoes. The two question types, EDQs and SWQs, differ with regard to the range of speaker attitudes they permit: EDQs seem to pose no restrictions on the speaker’s stance toward the aspect being echoed, whereas SWQs strongly favor a negative affective or epistemic stance. This property may be tied to

89

98 Kiss the semantics of the two question types: answering a wh-​question requires more effort from the addressee than answering a polar question, and, therefore, questions that only require a choice between two alternatives may be considered less face-​threatening and hence more polite (cf. Levinson 2012). Lastly, the different syntactic forms of the two question types give rise to a further asymmetry. Well-​ known examples from the literature show that EDQs in English can echo any speech act by adding a sentence-​final rise. SWQs, however, are more restricted in this respect as well. Their form as wh-​ questions does not allow for verbatim echoes of interrogative or imperative clauses. Therefore, I have made a distinction between metalinguistic and meta-​ conversational echoes, with the latter being an echo of the previous speaker’s commitment to the (appropriateness of the) latest discourse move. Both EDQs and SWQs can be meta-​conversational echoes, but only EDQs lend themselves freely to a metalinguistic echoing use. In a nutshell, the present study concludes that whatever can be meant can be echoed. If it is true that “whatever can be meant can be said” (cf. Searle 1969) and whatever can be said can be echoed, it follows that whatever can be meant can be echoed. This seems to be true for both EDQs and SWQs, even though the latter are more restricted in their potential to echo. Looking at meta-​ conversational echoes can be beneficial in a two-​ way learning process for pragmaticists. First, given what we know about common ground management, we can learn about SWQs by examining the contexts they feature in. Second, assuming that whatever can be meant can be echoed, we can, by meta-​conversational echoes, reveal the subtle and implicit contextual changes a certain speech act brings about in a discourse. Learning about the context-​change potential of echoing utterances in general would bear important implications to commitment-​based models of discourse (cf.  Farkas & Bruce 2010; Malamud & Stephenson 2015; Farkas & Roelofsen 2017; Heim 2019). Commitments are typically defined as a speaker’s publicized propositional attitudes, a notion that has been challenged here by widening it to include commitment to the appropriateness of the utterance act and the illocutionary act. The present chapter did not present an analysis of EDQs or SWQs in a formal framework, as the intuitions presented here first need empirical support gathered in a systematic way. Corpus studies and psycholinguistic and phonetic experiments examining intonation contours in these questions are some ways that have already been explored (cf. e.g. Bartels 1999; Jeong 2018; Heim 2019). The aim of this work is to serve as a further point of reference for future work in this vein.

Notes 1 I thank Ruth Maddeaux, Dan Milway, Andrew Peters, and Heather Stephens for providing judgments, as well as the audience of AMPRA-4 and the reviewers of this chapter for their helpful comments. 2 As for the relation between verbatim echoes and quotes, the reader is referred to Noh (2000) and Bublitz (2015).

9

Whatever Can Be Meant Can Be Echoed 99 3 In some cases, the pronoun used in a verbatim echo may also change due to context-​ shifting (a change in speaker–​addressee roles) in echo questions. Have you read “Great Expectations”? can be echoed as having either the 1st or 2nd person singular pronoun (Have I read…?/​Have you read…?). Here I treat both as literal/​verbatim echoes. 4 I here adhere to Portner’s (2018) dynamic pragmatic model of speech acts, where commands assign properties (P) to an interlocutor’s individual to-​do lists.

References Asher, Nicholas and Alex Lascarides. 2003. Logics of Conversation. Cambridge: Cambridge University Press. Bartels, Christine. 1999. The Intonation of English Statements and Questions. A Compositional Interpretation. New York/​London: Garland Publishing. Blakemore, Diane. 1994. Echo questions: A pragmatic account. Lingua, 94: 197–​211. doi: 10.1016/​0024-​3841(94)90009-​4 Brown, Penelope and Stephen C. Levinson. 1987. Politeness. Some Universals in Language Usage. Cambridge: Cambridge University Press. Bublitz, Wolfram. 2015. Introducing quoting as a ubiquitous meta-​ communicative act. In: Arendholz, Jenny, Wolfram Bublitz and Monika Kirner-​Ludwig (eds.), The Pragmatics of Quoting Now and Then. Berlin: De Gruyter. 1–​28. Cheung, Lawrence Y. 2009. Negative wh-​construction and its semantic properties. Journal of East Asian Linguistics, 18: 297–​321. doi: 10.1007/​s10831-​009-​9051-​2 Clark, Herbert H. 1996. Using Language. Cambridge: Cambridge University Press. Du Bois, John W. 2007. The stance triangle. In: Englebretson, Robert (ed.), Stancetaking in Discourse. Subjectivity, Evaluation, Interaction. Amsterdam/​Philadelphia, PA: Benjamins. 139–​182. Farkas, Donka F. and Kim Bruce. 2010. On reactions to assertions and polar questions. Journal of Semantics, 27: 81–​118. doi: 10.1093/​jos/​ffp010 Farkas, Donka F. and Floris F. Roelofsen. 2017. Division of labor in the interpretation of declaratives and interrogatives. Journal of Semantics, 34: 237–​289. doi: 10.1093/​ jos/​ffw012 Geurts, Bart. 2019. Communication as commitment sharing: Speech acts, implicatures, common ground. Theoretical Linguistics, 45(1): 1–​30. doi: 10.1515/​tl-​2019-​0001 Gunlogson, Christine. 2003. True to Form: Rising and Falling Declaratives as Questions in English. New York: Routledge. Gunlogson, Christine. 2008. A question of commitment. Belgian Journal of Linguistics, 22: 101–​136. doi: 10.1075/​bjl.22.06gun Gyuris, Beáta. 2019. Thoughts on the semantics and pragmatics of rising declaratives in English and of their Hungarian counterparts. In: Gyuris, Beáta, Katalin Mády and Gábor Recski (eds.), K + K = 120. Papers Dedicated to László Kálmán and András Kornai on the Occasion of Their 60th Birthdays. Budapest: Research Institute for Linguistics, Hungarian Academy of Sciences. 247–​280. Heim, Johannes. 2019. Commitment and engagement: The role of intonation in deriving speech acts. Doctoral thesis, University of British Columbia, Vancouver, Canada. Horn, Laurence R. 1985. Metalinguistic negation and pragmatic ambiguity. Language, 61: 121–​174. Horn, Laurence R. 1989. A Natural History of Negation. Chicago: University of Chicago Press.

01

100 Kiss Iwata, Seizi. 2003. Echo questions are interrogatives? Another version of a metarepresentational analysis. Linguistics and Philosophy, 26: 185–​254. Jaffe, Alexandra. 2009. Introduction. The sociolinguistics of stance. In: Jaffe, Alexandra (ed.), Stance. Sociolinguistic Perspectives. Oxford: Oxford University Press. 1–​29. Jeong, Sunwoo. 2018. Intonation and sentence type conventions: Two types of rising declaratives. Journal of Semantics, 35(2): 305–​356. doi: 10.1093/​semant/​ffy001 Kiesling, Scott F. 2009. Style as stance. In: Jaffe, Alexandra (ed.), Stance: Sociolinguistic Perspectives. Oxford: Oxford University Press. 171–​194. Kiss, Angelika. 2017. Meta-​ conversational since when-​ questions and the common ground. In: Petukhova, Volha and Ye Tian (eds.), Proceedings of the 21st Workshop on the Semantics and Pragmatics of Dialogue. Saarbrücken: Universität des Saarlandes.  38–​47. Levinson, Stephen C. 2012. Interrogative intimations: On a possible social economics of interrogatives. In: De Ruiter, Jan P. (ed.), Questions: Formal, Functional and Interactional Perspectives. Cambridge: Cambridge University Press. 11–​32. Malamud, Sophia and Stephenson, Tamina. 2015. Three ways to avoid commitment: Declarative force modifiers on the conversational scoreboard. Journal of Semantics, 32(2): 275–​311. doi: 10.1093/​jos/​ffu002 Noh, Eun-​Ju. 1998. Echo questions: Metarepresentation and pragmatic enrichment. Linguistics and Philosophy, 21: 603–​628. Noh, Eun-​Ju. 2000. Metarepresentation. A Relevance-​Theory Approach. Amsterdam/​ Philadelphia, PA: Benjamins. Ochs, Elinor and Bambi Schieffelin. 1989. Language has a heart. Text 9(1):  7–​25. Pierrehumbert, Janet and Julia Hirschberg. 1990. The meaning of intonational contours in discourse. In: Cohen, Philip R., C. Jerry Morgan and Martha E. Pollack (eds.), Intentions in Communication. Cambridge, MA: Bradford Books, MIT Press. 271–​311. Portner, Paul. 2018. Commitment to priorities. In: Fogal, Daniel, Daniel W. Harris and Matt Moss (eds.), New Work on Speech Acts. Oxford: Oxford University Press. 296–​316. Poschmann, Claudia. 2008. All declarative questions are attributive? Belgian Journal of Linguistics, 22: 247–​269. doi: 10.1075/​bjl.22.12pos Schlöder, Julian, Ellen Breitholtz and Raquel Fernández. 2016. Why? In: Hunter, Julie, Mandy Simons and Matthew Stone (eds.), Proceedings of the 20th Workshop on the Semantics and Pragmatics of Dialogue. New Brunswick, NJ: Rutgers University. 5–​14. Searle, John R. 1969. Speech Acts. Oxford: Oxford University Press. Searle, John R. and Daniel Vanderveken. 1985. Foundations of Illocutionary Logic. Cambridge: Cambridge University Press. Stalnaker, Robert. 1978. Assertion. Syntax and Semantics, Volume 9, New  York: Academic Press. 315–​332. Van der Sandt, Rob A. 1991. Denial. In: Dobrin, Lise M., Lynn Nichols and Rosa M. Rodriguez (eds.), Papers from the 27th Regional Meeting of the Chicago Linguistic Society, Vol. 2: The Parasession on Negation. Chicago: Chicago Linguistic Society. 331–​344.

1 0

6  The Pragmatics of Japanese Quotative Constructions A Comparative Study between Japanese and Japanese Heritage Language Families Kiyono Fujinaga-​Gordon

1 Introduction1 According to the American Community Survey, 21.5 % of the population of children in the United States aged 5 to 7 speak a language other than English at home (U.S. Census Bureau, 2012–​2016 American Community Survey 5-​Year). In immigrant communities, heritage language acquisition is characterized “by normal exposure to the parental (i.e. heritage) language in the first few years of life, followed by an abrupt shift to the majority language when formal schooling begins” (O’Grady et al. 2011). While the majority of heritage studies have focused on languages such as Spanish or Chinese, both of which have large communities in the United States, research on Japanese heritage families is so far limited to the work of Kondo-​ Brown (2005). Kondo-​Brown’s research focuses on Hawaiian Japanese communities where most of the current generation of heritage speakers use English at home. The present study focuses on Japanese heritage children’s (ages 0 to 12) language development in upstate New York, where there are no large Japanese immigrant communities as in Hawaii or California. These children are second generation, experiencing a strong Japanese language input from their parents. Moreover, research on heritage language acquisition tends to focus on the macro-​social aspects of language use such as language attitudes or ideology, neglecting grammar acquisition. My research is based on the premise that linguistic constructions encode cultural conventions (Goldberg 2006; Kay & Fillmore 1999). Thus, I  focus on the choice of linguistic constructions encoding certain interactional meanings, the use of quotative constructions in Japanese. It is worth examining the Japanese quotative constructions because the pragmatic functions of quotatives have been observed and discussed cross-​ linguistically (Okamoto & Ono 2008; Buchstaller & D’Arcy 2009; Fujii 2013) and because this study will be an important step towards understanding the grammar acquisition in relation to pragmatic acquisition.

2 0 1

102 Fujinaga-Gordon By comparing Japanese heritage children in the United States and Japanese monolingual children in Japan with regard to their input and output of quotative constructions, the present study investigates whether the children in the two groups (henceforth: heritage vs Japanese) show differences in the use of quotative constructions in family discourse between caregivers and children. The present chapter is structured as follows: Section 2 reviews previous studies regarding Japanese quotative constructions, the acquisition of quotative constructions, and Schieffelin & Ochs’s Language Socialization Theory (1986). Section 3 introduces the design of the study and the participants’ background. Section 4 presents the quantitative and qualitative observations of Japanese quotative constructions used at home, and discusses relevant findings.

2  A Review of Previous Studies 2.1  Japanese Quotative Constructions In Japanese, the quotative particle to (or its colloquial variant tte ) marks a quoted clause. Example (1) demonstrates that, even though direct and indirect quotations are both indicated by the quotative marker to, they are not formally distinguishable in Japanese. (1)

Direct versus indirect quotative constructions a. Midori wa  “Biru  wa manekareteimasen”  to itta. TOP  TOP  is.not.invited.POL QUO said. ‘Midori said, “Bill is not invited.” ’ b. Midori wa Biru wa manekareteinai to itta. TOP TOP is.not.invited QUO said. ‘Midori said that Bill was [Lit. is] not invited.’ (Hasegawa 2015: 345)

(1a) is a direct quotation and (1b) is an indirect quotation: we know so because (1a) contains the addressee-​ oriented polite verbal form manekareteimasen, whereas (1b) does not contain such an addressee-​oriented linguistic form. In (1a), Bill said manekareteimasen with these exact words to Midori and Midori reported that directly to the speaker. (1b), on the other hand, represents Midori’s summary of what Bill said to her about Bill not being invited. The quoted sentence is “typically followed by an information-​processing verb having to do with saying, writing, thinking, judging, or the like” (Martin 1975: 996). While Japanese quotatives can be used with verbs of information-​ processing as is the case in most other languages, Japanese quotative constructions are particular in that a quoted clause may also occur with other kinds of verbs, as in (2):

3 0 1

Japanese Quotative Constructions 103 (2) Quotative constructions without verbs of saying/​thinking a. Midori wa  bakabakashii  to tachiagatta. TOP ridiculous  QUO stood.up ‘Midori stood up, (saying) “(that’s) ridiculous.” ’ b. Midori wa  shinjirarenai  to  watashi  o  mitsumeta. TOP  cannot.believe  QUO  I  ACC  gazed. ‘Midori gazed at me, (thinking that what I had just said) was unbelievable.’ c. Midori wa  kaette  kudasai  to  doa  o  shimeta. TOP  go.home  please  QUO  door  ACC  closed. ‘Midori said “Please go home” and shut the door.’ (Hasegawa 2015: 347) All the examples in (2) demonstrate that Japanese quotative constructions can be made with verbs such as stand up, gaze, and shut (the door) with the guide of pragmatics. Although the examples in (2) do not contain information-​ processing verbs, we can still understand each phrase –​bakabakashii ‘ridiculous’, kaette kudasai ‘please go home’, and shinjirarenai ‘unbelievable’  –​ as quoted elements. Hasegawa (2015) accounts for such variations of quotative constructions as implicatures: the quotative particle to is so strongly tied to a verb of saying/​thinking that the saying/​thinking part can conventionally be implied without mentioning it. I  argue further that the quoted phrases  –​ bakabakashii ‘ridiculous’, kaette kudasai ‘please go home’, and shinjirarenai ‘unbelievable’ –​can also be understood as adverbials, where they are not exact quotations, but rather an approximation of the inner speech that co-​occurs with the event denoted by the main verbs. Because of the peculiar nature of Japanese quotative construction, previous studies have argued about the nature of to-​marked clauses. While the majority take the stance of treating to-​marked clauses as a complement of the main verb, Fujita (2000) claims that they are adverbial. Fujii (2013) treats the following quotative constructions (3) to (6) as a family of polysemous constructions. She identifies (3) as the complement of the main verb; (4) and (5) as evoked by the noun in the main clauses –​namida ‘tear’ and tegami ‘letter’ respectively–​; and (6) as a clause-​external adverbial. (3) Ichiroo wa [wakarimashita]       to    it-​ ta.      TOP [(he).has.understood.(it)]   QUO   say-​ PAST ‘Ichiro said that (he) has understood (it).’ (4) Ichiroo wa   [wakatta]       to  namida o    nagashi-​ ta.    TOP  [(he).has.understood.(it)] QUO tear  ACC shed-​ PAST ‘Ichiro cried (feeling/​saying) that (he) has

4 0 1

104 Fujinaga-Gordon (5) Ichiroo wa  [wakatta]       to  tegami o       TOP  [(he).has.understood.(it)] QUO letter  ACC  ‘Ichiro sent a letter (saying) that (he) has understood (it).’

okut-​ ta. send-​PAST

(6) Ichiroo wa  [wakatta]  to    tobide-​te    it-​ta   TOP  [(he).has.understood.(it)]  QUO  jump.out-​te  go-​past ‘Ichiro jumped out (of the room) (saying) that (he) has understood (it).’ (Fujii, 2013: 295–​296) Suzuki (2007) discusses the process of (inter)subjectification of quotative construction with tte, following Hopper and Traugott’s (2003) definition of subjectification and intersubjectification, which refers to a mechanism whereby meanings come over time to encode or externalize the SP/​W1’s perspectives and attitudes as constrained by the communicative world of the speech event, rather than by the so-​called ‘real-​world’ characteristics of the event or situation referred to. (Hopper & Traugott 2003 : 126) Hopper and Traugott (2003 ) define intersubjectification as “the semasiological process whereby meanings come over time to encode or externalise implicatures regarding speaker/​writer’s attention to the ‘self’ of addressee/​ reader in both an epistemic and a social sense” (Hopper & Traugott 2003: 129f.). While both subjectification and intersubjectification are driven by the speaker’s communicative needs, the crucial difference between subjectification and intersubjectification is the speaker’s awareness of the addressee. First, we can observe that the quality of the quoted element has diachronically become less and less utterance-​like. For instance, in type (i) tte-​marked clauses began as utterances, and later started to include thoughts or sounds. In type (ii), tte-​ marked clauses were originally used to introduce an entity or situation rather than pure utterances. This strongly indicates the subjectifiation of quotatives, where a speaker presents an impression of an event, rather than reproducing someone else’s utterance or sounds. The main verb associated with the quotative tte is no longer limited to verbs of saying or thinking as we observed in the case of to in (2). Moreover, quotative constructions in Japanese may occur without a main verb. In such cases, the Japanese quotative marker tte occurs utterance-​finally and the whole utterance with tte at the end can indicate a joking tone (cf. Suzuki 2007). Furthermore, I  argue that the phrase quoted before the quotative particle tte often refers to collective viewpoints about a certain situation, namely conventionally experienced feelings that others in the community may also share. This is what makes this joking tte intersubjective. Suzuki establishes four types of usage in regard to the Japanese quotative particle tte: Type (i): We find tte immediately at the end of an utterance (later including thought/​ sound). Currently this is the typical function of tte

5 0 1

Japanese Quotative Constructions 105 known as a quotative complementizer, often followed by the verb iu insert Japanese ‘say’. Type (ii): tte also occurs immediately before or after a nominal, introducing a label of an entity/​situation or describing an unfamiliar entity/​ situation. Type (iii): Recently, we find tokens of tte occurring in final position in an utterance. There is no “main” verb associated with the tte clause in this case. Utterance-​finally, tte expresses a variety of pragmatic information, such as ‘I heard that…’ ‘I’ am telling you’. ‘Just kidding’ and so forth. Type (iv): We also find tte as part of a connective phrase of repair, tte iu ka in present-​day Japanese’. We find it in utterance-​internal, final, as well as initial position. (Suzuki 2007: 209–​210) Japanese quotative constructions have emerged from Type (i) >(ii) >(iii) >(iv). The quotation element in Type (i) always accompanies a main verb, typically a verb of saying/​thinking. However, later in the development in Type (iv), the quotation element no longer requires a main verb and becomes an independent clause on its own. Suzuki also points out that, although the order of emergence is from Type (i) >(ii) >(iii) >(iv), the discourse-​motivated development is not from Type (i) >(ii) >(iii) >(iv), but rather Type (i) provides the functional and structural basis for other variations. She argues that quotative constructions have two properties, “reproduction” and “distance” that contribute to the development of the variations, the latter being the abstract extension of the first property (2007 : 210). Type (i) inherently represents the “reproduction” properties, namely quoting someone else’s utterance as exactly as possible. The transition from Type (i) to Type (iii) involves the “distance” property is at work because citing somebody else’s utterance as evidential is a good way of the speaker distancing themself without making a commitment to the quoted contents (Suzuki 2007: 212). While Suzuki investigated adults’ conversation data, I focus on conversations between caregivers and children. Such contexts will be useful to observe how caregivers use quotative constructions to navigate/​help the children to act in socially desirable ways. 2.2  On the Acquisition of Quotative Markers and Language Socialization The quotative marker tte is one of the earliest grammatical forms to emerge in Japanese children’s language acquisition (Clancy 1985). Clancy speculates that the early acquisition of quotative constructions depends on the high frequency of quotative constructions in the mother’s speech. According to Clancy’s observations, Japanese mothers frequently use quotative constructions in order to draw attention to a sound or somebody else’s speech and tell the children

6 0 1

106 Fujinaga-Gordon what to say in response to such stimuli. Interestingly, Clancy observed that Japanese children (aged 1;11 to 2;11)2 acquire the sentence-​final tte quotative constructions (cf. (7)) earlier than the full-​sentence counterparts with the verbs of saying and thinking (cf. (1)). (7)

Sentence-​final quotative construction Arigatoo tte. thank.you QUO ‘Thank you’ (Clancy 1990: 436)

Clancy argues that example (7) can be interpreted as the mother prompting the child to say “thank you” to a third party, or repeating somebody else’s utterance, or interpreting somebody else’s actions as gratitude and communicating that to her child. As the early acquisition of quotative constructions in Japanese seems to have both a linguistic and a cultural basis, in that they are frequently found in child-​directed speech and the Japanese cultural code is to conform to others’ needs, this study applies the Language Socialization Theory (Schieffelin & Ochs 1986). Language Socialization asserts that: (i) the older members in the community socialize the novice into the community practices by using certain discursive linguistic patterns; and (ii) novice members in the community socialize into their culture by using the language. The mastery of particular language expressions demonstrates the social competence of the speaker. In terms of (i), Suzuki (1999) looked at the use of the so-​called aspectual suffix -​chau used by a Japanese mother in a recorded interaction with her two-​year-​old son. The suffix -​chau originates from the lexical verb shimau ‘to put away’ but today is used in its grammaticalized shape, i.e. as an aspectual marker as well as a pragmatic marker expressing a sense of regret. For instance, Example (8) demonstrates the suffix use of -​chau in the mother-​child interaction, where the mother is calling an affective attention from the child because the child is banging a toy against the floor. (8)

hora hora koware-​chau yo. look look break-​chau SF ‘Look, look, (the phone) will break-​chau.’

Suzuki argues that Japanese caregivers’ use of -​chau plays an important role as an “affective marker” when expressing their own affect and when attributing affect to children or third parties in order to instruct children to pay attention in emotionally and socially significant situations. Similarly, Burdelski (2015) examined the use of quotatives by Japanese mothers in caregiver-​child triadic interactions and claims that Japanese mothers use quotatives to help children navigate interpreting the communicative behavior of others. Clancy (1990) discusses the use of Japanese mothers’ monologues along with such use of quotatives as one of the implicit ways of Japanese mothers fostering

7 0 1

Japanese Quotative Constructions 107 Japanese children into the Japanese “mind-​reading” style of communication. She contends that young Japanese children, even at the age of two, seem to be aware –​through their mothers’ indirect mode of teaching –​of what is socially preferred. For instance, young Japanese children seem to understand that their mothers’ monologues must be interpreted as requests rather than undirected speech, which is why the children learn to respond accordingly. This relates to both assertions of the Language Socialization Theory: First, the caregiver (mother) prompts the child to pick up certain behaviors that commonly occur in society; secondly, the child’s ability to comply with their mother’s prompt indicates their social competence. As for assertion (ii) of the theory, Platt (1986) observed Samoan children’s use of imperatives with certain motion verbs. In traditional Samoan society, people were stratified into a strict hierarchy where people of higher status are associated with less motion-​related activities whereas people in lower status are associated with higher motion-​related activities such as coming or bringing. Therefore, the proper use of motion verbs in relation to speaker and addressee when making a request is required. Platt found that children aged two would already take into account the addressee’s social status when using motion verbs and interpreted this as evidence of the children’s social awareness and pragmatic competence. Similarly, Brown (2007) observed that, in Tzeltal, a Mesoamerican language, children would use the benefactive construction. He found that Tzeltal children at the age of two already knew culture-​specific ways of using this construction: someone does something to some other thing to help/​hurt someone else such as yixlanbet laso ‘He played with (your) rope for/​on you’ or lo’ben-​tomut antun o ‘Antun ate me (my) egg.’ Brown argues that the benefactive constructions bear culture-​specific meanings in the sense of what constitutes things one would want to happen as opposed to things that are not desirable. What is amazing about Tzeltal children is that they express these benefactive meanings in spite of their morpho-​syntactic complexity. To my knowledge, the majority of research into Japanese language use and acquisition in connection with Language Socialization Theory has focused on point (i), i.e. on how caregivers teach children social norms and expectations through the use of language, rather than on (ii), i.e. how children produce linguistic expressions as indicators of their social competence. The fact that there has been a much stronger interest of research in (i) than (ii) results in the trends in previous studies to treat Japanese children as passive learners who merely acquire knowledge and receive assistance from their caregivers.

3  The Design of the Study 3.1  Participants Participants’ names have been changed to pseudonyms for the publication of results. Their birth years and birth countries are indicated in parentheses. Four families in the United States and three families in Japan participated in

8 0 1

108 Fujinaga-Gordon Table 6.1 Summary of participant pool Family ID

Residence

Father

Mother

Chi 1

Chi 2

Chi 3

17I

US

17F

US US US

17A

JA JA

17S

JA

Mike (2013)/​US Maru (2014)/​JA Koharu (2016)/​JA Take (2015)/​JA

Will (2017)/​US

17O

Ohana (2008)/​JA Mizu (2006)/​JA Kuro (2013)/​US Maria (2009)/​US Shin (2012)/​JA Dai (2012)/​JA Yama (2013)/​JA

Chuhana (2012)/​JA Tori (2013)/​US

18H

Hanako (1979)/​JA Kate (1977)/​US Ao (1973)/​JA Chika (1977)/​JA Natsu (1989)/​JA Haru (1989)/​JA Aki (1989)/​JA

Kohana (2013)/​JA

17M

Taro (1979)/​JA Jiro (1974)/​JA Saburo (1971)/​JA John (1981)/​US Naoki (1987)/​JA Hayato (1986)/​JA Yuu (1990)/​JA

the study,3 which amounts to 29 participants taking part in this study (see Table 6.1). The 17I family moved to Buffalo, NY, from Japan in 2013, because the father, Taro, had been offered a job in the United States. They indicated that they were to stay in the United States only temporarily, but also that they were not planning to return to Japan. They had three daughters, the youngest of whom, Kohana, was born in Buffalo. The oldest daughter, Ohana, went to a local primary school in Buffalo and was fluent in both Japanese and English. Because Ohana was already 5 years old and had attended first grade in primary school in Japan before the 17I family came to the United States, I treat her data as a Japanese datapoint for this study in comparison to the heritage child of the same age, Maria, in the 18H family living in the United States. The middle daughter, Chuhana, had just started school at the time of data collection. She spoke fluent Japanese. Her English was not fluent so she was placed in the ESL class at school. Kohana only spoke Japanese then. In the 17F family, the father, Jiro, is Japanese and the mother, Kate, is American. They met in Japan when Kate was teaching English in Tokyo. They had their first daughter, in Japan but came back to the US as soon as Mizu was born. The second daughter, Tori, was born and grew up in the US. In the 17M family, the father, Saburo, and the mother, Ao, both originally came from Tokyo, Japan. They each came to the US for their doctoral studies and met there. After graduation, they did not go back to Japan, but instead got jobs and started a family: their son, Kuro, was born in the US. In the 18H family, everyone but the Japanese mother, Chika, was born and grew up in the US. Chika met her husband John in Japan and moved to the US after they got married. All participants in Japanese families in this study are Japanese monolingual speakers. They all live in Oita-​prefecture.

9 0 1

Japanese Quotative Constructions 109 3.2  Methodology Based on the Heritage Language Variation and Change Project (http://​ projects.chass.utoronto.ca/​ngn/​HLVC/​ 0_​0_​home.php), the participants were interviewed in order to elicit information about their biological and linguistic backgrounds, their language attitudes, and their socialization situations. In my study, the participants used their own smartphones to record their family interactions at home. The smartphones were placed in a location in each family’s house that the researcher and the families agreed on as being appropriate to film the family interactions. I  provided the families with selfie sticks so they could hold their smartphones stable during the recordings. Because my study focused on family interactional data in a natural setting, I was not present for the recordings. Every two weeks, the participants’ interactions were recorded for one hour and they sent their recordings to the researcher. The participants could choose what time of a certain day they wanted to film, while having to make sure that every family member was present at the house and that the recording could be done without any interruptions during the one hour designated for the recording. The participants were informed that they could drop out of the study at any time they wished and/​or were free to ask the researcher to delete the data they were not comfortable with. Overall, I  transcribed 10 hours of recordings from each family using ELAN (https://​tla.mpi.nl/​tools/​tla-​tools/​elan), obtaining a total of 70 hours of recordings. The transcribed data on ELAN were saved as text files, which I then searched for tokens of quotative constructions. These tokens were coded for: a) uses of tte (colloquial) or to (formal); b) who produced the utterance; c) how old the speaker was; d) the element preceding the quotative marker; e) the element following the quotative marker.

4  Results and Discussion Four adult speakers did not produce any quotative constructions at all. Additionally, two adult speakers, the mother, Kate, in the 17F family and the father, John, in the 18H family, were excluded from the analysis because they mainly spoke English and did not produce the target constructions. The father, Hayato, in the 17O family did not produce any quotative construction because he was not around during recordings because of his work commitments. Yuu, father in the 17S family, did not produce any constructions, because he did not speak much in general. Three children did not produce any quotative constructions either. Will, in the 18H family, was 0;9–​1;8 during the period of the recordings, and did not produce any words either in English or in Japanese because he was not yet in the phase of producing meaningful words. His older brother Mike, who was 3;4–​4;3 during the recordings, was dominant in English and rarely produced Japanese utterances. The youngest daughter in family 17F, Tori

0 1

110 Fujinaga-Gordon Table 6.2 Total number of occurrences of to/​tte-​marked quotative constructions Family ID

Residence

Father

Mother

Chi 1

Chi 2

Chi 3

17I

US

17F

US US US

17A

JA JA

17S

JA

to (0) tte (1) to (2) tte (17) to (0) tte (0) to (0) tte (2)

to (0) tte (0)

17O

to (2) tte (8) to (0) tte (2) to (5) tte (61) to (2) tte (13) to (3) tte (52) to (1) tte (43) to (4) tte (21)

to (2) tte (43) to (0) tte (0)

18H

to (9) tte (77) LI English Speaker to (34) tte (129) to (9) tte (92) to (15) tte (133) to (19) tte (100) to (2) tte (30)

to (0) tte (10)

17M

to (11) tte (71) to (9) tte (24) to (13) tte (80) LI English Speaker to (11) tte (56) to (0) tte (0) to (0) tte (0)

(4;11–​5;10), also did not produce any quotative constructions. Like Mike in family 18H, Tori spoke mostly in English. In the Japanese families, Koharu (1;6–​2;5) in family 17O did not produce any quotative constructions due to the fact that she was still at an early language developmental age. Table 6.2 displays the number of occurrences of to/​tte-​marked quotative constructions in my data. Other than those speakers who did not produce any quotative constructions, all other speakers with no exception produced more tte than to forms. It is also noteworthy that some of the youngest children participating in this study did not produce any quotative constructions with to, although they did produce a few tte-​marked quotative constructions. This makes sense because to is a formal variant while tte is a colloquial one, as previous studies pointed out (Hasegawa 2015); since the data is composed of conversations between family members, more colloquial and less formal speech styles are expected. Figures  6.1–​6.4 summarize what comes before [preceding] and follows [postcedent] each quotation marker to and tte. Note that what precedes the quotative marker is categorical, i.e. [clause], [mimetic]4 or [noun], while the quotative marker is usually followed by verbs, or in a few cases, adjectives. The results suggest that, while the heritage children used to-​marked quotative constructions both as a complement of the main verb as in the [clause] + the verbs of thinking (cf. (1)), and as a clause-​external adverbial as in [mimetic] + quotative constructions (cf. (5)), Japanese children preferred to-​ marked quotative constructions only as the complement of main verbs (cf. (6)). To-​marked constructions show strong preference for [clause] + omou (‘think’) patterns for all speaker categories, i.e. US caregivers, Japanese caregivers, US children, and Japanese children:

1

Japanese Quotative Constructions 111 to

US/​caregivers

JA/​caregivers

Preceding

Postcedent

Preceding

clause (78) 80%

omou ‘think’ (78) 79%

mimetic (12) 12% direct quotation (5) 5% noun (3) 3%

zero (9) 9% iu ‘say’ (4) 4%

clause (40) omou ‘think’ (40) 91% 91% noun (3) 7% zero (2) 5% mimetic iu ‘say’ (1) 2 % (1) 2% suru ‘do’ (1) 2%

Total (98) 100%

yomu ‘read’ (1) 1% afureru ‘overflow’ (1) 1% ireru ‘put’ (1) 1% kuttsukeru ‘glue’ (1) 1% suteru ‘throw away’ (1) 1% taberu ‘eat’ (1) 1% tsukeru ‘attach’ (1) 1% (98) 100%

(44) 100%

Postcedent

(44) 100%

Figure 6.1 Summary of to-​marked quotative constructions uttered by caregivers (9)

Chika: ano hito yaser-​eba mouchotto kakkoii to omou yo That person lose.weight-​ more handsome QUO think SF COND ‘(I) think (he) will be more handsome if he loses weight.’ (18H: 2018-​09-​01)

This is the most prototypical usage of quotative constructions as prescribed in previous studies where the to-​marked clause serves as the complement of the verbs of thinking (cf. (1) above; Hasegawa 2015). A relatively high frequency of mimetic words preceding to was observed in both US caregivers and children’s data but was absent in Japanese caregivers and children’s conversations. In (10) and (11), we see used to describe the manner of the action designated by the main verb rather than being the complement of the main verb. Hence, these to-​ marked clauses are adverbial in nature (cf. (11)). In Example (10), the mother, Hanako, is persuading her youngest daughter Kohana to eat some seaweed. Kohana is reluctant to eat the seaweed. (10) Hanako: hijiki tabete pakun to? eat mimetic  QUO ‘Eat up your seaweed.’ (17I: 2018-​08-​18)

2 1

112 Fujinaga-Gordon tte

US/​caregivers

JA/​caregivers

Preceding

Postcedent

Preceding

Postcedent

noun (168) 36% direct quotation (121) 27% clause (117) 26%

zero (221) 49% iu ‘say’ (142) 31%

clause (151) 48% noun (100) 32%

zero (175) 56% iu ‘say’ (87) 28%

kaku ‘write’ (46) 10% naru ‘become’ (9) 2% yaru ‘do’ (6) 1% suru ‘do’ (5) 1% omou ‘think’ (4) 0.8% shiru ‘know’ (3) 0.6% deru ‘exit’ (2) 0.4% naku ‘cry’ (2) 0.4% kikoeru ‘hear’ (2) 0.4% yomu ‘read’ (2) 0.4% arau ‘wash’ (1) 0.25% ase ga detekuru ‘sweat’ (1) 0.25% hagareru ‘fall off’ (1) 0.25% iku ‘go’ (1) 0.25%

mimetic (37) 12%

kaku ‘write’ (8) 3%

direct quotation (24)8%

suru ‘do’ (7) 2%

mimetic (46) 10% zero (3) 1%

akeru ‘open’ (2) 1% kiku ‘listen to’ (2) 1% ireru ‘put’ (2) 1% agaru ‘raise’ (1) 0.07% aru ‘be’ (1) 0.07% deru ‘exit’ (1) 0.07% hanasu ‘chat’ (1) 0.07% haru ‘attach’ (1) 0.07% kimeru ‘decide’ (1) 0.07% miru ‘look’ (1) 0.07%

kureru ‘give to me’ (1) 0.25% kau ‘buy’ (1) 0.25% mawaru ‘go around’ (1) 0.25% miru ‘look’ (1) 0.25% oriru ‘get off’ (1) 0.25% wakaru ‘understand’ (1) 0.25% yobu ‘call’ (1) 0.25% soudan suru ‘consult’ (1) 0.25%

Total

(455) 100%

(455) 100%

omou ‘think’ (6) 2% shiru ‘know’ (5) 2% kangaeru ‘wonder’ (4) 1% naru ‘become’ (4) 1% yaru ‘do’ (3) 1%

nakidasu ‘burst into tears’ (1) 0.07% nomu ‘drink’ (1) 0.07% suwaru ‘sit’ (1) 0.07% taberu ‘eat’ (1) 0.07% watasu ‘pass’ (1) 0.07%

(312) 100%

yomu ‘read’ (1) 0.07% yasashii ‘kind’ (1) 0.07% (312) 100%

Figure 6.2 Summary of tte-​marked quotative constructions uttered by caregivers

3 1

Japanese Quotative Constructions 113 to

US/​children Preceding

Postcedent

Preceding

Postcedent

clause (7) 64%

omou ‘think’ (7) 64% iu ‘say’ (1) 9% otosu ‘drop’ (1) 9% suru ‘do’ (1) 9% zero (1) 9% (11) 100%

clause (9) 90%

omou ‘think’ (4) 40% zero (4) 40% deru ‘exit’ (1) 10% suru ‘do’ (1) 10%

mimetic (4) 36%

Total

JA/​children

(11) 100%

mimetic (1) 10%

(10) 100%

(10) 100%

Figure 6.3 Summary of to-​marked quotative constructions uttered by children

The mimetic sound pakun emphasizes the sound of an object that has completely gone from inside a mouth. Hanako encourages her daughter Kohana to just eat up the seaweed without chewing it and hence tasting the flavor of it. The quotative with mimetic sound was used by the mother to encourage the child to empty her plate. Heritage children like Kuro (5;2) In 17M family used to-​marked quotative constructions with mimetic words. In (11), Kuro is reporting to his father that he saw Hotokeakabane (an animation character) for real in their garden. (11) Kuro: Neenee  papaa Kuro-​chan ne anone hiruma    asoko ni  ne hotokeakabane ga Hey  daddy Kuro-​dim SF like in.the.afternoon there LOC SF  NOM pyuun  to  ita mimetic  QUO was ‘Hey daddy, Kuro-​chan (saw) Hotokeakabene jumping about right there in the afternoon.’ (17F: 2018-​05-​24)

However, heritage children’s usage of mimetic quotative constructions is different from their caregivers in that they do not use them interpersonally as in (10). In (11), Kuro describes the way Hotokeakabane jumps by using mimetic + to-​marked quotative construction. Tte-​marked quotative constructions were predominantly used without accompanying main clauses. This tendency turned out to be true of all the speaker categories  –​US caregivers, Japanese caregivers, US children, and Japanese children. This suggests that the tte-​marked quotative construction has gone through a radical intersubjectification process as discussed in (Suzuki, 2007) and all the speakers are aware of the fact tte-​marked quotatives are used exclusively as clause-​end particles. Verbs of saying were preferred, following [zero]. This was consistent for all the speaker groups. In (12), the mother, Aki, in 17S family spoke on the phone

4 1

114 Fujinaga-Gordon tte

US/​children

JA/​children

Preceding

Postcedent

Preceding

Postcedent

direct quotation (55) 36% noun (51) 34% clause (24) 16% mimetic (20) 13%

zero (53) 35%

clause (40) 31%

zero (76) 60%

iu ‘say’ (50) 33% kaku ‘write’ (24) 16%

iu ‘say’ (33) 26% yaru ‘do’ (8) 6%

zero (2) 1%

yattsukeru ‘defeat’ (3) 2% dekiru ‘can’ (2) 1% shiru ‘know’ (2) 1%

noun (40) 31% mimetic (33) 25% direct quotation (13) 10% zero (1) 3%

(127) 100%

(127) 100%

yaru ‘do’ (5) 3%

suru ‘do’ (2) 1% ageru ‘give’ (1) 0.1% hayai ‘fast’ (1) 0.1% kuru ‘come’ (1) 0.1% naru ‘become’ (1) 0.1%

Total

(152) 100%

mawaru ‘go around’ (1) 0.1% ochiru ‘drop’ (1) 0.1% okoru ‘upset’ (1) 0.1% taberu ‘eat’ (1) 0.1% omou ‘think’ (1) 0.1% utau ‘sing’ (1) 0.1% (152) 100%

naru ‘become’ (3) 2% oreru ‘break’ (3) 2% suru ‘do’ (2) 2% jampu suru ‘jump’ (1) 0.28% kikoeru ‘hear’ (1) 0.28% kokeru ‘fall’ (1) 0.28% shinu ‘die’ (1) 0.28% tobu ‘fly’ (1) 0.28% wakaru ‘understand’ (1) 0.28% warau ‘laugh’ (1) 0.28%

Figure 6.4 Summary of tte-​marked quotative constructions uttered by children

with the father, Yuu. Aki told her two sons, Yama (4;4) and Take (2;2) that their father was on his way back home from work. Aki tells them that their father will scold them if they do not tidy up all their toys. (12) Aki:

taiheen kaettekuru tte. naosanaku-​chaa. gee come.home QUO put.away-​have.to ‘Gee, (he said) (he is) on his way. (We) have to put (these things) away!’ (17S:2017-​11-​19)

5 1

Japanese Quotative Constructions 115 This use of the quotative construction with tte is traditionally considered reportative or hearsay (Suzuki 2007). In our example, it is indeed reportative in that Aki conveys to the children what she her husband Yuu told her. However, the use of the quotative here is more pragmatically motivated than a report: the mother, Aki, uses the authoritative voice of the father, Yuu, to discipline the children. Japanese children in Japan also frequently used tte quotative constructions without main verbs. In (13), Maru (3;1) is asking her mother Natsu to take her picture with Natsu’s iPhone. However, Natsu is engaged in a conversation with her husband Naoki. Maru has been asking her mother repeatedly to take her picture, but they are not paying attention to her. (13) Maru:

ochachin tte. photo  QUO ‘(I am saying) (take my) photos.’ (17A: 2017-​11-​12)

Here, in a somewhat frustrated fashion, Maru quotes her own speech to her parents that they should pay attention to what she is saying. Next after the dominant use of quotative constructions marked by [zero], the speakers in all groups –​US caregivers, Japanese caregivers, US children, and Japanese children –​used the quotative constructions with the verb of saying. The heritage children group, however, unlike the other groups, showed no significant preference towards [zero] and the verb of saying. This indicates that Japanese heritage children do not use quotative constructions intersubjectively as in Type (iii) and Type (iv) in Suzuki (2007)’s classification, even though this is the most common usage among all the speakers –​US caregivers, Japanese caregivers, and Japanese heritage children. Example (14) shows how a Japanese child Dai (5;3) used the [direct quotation] + iu ‘say’ verb quotative construction as in (1a). (14) Dai: Koharu-​chan  mo  kakete tte yutteru yo chottodake. Koharu-​DIM too give.me  QUO  is.saying  SF a.bit ‘Koharu-​chan is saying “give me too” a little bit.’ (17O: 2017-​12-​24) In Example (9), the 17O family is having dinner together. The child Koharu (1;11) is too small to communicate with others verbally. Dai (5;3) noticed that his younger sister Koharu wanted to have more toppings on her rice from her gestures. Because she cannot produce a meaningful utterance, he translates her wishes to their mother using a quotative construction. Previous literature often discusses the role of Japanese mothers as a mediator between the children and others (Burdelski 2015), thus portraying mothers as primary agents in Language Socialization. However, by looking at family interactions closely,

6 1

116 Fujinaga-Gordon my study found many cases where children also initiate socialization with caregivers as in (14). Next down the list of frequency of use, the high frequency of the verb kaku ‘write’ following iquotative constructions is noticeable only for heritage family groups. This is probably because there are many interactions between the US caregivers and US children discussing the spelling of Japanese words. Japanese heritage children often have a hard time spelling. This quotative construction corresponds to Type (ii) in (Suzuki, 2007), where they are accompanied by nominal elements preceding a tte-​marked quotative as in (15). (15) Kuro: Kakatakana no “ji” tte douyatte kaku n da POSS QUO how write NMNL COP ‘How was I supposed to write Katakana “ji”?’

kke? Q

(17M: 2017-​01-​18) When the caregivers use tte-​marked quotative constructions with the verb kaku ‘write’, it is typically the case that either they are reading letters the children cannot read or instructing the children how to write a response properly. In (16), the father Jiro and the daughter Tori (5;2) are writing a birthday card to Tori’s teacher from Japanese weekend school. Because Tori is reluctant to write, Jiro is asking her just to sign her name. (16) Jiro:  ne, ne,  “Tori”  tte kai-​te kureru? SF SF QUO  write-​te  give ‘Please Please. Can you (at least) write “Tori” for me?’ (17F: 2018-​03-​03) When the Japanese heritage children use the tte-​marked quotative construction with verb kaku ‘write’, they are typically asking for spellings: (17) Kuro: Kakatakana no “ji” tte douyatte kaku n da kke? POSS QUO how write NMNL COP Q ‘How was I supposed to write Katakana “ji”?’ (17M: 2017-​01-​18)

Compared to [zero], the verb iu ‘say’, and kaku ‘write’, other types of verbs following the quotation marker in quotative constructions are very low in frequency. Example (18) shows a low frequency use of quotative construction with the verb yattsukeru ‘defeat’. (18)

Kuro: tabun ne waru kyooryuu o yattsukeru tameni doon tte maybe SF bad dinosaur ACC defeat in.order. to mimetic QUO baaan tte yattsukeru n janai? mimetic QUO defeat NMNL TAG ‘Maybe in order to defeat bad dinosaurs, (they go) , I  hink?’ (17M: 2018-​04-​28)

7 1

Japanese Quotative Constructions 117 The so-​called light verbs suru ‘do’ and naru ‘become’ were observed in all groups, but were not used frequently. Surprisingly, the verbs of thinking were rarely used with the tte-​marked quotative construction among all the groups (see Figures 6.2 and 6.4), although the to-​marked quotative construction was exclusively used with verbs of thinking as the main verb (cf. Figures 6.1 and 6.3). This suggests that to-​marked quotative constructions are more associated with internal thoughts while tte-​marked constructions are more associated with speech. Moreover, the tte-​marked construction seems to be pragmaticalized to present an authoritative voice as in (12) or approximate unfamiliar objects as in (17).

5  Conclusion Traditionally, studies in Language Socialization have regarded children as passive participants in society who mainly receive cultural lessons through language practice. My study, however, discusses cases where children also actively participate in socialization with the use of quotative constructions. The study found an overall tendency that to-​marked quotative constructions tend to be associated with internal thoughts while tte-​marked quotative constructions tend to be associated with speech. All speaker groups –​US caregivers, Japanese caregivers, US children, and Japanese children –​seem to be aware of such an overall distribution. However, there were some differences observed in heritage families, especially heritage children regarding the usage of quotative constructions. First, both US caregivers and US children frequently used the to-​marked quotative construction with mimetic expressions preceding it. This was not observed in the Japanese group. Although it is not clear why US groups produced such quotative constructions together with mimetic expressions, it was suggested that for US groups to-​marked quotative clauses exist both as the complement of a main verb such as omou ‘think’ as well as a clause-​external adverbial as observed in the combination of mimetic words. Second, while all the other speakers exclusively preferred tte-​marked quotative constructions without any main verbs, US heritage children did not show such a preference. I claimed that the tte-​marked quotative constructions without the main verb are the most pragmaticalized forms, communicating various interactional meanings. Although US heritage children produced a relatively high frequency of tokens such as [zero] quotative constructions (35%), the fact that they did not use this form exclusively like the other speakers might suggest that they are not aware of some interactional meanings strongly associated with the use of such quotative constructions. From the Language Socialization point of view, this difference could be interpreted as the heritage Japanese children’s lack of certain communicative competence indicated by the use of the utterance-​final quotative tte. Alternatively they might have used different linguistic strategies to achieve the same communicative goal. Future research is needed to investigate longitudinal development in variations between Japanese heritage children and Japanese children to see whether the difference observed in this study will be dismissed in the course of language development or remain.

8 1

118 Fujinaga-Gordon

Notes 1 I would like to thank to my participants whose family interactions I always had fun transcribing. Without their active participation and moral support, this study would not have been possible. This study was funded by the Department of Linguistics at the University at Buffalo (SUNY) and the Mark and Diamond Research Fund. 2 One year and eleven months to two years and eleven months old. 3 As of 2018 November. 4 Mimetics are sound-​symbolic words that are found in many languages but the Japanese language is particularly known for having a rich mimetic vocabulary (Akita & Usuki 2016: 246).

References Akita, Kimi and Takeshi Usuki. 2016. A constructional account of the ‘optional’ quotative marking on Japanese mimetics. Journal of Linguistics, 52(2): 245–​275. Brown, Penelope. 2007. Culture-​specific influences on semantic development: Acquiring the Tzeltal ‘benefactive’ construction. In: Barbara, Pfeiler (ed.), Learning Indigenous Languages: Child Language Acquisition in Mesoamerica. Berlin: Mouton de Gruyter. 119–​154. Buchstaller, Isabelle and Alexandra D’Arcy. 2009. Localised globalisation: A multi-​ local, multivariate investigation of quotative ‘be like’. Journal of Sociolinguistics, 13: 291–​331. Burdelski, Matthew. 2015. Reported speech as cultural gloss and directive: Socializing norms of speaking and acting in Japanese caregiver-​child triadic interaction. Text and Talk, 35(5): 575–​595. Clancy, Patricia M. 1985. The acquisition of Japanese. In: Slobin, Dan I. (ed.), The Crosslinguistic Study of Language Acquisition. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. 373–​524. Clancy, Patricia M. 1986. Acquiring communicative style in Japanese. In: Schieffelin, Bambi. B. and Elinor Ochs (eds.), Language Socialization Across Cultures. New York: Cambridge University Press. 213–​250. Clancy, Patricia. M. 1990. Acquiring communicative style in Japanese. In: Scarcella, R., E. Anderson, and S. Krashen (eds.), Developing Communicative Competence in a Second Language. New York: Newbury House Publishers. 27–​34. Fujii, Seiko. 2013. A corpus-​based analysis of adverbial uses of the quotative TO construction: Speech and thought representation without speech or thought predicates. In: Giriko, Mikio, Naonori Nagaya, Akiko Takemura and Timothy J. Vance (eds.), Japanese/​Korean Linguistics, Volume 22. Stanford, CA: CSLI Publication. 293–​305. Fujita, Yasuyuki. 2000. Kokugo in'yo kobun no kenkyu (A Study of Quotative Constructions). Osaka: Izumi Shuppan. Goldberg, Adele E. 2006. Constructions at Work: The Nature of Generalization in Language. Oxford: Oxford University Press. Hasegawa, Yoko. 2015. Japanese: A Linguistics Introduction. Cambridge: Cambridge University Press. Hopper, Paul J. and Elizabeth Closs Traugott. 2003. Grammaticalization. Cambridge: Cambridge University Press. Kay, Paul and Charles J. Fillmore. 1999. Grammatical constructions and linguistic generalizations: The What’s X doing Y? construction. Language, 75(1):  1–​33.

9 1

Japanese Quotative Constructions 119 Kondo-​Brown, Kimi. 2005. Differences in language skills: Heritage language learner subgroups and foreign language learners. The Modern Language Journal, 89(4): 563–​581. Martin, Samuel E. 1975. A Reference Grammar of Japanese. New Haven, CT: Yale University Press. O’Grady, William, Hye Young Kwak, On Soon Lee and Miseon Lee. 2011. An emergentist perspective on heritage language acquisition. Studies in Second Language Acquisition, 33(2): 223–​245. Okamoto, Shigeko and Tsuyoshi Ono. 2008. Quotative tte in Japanese: Its multifaceted functions and degrees of “subordination”. In: Laury, Ritva (ed.), Crosslinguistic Studies of Clause Combining: The Multifunctionality of Conjunctions. Amsterdam: Benjamins. 205–​230. Platt, Martha. 1986. Social norms and lexical acquisition: A study of deictic verbs in Samoan child language. In: Schieffelin, Bambi B. and Elinor Ochs (eds.), Language Socialization across Cultures. Cambridge: Cambridge University Press. 127–​151. Schieffelin, Bambi and Elinor Ochs. 1986. Language socialization. Annual Review of Anthropology, 15: 163–​191. Suzuki, Ryoko. 1999. Language Socialization through morphology: The affective suffix -​CHAU in Japanese. Journal of Pragmatics, 31(11): 1423–​1441. Suzuki, Ryoko. 2007. (Inter)subjectification in the quotative tte in Japanese conversation. Journal of Historical Pragmatics, 8(2): 207–​237. U.S. Census Bureau. 2018. Retrieved from https://​data.census.gov/​cedsci/​table?y=2018 &d=ACS%205-​Year%20Estimates%20Data%20Profiles&tid=ACSDP5Y2018. DP02&hidePreview=true

0 2 1

7  Managing Epistemic Asymmetry through Dialogic Resonance in Therapy Interactions Rong Lei

1 Introduction1 Epistemics has been referred to as “the knowledge claims that interactants assert, contest, and defend in and through turns at talk and sequences of interaction” (Heritage 2013: 370). The coordination of epistemics and the management of epistemic asymmetry are ubiquitous in human communication (Heritage 2012). Therapy interaction involves the negotiation of “epistemics of experience” and “epistemics of expertise” (Ekberg & LeCouteur 2015: 22), and while the former is concerned with the personal experiences and feelings of clients, the latter focuses on the medical and professional knowledge of therapists. Epistemic asymmetries, which regularly occur as the interaction unfolds, could halt or derail the progression of the interaction and therefore pose interactional problems: during the course of therapy talk, therapists need to manage these asymmetries. Since its inception (Du Bois 2007), dialogic resonance has been employed in interactional management, such as stance management (Siromaa 2012; Nir 2017), topic management (Laury 2005), and inter-​subjectivity management (Nir et al. 2014). Although dialogic resonance has been found in the management of stance or epistemic related issues (Dori-​Hacohen 2017; Nir & Zima 2017; Escobar 2011), there are at least two research gaps which require further exploration. First, insufficient research has been conducted on resonance in naturalistic interactions, as mentioned in Ochs et al. (2004) and by Sterponi and Shankey (2014). It is therefore necessary to “target more naturalistic environments of person-​to-​person engagement” and “to look at how dialogic resonance works in everyday life where conditions are more dynamic and less predictable” (Du Bois et al. 2014: 437). Secondly, to the best of my knowledge, little attention has been paid to investigating the pragmatic motivations that drive the use of dialogic resonance in managing interactional contingencies, that is, both the circumstances under which therapists will resonate and the part of the information which will undergo resonance, and how this interactional process is explained pragmatically. Referring to existing literature (Du Bois 2007, 2014), resonance is currently defined as an interactional resource from which the speaker reproduces the selected aspects of a prior utterance, with the intention of highlighting the

1 2

Managing Epistemic Asymmetry 121 relevant information pertaining to the recipient’s cognitive effort. The aim of this chapter is to examine how dialogic resonance is used to manage epistemic asymmetry and facilitate epistemic symmetry. The following three research questions are addressed: first, Sections 4.1.1 and 4.2.1 address the question of how epistemic asymmetry is manifested and developed in therapy talk; second, 4.1.2 and 4.2.2 deal with how dialogic resonance is utilized to manage these asymmetries and how relevance comes into play in the process; third, Section 5.2 discusses how we account for the use of dialogic resonance in managing epistemic asymmetry from a pragmatic perspective.

2  A Review of the Literature 2.1  Previous Studies on Epistemic Asymmetry In recent years, there has been an abundance of studies into epistemics and related areas. Epistemics, which concerns the territories of knowledge, distinguishes between a relatively unknown position relative to others (or K-​) and a more knowledgeable position (or K+). This imbalance in knowledge status, referred to as epistemic asymmetry, or epistemic incongruence, or epistemic discrepancy, can be indicated through a number of different resources, for example, assessments and second assessments. A downgraded first assessment, if syntactically expressed via a tag question by the speaker, implicates a putatively secondary access to the referent relative to the co-​participant, thereby indicating epistemic asymmetry (Heritage & Raymond 2005). Drawing on data from a Spanish context, García-​Ramón (2018) claims that epistemic asymmetry can be signaled via uy-​responses. By using the particle uy in agreement sequences, the speaker implicates the previous turn is problematic in some way, he/​she then upgrades the claim embodied in the previous turn with further information or evidence provided. In addition to the practice of assessment, actions such as advice delivery and questioning also signal epistemic asymmetry. In advice delivery, the advisor “describes, recommends or otherwise forwards a preferred course of future action” (Heritage & Sefi 1992: 368). According to Waring (2007), advising involves a series of human actions such as giving an immediate correction or providing a long-​term solution. In so doing, the advisor assumes a higher epistemic status of the issue under discussion and recommends a solution to the recipient (Heritage & Sefi 1992). Therefore, the advice-​giving sequences implicate epistemic asymmetry. Questioning is a practice by which the questioners assign themselves a subordinate epistemic primacy with regard to the recipients. The act of questioning invokes a claim that the questioner lacks certain information (or certainty about it), and at the same time he/​she assumes that the recipient has the information (or is likely to have it) (Heritage & Raymond 2012). Question types such as wh-​questions and yes-​no-interrogatives presume a basic epistemic asymmetry in that the speaker is less informed than the recipient about the matter being discussed (Sidnell 2012).

21

122 Lei An I-prefaced expression is another resource through which epistemic asymmetry can be pointed out. Drawing on data from doctor-​patient interactions, Lindström and Karlsson (2016) report that the use of the epistemic disclaimer ‘I don’t know’ may not only display a speaker’s lack of relevant knowledge in a certain context, but can also signal interactional problems that are contingent on the epistemic asymmetries between the participants. The Finnish ‘mä tiedän’ (‘I know’) response, when used in responsive sequences, indicates epistemic asymmetry: the first part is an informing type turn which implicates an unknowing (or less knowing) recipient. With the mä tiedän response, the speaker resists the prior implication and highlights the epistemic asymmetry that exists at that moment (Vatanen 2018). Some other practices, such as ‘I thought’-​initiated turns (cf. Smith 2013) are used to address discrepancies between mutual knowledge. For instance, if speaker A informs speaker B that “Mike went to Japan yesterday” in turn 1, speaker B responds with “I thought you were going” in turn 2. After this, speaker A explains “No, I changed my mind” in turn 3. This example shows that an ‘I thought’- turn can mark a discrepancy between the shared knowledge between these two speakers. The literature mentioned so far has reviewed the way in which epistemic asymmetries are signaled in interactions  –​how they are managed, however, is another research focus. In prior studies on the management of epistemic asymmetry, the medical setting has already been a very important interactional context for investigation. Lehtinen (2007), for instance, examines the ways in which doctors deal with clients’ discrepant proposals. These proposals are turns in which the client presents information that is potentially different from that given by the doctor, for instance, the client may begin her discrepant proposal with the particles but still. When such discrepancies occur, the doctor merges his proposal with the client’s and organizes them into a single coherent proposal. At the same time, the doctor provides evidence for his/​her original proposal. According to other studies, the ways in which doctors deal with such discrepant information may vary from “virtual inattention” (Peräkylä 2002: 244) to uptake by returning to medical interview and physical examination (Lehtinen & Kääriäinen 2005). The interactional context that is the focus of this study differs from the general medical contexts mentioned above. In therapy talk, it is not sufficient for therapists to merely listen to their clients, they also have to respond to their troublesome experiences and reshape them in various ways. This reshaping process is crucial for helping the client obtain a new insight into their problem, thereby enabling them to interpret their problematic experience in a different manner. This process is thus of great significance in the recovery and well-​being of clients (Weiste et al. 2016). 2.2  Previous Studies on Dialogic Resonance Dialogic resonance is derived from dialogic syntax theory (Du Bois 2007, 2014). Within this theory, dialogic resonance is defined as “the catalytic activation of affinities across utterances” (Du Bois 2014: 359). Nir et al. (2014)

3 2 1

Managing Epistemic Asymmetry 123 associate dialogic resonance with the ways in which speakers rely on previous utterances when conducting their current conversation. By recycling words, phrases, syntactic structures, or prosodic patterns, as well as other language resources invoked by previous interlocutors, considerable structural equivalences across speakers and turns can occur (Kärkkäinen 2006). To elaborate on the key aspects of the mapping between forms that resonate across utterances in a more visible way, a diagraph (i.e. the emergent structure created by language users) is proposed and instantiated by Du Bois (2007: 160–​161), which is commonly referred back to and employed in younger studies (Du Bois 2014: 362; Maschler & Nir 2014). Resonance can be classified into syntactic resonance and semantic resonance (Maschler & Nir 2014). Another distinction is made between self-​resonance and other resonance (Maschler & Nir 2014; Brône & Zima 2014). The former, namely self-​resonance, involves recycling the speaker’s own utterance, while the latter refers to the recycling of the co-​participant’s words. It should be noted that these two classifications can be used together, that is, an interlocutor can semantically self-​resonate –​in this case, both self-​resonance and semantic resonance would be involved. A further point worth noting is that resonance does not necessarily mean exact structural equivalence, but may also involve expansion. Whatever form assigned, a speaker intentionally activates resonance to convey “pragmatic meanings and pragmatic stance among socially positioned agents” (Zima et al. 2009: 3). In other words, interlocutors do not merely echo each other or echo their own words, but exploit the potential meaning of their co-​interlocutor’s utterance or their own words to achieve the desired pragmatic effects (Sakita 2006; Du Bois 2014). The pragmatic function and effects of dialogic resonance have been explored in earlier studies. By examining the formulation of political talk on Israeli radio networks, Nir et al. (2014) found that dialogic resonance, in the form of lexical and syntactic resonance, is used as a stance-​taking device by a caller-​in to reject the host’s utterances throughout the interaction. Drawing on data from French parliamentary debates, Zima et al. demonstrate that dialogic resonance can be activated through “explicit repetition of linguistic form and implicit echoing of semantic-​pragmatic meanings” (2009: 1). They found that dialogic resonance serves a dis-​associative pragmatic function, whereby socio-​political stances and power relationships will be negotiated. Du Bois et al.’s (2014) study is one of the few to introduce a new methodological approach for investigating the pragmatic function of dialogic resonance. They devise a controlled study to capture the dynamic one-​ on-​one conversations between an adult interviewer and autistic children. They found that, through the use of resonance, frame resonance for instance, autistic children were able to indicate and experience moments of communicative engagement. A number of studies into e.g. artificial language, monologue narrative discourse, and the discursive behaviors of children and adolescents with autism, have investigated the use of dialogic resonance as a resource for stance-​taking and intersubjective involvement. However, few studies have focused on the

4 2 1

124 Lei interactional process of dialogic resonance, namely, how interactional and contextual factors influence its use and the communicative functions it performs. To bridge this gap, the present study utilizes data obtained from naturally occurring therapeutic talk with the intention of scrutinizing how dialogic resonance functions in this institutional context, an area that has overall been neglected in previous studies.

3  Methodology and Data 3.1  Data, Participants, and Settings The data to be analyzed in this study is drawn from 22 individual therapy sessions, amounting to 20 hours of audio recordings. Each unit of therapy talk features one client and one therapist, which amounts to a total of two therapists and seven clients being involved in the current study. A relationship-​ focused integrative therapeutic approach, incorporating the most salient psychotherapeutic theories within the field, was adopted by the therapists. The relationship-​ focused integrative therapeutic approach claims that “change comes and remains solid as clients make their own meaning out of their own internal and external experiences” (Clark 1996: 313–​314). The clients who agreed to participate in this study were all suffering from moderate symptoms of depression due to mental health problems pertaining to domains such as relationships (e.g. relationships with significant others, such as their spouse) and emotional well-​being (e.g. workplace-​related stress or depression resulting from previous traumas). The recordings were collected in a Chinese hospital between April 2017 and May 2018 by the therapists. The two excerpts analyzed in Section 4 are taken from two separate therapy sessions which were conducted by two different male therapists with seven and fifteen years of professional experience, respectively. Permission for data recording and data use was obtained from each participant prior to the recording of each therapy session. The sessions were audio recorded by the therapists, and were then transcribed by the researcher according to the Jeffersonian transcription system (Jefferson 2004), after which they were analyzed by using a conversation analysis approach. Any information that could reveal a participant’s identity or personal details was anonymized by the researcher. The criteria used for identifying resonance are based upon Du Bois’s work (2007) and paired relationships between utterances are represented via diagraph(s) also following Du Bois (cf. 2007: 160–​161, 2014: 362). 3.2  Institutional Features of Therapy Talk Therapy interaction as a specific activity type shares similarities with other types of medical interaction with regard to interactional asymmetry, turn-​ taking management, and the collaborative dimension of interaction management. However, the fact that the kind of therapy interaction the present study is addressing deals with mental rather than physical health problems renders this

5 2 1

Managing Epistemic Asymmetry 125 material and the circumstances in which it has been collected particular: after all, when compared with non-​chronic physical illness, which may suddenly afflict the patient and would then be sought to be cured as immediately as possible so to bring the patient swift relief, mental health problems generally are characterized by chronicity, thus last for a longer period of time (Sarangi 2000) and require specific, multilateral, and often longitudinal treatment. Generally, it is only after a problem has been developing for a while that a client feels compelled to see a therapist. Therefore, from this perspective, therapy talk constitutes a particularly interactional genre with two distinctive features in this specific study design: First, a standard therapy session in China (where our data was collected) lasts for at least 50 minutes, with generally more time being devoted to each individual client as compared to other medical interactions. Given that clients obtain epistemic priority and epistemic authority with regard to their personal experiences and feelings, they may resist accepting the therapists’ epistemic claim through the action of “disagreeing, rejecting or providing a relevant answer” (Muntigl 2013: 18). Moreover, the time allocated to each therapy session is dedicated to a preferably thorough discussion of the client’s mental illness. Second, the therapeutic value of client-​centeredness empowerment, i.e. clients are considered as “having expertise in the specifics of their own situation and in determining the fit of any proposed solution to their problem” (Butler et al. 2010: 267), is generally accepted and acknowledged in therapeutic practice and aims to promote self-​directness. Hence, clients are supposed to develop their own abilities and viewpoints in finding solutions to their problems rather than the therapist having to explicitly propose what the client might or should do. Consequently, this therapeutic philosophy affects both the way in which a therapist’s utterance will be designed and how a client’s statements will be responded to.

4  Interactional Unfolding of Dialogic Resonance in Therapy Interaction This section zooms in on two episodes extracted from the collected dataset to demonstrate and discuss how dialogic resonance is employed by therapists in dealing with epistemic asymmetries that occur during a therapist-​client interaction. The way in which epistemic asymmetry is manifested and developed will be presented first, as this serves as the foundation for the initiation of dialogic resonance; then, the manner in which these asymmetries are managed through the use of dialogic resonance will be discussed. Two types of dialogic resonance feature in these episodes, with Excerpt  1 focusing on both self-​resonance and other-​resonance, and Excerpt 2 concentrating solely on self resonance. These two types of resonance, which have different positions in the conversation and occur in different forms, function as explicit/​implicit markers of epistemic asymmetry at particular moments in the conversation, thereby invoking an extra processing effort on the hearer’s part and contributing to the achievement of epistemic symmetry.

6 2 1

126 Lei 4.1  Managing Epistemic Asymmetry through Self-​Resonance and Other-​Resonance: Excerpt 1 In Excerpt 1, the conversation features a client who is a student at a provincial high school, suffering from depression as a result of a particular sudden event in his life. As he had previously been doing well at school, obtaining positive social recognition from both teachers and peers, the said sudden event brought about a loss of control and self-​confidence. The co-​participants are discussing ‘a kind of feeling’ that is perceived as self-​confidence by the client. Excerpt 1, Part 1 C refers to the client, T refers to the therapist 01 C: 02 T: 03 C: 04 05 T:

06 C: 07 08 T:

09 C:

£但是我还想问£ 我最开始的那个问题= £ But I still want to ask £ my initial question= =嗯？ =Uhm？ 嗯::（1.0）£这种感觉£:: >是自信吗