Quotatives: Cross-linguistic and Cross-disciplinary Perspectives 9789027239051, 9789027274793

Table of contents :
Quotatives
Editorial page
Title page
LCC data
Table of contents
Authors’ biographies
Preface: Introductory remarks on new and old quotatives
1. Introduction
2. New and old ways of introducing quotation
3. Situating the volume
References
Part I. Discourse perspectives
Impersonal quotation and hypothetical discourse
1. Introduction
2. Methodology and data
3. Analysis: Hypothetical discourse in conversation
3.1 Modeling discourse
3.2 Claim-backings
3.3 Fictitious humorous stories
4. Concluding discussion
References
Appendix: Transcription Conventions
By three means: The pragmatic functions of three Norwegian quotatives
1. Introduction
1.1 Aim and scope
1.2 Theoretical framework
1.3 Data and methodology
2. Bare
2.1 Bare as a discourse marker
2.2 Quotative bare
3. Liksom
3.1 Liksom as a discourse marker
3.2 Quotative liksom
4. Sånn
4.1 Sånn as a discourse marker
4.2 Quotative sånn
5. Distribution across age groups
6. Combinations of the markers
7. Conclusions
References
Appendix: Transcription Conventions
Part II. Typological perspectives
Minds divided: Speaker attitudes in quotatives
1. Introduction
2. Speaker attitudes and quotative constructions
3. Grammatical categories in quotative constructions
3.1 Modality and evidentiality
4. Speaker attitudes and reported messages: A sample study
4.1 Introduction
4.2 Overview of the results
4.3 Number and functions of quotative constructions
4.4 Categories in the de dicto domain
4.5 Evidentiality in quotation
4.6 Modal categories
4.7 Illocution and some remaining observations
4.8 Summary of the preceding sections
5. A method for the typological analysis of speaker attitudes in quotative constructions
6. Some closing remarks: Semantics and pragmatics
References
Thetic speaker-instantiating quotative indexes as a cross-linguistic type
1. Quotative indexes from a typological perspective
1.1 Functional properties
1.2 Morphosyntactic typology
1.3 Non-clausal speaker representation
2. Speaker-instantiating quotative indexes: The data
2.1 Historical cases
2.2 Modern cases
3. Speaker-instantiating quotative indexes as a cross-linguistic type
References
Part III. Functional and formal perspectives
On the characteristics of Japanese reported discourse: A study with special reference to elliptic quotation
1. Introduction
2. Characteristics of reported discourse in Japanese
2.1 Alleged continuity between direct and indirect reported discourse
2.2 Syntactic realization of a quote: Complement or modifier?
2.3 Elliptic quotation
3. Varieties of elliptic quotation
3.1 Suspensive QPE
3.2 Sentence-final QPE
3.3 Extensions of sentence-final QPE
3.4 QPE in a noun-modifying construction
3.5 QPE in topic phrases
3.6 QPE in concessive constructions
4. Conclusion
Acknowledgments
References
Appendix: Sources of data
Quotative go and be like: Grammar and grammaticalization
1. Introduction
2. Problems with item-based accounts
3. Solving the puzzle: An interclausal type of complementation
3.1 Problems in the verbal complementation analysis
3.2 An alternative proposal: Interclausal complementation
4. Relocating the initial innovation: ‘Imitation clauses’ constructionally apprehended as ‘reporting clauses’
4.1 The imitative meaning of be like and go clauses
4.2 A constructional interpretation of the imitation-reporting analogy
4.3 I’m like, he went, this is me in a network of constructions
5. Revisiting the question of grammaticalization
5.1 Quotative be like and go: Grammaticalization or constructionalization?
5.2 Extensions of be like and go quotatives
6. Where do we go from here? Current and future developments
7. Conclusion
References
Quotation in sign languages: A visible context shift
1. Introduction
2. Some aspects of sign language grammar
3. Verb agreement in sign languages
4. Formal aspects of role shift in sign languages
4.1 Methodology and data
4.2 Role shift: Direct or indirect speech?
4.3 Nonmanual markers for role shift
4.4 Interaction of nonmanuals for role shift in DGS
5. Functional aspects of role shift in sign languages
6. Reported speech in spoken and sign languages
7. Conclusion
References
Part IV. Language variation and change
Performed narrative: The pragmatic function of this is + speaker and other quotatives in London adolescent speech
1. Introduction
2. The study
3. The Results
4. Be like
5. This is + speaker
6. Pragmatic functions of new quotatives
7. Discussion and conclusions
References
Appendix 1: Transcription Conventions
Dutch quotative van: Past and present
1. Introduction
2. The structural profile of quotative van
2.1 Non-quotative use of van
2.2 Quotative van
3. Sociolinguistic and stylistic aspects of quotative van
4. The diachronic development of van
5. Conclusion
References
Appendix: Glossary of specialist terms for research in quotation
Author index
Index of terms

Citation preview

Quotatives

Converging Evidence in Language and Communication Research (CELCR) Over the past decades, linguists have taken a broader view of language and are borrowing methods and findings from other disciplines such as cognition and computer sciences, neurology, biology, sociology, psychology, and anthropology. This development has enriched our knowledge of language and communication, but at the same time it has made it difficult for researchers in a particular field of language studies to be aware of how their findings might relate to those in other (sub-)disciplines. CELCR seeks to address this problem by taking a cross-disciplinary approach to the study of language and communication. The books in the series focus on a specific linguistic topic and offer studies pertaining to this topic from different disciplinary angles, thus taking converging evidence in language and communication research as its basic methodology. For an overview of all books published in this series, please see http://benjamins.com/catalog/celcr

Editors Marjolijn H. Verspoor University of Groningen

Wilbert Spooren

Vrije Universiteit Amsterdam

Advisory Board Walter Daelemans

Leo Noordman

Cliff Goddard

Martin Pütz

University of Antwerp University of New England

Tilburg University University of Koblenz-Landau

Roeland van Hout

Radboud University Nijmegen

Volume 15 Quotatives. Cross-linguistic and cross-disciplinary perspectives Edited by Isabelle Buchstaller and Ingrid Van Alphen

Quotatives Cross-linguistic and cross-disciplinary perspectives Edited by

Isabelle Buchstaller Leipzig University

Ingrid Van Alphen University of Amsterdam

John Benjamins Publishing Company Amsterdam / Philadelphia

8

TM

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

Library of Congress Cataloging-in-Publication Data Quotatives : cross-linguistic and cross-disciplinary perspectives / edited by Isabelle Buchstaller, Ingrid Van Alphen. p. cm. (Converging Evidence in Language and Communication Research, issn 1566-7774 ; v. 15) Includes bibliographical references and index. 1. Quotation. 2. Speech acts (Linguistics) 3. Semantics. 4. Intercultural communication. I. Buchstaller, Isabelle. II. Alphen, Ingrid Van. P302.814Q68 2012 401’.41--dc23 2012000582 isbn 978 90 272 3905 1 (Hb ; alk. paper) isbn 978 90 272 7479 3 (Eb)

© 2012 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

Table of contents Authors’ biographies Preface: Introductory remarks on new and old quotatives Isabelle Buchstaller and Ingrid Van Alphen

vii xi

part i.  Discourse perspectives Impersonal quotation and hypothetical discourse Andrea Golato By three means: The pragmatic functions of three Norwegian quotatives Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig

3

37

part ii.  Typological perspectives Minds divided: Speaker attitudes in quotatives Stef Spronck Thetic speaker-instantiating quotative indexes as a cross-linguistic type Tom Güldemann

71

117

part iii.  Functional and formal perspectives On the characteristics of Japanese reported discourse: A study with special reference to elliptic quotation David Y. Oshima and Shin-ichiro Sano

145

Quotative go and be like: Grammar and grammaticalization Lieven Vandelanotte

173

Quotation in sign languages: A visible context shift Annika Herrmann and Markus Steinbach

203



Quotatives

part iv.  Language variation and change Performed narrative: The pragmatic function of this is + speaker and other quotatives in London adolescent speech Sue Fox

231

Dutch quotative van: Past and present Peter-Arno Coppen and Ad Foolen

259

Appendix: Glossary of specialist terms for research in quotation

281

Author index Index of terms

291 293

Authors’ biographies Isabelle Buchstaller is professor of varieties of English at Leipzig University. Her research interests range from language variation and change and dialectology to innovative methods for data collection and analysis. Areas she has recently explored are dialectal (morpho-)syntactic forms such as adverbial clauses, intensifiers and the Northern subject rule. She has published extensively on the social, linguistic and socio-psychological aspects of the new quotatives in a range of English varieties and she is currently writing a book on reported speech to be published with Wiley-Blackwell. Peter-Arno Coppen is professor of pedagogics in language, history and culture at the Center for Education Research. Having graduated in 1981 in Chomskyan linguistics, he worked for quite some time within the field of computational linguistics. His Ph.D. (1991) examined the syntax and semantics of the Dutch NP specifier. Although he has always been interested in Dutch syntax, in line with his current position of head of the Nijmegen institute of teacher education, his focus is now on grammar education. He was editor of the Dutch linguistic journal Gramma/TTT (1985–1999) and he is currently editor of the popular linguistics magazine Onze Taal. Ad Foolen is assistant professor in the Linguistics Department at the Radboud University, Nijmegen. His research focuses on pragmatic markers and expressive constructions from a pragmatic and cognitive linguistic perspective. His interest in direct speech and quotative markers arose from that of expressive language use. In 2000 he co-edited, with Frederike van der Leek, Constructions in Cognitive Linguistics (Benjamins) and in 2002, with Ton van der Wouden and Piet van de Craen a special volume on Particles for the Belgian Journal of Linguistics. Ad Foolen is secretary/treasurer of ICLA, the International Cognitive Linguistics Association. Sue Fox is a research fellow at Queen Mary, University of London. Her primary research interest is in variationist sociolinguistics as it relates to contemporary language variation and change in English. She gained her Ph.D. from the University of Essex in 2007, writing a thesis which explored immigration from Bangladesh and its impact on the London/Cockney dialect. Her subsequent work on

 Quotatives

ESRC-funded projects Linguistic Innovators: The English of Adolescents in London and Multicultural London English has focused mainly on the social and historical contexts that have led to the variety of English that is spoken in London today. Andrea Golato is an associate professor in the Department of Germanic Languages and Literatures at the University of Illinois at Urbana-Champaign. Her research interests lie primarily in conversation analysis, specifically in the area of grammar in interaction and culture and communication. The overarching themes she has explored are cross-cultural comparisons of various actions and specific linguistic forms and their functions in interaction. In addition to having published articles on reported discourse, she has published on compliments in German, including a book with John Benjamins Press entitled Compliment and Compliment Responses: Grammatical Structure and Sequential Organization. Tom Güldemann is professor of African languages at the Humboldt University, Berlin. His research interests range from language typology, historical linguistics and language documentation and description to information structure and the syntax-discourse interface. He has a particular interest in languages subsumed under “Khoisan” and the Bantu family. Among his publications is Quotative Indexes in African Languages: A Synchronic and Diachronic Survey (2008, Mouton de Gruyter), which deals with the synchrony and diachrony of quotative indexes and reported discourse as a whole from a typological perspective. Annika Herrmann is a research assistant at the German Department of the GeorgAugust-University, Göttingen. She defended her Ph.D. (entitled Sentence Types: Variation and Interpretation) at the Goethe-University of Frankfurt’s post-graduate program in 2010. Since January 2010 she has been in charge of the Sign Language Lab at the University of Göttingen. She is interested in theoretical and experimental sign-language linguistics, mainly examining interfaces of grammar, including the form and function of nonmanuals and the syntax-prosody interface. Her research interests concern the processing and the diachronic development of sign languages, sign-language quotation and agreement marking. Ingrid Kristine Hasund is associate professor of linguistics at the Department of Nordic and Media Studies at Agder University and a senior researcher at the Department of children’s and adolescents’ mental health at the Hospital of Southern Norway in Kristiansand. Her research deals with English and Norwegian teenage language. She has published several articles and books on the topic, with particular emphasis on discourse markers, slang and taboo words. She is also involved in research on conflict work and preventive mental health work in kindergartens.



Authors’ biographies 

Toril Opsahl works as a researcher and editor on the project Norsk Ordbok 2014 (Norwegian Dictionary 2014) at the University of Oslo. Her Ph.D. thesis (2010) examined structural characteristics associated with Norwegian spoken among youth in multiethnic areas in Oslo. Between 2006 and 2009 she was part of the national research project UPUS (Utviklingsprosesser i urbane språkmiljøer – ‘Linguistic development in urban settings’), where she contributed to research and publications covering various aspects of linguistic practices among adolescents in multiethnic areas in Oslo. In 2002 she wrote her Master thesis in Scandinavian linguistics on the pragmatic marker bare. David Y. Oshima is an associate professor in the Department of International Communication at Nagoya University. His fields of specialization are semantics, pragmatics, syntax, and Japanese linguistics. He earned his Ph.D. in linguistics from Stanford University in 2006, writing a thesis on the semantics of propositional attitude reports with special reference to perspective- and deixis-related phenomena within reported discourse. His current research interests lie primarily in discoursemanagement, information structure, and perspective-taking in language. Shin-ichiro Sano is a visiting associate professor in the Department of Language Sciences at International Christian University in Japan. He received his Ph.D. in linguistics from Sophia University in 2009. His research interests include language variation and change, phonetics, phonology, morphology, and corpus linguistics. He is currently working on ongoing language changes in Japanese involving the conjugation paradigm, on allophonic distribution, adjectives, and negative polarity items, using multivariate statistical techniques and Optimality Theory. Stef Spronck is a Ph.D. candidate at the Australian National University. He works on quotation and stance taking in the Australian Aboriginal language Ngarinyin (Kimberley region, Western Australia). He previously completed BAs in general and Slavic linguistics (Russian) and an MA in theoretical linguistics at the University of Amsterdam. His Ph.D. research forms part of the project Social cognition and language: The design resources of grammatical diversity. Apart from stance, Australian Aboriginal languages and reported speech and thought, Stef ’s research interests include typology, the relation between sociality and grammar, and the linguistic work of the Mikhail Bakhtin circle. Markus Steinbach is professor of linguistics at the German Department of the Georg-August-University, Göttingen. His research is concerned with the influence of language modality (spoken or sign languages) on language structure, development and processing. The focus of his research is on the interaction of syntax, semantics and pragmatics, experimental linguistics and sign language linguistics. Since 2010 he is Principal Investigator at the Courant Research Center on a project



Quotatives

entitled ‘The Multi-layered Text Protocol: Micro and Macro Level Structures in Written Discourse’. He is a member of the editorial boards of the journals Zeitschrift für Sprachwissenschaft and Sign Language & Linguistics. Jan Svennevig is professor of linguistics at the University of Oslo. He has published several studies on Norwegian discourse markers. In addition, his research deals with interaction in workplace meetings, focusing on leadership and chairing activities. He has also studied conversational strategies for establishing mutual understanding in second language conversations in social welfare consultations, with a focus on patterns of repetition and repair. He is the author of a book-length study on social interaction among new acquaintances: Getting acquainted in Conversation. A Study of Initial Interactions (Benjamins, 1999). Ingrid Van Alphen is assistant professor of linguistics at the University of Amsterdam and senior researcher at the Amsterdam Center of Language and Communication (ACLC). Her research interests are language and gender, sociolinguistics, semantics and pragmatics. Her Ph.D. thesis (1999) investigated gender-based lexical-grammatical patterns and speech acts in Dutch adolescent conversations. She has also published on the interactional power of questions. In 2006 she initiated the Dutch Van-club: A group of linguists doing research into the Dutch quotative marker ‘van’. Her current research project combines her interest in questions and quotatives: The use of direct speech in questioning and the use of questions in direct reported speech. Lieven Vandelanotte is associate professor of English linguistics at the University of Namur (Belgium) and affiliated researcher in the Functional Linguistics Leuven research group at the University of Leuven. He is interested in English grammar, including prenominal adjectives, modified proper names, and the structural, deictic and semantico-pragmatic properties of different reported speech constructions. In 2009 he published the monograph Speech and Thought Representation in English: A Cognitive-Functional Approach. He co-edited, with Kristin Davidse and Hubert Cuyckens, the volume Subjectification, Intersubjectification, and Grammaticalization.

preface

Introductory remarks on new and old quotatives Isabelle Buchstaller and Ingrid Van Alphen Leipzig University and University of Amsterdam

1. Introduction When William Labov was asked before the opening of the 17th Sociolinguistics Symposium in Amsterdam if there were still any linguistic changes that surprised him, he mentioned “a new form in English that popped up fifteen or twenty years ago to introduce direct reported speech. Quite suddenly the old way of using I say or He goes was replaced by I’m like and he’s like”. Labov pointed out that this form “has penetrated as far as Australia and that it has by now become the way to start a quotation. This change went at lightning speed. We don’t know why” (2008: 93).1 Crucially, while like is certainly the most robust and geographically widespread quotative innovation, during the last few decades the literature has picked up on a range of novel forms in English (as in 1–3). (1) She went/was like/was all/was git “That’s great!” (Butters 1980, 1982; Waksler 2001; Norten 2008, see also Vandelanotte this volume) (2) This is me “That’s great!” (Cheshire and Fox 2007; Fox, this volume) (3) I was sitting there “That’s great!” (Stein 1990) Similar innovations have been reported in a wealth of languages across the world, including Dutch, Hebrew and Japanese (as in 4–6), but also German, Swedish, Russian, Greek, Norwegian, Icelandic, Portuguese, Spanish, Italian, and many others. 1. The interview was printed in the Dutch linguistics magazine ‘Onze Taal’ (Our Language) in March 2008. The Dutch wording was as follows: “Zo’n vijftien of twintig jaar geleden dook er een nieuwe vorm in het Engels op om de directe rede aan te kondigen. Vrij plotseling veranderde de vertrouwde manier met I say of he goes in I’m like en he’s like. Tot in Australië is deze vorm doorgedrongen. Vandaag de dag is het dé manier om een citaat te beginnen. Die verandering is bliksemsnel gegaan. We weten niet hoe het komt”. [Translation IVA]

Isabelle Buchstaller and Ingrid Van Alphen



(4) Toen had ik zoiets van “geen woorden voor.” [Then I had something like “no words for it.”]  (Dutch: Van Alphen 2008; see also Coppen and Foolen, this volume) (5) Hu pitóm ómer li kazé “eh ...titxatní ítì?” (Hebrew: Maschler 2001) [He suddenly says to me like this “eh... will you marry me?”]

(6) Demo, Hiroshi-wa “Ore-ni-wa kankei-nai” mitai-na [But Hiroshi was like, “That’s not my business.”] (Japanese: Oshima p.c.)

Partly due to the proliferation of these innovative forms, quotation – old and new ways of reporting speech, thought, attitudes or physical activity – has become a hot topic, not only in the academic world but also in venues aimed at the general public, especially in the realm of the World Wide Web. And even though the stereotypes and attitudes attached to the novel quotative forms are not necessarily equivalent across the different linguistic communities in which they have emerged, their perceptual lore is often far from positive. In the USA, for example, the use of be like has been described as “‘vacuous’, ‘silly’, ‘airheaded’, ‘California’” and typically used by teenage girls (Blyth et al. 1990: 224; see also Buchstaller 2006a; Dailey-O’Cain 2000; D’Arcy 2007). Go, on the other hand, seems to be “indicative of uneducated, lower-class males”, and tends to be considered a blue-collar feature (Blyth et al. 1990: 224; see also Tagliamonte and Hudson 1999: 160). In the Netherlands as well as in the Dutch speaking parts of Belgium, the new quotative van regularly gets the top prize for the worst “newcomer” in the language (Van Alphen 2006: 29), which is surprising since its increased popularity started around 1975 and, as such, it can hardly be called new (Van Alphen 2006, 2008).2 In Israel, a whole generation has been named after their use of new Hebrew quotatives (and discourse markers) – the “kaze-ke’ilu- generation” – and the use of these innovative forms is stereotyped as “non-commital”, a trait that is associated primarily with the language of adolescents (Maschler 2002: 245; Ziv 1998). These negative attitudes do not seem to prevent speakers in a large and diverse set of speech communities from using new (and old) ways for recreating one’s own or other people’s words, thoughts or activities. Thus, cross-linguistic evidence suggests that a range of typologically related and unrelated languages are currently undergoing a parallel development: lexical material with non-reportative semantics is being recruited to introduce speech, thought, activities or attitudes, thereby intruding into the pool of existing quotation strategies. The fact that this development is happening across language families raises a number of important questions. Why do speakers use these 2. In 2010 it was again elected as number one linguistic item for vaagtaal ‘vague language’ (http://vaagtaal.nl/2010/10/ik-heb-zoiets-van-vaagste-vaagtaal-van-2010/).



Preface:  Introductory remarks on new and old quotatives

innovative forms? What, if anything, do these strategies have in common? Is it possible to postulate any generalizations regarding the formal properties, functional load and social distribution of these newcomer variants? Moreover, we need to bear in mind that in every language for which the innovative quotatives have been reported, the newcomers are incorporated into an already existing system of speech and thought reporting. This begs the question how – and indeed whether – we can conceptualize quotation in a way that encompasses older majority forms as well as intruding, incipient and minority forms. Can we establish any common ground that unites newcomer forms with already firmly established, ‘old’ quotatives?3 What do speakers do when they report? And what kind of textual, interpersonal and grammatical functions do quotative forms – older as well as newer ones – fulfill? More generally, as this volume endeavors to explore, can we depict any cross-linguistic tendencies as regards the form, function and social use of quotation?

2. New and old ways of introducing quotation Interestingly, the lexical elements that have recently taken on this quotative function largely derive from a limited number of source constructions. Table 1 shows the main semantic bases for new ways of reporting speech, thought and attitudes (see also Buchstaller and Van Alphen 2008). Note that this list, which we have collated on the basis of reports in the literature as well as discussions with colleagues and informants of the languages represented, does not aim to provide a complete and exhaustive inventory of innovative quotatives. Rather, it seeks to illustrate the preponderance of certain types of source constructions to be recruited as innovative quotations and also (to a certain extent) their typological spread.4

3. For the purposes of this introduction, ‘old’ should be taken to mean more traditional quotative markers that speakers do not perceive as innovative, that have been reported in the literature for some time and/or have developed into conventionalized fixed expressions within the quotative system. 4. Note that quotative markers can obviously collocate, resulting in constructions such as He went like this “Quote”, which combines a motion verb, a maker of similarity as well as a deictic marker. At the same time, one item can combine several semantic traits, such as Hebrew kaze ‘like this’, which combines the features deixis and similarity.





Isabelle Buchstaller and Ingrid Van Alphen

Table 1 (New) quotatives and their semantic sources5 i. Comparative (similarity/approximation): Afrikaans soos ‘so + as’, Čzech jako (že) ‘as’, Buang (na) be ‘like’, Danish ligesom ‘like + as’, Dutch van ‘like’, English like, Estonian nagu ‘like’, Finnish niinku (niin kuin) ‘as if ’, ihan et ‘like’, French comme ‘like’, genre ‘kind (of)’, style ‘style’, Frisian fan ‘like’, Greek tipou ‘type’, Hebrew ke’ilu ‘as if ’, kaze ‘like + this’, Italian tipo ‘type’, come ‘like’, stile ‘style’, genere ‘kind’, Japanese mitai-na ‘like’, Croatian kao ‘like’, tipa ‘type’, Swedish typ ‘type’, liksom ‘like + as’, Norwegian typ ‘type’, liksom ‘like + as’, Polish typu ‘type’, Portuguese tipo ‘type’, Brazilian Portuguese tipo + assim ‘type + so’, Russian tipa ‘type’, Spanish como ‘like, as’, Thai bæ:p ‘like’. ii. Demonstrative deictic: Afrikaans soos ‘so + as’, Čzech na to ‘on this’, Danish sådan ‘such + like + this’, Dutch zo ‘so’, Estonian nii et ‘so that’, London English this/here is NP, Finnish et (tä) ‘that’, German so ‘so’, Hebrew kaze ‘like + this’, Croatian ono ‘that’, ono kao ‘that + like’, (Brazilian) Portuguese (tipo+) assim ‘(type+) so’, Russian takoij ‘such + like + this/that’, Spanish asi ‘so’, Norwegian sånn ‘such + like + this/ that’, Swedish såhär (sär) ‘such + like + this/that’. iii. Quantifiers: Danish bare ‘just, only’, Dutch helemaal ‘all’, English all, Estonian täiega ‘totally’, Finnish vaa(n) ‘just’, Icelandic bara ‘just, only’, Norwegian bare ‘just, only’, Swedish ba(ra) ‘just, only’. iv. Generic verbs of motion and action: English go6, Dutch komen ‘to come’, Greek kano ‘do’, Puerto Rican Spanish hacer ‘do, make’. Table 1 reveals that the source for the vast majority of these new quotatives – in the Indo-European language family but also in typologically unrelated languages such as Estonian and Japanese – are lexical items that denote comparison, similarity or approximation. This is hardly surprising given that reported speech in spontaneously 5. Sources/References: Buchstaller (2004); Fleischman (1999); Golato (2003); Meyerhoff (2002); Rathje (2011); Siamou (2004). We would like to thank the following people, who kindly supplied us with new quotative forms: Ekaterina Bobyleva, Natalia Chelstowska, Alessandra Corda, Hjörtur Einarsson, Kees Hengeveld, Mie Hiramoto, Wesley Jantjies, Preena Kangkun, Fivos Karalis, Katrin Kaun, Radovan Lučić, Miriam Meyerhoff, Hannele Buffy Nicolson, Laura Riojo, Konrad Rybka, Sumittra Suraratdecha, Aaron Tsang and Kazumi Yoshihara. 6. The use of the lexeme go for quotation in English has been attested in quotative-introductory function from 1791 onwards (Cowper Reddy “and his noble heart goes pit-a-pat”, OED). However, it was until recently restricted to onomatopoeic sound effects and non-linguistic material and only broadened its functional scope to include full quotative function in the last 30 years or so. Quotative go with reported speech was first mentioned by Butters (1980) in the US.



Preface:  Introductory remarks on new and old quotatives

occurring conversation is almost never word-for-word, verbatim reproduction. Indeed, any form of spontaneous oral quotation is inevitably compromised by the reporters’ accent, style, prosody and, importantly, memory, and is thus nothing more than an approximation of the original speech act (Buchstaller 2004; Clark and Gerrig 1990; Macaulay 1987). Note that the reporters’ inability to reproduce the original material is even more obvious when quoting non-linguistic features, such as gestures, sounds, body stances, and facial expressions. By using lexical material with comparative/similative semantics (Haspelmath and Buchholz 1998), speakers acknowledge and even highlight the approximative value of the quotation and thereby shield themselves from potential criticism regarding the inexact nature of the reproduction (see Tannen’s 1986 notion of ‘constructed dialogue’). The epistemic hedging function of these types of quotatives has been pointed out by a range of scholars (Buchstaller 2004; Pascual 2002; Schourup 1982a; see also Sweetser 1991: 28 on the frequent link between items that signal physical likeness and epistemic stances). Indeed, this hedging function is especially useful for the reporting of stance, feelings or attitudes, opinions or point of view (Buchstaller 2011; Jones and Schieffelin 2009; Romaine and Lange 1991; Van Alphen 2006, 2009). Hence, elements with comparative semantics give speakers the opportunity to show that quotes function not as the exact depiction of an individual speech act in a particular situation, but rather as a typification of a situation, a group of people or an individual (Koven 2001; Mazeland 2006; Hasund, Opsahl and Svennevig, this volume). For example, Koven (2001: 550) points out that speakers report performances of prototypical or representative speakers, utterances and thoughts in order to maintain or highlight these types within the speech community. The second major source of innovative quotative forms is lexical items that have demonstrative or deictic function. Note that this semantic/pragmatic field of source constructions supports a line of research that considers quotations as “demonstrations” of an original act which enable the “hearer to SEE for himself what it is, that is to say, in a way, [the person quoting] shows this content” (Clark and Gerrig 1990: 802; see also Fox, this volume). The deictic pointing effect of these lexemes thus focuses the hearer’s attention onto the quotation and allows the performative aspects of the enactment to take center stage. Indeed, ever since Wierzbicka’s (1974: 282) ‘theatre metaphor’, much of quotative research has sought the function of quotation in the provision of entertainment, drama and the marking of climactic points in the narration. In such story-telling situations, reporters might recruit deictic pointers in order to indicate changes of  ‘voice’ (Bakhtin 1986; Vološinov 1973) and, by doing so, signal shifts in speaker role. By referring away from themselves, reporting speakers can thus move entirely into the background (or literally aside, in the case of sign languages; see Steinbach and Herrmann, this volume) and simply point to the actors (or their speech acts) within the space of the reported narrative.





Isabelle Buchstaller and Ingrid Van Alphen

As such, the clear role demarcation as signaled by a shift between the conversational space of the reporter and that of the quoted speaker can be harnessed for a range of rhetorical goals, such as demonstration of objectivity, claims backing and argumentation (Antaki and Leudar 1990; Clift 2006; Golato, this volume). The third source for new quotatives involves elements that have quantificational semantics; they move the constituent over which they scope up or down an imaginary scale. Interestingly, all source constructions we have come across for innovative quotatives are situated at either ends of the scale, either a maximum (such as English all, Estonian täiega ‘totally’) or a minimum (as in Norwegian bare ‘just’). Güldemann (2008: 362) speculates that these semantic opposites are chosen because of their potential foregrounding function. Indeed, the use of quantifiers in quotative constructions seems to bear on the epistemic stance and attitudinal position speakers tend to take towards the quotation. By using elements with boosting function, for example, speakers can upgrade the evidential value of the report, portraying themselves as reporting ‘first hand’ information (Clift 2006), as fully committed to the accuracy or the appropriateness of the quotation, or as emotionally involved (often superimposing evaluative features; see Bucholtz 2004; Spronck, this volume, Hasund et al., this volume). On the other hand, by using downtowners, speakers might want to show minimal commitment to the form or occurrence of the quote (see Maschler’s 2002 discussion of the ke’ilu generation; Lucy 1993b) or point out that they report habitual occurrences (rather than hot news). The lexemes grouped into the fourth source in Table 1 can be described as generic verbs of motion and action, such as come and go, do and make, the use of which in quotative constructions has been interpreted in the past via a number of metaphors. Sweetser (1987) points out that the lexical fields of physical motion, action, location and of mental states and speech acts are metaphorically connected. Within the realm of the general metaphor ‘a conversation is a journey’ (Lakoff 1987), reported discourse has been interpreted as a traveling message from a sender to an addressee (the ‘conduit metaphor’, cf. Foley 1997; Lakoff 1987; Lakoff and Johnson 1980). It should therefore not be surprising to find motion verbs being recruited for the expression of telicity, ablativity or allativity, notions which figure prominently in the way we conceptualize conversation. Another metaphor we can draw upon in order to explain the choice of generic action verbs as quotative source constructions is the fact that conversation can be ‘work’ – an achievement that has to be accomplished and maintained. In particular the literature on language and gender has pointed out the vast amounts of conversational work women do in a range of different socio-cultural settings (Fishman 1983; Ochs and Taylor 1996; West and Zimmerman 1987). We can thus conceptualize the reproduction of speech acts as ‘doing’ reporting, and hence as a conversational task the speaker is performing.



Preface:  Introductory remarks on new and old quotatives

In sum, the above types of source constructions for novel quotative variants can be summarized as quotations as approximations and typifications (like), quotations as demonstrations (so/this), quotations as pointers, evidentiality/attitudinal markers (all/just) and quotations as movement or achievement (go/make). Crucially, the proposed taxonomy of innovative quotative forms is not at odds with research that reports on the semantic roots of older, existing quotatives in a number of languages. Indeed, in a large overview of quotative structures in African languages, Güldemann (2008) concludes that elements that are used in quotative constructions originate far less frequently in speech verbs than is commonly assumed: “The overall picture is quite unequivocal for the sample: the ratio of speech verbs vs. other items regarding their recruitment for regular non-lexical function in QIs [quotative indexes, eds.] and beyond is almost 1:5” (2008: 372). Below, we will briefly discuss the range of quotative source constructions (other than verbs of saying/speaking/sounding, perception or cognition) that has been reported in the typological literature.7 Similar to our overview of innovative quotative forms we presented above, this discussion – which is heavily indebted to Güldemann (2001, 2008) – does not aim to be exhaustive, but merely illustrative of the main typologically attested sources of quotation.8 i. Markers of similarity, approximation and manner Schourup (1982b: 32) gives data from Sierra Miwok, Lahu and Raluana, in which lexemes with comparative-similative semantics function simultaneously as, what he calls, “evincive”. Buchstaller (2004) and Meyerhoff (2002) cite Buang (na) be and Bislama olsem, both of which can be glossed as ‘like.’ Güldemann (2008: 321) contributes further examples from African languages, citing “sε ‘be like’ in at least some dialects of the Akan cluster (Lord 1993: 151–184), kε ‘like’ in the Kode dialect of Baule (Lord 1993: 201)” as well as er ‘like’ in Tiv. Thus, a wealth of cross-linguistic evidence supports the semanticpragmatic link between markers of similarity/approximation and quotation. ii. Equational and inchoative verbs such as be and become Clark and Gerrig (1990: 772) point out that the English copula is frequently used to frame reported speech: “The use of the copula [i.e. be as a quotative] 7. We will also not be concerned with paralinguistic verbs (Ware 1993: 169) such as smile, nod and with performance verbs such as throw, put etc. 8. In this discussion, we are focusing on elements that are (part of) the quotative frame which indexes – either cataphorically or anaphorically – reported discourse. However, we must not forget to mention the occurrence of ‘zero quotations’ (Mathis and Yule 1994; Yule 1993), in which reported material is not indexed via any lexical frame. This form of reported speech (or thought) is often found in highly performative narratives in which several characters are animated, often in opposition to one another.





Isabelle Buchstaller and Ingrid Van Alphen

alone (...) is a form we have recorded many instances of ” (see also Macaulay 2001 for Glaswegian Scots; Cameron 1998 for Spanish). Similarly, Güldemann (2008: 306) reports that in Bedauye, an, which is apparently identical with one of several equational verb stems, can also occur in quotative constructions. iii. Generic performance or action verbs such as do9 A number of researchers have pointed out that quotative constructions can consist of semantically generic verbs, which basically mean make or do. Tannen (1986), for example, pointed out that Greek kano can be used to enquote reported speech and German machen is regularly recruited to introduce more mimetic quotes such as gestures and sounds. Güldemann (2008: 310) points to a colloquial Hebrew quotative construction which is based on the verb asot ‘do’ (Zuckermann 2006). He also mentions quotation introduction in Pastaza Quechua via rana ‘do, make’ and in Biron via ye ‘do’. iv. Motion or transfer verbs such as go English quotative go has been described widely in the literature. However, note that – apart from some reports in Germanic languages, such as Dutch komen ‘come’, and German gehen ‘go’ – motion verbs tend not to be amongst the widely recruited source constructions for quotation. Typologically, this construction is rare. Indeed, the only non-Indo-European language in which a motion verb has been clearly attested as having grammaticalized into a quotative construction is Dongala án ‘go, become’ (cf. also Armbruster 1960). v. Deictics, foregrounding devices and presentationals Historical cases of constructions with deictic or presentative meaning becoming recruited for quotation introduction are Sanskrit iti ‘thus’ (Hock 1982) and Old Georgian (rame)tu/vitarmed (which both mean ‘thus’; see Güldemann 2008: 321). In terms of contemporary spoken languages, Milroy and Milroy (1977: 54) mention the construction Here’s me in their Belfast data. Other deictic elements used cross-linguistically are Shona ti ‘be like this’ (Güldemann 2002), Bengali emon kOra ‘do thus’ (van der Wurff 1996: 270) and Plains Cree itwê ‘thus’ (Blain and Déchaine 2007). Indeed, the Germanic languages provide several well-known cases where deictics and presentationals have developed into conventionalized complementizers (for example German dass, Faroese tadh or English that, see Lockwood 1968). Güldemann (2002) further points out that quoted gestures tend to be indexed by manner deictics in languages such as Ilokano (nga kas, Streeck 1994: 255–256, see also Japanese koo ‘thus’, Streeck 1988: 71–74). Note that speakers can also recruit focus or foregrounding 9. Güldemann (2008: 310) points out that in some cases an “account like ‘do’ etc. may (...) be largely a descriptive confession that any monosemous assessment remains elusive due to the item’s extreme semantic-functional versatility”.



Preface:  Introductory remarks on new and old quotatives

markers to introduce reported speech. In Tumbuka the foregrounding particle waka ‘only, just’ (Childs 1994: 187) can be used. vi. Speaker or adressee-referring pronominals10 Güldemann (2008: 116) points out that a quotative construction can “consist of nothing but a nominal referring to a participant in the reported speech event. This is generally the (...) [speaker], but it is also possible that the (...) addressee is encoded in addition or exclusively.” We have found structures such as and he “quote” in Dutch, British and US American Englishes. Some languages, such as Tikar, even appear to have a more grammaticalised strategy, relying on a “series of special quotative pronouns, [which] provide the essential constituent of the reduced [quotation].” (Güldemann 2008: 371). A comparison between the two inventories – for new and for old quotatives – reveals that there is indeed a considerable overlap in terms of the sources of older quotative indexes and the list of recently attested innovative quotations we collected in Table 1. The congruence is especially pertinent as regards the recruitment of markers of similarity for quote introduction. Newer as well as older quotatives also have their origins in generic verbs of equation, motion and action, and markers of focus and presentation. Thus, as regards the source constructions for quotative forms, the new quotatives are recruited from similar semantic bases as the old quotatives. Ergo, while the sudden proliferation of quotative forms seems to be a new development, the sources from which these innovative forms are being recruited are certainly not new at all. 3. Situating the volume Quotation is a well-researched phenomenon. Research on reporting other speakers’ words, sounds, facial/gestural movement or thoughts (or one’s own at another point in time) has been conducted in disciplines as diverse as: – anthropology and ethnology (Besnier 1993; Boeder 2003; Briggs 1992; Hymes 1975; Lucy 1993a,b), – bilingualism (Koven 2001, 2007), – discourse analysis, pragmatics and literary stylistics (Bakhtin 1986; Bolden 2004; Fleischman and Yaguello 2004; Goffman 1981; Golato 2003; Holt and Clift 2007; the papers in Johnstone 1994; Jones and Schieffelin 2009; Leech and Short 1981; Mathis and Yule 1994; Myers 1999; Polanyi 1982; Thompson 1996; Yule 1993; Yule and Mathis 1992), 10. Note that Güldemann (2008) also mentions quote-referring pronominals, which are pronominals with quotative function. We will not discuss this type of quotative source in detail.





Isabelle Buchstaller and Ingrid Van Alphen

– sociolinguistics (Blyth et al. 1990; Buchstaller 2006a, 2011; Buchstaller and D’Arcy 2009; Cameron 1998; Cukor-Avila 2002; Erickson 1995; Ferrara and Bell 1995; Foolen 2006; Foolen et al. 2006; Mazeland 2006; Singler 2001; Tagliamonte and D’Arcy 2007; Tagliamonte and Hudson 1999; Van Alphen 2006, 2008; Vincent and Dubois 1996), – cognitive linguistics (Chafe 1994; Clark and Gerrig 1990; Foolen 2008; Pascual 2002, 2006; Sanders and Redeker 1996; Schiffrin 1981; Tannen 1986, 1988), – social psychology (Buchstaller 2006a; Dailey-O’Cain 2000; D’Arcy 2007), – functional grammar (the papers in Janssen and van der Wurff 1996; Mayes 1990; Maynard 1996), – historical linguistics and language change (Buck 1915; Deutscher 2000; Kammerzell and Peust 2002), – kinesics and gesture research (Kendon 1980, 1992; Streeck 1988, 1993, 1994; Wilcox 2004), – language typology (Coulmas 1985, 1986; Fleischmann 1999; Güldemann 2008; Klamer 2003; Li 1986; Massamba 1986; von Roncador 1988; and the papers in Coulmas 1986 and Güldemann and von Roncador 2002), – phonetics and intonation (Fonagy 1986; Klewitz and Couper-Kuhlen 1999; Kvavik 1986), – sign language research (Padden 1986; Wilcox 2004) as well as – syntax and semantics (Declerk and Tanaka 1996; Munro 1982; Partee 1973; Steever 2003; Vandelanotte 2009; Vandelanotte and Davidse 2009; Wierzbicka 1974).11 Previous research has yielded a rich and diverse knowledge-base of the phenomenon. The consolidation of findings between different strands of investigation, however, has not advanced at the same speed as research in individual disciplines. And so, to date, there has been relatively little cross-fertilization between the diverse (sub-)disciplines engaged in quotation research. Moreover, given that the investigation has been conducted on the basis of a typologically diverse range of languages or even language families, a fair share of research on the phenomenon ‘quotation’ remains within the specialist literature and is relatively inaccessible to the broader research community. What is needed is a consolidation of findings across sub-disciplines and languages in order to seek generalizations about the linguistic properties, social uses and pragmatic functions of quotation.

11. The list by no means claims to be exhaustive. It only seeks to demonstrate the breadth of inquiry into the phenomenon. See Cameron (1998) for a similar list; consider also Güldemann and von Roncador (2002) for a comprehensive bibliography. Obviously a great deal of research has been published since then.



Preface:  Introductory remarks on new and old quotatives

The present collection aims to take a step towards a more comprehensive understanding of the phenomenon ‘quotation’ by connecting research based on a wide array of genetically unrelated languages and from a range of linguistic subdisciplines. Publishing a selection of key materials in one volume and explicitly flagging up connections and convergences across and between chapters, concepts and findings we hope will ultimately lead to a more holistic perspective on quotation in all its aspects. The volume developed out of the workshop New perspectives on new (and old) quotatives, organized by the two editors and held at the Sociolinguistics Symposium 17 in Amsterdam, the Netherlands, 3–5 April 2008. In addition to selected papers presented at the workshop, we invited contributions from scholars who spearhead research on quotatives in a number of languages in order to capture the cross-linguistic similarities and differences within this linguistic domain. This collection, which aims to investigate both ‘old’ and ‘new’ quotatives, is in some ways a sequel to three highly successful edited volumes, the relatively old Coulmas (1986) Direct and Indirect Speech, Janssen and van der Wurff (1996) Reported speech: Forms and Functions of the Verb and the more recent Güldemann and von Roncador (2002) Reported Discourse: A Meeting Ground for Different Linguistic Domains, all of which report mainly on functional-typological work. We follow their lead in aiming at a broad typological spread of languages. Indeed, a defining feature of this book is the fact that it is based on a typologically diverse range of languages/language families, some of which remain under-investigated with respect to quotation, such as German Sign Language (Steinbach and Herrmann) and Japanese (Oshima and Sano). Two chapters in this volume aim to establish crosslinguistic tendencies in the forms and functions of quotation in a range of typologically unrelated languages (Güldemann and Spronck). The following languages are discussed in depth in this collection: Dutch (modern and historical), different varieties of English, including multi-ethnic London English, German, German Sign Language, Japanese and Norwegian. Furthermore, the chapters address quotation in a range of Western and non-Western languages, which are in alphabetical order: Old Akkadian, Bininj Gun-Wok, Bulgarian, Croatian, Danish, Duna, Old Egyptian, Evenki, Ewe, Georgian, Hausa, Jarawara, Kanuri, Koasati, Kolyma, Korean, Koromfé, Kunama, Kwaza, Lamang, Lao, Lele, Lezgian, Manambu, Ndyuka, Ngarinyin, Portuguese, Semalai, Serbian, Shona, Spanish, Swedish, Tariana, Tikar, Tonga-Imhambane, Toqabaqita, Tubu, Turkish, Tuvaluan, Wari, Warrwa, WestGreenlandic, Yukaghir and Zapotec Mayan. An aspect that sets this collection apart from earlier edited works on reported speech is that our collection aspires to give a common platform to thematically coherent but theoretically diverse research conducted within the field of linguistics broadly conceived. Hence, in line with the remit of the series “Converging





Isabelle Buchstaller and Ingrid Van Alphen

Evidence in Language and Communication”, the present volume draws on experts from a wide range of (sub-)disciplines in order to investigate quotation from different disciplinary angles. The scholars that have contributed to this volume come from approaches ranging from conversation analysis and language variation and change to usage-based approaches such as grammaticalization and construction grammar and to more formal strands of linguistic enquiry. Indeed, the scholars represented here have not been previously published in the same venue and were often not aware of each other’s findings and their mutual relevance. The benefits of this kind of synthetic approach have been pointed out by Meyerhoff (2002); the choice to approach a phenomenon such as ‘quotation’ from a number of different perspectives stems not from a “lack of conviction in any method or theoretical framework, but rather out of strong conviction that the full picture [of the phenomenon] ... requires explanations that eschew existing orthodoxy and assumptions of excessive modularity in the grammar” (ibid: 356). As such, research that is maximally but selectively inclusive has the potential to result in a picture that is maximally encompassing. Indeed, we would like to argue that the chapters included in this volume are mutually enriching and generate a wider and more nuanced knowledge-base than monolingual or single disciplinary approaches could deliver. This is because findings regarding the formal, functional or social properties of quotation can have important repercussions and indeed push forward the investigatory process in other fields of inquiry. As such, all contributors were encouraged to look beyond the borders of their own fields, which results in synergies and mutual relevancies. Note that, while this volume is fundamentally cross-disciplinary, the contributions are bound together by a number of over-reaching premises: an important methodological underpinning is that the contributors aim to base their analyses – as much as possible – on real-occurring data rather than on constructed examples. As such, the chapters in this collection contain a wealth of real-occurring instances of reported activity, drawn from corpora, participant observation or other types of data collection in spoken, signed or written languages (alongside data drawn from grammars and constructed examples). Furthermore, the focus of all chapters is the investigation of direct reporting (of speech, thought, gesture or sounds) rather than indirect speech. In addition, all contributions to this volume aim to relate their analysis to broader, typological considerations, drawing on other related or unrelated languages and taking into account converging evidence. A final theoretical viewpoint that pervades this volume is that all authors work within a poly-functional framework, discussing a wide range of pragmatic and interactional functions of quotative constructions. Several foci of research can be identified in the book and we have grouped the papers according to four sub-fields for easier orientation. However, the overall aim



Preface:  Introductory remarks on new and old quotatives



of this collection is for these fields to be mutually enriching, leading to a fuller and more comprehensive understanding of the phenomenon ‘quotation’ in both spoken and signed languages. Thus, broadly constructionist approaches, represented here by for example Vandelanotte, who aims to compare and contrast the structure of older forms of reporting with innovative quotations, can benefit from work on internal and external constraints on language variation and change, both synchronically (as represented by Fox) as well as diachronically (see Coppen and Foolen). Other convergences can be found between discourse analytical and functional typological research. For example Golato’s and Hasund, Opsahl and Svennevig’s work demonstrates how discourse strategies become routinized as linguistic structure, and therefore overlaps in fruitful and beneficial ways with research within the field of typology, presented here by Güldemann or Spronck. Formal investigations of quotative constructions, as demonstrated by Oshima and Sano and Herrmann and Steinbach, can both shed light and draw fruitfully on usage-based explorations of quotative structures both cross-linguistically (Güldemann) and within one linguistic system (Vandelanotte). Many other connections link the contributions in this volume and we invite the reader to detect these relationships. The edited volume explicitly takes an interdisciplinary stance, creating a rich tapestry of connected research from a range of languages and sub-disciplines, which we hope will deepen our knowledge of the phenomenon ‘quotation’. The cross-cutting nature of the volume makes the findings broadly available and relevant. Furthermore, the present collection showcases some of the most innovative and influential work on quotation that is currently being done in a range of disciplines, presenting new findings, such as an innovative approach to the patterning of be like, insight into the linguistic and social reality of a new quotative in London (Fox), the development of strategies for reporting since the 17th century in Dutch (Coppen and Foolen) and the grammar of quotation in sign languages (Hermann and Steinbach). Other pioneering results are found in Golato, who discusses the interpersonal functions of under-reported quotative strategies in German, and Oshima and Sano, who aim to show that quotation in Japanese is much less special than previously claimed. Last but not least, Güldemann and Spronck offer typological frameworks for the types of speech, thought and activity reporting found across a wide range of languages. References Antaki, Charles and Leudar, Ivan. 1990. “Claim-backing and other explanatory genres in talk.” Journal of Language and Social Psychology 9: 279–292. Armbruster, C. H. 1960. Dongolese Nubian, a Grammar. Cambridge: Cambridge University Press.



Isabelle Buchstaller and Ingrid Van Alphen Bakhtin, Mikhail M. 1986. Speech Genres and Other Late Essays. Translation by Vern W. McGee. Austin, Texas: University of Texas Press. Besnier, Niko 1993. “Reported speech and affect on Nukulaelae Atoll.” In Responsibility and Evidence in Oral Discourse, Studies in the Social and Cultural Foundations of Language 15, Jane H. Hill and Judith Irvine (eds), 161–181. Cambridge: Cambridge University Press. Blain Eleanor and Déchaine, Rose-Marie. 2007. “Evidential types: Evidence from Cree dialects”. International Journal of American Linguistics 73(3): 257–291. Blyth, Carl, Recktenwald, Sigrid and Wang, Jenny. 1990. “I’m Like, ‘Say What?!’: a new quotative in American oral narrative”. American Speech 65(3): 215–227. Boeder, Winfred. 2003. “Speech and thought representation in the Kartvelian (South Caucasian) languages.” In Reported Discourse. A “Meeting Ground for Different Linguistic Domains.” Typological Studies in Language, Tom Güldemann and Manfred von Roncador (eds), 3–48. Amsterdam, Philadelphia: Benjamins. Bolden, Galina. 2004. “The quote and beyond: Defining boundaries in conversational russian.” Journal of Pragmatics 36: 1071–1118. Briggs, Charles. 1992. “‘Since I am a woman, I will chastise my relatives.’ Gender, reported speech, and the (re)production of social relations in Warao ritual wailing.” American Ethnologist 19: 337–361. Bucholtz, Mary. 2004. “From stance to style: Innovative quotative markers and youth identities in discourse.” Paper presented at the Sociolinguistic Perspectives on Age, New York University, New York. Buchstaller, Isabelle. 2004. The Sociolinguistic Constraints on the Quotative System – US. English and British English compared. Unpublished doctoral dissertation, Edinburgh University. Buchstaller, Isabelle, 2006a. “Social stereotypes, personality traits and regional perception displaced: Attitudes towards the ‘new’ quotatives in the U.K.” Journal of Sociolinguistics 10(3): 362–381. Buchstaller, Isabelle. 2006b. “Diagnostics of age graded linguistic behavior.” Journal of Sociolinguistics 10(1): 3–30. Buchstaller, Isabelle. 2011. “Quotations across the generations: A multivariate analysis of speech and thought introducers across 5 generations of Tyneside speakers.” Corpus Linguistics and Linguistic Theory 7: 59–92. Buchstaller, Isabelle and D’Arcy, Alexandra. 2009. “Localised globalisation: A multi-local, multivariate investigation of be like.” Journal of Sociolinguistics 13: 291–331. Buchstaller, Isabelle and Van Alphen, Ingrid. 2008. “New perspectives on new (and old) quotatives.” Introduction to the workshop 146. Sociolinguistics Symposium 17, April 3–5, 2008, Amsterdam. Buck, Carl. 1915. “Words of speaking and saying in the Indo-European languages.” American Journal of Philology 36(1):1–18, (2):125–154. Butters, Ronald. 1980. “Narrative go ‘Say’.” American Speech 55: 304–07. Butters, Ronald. 1982. “Editor’s note [on be like ‘think’].” American Speech 57: 149. Cameron, Richard. 1998. “A variable syntax of speech, gesture, and sound effect: Direct quotations in Spanish.” Language Variation and Change 10: 43–83. Chafe, Wallace. 1994. Discourse, Consciousness and Time. Chicago and London: University of Chicago Press. Cheshire, Jenny and Fox, Susan. 2007. “‘This is me’: An innovation in waiting and other quotative use among adolescents in London.” Paper presented at The International Conference on Language Variation in Europe (ICLaVE4), Cyprus, June.



Preface:  Introductory remarks on new and old quotatives Childs, G. Tucker. 1994. “African ideophones.” In Sound Symbolism, Leanne Hinton, Johanna Nichols and John J. Ohala (eds), 178–204. Cambridge: Cambridge University Press. Clark, Herbert and Gerrig, Richard. 1990. “Quotations as demonstrations.” Language 66: 764–805. Clift, Rebecca. 2006. “Indexing stance: Reported speech as an interactional evidential.” Journal of Sociolinguistics 10: 569–595. Coulmas, Florian. 1985. “Direct and indirect speech: General problems and problems of Japanese.” Journal of Pragmatics 9: 41–63. Coulmas, Florian. 1986. “Nobody dies in Shangri-La: Direct and indirect speech across languages.” In Language and Linguistics: The Interdependence of Theory, Data, and Applications, Deborah Tannen and James E. Alatis (eds), 140–163. Washington, DC: Georgetown University Press. Cukor-Avila, Patricia. 2002. “She say, she go, she be like: Verbs of quotation over time in African American Vernacular English.” American Speech 77(1): 3–31. Dailey-O’Cain, Jennifer. 2000. “The sociolinguistic distribution and attitudes towards focuser like and quotative like.” Journal of Sociolinguistics 4(1): 60–80. D’Arcy, Alexandra 2007. “‘Like’ and language ideology: Disentangling fact from fiction.” American Speech 82: 386–419. Declerk, Renaat and Tanaka, Kazuhiko. 1996. “Constraints on tense choice in reported speech.” Studia Linguistica 50: 283–301. Deutscher Guy, 2000. Syntactic Change in Akkadian: The Evolution of Sentential Complementation. Oxford: Oxford University Press. Erickson, Mats. 1995. “A case of grammaticalization in Modern Swedish: The use of ba in adolescent speech.” Language Sciences 17: 19–48. Ferrara, Kathleen and Bell, Barbara. 1995. “Sociolinguistic variation and discourse function of constructed dialogue introducers: The case of be + like.” American Speech 70(3): 265–290. Fishman, Pamela. 1983. “Interaction: The work women do.” In Language, Gender and Society, Barrie Thorne, Cheris Kramarae and Nancy Henley (eds), 89–101. Rowley, MA: Newbury House. Fleischman, Suzanne 1999. “Pragmatic markers in comparative perspective. A contribution to cross-language pragmatics.” Paper presented at Pragma 99, Tel Aviv, June. Fleischman, Suzanne and Yaguello, Marina. 2004 “Discourse markers across languages,” Carol Lynn Moder and Alda Martinovic-Zic (eds), Discourse Across Languages and Cultures, 129–148. Philadelphia, Pennsylvania: John Benjamins. Foley, William A. 1997. Anthropological Linguistics. Oxford: Blackwell. Fonagy, Ivan. 1986. “Reported speech in Hungarian.” In Direct and Indirect Speech, Florian Coulmas (ed.), 333–260. Berlin: Mouton de Gruyter. Foolen, Ad. 2006. “Het performatieve van opnieuw beschouwd.” (Performative van revisited) In Artikelen van de Vijfde sociolinguïstische conferentie, Tom Koole, Jacomine Nortier and Bert Tahitu (eds), 163–174. Eburon: Delft. Foolen, Ad. 2008. “New quotative markers in spoken discourse.” In Empirische Forschung und Theoriebildung. Festschrift für Norbert Dittmar zum 65. Geburtstag, Bernt Ahrenholz, Norbert Dittmar and Ursula Bredel (eds), 117–128. Frankfurt am Main: Peter Lang. Foolen, Ad, Van Alphen, Ingrid, Hoekstra, Erik, Lammers, Henk, Mazeland, Harrie and Pascual, Esther. 2006. “Het quotatieve van. Vorm, functie en sociolinguistische variatie.” (Quotative van. Form, function and sociolinguistic variation). In Toegepaste Taalwetenschap in Artikelen, nr 76. Thema’s en trends in de sociolinguistiek 5(2): 137–149. [Special





Isabelle Buchstaller and Ingrid Van Alphen

issue (of Journal for Applied Linguistics) about themes and trends in sociolinguistics in the Netherlands 2006] Goffman, Erving. 1981. Forms of Talk. Philadelphia, Pennsylvania: University of Pennsylvania Press. Golato, Andrea. 2003. “Self-quotation in German: Reporting on past decisions.” In Reported Discourse: A Meeting Ground for Different Linguistic Domains, Tom Gueldemann and Manfred von Roncador (eds), 49–70. Amsterdam: Benjamins. Güldemann, Tom. 2001. Quotative Constructions in African Languages: A Synchronic and Diachronic Survey. Habilitation Thesis: Institute für Afrikanistik, Universität Leipzig. Güldemann, Tom. 2002. “When ‘say’ is not say: The functional versatility of the Bantu quotative marker ti with special reference to Shona.” In Reported Discourse: A Meeting Ground for Different Linguistic Domains, Typological Studies in Language 52. Tom Güldemann and Manfred von Roncador (eds), 253–287. Amsterdam: John Benjamins. Güldemann, Tom. 2008. Quotative Indexes in African Languages: A Synchronic and Diachronic Survey. Empirical Approaches to Language Typology 34. Berlin: Mouton de Gruyter. Güldemann, Tom and von Roncador, Manfred. (eds). 2002. Reported Discourse: A Meeting Ground for Different Linguistic Domains. Typological Studies in Language 52. Amsterdam: John Benjamins. Haspelmath, Martin and Buchholz, Oda. 1998. “Equative and similative constructions in the languages of Europe.” In Adverbial Constructions in the Languages of Europe, Johan van der Auwera and Dónall P. Ó Baoill (eds), 277–334. Berlin, New York: Mouton de Gruyter. Hock, Hans Heinrich. 1982. “The Sanskrit quotative: A historical and comparative study.” Studies in the Linguistic Sciences 12: 37–96. Holt, Elizabeth and Clift, Rebecca (eds). 2007. Reporting Talk: Reported Speech in Interaction. Cambridge: Cambridge University Press. Hymes, Dell. 1975. “Breakthrough into Performance.” In Folklore: Performance and Communication, Dan Ben-Amos and Kenneth Goldstein (eds), 11–74. The Hague: Mouton de Gruyter. Janssen, Theo and van der Wurff, Wim (eds). 1996. Reported Speech: Forms and Functions of the Verb. Amsterdam: John Benjamins. Johnstone, Barbara (ed.). 1994. Repetition in Discourse: Interdisciplinary Perspectives. Vol. 1 and 2. Norwood, New Jersey: Ablex. Jones, Graham and Schieffelin, Bambi. 2009. “Enquoting voices, accomplishing talk: Uses of be + like in Instant Messaging.” Language & Communication 29(1): 77–113. Kammerzell, Frank and Peust, Carsten. 2002. “Reported speech in Egyptian: Forms, types, and history.” In Reported Speech: A Meeting Ground for Different Linguistics Domains, A. Güldemann and Manfred von Roncador (eds), 289–322. Amsterdam: Benjamins. Kendon, Adam. 1980. “Gesticulation and speech: Two aspects of the process of utterance.” In The Relationship of Verbal and Nonverbal Communication. Contributions to the Sociology of Language, Mary Key (ed.), 207–227. The Hague: Mouton de Gruyter. Kendon, Adam. 1992. “Some recent work from Italy on ‘Quotable gestures’”. Journal of Linguistic Anthropology 2: 92–108. Klamer, Miriam. 2003. “‘Report’ constructions in Kambera (Austronesian)”. In Reported Speech as a Meeting Ground for Different Linguistics Domains, Tom Güldemann and Manfred von Roncador (eds), 323–340. Amsterdam: Benjamins. Klewitz, Gabriele and Couper-Kuhlen, Elizabeth. 1999. “Quote-Unquote? The role of prosody in the contextualization of reported speech sequences.” Journal of Pragmatics 4: 459–485.



Preface:  Introductory remarks on new and old quotatives  Koven, Michèle. 2001. “Comparing bilignuals’ quoted performances of self and others in tellings of the same experience in two languages.” Language in Society 30: 513–558. Koven, Michèle. 2007. Selves in Two Languages. Bilinguals’ Verbal Enactments of Identity in French and Portuguese. Amsterdam: Benjamins. Kvavik, Karen H. 1986. “Characteristics of direct and reported speech prosody: Evidence from Spanish.” In Direct and Indirect Speech, Florian Coulmas (ed.), Berlin: Mouton de Gruyter. Lakoff, George. 1987. Women, Fire, and Dangerous Things. Chicago: University of Chicago Press. Lakoff, George, and Johnson, Mark. 1980. “The metaphorical structure of the human conceptual system.” Cognitive Science 4(2): 195–208. Leech, Geoffrey and Short, Michael. 1981. Style in Fiction: A Linguistic Introduction to English Fictional Prose. London, Longman. Li, Charles. 1986. “Direct speech and indirect speech: A functional study”. In Direct and Indirect Speech, Florian Coulmas (ed.), 29–45. Berlin: Mouton de Gruyter. Lockwood, W.B. 1968. Historical German Syntax. Oxford: Clarendon Press. Lord, Carol. 1993. Historical Change and Serial Verb Constructions. Amsterdam: Benjamins. Lucy, John (ed.). 1993a. Reflexive Language: Reported Speech and Meta-Pragmatics. Cambridge: Cambridge University Press. Lucy, John. 1993b. “Reflexive language and the human disciplines.” In Reflexive Language: Reported Speech and Metapragmatics, John Lucy (ed.), 9–32. Cambridge: Cambridge University Press. Macaulay, Ronald K. 1987. “Polyphonic monologues: Quoted direct speech in oral narratives.” IPRA Papers in Pragmatics 1:1–34. Macaulay, Ronald. 2001. “You’re like ‘why not?’ The quotative expression of Glasgow adolescents.” Journal of Sociolinguistics 5(1): 3–21. Maschler, Yael. 2001. “Veke’ilu haragláyim sh’xa nitka’ot bifním kaze (‘and like your feet get stuck inside like’): Hebrew kaze (‘like’), ‘ke’ilu (‘like’), and the decline of Israeli dugri (‘direct’) speech.” Discourse Studies 3: 295–326. Maschler, Yael. 2002. “On the grammaticization of ke’ilu (‘like’, lit. ‘as if ’) in Hebrew talk-in-interaction.” Language in Society 31(2): 243–276. Massamba, Daniel. 1986. “Reported Speech in Swahili.” In Direct and Indirect speech, Florian Coulmas (ed.), 99–120. Berlin: Mouton de Gruyter. Mathis, Terri and Yule, George. 1994. “Zero quotatives.” Discourse Processes 18: 63–76. Mayes, Patricia. 1990. “Quotation in spoken English.” Studies in Language 14: 325–363. Maynard, Senko. 1996. “Multivoicedness in speech and thought representation: The case of selfquotation in Japanese.” Journal of Pragmatics 25: 207–226. Mazeland, Harrie. 2006. “‘VAN’ as a quotative in Dutch: Marking reported speech as a typification.” In Artikelen van de Vijfde Sociolinguïstische Conferentie, Tom Koole, Jacomine Nortier and Bert Tahitu (eds), 354–365. Delft: Eburon. Meyerhoff, Miriam. 2002. “All the same? The emergence of complementizers in Bislama.” In Reported Discourse. A Meeting Ground for Different Linguistic Domains. Tom Güldemann and Manfred von Roncador (eds), 341–362. Amsterdam: Benjamins. Milroy, Lesley and Milroy, Jim. 1977. “Speech and context in an urban setting.” Belfast Working Papers in Language and Linguistics 2: 1–85. Munro, Pamela. 1982. “On the transitivity of say-verbs.” In Studies in Transitivity. Syntax and Semantics, Paul Hopper and Sandra Thompson (eds), 301–318. New York, San Francisco, London: Academic Press.

 Isabelle Buchstaller and Ingrid Van Alphen Myers, Greg. 1999. “Unspoken speech: Hypothetical reported discourse and the rhetoric of everyday talk.” Text 19(4): 571–590. Norten, Caroline. 2008. “The quotatives system in Newcastle, Gateshead and Sunderland.” Unpublished term paper, Newcastle University. Ochs, Elinor and Taylor, Carolyn 1996. “‘The father knows best’ dynamic in family dinner narratives.” In Gender Articulated: Language and the Socially Constructed Self, Kira Hall (ed.), 97–121. London: Routledge. Padden, Carol A. 1986. “Verbs and role-shifting in American Sign Language.” In Proceedings of the Fourth National Symposium on Sign Language Research and Teaching, Carol A. Padden (ed.), 44–57. Silver Spring, Maryland: National Association of the Deaf. Partee, Barbara. 1973. “The syntax and semantics of quotation.” In A Festschrift for Morris Halle, Stephen Anderson and Paul Kiparsky (eds), 410–418. New York: Rinehart and Winston. Pascual, Esther. 2002. Imaginary Trialogues: Conceptual Blending and Fictive Interaction in Criminal Courts. Utrecht University Dissertation Series 68. Utrecht, The Netherlands: Landelijke Onderzoekschool Taalwetenschap (LOT). Pascual, Esther. 2006. “Fictive interaction within the sentence: A communicative type of fictivity in grammar.” Cognitive Linguistics 17 (2): 245–267. Polanyi, Livia. 1982. “Linguistic and social constraints on storytelling.” Journal of Pragmatics 6 (5–6): 509–524. Rathje, Marianne. 2011. “Quotations and quotatives in the speech of three Danish generations.” In: Frans Gregersen, Jeffrey Parrott and Pia Quist (eds.), Language Variation – European Perspectives III: Selected papers from the 5th International Conference on Language Variation in Europe (ICLaVE 5), 71–82, Amsterdam: Benjamins. Redden, James E. 1979. A Descriptive Grammar of Ewondo. Occasional Papers on Linguistics 4. Carbondale: Department of Linguistics, Southern Illinois Romaine, Suzanne and Lange, Deborah. 1991. “The use of like as a marker of reported speech and thought: A case of grammaticalization in progress.” American Speech 66: 227–279. Sanders, José and Redeker, Gisela. 1996. “Perspective and representation of speech and thought in narrative discourse.” In Spaces, Worlds, and Grammar, Gilles Fauconnier and Eve Sweetser (eds), 290–317. Chicago, University of Chicago Press. Schiffrin, Deborah. 1981. “Tense variation in narrative.” Language 57: 45–62. Schourup, Lawrence. 1982a. Common Discourse Particles in English Conversation. Working Papers in Linguistics 28, Ohio State University, Department of Linguistics. Schourup, Lawrence. 1986. “Quoting with go ‘say’.” American Speech 57: 149. Siamou, Mariza. 2004. A ‘new’ quotative in Greek? Unbublished Master’s Thesis, Institute for Linguistics, University of Amsterdam. Singler, John. 2001. “Why you can’t do a VABRUL study of quotatives and what such a study can show us.” University of Pennsylvania Working Papers in Linguistics 7(3): 257–278. Steever, Stanford B. 2003. “Direct and indirect discourse in Tamil.” In Reported Discourse: A Meeting Ground for Different Linguistic Domains. Tom Güldemann and Manfred von Roncador (eds), 91–108. Amsterdam: Benjamins. Stein, Ellen. 1990. “I’m sittin’ there: Another new quotative?” American Speech 65: 303. Streeck, Jürgen. 1988. “The significance of gesture: How it is established.” Papers in Pragmatics 2(1–2): 60–83. Streeck, Jürgen. 1993. “Gesture as communication I: Its coordination with gaze and speech.” Communication Monographs 60: 275–299.



Preface:  Introductory remarks on new and old quotatives



Streeck, Jürgen. 1994. “Gesture as communication II: The audience as co-author.” Research on Language and Social Interaction 27(3): 239–267. Sweetser, Eve. 1987. “Metaphorical models of thought and speech: A comparison of historical directions and metaphorical mappings in the two domains.” In Proceedings of the Thirteenth Annual Meeting of the Berkeley Linguistics Society, Jon Aske, Natasha Beery, Laura Michaelis and Hana Filip (eds), 446–459. Berkeley, California. Sweetser, Eve. 1991. From Etymology to Pragmatics. Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. Tagliamonte, Sali and D’Arcy, Alexandra. 2007. “Frequency and variation in the community grammar. Tracking a new change through the generations.” Language Variation and Change 19: 199–217. Tagliamonte, Sali and Hudson, Rachel. 1999. “Be like et al. beyond America: The quotative system in British and Canadian youth.” Journal of Sociolinguistics 3: 147–172. Tannen, Deborah. 1986. “Introducing constructed dialogue in Greek and American conversational and literary narrative.” In Direct and Indirect Speech, Florian Coulmas (ed.), 311–332. Amsterdam: Mouton de Gruyter. Tannen, Deborah. 1988. Connecting Observation and Understanding. Lectures from the 1985 LSA/TESOL and NEH Institutes. Vol. XIX of Discourse Processes. Norwood, New Jersey: Ablex Publishing. Thompson, Geoff. 1996. “Voices in the text: Discourse perspectives on language reports.” Applied Linguistics 17: 501–530. Van Alphen, Ingrid. 2006. “Ik had zoiets van ‘doei’. Interactioneel sociolinguïstische aspecten van van-citaties.” [I was like ‘bey’. Interactional sociolinguistic aspects of van-quotations.] In Artikelen van de Vijfde sociolinguïstische conferentie, Tom Koole, Jacomine Nortier and Bert Tahitu (eds), 29–42. Delft: Eburon. Van Alphen, Ingrid. 2008. “Het zijn niet mijn woorden hoor’. Gender and (pseudo-)citaten.” [These are not my words hey, Gender and pseudo-quotations]. In Tijdschrift voor Genderstudies, Themanummer Taal, [Journal for Genderstudies, Special Issue on Language], 11 (1): 37–52. Van Alphen, Ingrid. 2009. “Quotatives in global perspective.” Paper given at the OAP Symposium, December 18th, Amsterdam Center of Language and Communication, University of Amsterdam. Vandelanotte, Lieven. 2009. Speech and Thought Representation in English: A Cognitive-functional Approach. Berlin/New York: Mouton de Gruyter. Vandelanotte, Lieven and Davidse, Kristin. 2009. “The emergence and structure of be like and related quotatives: A constructional account.” Cognitive Linguistics 20: 777–807. van der Wurff, Wim. 1996. “Sequence of tenses in English and Bengali.” In Reported Speech, Theo Janssen and Wim van der Wurff (eds), 261–286. Vincent, Diane and Dubois, Sylvie. 1996. “A study of the use of reported speech in spoken language.” In Sociolinguistic Variation: Data, Theory, and Analysis, Jennifer Arnold, Renée Blake, Brent Davidson, Scott Schwenter and Julie Solomon (eds), 361–374. Stanford, California: CSLI. Vološinov, Valentin N. 1973. Marxism and the Philosophy of Language. Cambridge, Massachusetts: Harvard University Press. von Roncador, Manfred. 1988. Zwischen Direkter und Indirekter Rede: Nichtwörtliche Direkte Rede, Erlebte Rede, Logophorische Konstruktionen und Verwandtes. Tübingen: Max Niemeyer.



Isabelle Buchstaller and Ingrid Van Alphen Waksler, Rachelle. 2001. “A new all in conversation.” American Speech 76(2): 128–138. Ware, Jan. 1993. “Quote formulae in The Final Diagnosis.” Journal of Translation and Text Linguistics 6(2): 161–178. West, Candice and Zimmerman, Don. 1987. “Doing gender.” Gender & Society 1: 125–151. Wierzbicka, Anna. 1974. “The semantics of direct and indirect discourse.” Papers in Linguistics 7: 267–307. Wilcox, Sherman. 2004. “Gesture and language; cross-linguistic and historical data from signed languages.” Gesture 4: 43–73. Yule, George. 1993. “Vera Hayden’s dilemma. Or the indirection in direct speech.” In Principles and Prediction: The Analysis of Natural Language, Mushira Eid and Gregory Iverson (eds), 233–242. Amsterdam: Benjamins. Yule, George and Mathis, Terrie. 1992. “The role of staging and constructed dialogue in establishing speaker’s topic.” Linguistics 30: 199–215. Ziv, Yael. 1998. “Hebrew KAZE as discourse marker and lexical hedge: Conceptual and procedural properties.” In Discourse Markers: Descriptions and theory, Andreas H. Jucker and Yael Ziv (eds.), 203–221. Amsterdam: John Benjamins. Zuckermann, Ghil’ad. 2006. “Direct and indirect speech in straight-talking Israeli.” Acta Linguistica Hungaria 53(4): 467–481.

part i

Discourse perspectives

Impersonal quotation and hypothetical discourse Andrea Golato

University of Illinois at Urbana-Champaign This chapter uses Conversation Analysis to study hypothetical reported discourse, that is, reportings consisting of discourse that is entirely fictitious. The paper demonstrates that hypothetical discourse in German is used for three different interactional functions: (a) to model the discourse of others, (b) to back or illustrate a claim in challenges, accounts or explanations, and (c) to (co-) construct humorous stories. In the first environment, the hypothetical discourse is produced in response to a prior turn (typically a complaint). In the latter two environments, the producer of the hypothetical discourse first utters an assessment which is then subsequently illustrated with hypothetical discourse. The paper discusses both the interactional functions and the turn design of hypothetical discourse, outlining how hypothetical discourse accomplishes actions different from those of other forms of reported discourse in German and English data. Keywords: Conversation Analysis, hypothetical discourse, claim backing, German

1. Introduction1 It is widely known that a good part of our daily interactions involve reporting the discourse of others (Bakthin 1981). In this chapter, the focus is on reported discourse of a particular kind, namely on hypothetical discourse (Haberland 1986). That is, it focuses on reportings consisting of discourse that is entirely fictitious and, thus, was never (and, at times, could never have been) uttered in a prior 1. I would like to thank Maria Egbert, Jürgen Streeck, Amanda Huensch and Peter Golato for comments on different versions of this chapter, and Matt Garley for identifying data samples. I am also grateful to the editors and two anonymous reviewers for their constructive feedback. All remaining errors are mine.



Andrea Golato

interaction. It includes ficticious or phantasized discourse as well as possible discourse (i.e. something one might have said in the past or may say in the future). The following excerpt contains an example of hypothetical discourse in lines 7–8. Ingo (I) is soon returning to Germany from an exchange year at an American university. His return is earlier than he had anticipated during the previous year, leaving him time for a summer job in Germany. Prior to this segment, Klaus (K) had asked Ingo if he would have the same summer job as in the previous year. Ingo denies this and then provides an account of why not, explaining that when his boss had asked him last year if he wanted to work for him again next year, Ingo had turned down this offer because he knew he would then be studying in the US. Of particular interest are line 6, where an assessment is made (see arrow a), and lines 7 and 8, where the hypothetical discourse occurs (see arrow b). Example 1: [Audio Ingo_2B_22_Arbeit]2  1 I: und da hab’ ich nein gesacht weil ich nach and then I said no because I am going  2  3 K:  4 I:  5 K:

amerika gehe und so. das[war echt total nett. to america and such. that[was totally nice. [ [ja un[d ähm: (0.5) a- jetz kann ichan[d uhm: (0.5) b- now I can[ [mhm. [uhum.

a →  6 I:

ich ich hab irgnwie so blödes ich ich habe irgendwie so blödes i i have+prs.1sg somehow such stupid i i somehow feel stupid

 b →  7

*animated voice, higher pitch* gefühl dann da wieda hinzuge:hn. .h *herr gefühl dann da wieder hin-zu-gehen *herr feeling then there again to-inf-go *mister then to return to them again. .h *mister

b →  8

pressler pressler (name) i pressler

ich möchte jetz doch arbeiten.* ich möcht-e jetzt doch arbeiten.* want.ind-1sg now partl work.inf i’d like to work now after all.*

2. See Section 2 for a description of the transcription conventions.



Impersonal quotation and hypothetical discourse  9

das sieht so: .hhh that looks so: .hhh

10 K: ja (mhm). yeah (uhum). 11 I: das is so ne schwächliche position insgesamt.= that is such a weak position on the whole.= 12

deswe:gn (.) hab ich da eigentlich keinen bock that’s way (.) i really don’t feel like

13

drauf. it.

By prefacing an utterance not addressed to his current conversational partner with the address term Herr Pressler ‘Mister Pressler’3 and by using animated voice and higher pitch, Ingo clearly presents lines 7 and 8 as directly reported discourse. However, this reported discourse is obviously hypothetical, as Ingo is not reporting on a conversation that has already taken place; instead, he is acting out a situation that would arise if he were to go back to his previous employer, rescinding his prior resignation. This hypothetical discourse illustrates the previous assessment ich hab irgnwie so blödes Gefühl dann da wieda hinzuge:hn. ‘I somehow feel stupid then to return to them’. The speaker is acting out a situation, thereby providing the co-participant with evidence to ‘see for himself ’ that the prior descriptive assessment he uttered was accurate; indeed, he receives agreement from his coparticipant (line 10). Reported discourse of this kind has been observed for spoken Danish (Haberland 1986), French (Fleischman and Yaguello 2004), Dutch (Coppen and Foolen, this volume), German (Breslauer 1996; Golato 2002; Günthner 1997; Vlatten 1997), British and American English (M. H. Goodwin 1990/1991; Holt 2007; Lauerbach 2003; Myers 1999; Tannen 1983; Winchatz and Kozin 2008), Maya (Hanks 1993), and Russian (Bolden 2004).4 The present chapter builds on the work of Golato (2002) and Vlatten (1997) and discusses both the function of hypothetical discourse in German interaction and the specific forms of the quotative constructions used to introduce hypothetical discourse into the talk. It argues that hypothetical discourse in German is used in three main interactional environments:

3. For ease of reading, quotes from the transcripts are presented with regular capitalization in the text. 4. For a discussion of hypothetical discourse in written texts, see Semino, Wynne and Short (1999).





Andrea Golato

i. to model discourse for others, that is, when speakers are acting out what one could say in a particular situation; ii. to back or illustrate a position a speaker has taken in arguments or accounts, as in Example (1) above; iii. to illustrate assessments, particularly in humorous stories. Admittedly, these are widely varied interactional environments but, as the analysis will show, environments ii. and iii. share the feature that speakers take a certain stance and subsequently use hypothetical discourse to illustrate or demonstrate the prior utterance; that is, this form of reported discourse is used when speakers are attempting to warrant a position they have taken or to support what they have said. These actions have been called “claim-backings” by Antaki and Leudar (1990: 279): “Claim-backing is the use of explanations to warrant the truth of what one has said, or the way one has said it.” In presenting evidence for a claim made, a speaker marks a turn as disputable and at the same time deals with this disputability by trying to eliminate it. Prior research has shown that speakers also employ actual (as opposed to hypothetical) reported discourse in disputable contexts (see, for example, Couper-Kuhlen 2007; Holt 1996). The current paper demonstrates the import of using hypothetical over actual reported discourse and discusses the speaker’s orientation to the fictitious nature of the talk. As mentioned above, hypothetical discourse is fictitious discourse and thus has never actually been uttered in any prior interaction. Yet, it is presented as quoted direct discourse. Prior research on reported discourse that purports to report on a prior interaction has shown that even these quoted materials are typically not a verbatim rendition of what has been said earlier (Bergmann 1987; Clift 2007; Coulmas 1985, 1986; Firle 1988; Lucy 1993; Pütz 1994). Thus, reported discourse in general is not authentic to the original utterance and would better be called “constructed” discourse (Tannen 1986).5 From a conversation analytic perspective, it is irrelevant whether direct or indirect quotations are entirely faithful to an original utterance. What matters is that direct quotation is cast in a way as if it had been uttered in this particular way, thus “verisimilitude . . . rests on the intended iconicity of RS [reported speech], that is a property of speech by which the narration of an event can count for participants as isomorphic with the model speech event” (Álvarez-Cáccamo 1996: 52). In other words, the speakers are doing being 5. Tannen (1986) uses the term constructed discourse for reporting actual talk and for reporting on hypothetical speech, while for Haberland (1986) the term constructed discourse is synonymous with hypothetical discourse. In order to avoid confusion, the present article refrains from using the term constructed discourse and instead uses the term hypothetical discourse for reports on fictitious interactions, and the term reported discourse for all other types of reportings on prior talk.



Impersonal quotation and hypothetical discourse

faithful (Bergmann 1987). When reading the existing literature on reported discourse, it is apparent that the faithfulness of the reported utterance tends not to be questioned by the interactants in everyday interaction. Moreover, speakers do not orient the design of the reported discourse to some ‘original utterance’ but to the local context and the current action of troubles-tellings, storytellings, or claimbackings. By using direct discourse and thus presenting the quote as a ‘faithful’ one, reporters are re-enacting or restaging an earlier utterance and thus casting the co-participants as witnesses of that earlier utterance (Baumann 1986; Clark and Gerrig 1990; Clift 2006; Fónagy 1986; Holt 1996; 1999: 513). In most cases, coparticipants do not only become witnesses to individual utterances, but to an entire scene acted out in front of them like a play in a theater (Anaxagorou 1994; Baumann 1986; Golato 2000; Holt 2000; Wierzbicka 1974). In this way, quotation becomes restaging (Yule and Mathis 1992). As can be seen from the data excerpt presented at the beginning of the chapter, hypothetical discourse is also presented as directly quoted speech. Yet, there is evidence from the linguistic features used in the quote and often also from the quotative construction itself that the quoted speech is fictitious. As I will show in this discussion below, co-participants can frequently be seen to orient to the fictitiousness of this discourse. The chapter is structured as follows: After a discussion of the methodology and the data, I present three contexts – modeling discourse for others, claim-backings, and fictitious humorous stories – in which hypothetical discourse is regularly used in the present corpus. For each context, I analyze one or more German data samples and compare the findings to previous work on hypothetical discourse in English. In the conclusion, I discuss the importance and implications of the findings. 2. Methodology and data The present chapter employs conversation analysis (CA) as its methodology. CA is a qualitative, empirical approach to the study of non-elicited talk-in-interaction, either in everyday or institutional talk. It seeks to discover the methods by which members of a society produce a sense of social order through their turns-at-talk and sequences of turns-at-talk. It further examines how language both creates and is created by social context. In addition, CA has the goal of revealing the systematicity of conversation and showing how patterns of communication unfold. Researchers in CA adopt an emic approach, that is, they take a member’s perspective when analyzing stretches of talk-in-interaction. The assumption is that because interactants design their talk for each other, a current speaker displays in their talk if and how they understood a prior turn and how they orient to it (Schegloff 1984: 37). Thus, an analyst can use a next turn to interpret the actions and meanings





Andrea Golato

of prior turns; put differently, analysts can make use of a “next-turn proof procedure” (Hutchby and Wooffitt 1998: 15). The data for this study come from a corpus of 35 hours of naturally occurring, mundane German conversation, and include both videotaped face-to-face interactions and audiotaped telephone conversations among friends and family. They were recorded between 1996 and 2002. Some additional German data were provided by other researchers (as indicated). The interactants are from different geographical regions in Germany, they belong to the middle or upper-middle class, are of both sexes, and are between 23 and 70 years of age.6 Twenty-eight sequences of talk in the data featured hypothetical discourse for a total of 37 times (i.e. in some instances, hypothetical discourse was used several times by the same speaker or by different speakers within the sequence of talk, as the discussion below will show). All sequences were transcribed according to the transcription conventions established by Gail Jefferson, as described in Atkinson and Heritage (1984, see also the Appendix). The German examples also contain an idiomatic translation in italics. In addition, the lines containing hypothetical discourse are glossed according to the Leipzig Glossing Rules. The lines that are glossed have four lines of transcript: the first is the German original rendered in Jeffersonian style transcription, which renders the speech as it sounds, including hesitations and inbreaths. The second line provides a transcription in standard German, the third contains the Leipzig gloss, and the fourth is an idiomatic translation. The data for the comparisons with English come mostly from existing publications on reported discourse, from the Santa Barbara Corpus of Spoken American English (Du Bois, Chafe, Meyer and Thompson 2000), or were provided by other researchers (as noted below). 3. Analysis: Hypothetical discourse in conversation In this section, I present a data-driven account of the three contexts in which hypothetical discourse can be found in the present corpus: namely, modeling discourse, claim-backings and fictitious hypothetical stories. The data are analyzed in terms of form and function of the hypothetical discourse while taking the larger context of the utterance into consideration. The German findings are then compared to prior work on English. 3.1

Modeling discourse

The corpus yielded four sequences with a total of four instances of hypothetical discourse in which participants can be seen to model discourse for others, that is, 6. The sociolinguistic variables of the speakers are listed to provide a clearer description of the data, but were not the focus of the analysis.



Impersonal quotation and hypothetical discourse

to show what one should or could have said in a given situation as displayed in the following excerpt. Anne (A), who is a pharmacist, has just complained that when a particular doctor’s office is asking her to hold the line, she can overhear long stretches of sensitive conversations in the background. Tini (T), who works for a different doctor’s office, has had a similar experience with them and commented that she did not have time to be left on hold. To this Anne responds in line 1. The turns of interest are line 2, in which a quotative construction is used (see arrow a) and line 3, in which the hypothetical talk occurs (see arrow b). Example 2 [Video DAA1 33:20]7 1 A: glaubst du denn dass ich zeit habe? do you think that I have time?

→ a 2 T:

du musst nu:r bölken ganz ge- laut du muss-t nur bölken ganz ge laut you must.ind-2sg only bellow.inf very (?) loud you o:nly have to bellow very loud

→ b 3 T:

APOTHEKE HIE::R! Apotheke hier! Pharmacy.sg here PHARMACY HE::RE!

4

(.)

5 T:

irgendwie [sowas musste sagn. irgendwie [so etwas muss-t du sagen somehow [like this must-2sg you say.inf something [like that you have to say. [

6 B: 7 H2:

[he he he he [hee [ [he he he ja? [he he he really?

In line 2, Tini gives Anne advice as to what she ought to do in situations where she is put on hold. Tini gives this advice by first describing how Anne ought to speak (i.e., laut bölken ‘bellow loudly’) and then acting this advice out, thereby modeling what to say and providing a demonstration of the former description. When there is no uptake from the co-participants, Tini’s next line of talk indicates that her prior talk is only one example of what one might say. The other co-participants overlap with laughter.

7.

I would like to thank Maria Egbert for making this data segment available to me.





Andrea Golato

In the four instances of this type of talk in the present corpus, the quotation formats make explicit that the talk serves as a model to the co-participants for what one might say in a particular situation. They contain modal verbs such as müssen ‘must’ or können ‘can’ and a verb of saying, and they are phrased as overt directions to the co-participant, thereby indicating that they should be taken as an example of what one might say.8 Hypothetical discourse serving the function of showing one’s co-participants what one should or could say has also been found in the American English discourse of some focus group participants, where participants use this strategy to model discourse for others (Myers 1999: 579–580). In the following excerpt, the focus group leader introduces a sorting task and exemplifies the type of talk that is expected from focus group participants: Example 3: (Myers 1999: 579) Mod: If everyone could join in and so that it’s,  either physically or what about this one where should this go

Note also that Goodwin (1990/1991) found that when young African-American girls respond to instigating stories, they illustrate what they will say to the respective party in the future. This can be seen in Example 4. Example 4: (M. H. Goodwin 1990/1991: 271) 101 Barbara: I better not see Kerry today. I’m a 102 say “Kerry I hear you was talkin’ ‘bout 103 me.”

Thus, both in English and in German, speakers use hypothetical discourse to model the discourse of what others ought to say or what oneself will say in the future. The small number of instances in German in which speakers exemplify discourse for others (or themselves) does not allow for a systematic comparison with the American English data in terms of how the modeling is accomplished (but see Myers 1999, Goodwin 1990/1991 for American English). The next section will analyze claim-backings, a context in which hypothetical discourse occurs very frequently in the German corpus. 3.2

Claim-backings

Prior research has shown that in order to assert the truthfulness of one’s claims, speakers can be seen to summon witnesses (i.e. actual people; see M. H. Goodwin 8. Future research may want to investigate what other formulations, apart from modals, might be used for this purpose.



Impersonal quotation and hypothetical discourse

1990). When it is impossible to summon a witness, interactants frequently invoke third parties or quote them in order to back a claim (M. H. Goodwin 1982, 1990; Heritage and Robinson 2006; Holt 1999; Potter 1996; Wooffitt 1992, 2001) or to provide evidence for what they said (Clark and Gerrig 1990; Clift 2006, 2007; Holt 1996, 2000, 2007). Tannen (1983) shows that, at times, speakers insert stories, and with them reported discourse, into the interaction in order to support a point they are making. Polanyi (1982: 166) calls stories of this kind “throwaway stories” because the only reason for telling them is to back a claim. In addition, CouperKuhlen (2007) showed for American and British English that speakers regularly quote the actual speech of others in the course of providing accounts and assessments in non-narrative contexts. By doing so, speakers provide a demonstration of the prior assessment or account, thereby substantiating and authenticating it. This can, of course, be done in German as well, as the following data segment shows. Here, Irene (I) is telling her friend that she called her advisor (Mr. Meier) when she found out that she was not selected for a job training program. The turns containing the hypothetical discourse are marked by arrows. Example 5: [Audio Oregon_2B_10.00] 1 I: isch hab nur dann (.) wie isch des erfahrn then only (.) when I found 2

hab- (.) danach dann mit dem herrn meier in out- (.) after that I then phoned mr meier in

3 → 4

rosenheim telefoniert und der war natürlisch rosenheim and obviously he was *breathy voice sischtlisch geschockt. (1.0) *WA::S. frau was frau what Ms. visibly shocked. (1.0) *WHAAT.Ms.

→ 5

blott.*>hat er dann zu mir gsacht.< Blott. hat er dann zu mir ge-sag-t. (name) have.aux he then to me ptcp-say-ptcp blott.>he then said to me.
kannst du nich beim kunden direkt< >can’t you change plans directly<  2 T: gleich [pläne ] when you [are at ] [ ]  3 R: [mh hehe] [mh hehe]  4 T: ändern? und[äh dann p- ] the custome[r’s? uh- then p- ] [ ]  5 B: [>ich könnte jetzt ] wenn ich [>i could now ]if i  6 B: jetzt< angenommen wenn des bei uns so now< suppose if this were as advanced  7 B: weit wäre dass das also jetzt .mhh äh with us that this well were already mhh  8 B: schon perfekt is es noch perfect is it not yet =  9 ?:

[nich.=es] is [it ]is [ ] [( )] [ ]





Andrea Golato 10 B: 11 ?: 12 R: 13 A:

noch im [entst]ehn sag [ich mal. still in[devel]opment let[’s say. [ ] [ [( ) ] [ [( ) ] [ [ [ha ha ha [ha ha ha ha .hhhe ha .hhhe

14 B: wenn das so weit wäre, dann würde ich if it were ready, then i would 15 B: sagen und dann also jetzt angenommen wir say and then suppose we 16 B: ham das super entwickelt,=die ham en plan developed that nicely,=they designed a 17 B: entworfen für diesen schönen gastronomie floorplan for this nice gastronomical 18 B: oder wie auch immer, .hh und dann äh or whatever, .hh and then uh 19 B: machen wir noch (.) plan ausdrucken aber we do still (.) print the plan but 20 B: ich nehm die diskette mit und computer, i take the diskette along and computer, 21 B: und dann kann man beim kunden dann and then you can sit at the customer’s 22 B: sitzen und kann dann vielleicht gleich and you can maybe immediately alter the 23 B: die perspektive innerhalb bildschirm perspective immediately on the 24 B: gleich dadurch laufen lassen. in den screen. in the 25 B: laden ne? kann man das ja. store right? One can do that right. 26 (.) 27 B: is ja auch [( ) it is also [( ) [

*animated, T moves



Impersonal quotation and hypothetical discourse 

→ b,c28 T:

→ c 29 T:

[oder er sacht *mensch das [oder er sag-t mensch das [or he say-3sg man that [or he says man i want right arm up, moves right arm to the right and left arm to the left ___________________|____________________ möcht ich en bisschen tiefer und das möcht-e ich ein bisschen tiefer und das want-1sg.ind i a bit lower and that that a bit lower and this should

→ c 30 T: soll brei*[ter sein. soll brei [ter sein should.3sg.ind wi[der be be wi [der. [ 31 B: [ja ja richtig .hh das könnte [yes yes right .hh one could 32 B: dann tu:n. nur das äh äh (.) erstens mal do that then. but that uh uh (.)first of all *B forms square with hands 33 B: is das ja nur *so: gro:oß der bildschirm*, it is only thi:s bi:ig the screen, 34 B: und wenn wir en plan machen und ich will and when we draw a plan and i want to 35 B: also da ne show machen, und will dann (.) well put on a show, and want to then (.) 36 B: dem also hier was zeigen, dann mach ich show him something here, then i draw 37 B: den maßstab eins zu zwanzich und dann the scale to twenty and then he

→ a 38 B:

kricht der einen plan von seinem krieg-t der einen plan von seinem get-3sg he a plan of his gets a floorplan of his

→ a 39 B:

toilettenladen der is (da brauchste) toilettenladen der ist (da brauchst du) toiletstore that is-3sg (there need-2sg you) toiletstore that is (there you need)



Andrea Golato

→ a 40 B:

wie der tisch. verstehste? dann mach wie der tisch. Versteh-st du? dann mach-e like the table.understand-2sg you? Then make-1sg like the table. get it? then i draw

→ 41 B:

ich ich ich two

→ b 42 B:

bisschen perschpektive, .hh bisschen perspektive, little bit of perspective bit of perspective, .hh and

dann frei hand dann frei hand dann free hand lines freehand

zwei striche mit nem zwei striche mit einem two lines with a with a little und dann und dann and then then

*higher pitch and breathy voice, holds on to table, leans over it gazes around tabletop → c 43 B: *BOOAAH schön OOHHH toll mein la- OOHH boah schön oh toll mein la- oh wow nice oh great my sto- oh WOOAAH nice OOHHH great my sto- OOHH

→ c,a 44 B:

toll.* .h das kann ich mit dem computer toll. das kann ich mit dem computer great that can-1sg i with the computer great.* .h that i can with the computer

→ a 45 B:

kommt nich rüber. komm-t nicht rüber. come-3sg not across. it does not get across.

46 T: ja gut. oke. da müssen [wir schon ] ne yes fine. okay. here we[have to put] on a [ ] 47 B: [verstehste?] [you know? ] 48 T: sho[w haben. ] dass man (.) das irgendwie sho[w. .] that you (.) project it [ ] 49 S/A: [da muß- ] [one mus- ] 50 T: an die wand wirft. ... on the wall....



Impersonal quotation and hypothetical discourse 

In lines 1 through 4, Tobias claims that CAD is advantageous by pointing out that it lets the user change floor plans immediately, even at a customer’s house. Bernhard tries to weaken this claim while simultaneously trying to support his own claim by turning Tobias’ claim into an unreal, counterfactual event through the use of a conditional angenommen ‘suppose’ (line 6) and the subjunctive II9 (line 14). By using angenommen, Bernhard opens up a hypothetical parallel world in which he creates a make-believe situation in which CAD is perfect. He describes what this perfect technology would allow him to do (lines 14–25): He would be able to take the plan on paper and on diskette to the customer. This is exactly the situation Tobias has described in lines 1 through 4: CAD would allow the salesman to change the floor plans when visiting the customer. Note that Tobias demonstrates that he is also visualizing this perfect situation by contributing to it; that is, in lines 28–30, Tobias is orienting to the hypothetical nature of the imagined situation by letting the imaginary customer come to life and allowing him to say something. Note that Tobias is using the quotative verb sagen ‘say’ in the present tense followed by the hypothetical dialogue in lines 28–30: mensch das möcht ich en bißchen tiefer und das soll breiter sein ‘man I want that a bit lower and this should be wider’. In other words, Tobias is producing an illustration of his earlier argument ‘being able to change plans’ in the form of hypothetical dialogue. It is clear that we are dealing with direct discourse because of the address term mensch ‘man’, which would not occur in indirect reported discourse (Coulmas 1986), and by the use of the first person pronoun which constitutes a speaker shift (er ‘he’ to ich ‘I’). We can therefore safely assume that the hypothetical, imaginary customer, and not Tobias, is supposed to be speaking these lines. Thus, in inventing talk of a hypothetical character, Tobias is changing footing (Goffman 1981). In Goffman’s (1981: 221) terms, he presents himself only as the “sounding box” but not as the author of the customer’s talk (although he actually is the author since he invented this character and his discourse). The customer’s talk becomes an illustration of the usefulness of CAD, and by presenting this illustration Tobias is able to back up his original claim. Bernhard agrees with it (line 31, in partial overlap with Tobias) and immediately uses the illustration for his own purposes. Stated differently, Bernhard and Tobias are co-constructing the grounds for their opposing arguments. Yet, while Tobias uses the hypothetical dialogue as support for his claim of the advantage of computer technology, Bernhard uses it as a launching point for stating and backing up his opposing opinion. He points out the disadvantages of a small laptop screen (lines 31–35) and continues his turn by saying that a floor plan 9. The subjunctive II in German is typically used for expressing unreal, hypothetical, or imaginary events and acts. It can also be used in indirect speech in order to mark the speaker’s doubt of the reported events; however, this usage is not as common (Duden 1985: 431).



Andrea Golato

drawn freehand to a much larger scale is far more advantageous in that it allows for a huge plan of a very small store (lines 38–40). Bernhard relates the effects of such a large-scale plan by enacting the fictional customer’s reaction, that is, Bernhard illustrates, and thus backs, his argument through hypothetical dialogue and bodily movement (lines 42–44). Note that the quotation is only preceded by und dann ‘and then’; however, it is very clear that it is the customer who is speaking here because Bernhard changes his voice quality. Other markers of direct speech are the change-of-state token OOHHH (Golato 2011; Heritage 1984, 2002; Holt 1996) and the assessment token BOAAH, which is uttered considerably louder than the rest and which could not occur in indirect discourse (Schourup 1983). Furthermore, gestures and bodily postures associated with the talk tend not to appear in indirect discourse. In this instance, Bernhard is acting out the customer: he holds on to the table, leans over it and gazes around it as if there were a large floor plan on the table. The co-participants – and we as analysts – become witnesses not only to reported hypothetical discourse but to an enactment of the (hypothetical) character; in effect, we become witnesses to a performance (Clark and Gerrig 1990; Holt 2007; Wierzbicka 1974). In line 44, Bernhard reverts back to his regular voice after the production of an inbreath, which here has the function of an unquote (Ebert 1986; Klewitz and Couper-Kuhlen 1999). Subsequently, Bernhard summarizes his illustration, relating it back to his claim and stating that such an effect could not be achieved with a computer. At this point, Tobias agrees (line 46) and the two speakers have established common ground.10 Thus, it can be argued that, similar to the prior example, the hypothetical quote has played a primary role in one of the interlocutors (Bernhard) winning the argument.11 In the examples above, speakers typically insert only snippets of dialogue (rather than full-blown stories)12 into their own talk to support their arguments. Hanks (1993) has observed something similar in the discourse of Mayas, who switch between real discourse and hypothetical discourse in order to reason through problematic situations, to provide accounts, or to make fun of each other. Similarly, Lauerbach (2003) notes that newscasters in British political interviews also present (and attack) stances of the opposing party as hypothetical discourse, as do some lawyers in their closing arguments in American English (Pascual 2006). In these scenarios, creating stretches of hypothetical discourse allows the interactants to 10. See Lee (2001: 39–42) for a literature review of the notion of common ground. 11. Note, however, that in this example, Tobias comes up with a face-saving strategy by describing a remedy to Bernhard’s last argument; the discussion between the two continues with Bernhard stating that he would prefer to do things other than learn to use computers (not displayed). 12. In Example 7, though, the speakers seem to be jointly creating a story.



Impersonal quotation and hypothetical discourse 

reason against the stance expressed in the hypothetical discourse. Antaki (1994: 186) observed for British English that speakers “set up different versions of what is the case in the world” when they are arguing or explaining. It seems that invoking hypothetical situations and hypothetical discourse is an ideal practice for doing so. When I started working on this particular category of reported discourse, I had expected to establish generalizations concerning the sequence in claim-backings. That is, I thought I would be able to see general trends, for example, that participants in a conversation first try to back their claims with factual and logical statements and premises, and only later in an argument turn to hypothetical situations as a claim-backing or illustration of their position. I had expected such generalizations because there is a cultural belief, inherent in the literature on formal and informal logic, that participants in a conversation link premises logically with each other in order to argue a point.13 However, this was not the case. This finding could be an idiosyncratic feature of my data since all conversations were harmonious ones among family members and friends which did not involve the discussion of very disputable or contentious topics. An area of future research would be to investigate the role of hypothetical discourse in conflict talk. So far, I have shown that in everyday conversation, hypothetical discourse is used as an illustration of an argument or an explanation in non-narrative settings. I will now broaden this discussion to other settings, specifically to classroom interaction. Given that in teaching interactions (such as in a classroom), the interactional achievement is to convey information and explanations to an audience, it should come as no surprise that hypothetical discourse can be frequently found in this setting. Indeed, this type of behavior was observed in the following data segment, where guest speakers were invited to visit a classroom in order to explain certain fields of business and economics. In the following example, guest speaker Werner (W) is trying to explain to advanced business-German students the function, duties and tasks of a controller in a company. Again, a claim is made in lines 9–10 (see arrow a) and is subsequently supported with a quotative construction (see arrow b) and hypothetical discourse (see arrow c). Example 8: [Video_SVW_B_064]  1 W: oke: .h und dann ist es ganz nett, wenn okay: . and then it is quite nice, if  2 W: man jetzt ne unabhängige abteilung hat, one has an independent department now,  3 (0.2)  4 W: die eben das ganze den mal which sorts through the whole 13. For a discussion of the relevant literature, see Antaki (1994) and Antaki and Leudar (1990).



Andrea Golato  5 W: durchleuchtet. schaut was wird thing. looks what is done  6 W: falsch gemacht. was könnte man besser wrong. what could be done  7 W: machen. better.  8

(.)

→ a  9 W:

u:nds dazu ist es wichtig, dass es nicht und dazu is-t es wichtig, dass es nicht and for that is-3sg it important, that it not a:nd for that it is important that it is

→ a 10 W:

jemand ist der in der Abteilung drin ist jemand is-t der in der Abteilung drin is-t someone is who in the department inside is not someone from within the department

→ b 11 W:

denn denn because because

es wird ja es wir-d ja it be-3sgprtl nobody will

keiner sagen (0.2) keiner sag-en nobody say-inf say (0.2)

*subdued tone of voice → c 12 W: *ah wir sind ziemlich schlecht. ah wir sind ziemlich schlecht. prtl we be-3pl.ind rather bad oh we are pretty bad. 13 S: he he → c 14 W:

[he [he [*subdued tone of voice [*wahrscheinlich liegt es an uns dass [ wahrscheinlich lieg-t es an uns dass [ probably lay-3sg.ind it with us that [probably it’s our fault that

→ c 15 W: das produkt nit 16

verkauft wird.* das produkt nicht verkauft wird the product not verkauft.ptcp be.3sg.aux.ind the product is not sold. *some students laugh and nod *(0.2)

17 W: also müßte das irgendwie extra sein. so this must somehow be extra.



Impersonal quotation and hypothetical discourse 

In this data segment, Werner is arguing the point that a company’s controlling department should be separate from the rest of the company and that a person checking on the company’s procedures should not work directly in the area they are checking on but rather work in another department. In order to back this point, Werner creates a situation in which a worker in a company blames themselves for the meager quality of the product (lines 12–15). Werner is labeling this situation as impossible (es wird ja keener sagen ‘nobody will say’), and it is clear that he is not quoting a real person but is inventing the discourse of a hypothetical speaker.14 The modal particle ja suggests that the information conveyed is assumed to be mutually shared (Möllering 2001). Clearly, Werner is presenting an impossible and ridiculous situation to the students, thus inserting humor into the interaction. Yet, the fact remains that this hypothetical discourse illustrates the point Werner had been making: namely, that a controller needs to work in an unabhängige Abteilung ‘an independent department’. After a short silence, in which the students claim understanding by nodding, Werner summarizes the point he was supporting with his hypothetical dialogue. When doing so, he shifts back to his regular voice, thereby indicating the end of the reported discourse. He goes on to explain how the controlling department is incorporated in the overall organization of the company (not shown in the transcript). Again, the hypothetical discourse constitutes a shift in footing (Goffman 1974, 1981) and allows the coparticipants (and the analyst) to judge the prior claim for themselves. Example (9) shows the use of hypothetical discourse in claim-backings in American English. Jim (J) and Al (A) are planning to attend a concert together for which the newspaper had listed the ticket prices for both lawn seats (i.e. for a spot on the lawn where concert attendees can spread a blanket, etc.) and reserved seats (i.e. assigned seats). When Jim called Ticketmaster to inquire about lawn seats, he was told that the computer would select the best available ones. Arrow a shows the assessment and arrow b the hypothetical discourse. Example 9: (Audio Concert tickets_lawnseats)15  1 J: so it sou::nds like .hhh that a:ll the seats  2 are rese:rved.  3 (1.0)  4 A: Even law:nsea:ts?  5 J: I dun[n  6 A: [How can you have reserved law- wait14. One of the editors pointed out that this might also be read as a model of ‘what not to say’. While this is true, the fact remains that the quotative construction and the hypothetical discourse serve to illustrate and back up the speaker’s prior claim. 15. I would like to thank Emanuel Schegloff for making this data segment available to me.



Andrea Golato

→ a  7 A:  8 J:  9 J: 10 A: → b 11 A: 12 J: → c 13 A: 14 A: 15 J: 16 J: 17 A: 18 J:

that makes n[o se:nse. [DI- I agree: it makes no [sense. [Nheh .hhh [Yer sitting- (.)= [But wh*slight staccato =*te:n feet from this po:st. (.) on the law:n,* what the helliztha:t. I dunno: I- I- don't- anyway I don't understa:nd what- (0.2) what is goin on over there, .hhhhhhh HHHHHHH >so whaddo we do:.< Well I dunno:. I ordered:two: sea:ts.

After displaying trouble in understanding Jim’s informing (see the silence in line 3, the repair initiator in line 4, and the self-interrupted challenge in line 6), Al assesses the information that all seats have been reserved as nonsensical in line 7. Hence, following Jim’s agreement in line 9, Al produces hypothetical discourse marked by a slight staccato. It is clear from this change of voice quality and pronoun shift, and from the content of the utterance that a shift in footing (Goffman 1974, 1981) is taking place and that Al is imagining and reporting what Ticketmaster would have to say if lawn seats could indeed be booked as assigned seats: Yer sitting ten feet from this post on the lawn. This quote allows both Al and Jim to hear how absurd such a statement would sound. The hypothetical discourse is thus an illustration of his prior assessment in line 7 (this makes no sense), and indeed Al negatively assesses it again following the hypothetical talk. Couper-Kuhlen (2007) and Clift (2007) discuss similar examples of speakers introducing reported discourse after an assessment to provide evidence for their assessment. The difference between their examples and the one above is that in the example above, hypothetical talk rather than actual talk is inserted, as the speakers are discussing a counterfactual situation while trying to make sense of an actual event. In this sense, Al and Jim are engaged in “thought experiments” of the same kind also observed in American focus group discussion (Myers 1999: 584), in which speakers use hypothetical discourse to discuss potential problems or limitations of ideas. Note that there was variation in terms of how the hypothetical discourse was introduced by the speakers. In seven instances, speakers employed zero quotatives (Mathis and Yule 1994), that is, they did not produce a conventionalized construction clause or marker that introduces the discourse (see Example 7 and 9). When a quotative construction was used, it did not refer to any specific person but to a very ‘general’, non-specific person, for instance es ‘it’ in da geht’s denn halt so ‘there it goes like this’ (see Example 6), man ‘one’ in man kann zum chef sagen ‘one can



Impersonal quotation and hypothetical discourse 

say to the boss’ (Example 8), or an unspecified client in oder er sacht ‘or he says’ (Example 7). These quotative constructions are particularly suited to the type of action the speakers are undertaking; given that they are indeed not citing a specific person – note the use of keener ‘nobody’ in Example 8 – but an invented, hypothetical one, it makes sense that the speakers are using either generic quotative constructions or zero quotatives. 3.3

Fictitious humorous stories

In the previous section, hypothetical discourse was inserted in non-narrative sequences. In Holt’s (2007) corpus, speakers of British English can be seen to employ hypothetical discourse in enactments that expand the initiation of a joke. Winchatz and Kozin (2008) analyze this interactional phenomenon in great detail for their American English corpus, labeling it “comical hypothetical”, that is a telling located between the genres of joke and storytelling (Winchatz and Kozin 2008: 402). In these comical hypotheticals, speakers invent imaginary situations that they take to the absurd. Within these imaginary situations, the interactants also shift footing and report on the speech of the characters within the story in the form of (invented) direct quotations. These comical hypotheticals allow speakers to display cultural and historical references, which is also true of other quotations and cultural typifications. Moreover, such stories bring “the interlocutors symbolically closer through their discursive co-creation and simultaneous utilization of the same cultural knowledge” (Winchatz and Kozin 2008: 402); that is, such stories have bonding functions. As is described elsewhere (Golato 2002; Vlatten 1997), German speakers also use hypothetical discourse for humorous tellings.16 My data show that, just as in Winchatz and Kozin’s (2008) data, speakers of German expand on a joke initiation. This joke initiation is typically done in the form of an assessment which is then illustrated through hypothetical discourse. One or more other co-participants then collaborate in the telling by creating additional hypothetical talk or by inventing additional characters and events. Example 10 illustrates this. Example 10: [Video GIS_A_448]  1 A: er arbeitet he is working  2

((dog barks once))

 3 A: bei einer computerfirma als ähm for a computer company as uhm  4 A: technical support. technical support. 16. For a discussion of how humor is used in conflict situations, see Norrick and Spitz (2008).



Andrea Golato  5

(.)

 6 A: und wurde da eingestellt mhe hi is ganz and he was employed mhe hi is quite  7 A: interessant eigentlich.= interesting actually.=  8 T: =ha =ha  9 A: er mußte da so‘n test machen, hat den he had to take some test there, passed 10 A: wohl auch bestanden. ((swallows)) war it i suppose. ((swallows)) but somehow 11 A: aber irgendwie der annahme .hh dass das he was under the impression .hh that the 12 A: ganze software support sein würde. .h whole thing would be software support. .h 13 A: u:nd was sich jetz aber als äh a:nd that turns out now to be a uh 14 A: tru:gschluss herausstellte dass er dann wrong assumption that he then he has 15 A: hinterher doch jetzt hardware sachen to do hardware stuff now after 16 A: machen muss.=er hat sich erst mal ein all.=the first thing he did was buy a 17 A: dickes buch gekauft? und ähm (.) gibt thick book? anduhm (.) is giving *smile voice 18 A: jetz am telefon irgendwelche (.) *netten now some (.) *friendly (.) advice* over

→ 1 19 A: (.)ratschläge*.=nehmen sie doch mal ihre nehm-en sie doch mal ihre take-2sg.imp you prtlprtl your the telephone.= why don’t you take your → 1 20 A:

zange(he) zange (laugh) pliers plie(h)rs and

und bre(h)che(h)n sie(h) das and brech-en sie das and break-2sg.imp you the brea(h)k(h) the



Impersonal quotation and hypothetical discourse 

→ 21 A:

li(h)nke .hhh blau(h)e tei(h)l herau(h)s. linke (breath) blaue teil heraus left blue part out le(h)ft .hhhblue(h) pa(h)rt ou(h)t.

22 A: .hhh he hehe .hhhh [ha ha he .hhh he hehe .hhhh [ha ha he [ → 2 23 R: [und dann [und dann [and then [and then

schmeißen schmeiß-en throw-2sg.imp you throw

→ 2 24 R:

sie ih(h)ren co(h)mpu(h)ter aus em sie ihren computer aus dem you your computer out the you(h)r co(h)mpu(h)ter out the

→ 2 25 R:

fenster. fenster window window.

26 A: .hhh .hhh

→ 3 27 S: 28 R: 29 S: 30 A: 31 R:

[nein das war ] [nein das war ] [no that be.3sg.pst.ind] [no that was ] [ ] [((clears throat)) ] [he hehe hi [hehehe hi [ [ha hahaha [hahahaha [ [mhemhemhemhemhe [mhemhemhemhemhe

→ 3 32 S: tut mir leid tu-t mir leid do-3sg me sorry i’m sorry 33 A:

das das the the

falsche teil. falsche teil wrong part wrong part.

[jetzt ist er abgestürzt. ] [jetzt ist er ab-ge-stürzt] [now is-3sg.indptcp-crash ] [now it crashed. ] [ ] [(verkehrtes schnell) ] [(wrong speed ) ]



Andrea Golato 34 A: .hhhh .hhhh

→ 4 35 B:

jetzt ist der computer aber abgestürzt. jetzt ist der computer aber ab-ge-stürzt now is-3sg.ind the computer prtlptcp-crash but now the computer crashed.

→ 4 36 B:

jetzt is alles schwarz. ja da jetzt ist alles schwarz ja da now be-3sg.ind all black yes prtl now everything is black. well then you

→ 4 37 B:

ham se die falsche sache abgeklemmpt. hab-en sie die falsche sache ab-ge-klemmt have-2sg.indthe wrong thing off-ptcp-take took off the wrong thing.

38 A: he he he he

→ 4 39 B:

ja dann kaufen sie bitte einen neuen. ja dann kauf-en sie bitte einen neuen yes then buy-2sg.imp you please a new yes then please buy a new one.

40 A: a he hehehe a he hehehe 41 R: °genau.° °exactly.°

→ 5 42 S: ich ka(h)nn

sie verbi(h)nden mit der ich kann sie verbinden mit der i can-1sg.ind you connect.inf with the ica(h)n conne(h)ct you with that

→ 5 43 S: 44 A: 45 R: 46 T:

abteilung. he [he he [he .hh abteilung department department. [hehe [he .hh [ [ [he he [he [hehe [he [ [ [mhe mhe[mhe [mhe mhe[mhe [ [is ja stark [that’s something



Impersonal quotation and hypothetical discourse  47 T: hat er keine chancen rüberzukommen? doesn’t he have a chance to come over?

After Annette explains in lines 1–17 that her boyfriend has a summer job for which he is not adequately prepared and that he is now trying to remedy this by reading and self-teaching, she goes on to state that he is now actually giving advice over the telephone (line 18). She assesses this advice as (.) nette (.) Ratschläge ‘(.) friendly (.) advice.’ This assessment is delivered with smile voice indicating that it is meant as an ironic remark. Note that Annette punctuates or highlights the assessment nette Ratschläge with a pause before and after uttering it. She then goes on to give an illustration of how she wants the assessment to be understood by ‘quoting’ her boyfriend through inventing discourse for him in lines 19–21. Both the content of this discourse and the laughter indicate that the assessment nette Ratschläge is not to be understood literally but as being clearly humorous. It is also apparent that her co-participants treat her utterance as an extreme and unreal illustration, not to be taken literally and one that will in all likelihood never be uttered. Ricki also constructs hypothetical discourse for Annette’s boyfriend (lines 23 and 24) which is interspersed with laugh tokens. Then another co-participant (Sybille) further develops the hypothetical discourse (line 27), which ends in laughter. Ricki and Annette join in the laughter (lines 30 and 31) and Sybille constructs further hypothetical discourse. Yet another co-participant, namely Bernhard, joins this co-construction of invented dialogue in lines 35–39, and in lines 42–43 Sybille constructs additional hypothetical discourse. Each of the co-participants in turn upgrades the hypothetical dialogue of the prior co-participant by being even more ridiculous. We can see that what started out as a short illustration of how to understand an assessment (lines 18 and 19 (.) netten (.) Ratschläge ‘(.) friendly (.) advice’) turns into a co-told hypothetical telling of somebody who gave wrongful advice to a customer, thereby breaking the customer’s computer so that the customer had to purchase a new one from the company. It is striking that several speakers can produce long stretches of hypothetical quotations – much longer ones than in other storytellings. Yet, this is not surprising for the following two reasons: firstly, invented stories of this kind have no ‘point to make’ other than to illustrate a former claim. Since there is no clear story line dictated from clear events that one has to adhere to, the co-participants need to jointly negotiate the end of the story. Secondly, as there is no real story line or event that one speaker witnessed, all participants are licensed co-tellers (or co-inventers) of the developing story.17 Myers (1999) also observed that hypothetical stories in group discussions in English can be expanded ad absurdum. We can observe that the sequence 17. I would like to thank one of the editors for pointing out that if the telling were presented with indirect discourse, it would not be humorous but would turn into a troubles-telling.



Andrea Golato

of additionally hypothetical discourse only comes to an end after Tobias’ assessment in line 46: is ja stark ‘that’s something’, which is positioned immediately following the hypothetical dialogue of others. Post-positioned assessments are a technique for displaying closure of a telling (Goodwin and Goodwin 1987). Yet, in this data segment, it is not quite clear whether the assessment is of the hypothetical story, or of the original telling by Annette that her boyfriend has a job as a computer consultant. It is noteworthy that the hypothetical discourse in fictitious stories in the present corpus is typically introduced with zero quotatives. This has also been observed for hypothetical discourse in humorous stories in English (see Holt 2007). Instead of using a quotative, speakers mark the beginning of the quote itself with shifts in deictic references, volume, voice, or pitch or with a short pause in which the speaker gets ready for an enactment (although this is not the case in Example 9 above where the context makes the change in footing clear). The use of zero quotatives is associated with creating an overall dramatic effect or a more dramatic expression of the speaker’s attitude (Mathis and Yule 1994); when co-participants co-construct additional reported speech (either hypothetical or real) with zero quotatives, they have been said to echo the attitude of the prior speaker (Mathis and Yule 1994). These features, which were observed for American English, also hold for the present German collection. Generally, in terms of the structural properties of hypothetical quotes, speakers tend to open up alternative worlds via the use of modal verbs or conditionals with verbs of saying when modeling the discourse of others. Overall, if the quote is encoded via a quotative construction, the beginning of the quote itself is sometimes also marked by a short pause in which the speaker gets ready for an enactment, or with shifts in volume, voice, or pitch (for similar observations, see Klewitz and Couper-Kuhlen1999; Vlatten 1997). The quote is always presented in direct discourse, not in indirect discourse. In other words, the invented quote is designed in the format just as if it were a genuine report on somebody else’s discourse (even if this assertion is negated: see keiner in Example 8, line 11). The quoted (yet invented) material is not limited to speech but can include nonverbal elements as well, such that the quote turns into an enactment. Of particular interest is the end of the quote. If a speaker invents quoted discourse, he or she also marks the end of their invented quote with the same features (i.e. shifts in voice, volume, or pitch) used by speakers who are quoting a real person within a storytelling (Golato 2000). Yet, even if the speaker marks the end of a hypothetical quote with an unquote, I have demonstrated that the interlocutors can collaboratively invent further quoted material so that the quote turns into an ongoing story invented by several co-tellers with different ‘voices’. I argue that this is a feature inherent in the nature of ‘invented’ stories; since they occur in alternative worlds, there is neither a principal storyline nor a principal storyteller. The floor is literally open to all.



Impersonal quotation and hypothetical discourse

4. Concluding discussion The findings of this chapter make a contribution both to the literature on reported discourse and to studies on argumentation. Argumentation and claim-backing in general have been studied in a variety of fields but with a focus on other devices and structures.18 Prior research has studied explanatory talk, specifically instances in which speakers back conversational moves, an activity which has also been labeled as accounting (Bergmann 1992; Heritage 1988; Heritage and Robinson 2006; Perakyla 1998; Pomerantz 1980; Wooffitt and Allistone 2008). The present chapter described one particular practice or device speakers have available in order to provide accounts for their statements: namely, claim-backing, defined here as speakers giving support or illustrations of something they have said. Through this practice, speakers substantiate and authenticate prior claims. The chapter’s main contribution is to the literature on reported discourse. It has shown that speakers of both English and German invent hypothetical situations replete with hypothetical discourse for three different functions: (i) modeling discourse for others, (ii) backing a claim in challenges or illustrating a claim in an account, and (iii) enacting humorous tellings. The starting point for a humorous story in the German data is always an assessment that is subsequently illustrated and enacted through the hypothetical discourse. Similarly in challenge situations and when providing accounts, speakers first make an assessment or claim which they then back or illustrate through the production of hypothetical discourse. Instead of calling upon witnesses or citing the actual words of others (see Couper-Kuhlen 2007), speakers create a hypothetical situation and characters, whose hypothetical talk is then ‘reported’ on in the current interaction. In addition, I have shown that, once a hypothetical situation has been created in support of one speaker’s argument, the co-participants can then use that situation to support their own claims. In both claim-backings and humorous stories, even the co-participants invent characters that are portrayed as talking within the hypothetical situation. In this fashion, the hypothetical world of reference is co-constructed by the members of the conversation. The practice of modeling discourse for others (i.e. articulating what one might or could say in a specific situation) is designed differently from the other two actions of claim-backings and telling humorous stories. It does not start out with an assessment or claim which then gets illustrated, but instead is produced in response 18. For a very detailed description of the work on explanatory talk and claim-backing in various schools of thinking, see the collection of essays in Antaki (1988), Antaki’s (1994) description of the state of the art in explaining and arguing, Antaki and Leudar’s (1992) argument model, and Billig’s (1996) work on arguing and thinking from a rhetoric perspective.





Andrea Golato

to a prior action by the co-participant (for example, in response to complaints or problem-tellings). There may also be other functions of hypothetical discourse in German and other languages other than the ones described here, but these three were the only ones that were identified in the present corpus of everyday conversations between friends and family. As noted earlier, the data for the present study consist of harmonious conversations and do not include conflict talk or highly sensitive topics. Data that include such topics and/or data taken from different speech exchange systems (such as courtroom interaction and classroom interaction which differ in how turntaking is organized) may yield other forms and functions of hypothetical discourse. As per my introduction, hypothetical discourse has not only been noted in British and American English and German, but also in other languages such as Danish (Haberland 1986), Dutch (Coppen and Foolen, this volume), French (Fleischman and Yaguello 2004), Maya (Hanks 1993), and Russian (Bolden 2004). Given that hypothetical discourse makes use of directly quoted discourse and that direct quotations are available to speakers in all languages, hypothetical discourse might be available in all languages. It remains to be seen, though, if the interactional functions for which it is employed are the same cross-linguistically. References Álvarez-Cáccamo, Celso. 1996. “The power of reflexive language(s): Code displacement in reported speech.” Journal of Pragmatics 25: 33–59. Anaxagorou, N. 1994. “Mood in the chronicle of Leontios Machairas.” In Themes in Greek Linguistics: Papers from the First International Conference on Greek Linguistics, Reading, September 1993, I. Philippaki-Warburton, Katarina Nicolaidis and Maria Sifianou (eds), 313–318. Amsterdam/Philadelphia: John Benjamins. Antaki, Charles. 1994. Explaining and Arguing: Social Organization of Accounts. London: Sage Publications. Antaki, Charles (ed.). 1988. Analysing Everyday Explanation. A Casebook of Methods. London, Newbury Park, Beverly Hills, New Delhi: Sage Publications. Antaki, Charles and Leudar, Ivan. 1990. “Claim-backing and other explanatory genres in talk.” Journal of Language and Social Psychology 9: 279–292. Antaki, Charles and Leudar, Ivan. 1992. “Explaining in conversation: Towards an argument model.” European Journal of Social Psychology 22: 181–194. Atkinson, J. Maxwell and Heritage, John (eds). 1984. Structures of Social Action. Studies in Conversation Analysis. Cambridge: Cambridge University Press. Bakthin, Mikhail M. 1981. “Discourse in the novel.” In The Dialogic Imagination, Mikhail M. Bakthin (ed.), 259–422. Austin, TX: University of Texas Press. Baumann, Richard. 1986. Story, Performance, and Event. Cambridge: Cambridge University Press. Bergmann, Jörg R. 1987. Klatsch. Zur Sozialform der diskreten Indiskretion [Gossip: On the social form of discrete indiscretion]. Berlin and New York: Walter de Gruyter.



Impersonal quotation and hypothetical discourse  Bergmann, Jörg R. 1992. “Veiled moralities: Notes on discretion in psychiatry.” In Talk at Work. Interactions in Institutional Settings, Paul Drew (ed.), 137–162. Cambridge: Cambridge University Press. Billig, Michael. 1996. Arguing and Thinking. A Rhetorical Approach to Social Psychology. Cambridge: Cambridge University Press. Bolden, Galina. 2004. “The quote and beyond: Defining boundaries of reported speech in conversational Russian.” Journal of Pragmatics 36(6), 1071–1118. Breslauer, Christine. 1996. Formen der Redewiedergabe im Deutschen und Italienischen. Heidelberg: Julius Groos Verlag. Clark, Herbert H. and Gerrig, Richard J. 1990. “Quotations as demonstration.” Language 66: 764–805. Clift, Rebecca. 2006. “Indexing stance: Reported speech as an interactional evidential.” Journal of Sociolinguistics 10(5): 569–595. Clift, Rebecca. 2007. “Getting there first: Non-narrative reported speech in interaction.” In Reporting Talk: Reported Speech in Interaction, Elizabeth Holt and Rebecca Clift (eds), 120–149. Cambridge: Cambridge University Press. Coulmas, Florian. 1985. “Direct and indirect speech: General problems and problems of Japanese.” Journal of Pragmatics 9(1): 41–63. Coulmas, Florian. 1986. “Reported speech: Some general issues.” In Direct and Indirect Speech, Florian Coulmas (ed.), 1–28. Berlin, New York, Amsterdam: Mouton de Gruyter. Couper-Kuhlen, Elizabeth. 2007. “Assessing and accounting.” In Reporting Talk: Reported Speech in Interaction, Elizabeth Holt and Rebecca Clift (eds), 81–119. Cambridge: Cambridge University Press. Du Bois, John W., Chafe, Wallace L., Meyer, Charles, and Thompson, Sandra A. 2000. Santa Barbara Corpus of Spoken American English, Part 1. Philadelphia, PA: Linguistic Data Consortium. Duden, Band 9, Richtiges und gutes Deutsch [Duden, vol. 9, Correct and good German]. 1985. (3rd, new rev. ed. Vol. 9). Mannheim: Bibliographisches Institut. Ebert, Karen. 1986. “Reported speech in some languages of Nepal.” In Direct and Indirect Speech, Florian Coulmas (ed.), 145–159. Berlin, New York, Amsterdam: Mouton de Gruyter. Ehlich, Konrad. 1987. “so – Überlegungen zum Verhältnis sprachlicher Formen und sprachlichen Handelns, allgemein und an einem widerspenstigen Beispiel.” In Sprache und Pragmatik. Lunder Symposium 1986, Inger Rosengren (ed.), 279–298. Stockholm: Almqvist and Wiksell. Firle, Marga. 1988. “Indirekte und erlebte Rede als textstilistische Phänomene.”” Zeitschrift für Germanistik 176–181. Fleischman, Suzanne and Yaguello, Marina. 2004. “Discourse markers across languages? Evidence from English and French.” In Discourse Across Languages and Cultures, Carol Lynn Moder and Aida Martinovic-Zic (eds), 129–147. Amsterdam and Philadelphia: John Benjamins. Fónagy, Ivan. 1986. “Reported speech in French and Hungarian.” In Direct and Indirect Speech, Florian Coulmas (ed.), 255–309. Berlin, New York, Amsterdam: Mouton de Gruyter. Gardner, Rod. 2002. When Listeners Talk: Response Tokens and Listener Stance. Amsterdam and Philadelphia: John Benjamins. Goffman, Erving. 1974. Frame Analysis. New York: Harper. Goffman, Erving. 1981. Forms of Talk. Philadelphia: University of Philadelphia Press. Golato, Andrea. 2000. “An innovative German quotative for reporting on embodied actions: Und ich so/und er so ‘and I’m like/and he’s like’.” Journal of Pragmatics 32: 29–54. Golato, Andrea. 2002. “Grammar and interaction: Reported discourse and subjunctive in German.” Zeitschrift für Sprachwissenschaft 22(1): 24–55.



Andrea Golato Golato, Andrea. 2011. “Appreciatory sounds and expressions of embodied pleasure used as compliments.” In Pragmatics of Society, Karin Aijmer and Gisle Anderson (eds), 359–390. Berlin: Mouton-de Gruyter. Goodwin, Charles. 1986. “Between and within: Alternative sequential treatments of continuers and assessments.” Human Studies 9(2/3): 205–218. Goodwin, Charles and Goodwin, Marjorie Harness. 1987. “Concurrent operations on talk: Notes on the interactive organization of assessments.” IPRA Papers in Pragmatics 1(1): 1–54. Goodwin, Marjorie Harness. 1982. “‘Instigating’: Storytelling as social process.” American Ethnologist 9: 799–819. Goodwin, Marjorie Harness. 1990. He-Said-She-Said. Bloomington and Indianapolis: Indiana University Press. Goodwin, Marjorie Harness. 1990/1991. “Retellings, pretellings, and hypothetical stories.” Research on Language and Social Interaction 24: 263–276. Günthner, Susanne. 1997. “The contextualization of affect in reported dialogues.” In The Language of Emotions, Susanne Niemeier and René Dirven (eds), 247–275. Amsterdam and Philadelphia: John Benjamins. Haberland, Hartmut. 1986. “Reported speech in Danish.” In Direct and Indirect Speech, Florian Coulmas (ed.), 219–253. Berlin, New York, Amsterdam: Mouton de Gruyter. Hanks, William F. 1993. “Metalanguage and pragmatics of deixis.” In Reflexive Language: Reported Speech and Metapragmatics, John A. Lucy (ed.), 127–157. Cambridge: Cambridge University Press. Heritage, John. 1984. “A change-of-state token and aspects of its sequential placement.” In Structures of Social Action. Studies in Conversation Analysis, J. Maxwell Atkinson and John Heritage (eds), 299–345. Cambridge: Cambridge University Press. Heritage, John. 1988. “Explanations as accounts: A conversation analytic perspective.” In Analalyzing Everyday Explanation: A Casebook of Methods, Charles Antaki (ed.), 127–144. London, Newbury Park, Beverly Hills, New Delhi: Sage. Heritage, John. 2002. “Oh-prefaced responses to assessments: A method of modifying agreement/disagreement.” In The Language of Turn and Sequence, Cecilia E. Ford, Barbara A. Fox and Sandra A. Thompson (eds), 196–224. Oxford: Oxford University Press. Heritage, John and Robinson, Jeffrey. 2006. “Accounting for the visit: Giving reasons for seeking medical care.” In Communication in Medical Care: Interactions between Primary Care Physicians and Patients, John Heritage and Douglas W. Maynard (eds), 48–85. Cambridge: Cambridge University Press. Holt, Elizabeth. 1996. “Reporting on talk: The use of direct reported speech in conversation.” Research on Language and Social Interaction 29: 219–245. Holt, Elizabeth. 1999. “Just gassing: An analysis of direct reported speech in a conversation between employees of a gas supply company.” Text 19: 505–537. Holt, Elizabeth. 2000. “Reporting and reacting: Concurrent responses to reported speech.” Research on Language and Social Interaction 33(4): 425–454. Holt, Elizabeth. 2007. “‘I’m eyeing your chop up mind’: Reporting and enacting.” In Reporting Talk: Reported Speech in Interaction, Elizabeth Holt and Rebecca Clift (eds), 47–80. Cambridge: Cambridge University Press. Hutchby, Ian and Wooffitt, Robin. 1998. Conversation Analysis. Principles, Practices and Application. Cambridge: Polity Press. Jefferson, Gail. 1984. “Notes on a systematic deployment of the acknowledgement tokens yeah and Mm hm.” Papers in Linguistics 17: 197–216.



Impersonal quotation and hypothetical discourse  Klewitz, Gabriele and Couper-Kuhlen, Elizabeth. 1999. “Quote-unquote? The role of prosody in the contextualization of reported speech sequences.” Pragmatics 9(4): 459–485. Lauerbach, Gerda. 2003. “Voicing and ventriloquizing – the constructive role of discourse representation in political interviews. Nicole Baumgarten, Claudia Böttger, Markus Motz, Markus and Julia Probst (eds), Übersetzen, Interkulturelle Kommunikation, Spracherwerb und Sprachvermittlung – das Leben mit mehreren Sprachen. Festschrift für Juliane House zum 60. Geburtstag.” Zeitschrift für Interkulturellen Fremdsprachenunterricht [Online] 8(2/3): 1–12. Lee, Benny P.H. 2001. “Mutual knowledge, background knowledge and shared beliefs: Their roles in establishing common ground.” Journal of Pragmatics 33: 21–44. Lucy, John A. 1993. “Reflexive language and the human disciplines.” In Reflexive Language: Reported Speech and Metapragmatics, John A. Lucy (ed.), 9–32. Cambridge: Cambridge University Press. Mathis, Terrie and Yule, George. 1994. “Zero quotatives.” Discourse Processes 18: 63–76. Möllering, Martina. 2001. “Teaching German modal particles: A corpus-based approach.” Language Learning and Technology 5(3): 130–151. Myers, Greg. 1999. “Unspoken speech: Hypothetical reported discourse and the rhetoric of everyday talk.” Text 19(4): 571–590. Norrick, Neal R and Spitz, Alice. 2008. “Humour as a resource for mitigating conflict in interaction.” Journal of Pragmatics 40(10): 1645–1822. Pascual, Esther. 2006. “Questions in legal monologues: Fictive interaction as argumentative strategy in a murder trial.” Text and Talk 26(3): 383–402. Perakyla, Anssi. 1998. “Authority and accountability: The delivery of diagnosis in primary health care.” Social Psychology Quarterly 61: 301–320. Polanyi, Livia. 1982. “Literary complexity in everyday storytelling.” In Spoken and Written Language: Exploring Orality and Literacy, Deborah Tannen (ed.), 155–170. Norwood, New Jersey: Ablex. Pomerantz, Anita. 1980. “Telling my side: ‘Limited access’ as a ‘fishing device’.” Sociological Inquiry 50(3–4): 186–198. Potter, Jonathan. 1996. Representing Reality: Discourse, Rhetoric, and Social Construction. London: Sage. Pütz, H. 1994. “Berichtete Rede und ihre Grenzen [Free subjunctive indirect discourse and its boundaries].” Nordlyd 22: 24–37. Schegloff, Emanuel. 1982. “Discourse as an interactional achievement: Some uses of uh huh and other things that come between sentences.” In Georgetown University Roundtable on Languages and Linguistics l98l. Analyzing Discourse: Text and Talk, Deborah Tannen (ed.), 7l93, Georgetown, Georgetown University Press. Schegloff, Emanuel A. 1984. “On some questions and ambiguities in conversation.” In Structures of Social Action. Studies in Conversation Analysis, J. Maxwell Atkinson and John Heritage (eds), 28–52. Cambridge: Cambridge University Press. Schourup, Lawrence C. 1983. Common Discourse Particles in English Conversation. Working Papers in Linguistics 28. Ohio: Ohio State University Department of Linguistic. Semino, Elena, Wynne, Martin and Short, Mick. 1999. “Hypothetical words and thoughts in contemporary British narratives.” Narrative 7(3): 307–333. Streeck, Jürgen. 2002. “Grammars, words, and embodied meanings: On the uses and evolution of so and like.” Journal of Communication 52(3): 581–596. Tannen, Deborah. 1983.“‘I take out the rock – dok!’: How Greek women tell about being molested (and creative involvement).” Anthropological Linguistics 25(3): 359–374.



Andrea Golato Tannen, Deborah. 1986. “Introducing constructed dialogue in Greek and American conversational and literary narrative.” In Direct and Indirect Speech, Florian Coulmas (ed.), 311–332. Berlin, New York, Amsterdam: Mouton de Gruyter. Vlatten, Andrea. 1997. Quotatives, Reported Speech, and Constructed Dialogue in Everyday German Conversation. Unpublished doctoral dissertation, University of Texas at Austin. Wierzbicka, Anna. 1974. “The semantics of direct and indirect discourse.” Papers in Linguistics 7: 267–307. Winchatz, Michaela R. and Kozin, Alexander. 2008. “Comical hypothetical: Arguing for a conversational phenomenon.” Discourse Studies 10(3): 383–405. Wooffitt, Robin. 1992. Telling Tales of the Unexpected: The Organization of Factual Discourse. London: Harvester. Wooffitt, Robin. 2001. “Raising the dead: Reported speech in medium-sitter interaction.” Discourse Studies 3(3): 351–374. Wooffitt, Robin and Allistone, Simon 2008. “Participation, procedure and accountability: ‘you said’ speech markers in negotiating reports of ambiguous phenomena.” Discourse Studies 10(3): 407–427. Yule, George and Mathis, Terrie. 1992. “The role of staging and constructed dialogue in establishing speaker’s topic.” Linguistics 30: 199–215.

Appendix: Transcription Conventions . Falling intonation ? Rising intonation , Slightly rising intonation ↑ Rising pitch in the next syllable/phrase : Lengthened speech = Latching speech - Cut off word underlined Stressed syllable CAPITALIZED Higher volume superscript zero Beginning and end of quieter speech (( )) Vocal effects accompanying speech [ Beginning of overlap of speech >< Speech faster than surrounding talk Speech slower than surrounding talk (number) Duration of silence in seconds (.) Micro pause (text) Transcriber’s best guess at what is being spoken ( ) Inaudible speech hhh Audible out-breath (laughing voice) .hhh Audible in-breath * and text in italics Nonverbal actions occurring without speech or at around the same time as the speech in the line immediately below

By three means The pragmatic functions of three Norwegian quotatives Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig University of Adger and University of Oslo

Whereas research in the Anglophone world has concentrated on the global spread of like, this paper sets out to comparatively discuss three Norwegian quotative markers bare ‘just’, liksom ‘like’ and sånn ‘such/like (this/that)’. Similarly to like, the Norwegian quotatives derive from pragmatic particles which express a range of pragmatic functions, such as epistemic hedge, focus marker and filler. However, these functions are more specialized in the three Norwegian markers, so that each of them is less multifunctional than like. Working within an ‘Interactional Linguistics’ framework, we show that quotative bare is a speaker-oriented intensifying marker, expressing the speaker’s affective stance. As a quotative, it is primarily used with dramatic or emphatic quotations, but may also be used as a neutral quotative. Liksom is a marker of comparison and similarity. When it functions as a quotative, it may be used as a speaker-oriented hedge, expressing epistemic stance of approximation/ uncertainty, or as a hearer-oriented marker, inviting the hearer to visualize the quoted material. Sånn is a speaker-oriented focus marker with a demonstrative or pointing function. In its quotative uses, it expresses epistemic stance of exactness/certainty. A quantitative comparison across three generations of speakers from Oslo shows that bare is the most frequently used quotative, and it also reveals clear generational differences, with adolescents being the primary users of all the quotatives. Keywords: Norwegian, quotatives, epistemic stance, intensifier, hedge



Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig

1. Introduction1 1.1

Aim and scope

The growing interest in quotative markers which has swept over the international linguistic landscape during the last decades has reached Scandinavia as well and quotative markers have received increasing interest from Scandinavian scholars. Several of these researchers have integrated descriptions of quotative functions in more general studies of discourse markers, both in Swedish (Eriksson 1997; Kotsinas 1994; Öqvist 2000) and Norwegian (Hasund 2003; Opsahl 2002; Opsahl and Svennevig 2007). One of the first scholars who investigated the quotative function of an innovative marker as a topic in its own right was Eriksson, in his 1992 and 1995 studies of Swedish ba(ra) ‘just’. More recently, Jönsson (2005) has described the grammar of quotative constructions in Swedish and Rathje (2009, 2011) has studied generational differences in the use of quotatives in Danish. The aim of this chapter is to present a description of three quotative markers in Norwegian: bare ‘just’, liksom ‘like’ and sånn ‘such/like (this/that)’. Whereas the two former forms have been studied previously by the authors (cf. above), the marker sånn has not yet been described as a quotative. All three are frequent in present-day spoken language, especially among younger speakers (cf. Section 6), amongst whom they have outnumbered traditional verbs of saying such as si ‘say’. The comparative perspective of this chapter will allow us to pinpoint similarities and differences between these markers. We will also compare them with equivalent markers in other Scandinavian languages. In Section 1.2 we will outline our theoretical framework and define some of the key words and concepts underlying our analysis. 1.2

Theoretical framework

The main function of quotative markers is to signal a shift from reporting to reported speech; they thus serve a discourse organizational function. Furthermore, quotative constructions invariably involve some expression of the speaker’s attitude to the events reported in the quotation (cf. Spronck, this volume) and also serve subjective and interpersonal functions. The Norwegian quotative markers analyzed herein also occur in non-quotative uses in which they belong to the wider category of discourse markers, just like the English quotative/discourse marker like (D’Arcy 2007; Hasund 2003; Romaine 1. We are grateful to the editors and two anonymous reviewers for valuable comments on previous versions of this chapter.



By three means 

and Lange 1991). We will therefore begin the analysis of each quotative marker presented here with a summary of their functions as discourse markers, that is, in non-quotative uses. Our claim is that some of these functions are also traceable in the quotative uses of the markers. This chapter is primarily a synchronic study of the pragmatic functions and social distribution of the markers investigated.2 We will consider functions that are primarily speaker-oriented, such as displays of the speaker’s affective and epistemic stance, as well as functions that are primarily hearer-oriented, such as appeals for involvement by the interlocutor. Speakers can express subjectivity via markers of epistemic and affective stance (cf. Kärkkäinen 2006; Spronck this volume). Epistemic stance concerns speakers’ commitment to what they are saying or how they display attitudes to knowledge of the affairs described. Discourse markers displaying epistemic stance include, for instance, the English hedge I think, conventionally used to reduce the speaker’s epistemic commitment (cf. Kärkkäinen 2003). Affective stance, on the other hand, concerns speakers’ involvement in the topic (Katriel and Dascal 1989) and involves evaluative devices, displays of emotion, dramatization etc. (Tannen 1984, 1989). An English discourse marker which has been described as primarily marking affective stance is just, for instance in the following utterance “Guy’s just such a dickhead” (cf. Erman 1997). In this example, the discourse marker functions as an intensifier and contributes to upgrading the speaker’s affective involvement in the presented evaluation. As we will see below, the Norwegian discourse marker bare ‘just’ can have a similar intensifying function, which is generally retained in its use as a quotative. However, a speaker’s stance-taking in conversation is never merely a unilateral expression of an internal state. It is always directed towards an interlocutor and thus embodies interactional functions related to the common activity being pursued by the interlocutors (Kärkkäinen 2006). Several discourse markers have as their main function to accommodate to the speakers’ and interlocutors’ epistemic and affective states. An example is the English marker you know, which may appeal for (explicit or implicit) acknowledgement by the interlocutor (cf. Schiffrin 1987). We will now briefly discuss our data and methodology, before we move on to an analysis of the three Norwegian quotative markers.

2. An in-depth description of the grammaticalization processes involved in the development of the various usages is beyond our scope. However, in our discussion of the possible grammaticalization that these markers have undergone, we will briefly refer to studies of equivalent markers in other languages.



Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig

1.3

Data and methodology

This study is situated within the field of ‘Interactional Linguistics’, which has as its main aim to integrate the description of lexical and grammatical resources within an account of the interactional practices used to perform social actions (cf. Lindström 2008; Ochs et al. 1996; Selting and Couper-Kuhlen 2001). This approach stresses the importance of using authentic conversational data and taking into account the interactional context in order to give an adequate description of the functional aspects of a given linguistic item, in our case a quotative marker. Hence, we base our study on conversational extracts and analyze the quotative markers in their activity context. In some cases, the examples presented in this chapter have been shortened due to exigencies of space, but the analysis is always based on a qualitative analysis of the sequential and interactional context in which they occur as well as on the acoustic realization of the marker in question. It is widely acknowledged that adolescents’ language may serve as a source of information about linguistic change (cf. for instance Buchstaller 2006; Eckert 1997: 52; Stenström et al. 2002; Tagliamonte and D’Arcy 2007), a point that led us to focus on the quotations we find in such language. Our main source of data is the Norwegian corpus of Oslo teenage language from the Scandinavian research project UNO (Språkkontakt og Ungdomsspråk i Norden ‘Language contact and teenage language in Scandinavia’, cf. Hasund 2003). The corpus consists of self-recorded conversations of 45 informants from different parts of Oslo. The informants are between 14 and 19 years of age. There were no researchers present during the recordings. As such, the recordings were organized by the informant themselves, resulting in conversations in dyads, triads and groups of informants. The corpus was collected from 1997–1998, and consists of 206,854 words of transcribed text. To support the underlying assumption(s) that we are dealing with new quotation markers and/or markers typical of teenagers’ speech, we also analyze crossgenerational data from the NoTa-corpus (Norsk Talespråkskorpus – Oslodelen ‘Norwegian Spoken Language Corpus, the Oslo part’). This corpus, collected between 2004 and 2006, consists of spoken data from 144 respondents from Oslo and the surrounding areas. The informants participate in semi-structured interviews and in dyadic conversations with (typically) a friend, a classmate or workmate. The researcher organized the recordings, and was present as an observer in the dyadic conversations. The corpus consists of approximately 790,000 words of transcribed text. The NoTa-corpus is evenly balanced regarding the respondents’ age, gender, place of residence and socio-economic background, and thus provides us with the opportunity to compare our teenagers’ speech with the speech of adults. Furthermore, since this second corpus was collected 6 years later than the first, it allows us to adopt a diachronic perspective.



By three means 

Table 1.  Overview of the UNO and NoTa corpora Corpus Speakers Hours Words

Type of data

Year of recording

Age

Region

UNO

N = 45

18

206.854

1997–1998 14–19

Oslo

NoTa

N = 144

90

790.000

Self-recorded conversations (dyads, triads and groups) Interviews and dyadic conversations

2004–2006 15–25, 26–50, 51+

Oslo

Both of the corpora contain data from Norway’s capital, Oslo, and its close surrounding areas. We do not claim to be able to make generalizations about ‘Norwegian’ per se, because dialectal diversity has been, and still is considerable in Norway. However, the Oslo varieties are closely related to the written Bokmål standard3 and are considered by many linguists and laypeople alike to be a standard language (Røyneland 2009: 12). The characteristics of the two corpora are summarized in Table 1. The transcription conventions are given in the appendix. All names in the transcripts are pseudonyms. We will now start our discussion of the pragmatic functions of the three new quotatives. In all three cases, we will first discuss the older discourse marker function and then move on to the quotative function. Examples from the UNO and NoTa corpora are marked with a file reference; unmarked examples are constructed for illustrative purposes. 2. Bare 2.1

Bare as a discourse marker

Bare’s semantic and pragmatic properties have received some attention in previous research on Norwegian (see for instance Fretheim 1980, 1999; Lie 1979; Opsahl 2002; Opsahl and Svennevig 2007). The discourse uses of bare can be seen as related to its nondiscourse uses as an adverb that is historically derived from an adjective meaning ‘bare’ or ‘naked’ (which, incidentally, shows the close relationship

3. There are two official written standards in Norway: Bokmål ‘book-language’ and Nynorsk ‘new-Norwegian’. In somewhat simplified terms, the ‘Bokmål’ standard can be traced back to Danish-influenced urban varieties and the ’Nynorsk’ standard to Ivar Aasen’s (1813–1896) compilations of rural dialects in the late 19th century.



Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig

with English).4 The adverb bare has mainly two functions, as illustrated in (1) and (2a and b): (1) Jeg liker bare sorte olivener. I like only black olives. I only like black olives. (2) a. Jeg har bare to søskenbarn. I have only two cousins. I only have two cousins. b. Han er bare student. He is just student. He is just a student. In (1) bare is used as a restrictive focus adverb with an exclusive meaning, ‘nothing else than’. In (2), on the other hand, it is used as a downgrading focus adverb indicating that something is low on a scale, either quantitatively (‘not more than’), as in (2a), or qualitatively (‘no[thing] better than’), as in (2b). In both cases bare focuses on an entity or process which is compared – in most cases indirectly so – to another entity or process we may call the referent entity/process (cf. Lee 1987 on English just). In (1) and (2a,b), the referent entities evoked are in the first case ‘green (or other available colors of) olives’ and in the second cases (a) ‘any other possible number of cousins higher than two’ and (b) ‘a more important (academic) occupation than student’. The downgraded meaning of the adverb bare is also visible in its ability to highlight the contrast between propositions, as in (3): (3) A: Hva er det som bråker? What is that which makes-noise? What’s that noise? B: Det er bare Petter som er på kjøkkenet. It is just Petter who is on kitchen It’s just Petter who is in the kitchen. Here in (3), bare implies that there is nothing ‘worse’ going on other than Petter being in the kitchen, indirectly pointing to a contrasting situation, for instance the presence of burglars (cf. Lie 1979: 42). The polysemy between the restrictive meaning illustrated in (1) and the downgrading meaning illustrated in (2a,b) may be considered the basis for the multifunctionality of the discourse marker as either a hedge or an intensifier (Opsahl and 4. The terms ‘discourse uses’ and ‘nondiscourse uses’ are taken from Romaine and Lange (1991: 244), who use the latter term to describe the “purely grammatical or syntactic” uses of like.



By three means 

Svennevig 2007). Indeed, as a discourse marker, bare can have an epistemic hedging function as well as an affective intensifying function, both of which can be seen as related to the nondiscourse uses of bare. We will now discuss these uses of bare in turn. As a hedge, bare reduces the speaker’s commitment to the proposition expressed. This function is often exploited for politeness reasons, for example in order to reduce the imposition of requests, as in (4): (4) Jeg bare lurte på om du hadde tid et øyeblikk? I just wondered on if you had time one moment? I was just wondering if you had a moment to spare? The use of several classic politeness strategies is present in this request, such as the choice of past tense and the expression et øyeblikk (literally ‘a glance of an eye’). The addition of bare further reduces the imposition of the request by downgrading the importance of the service requested. The function of bare as a hedge is considered as derived from its use as a downgrading focus adverb illustrated in (2a,b) and (3). Its intensifying function, on the other hand, is considered to be derived from its use as a restrictive focus adverb, illustrated in (1) ‘only black olives’, where bare has an exclusive meaning. When the Norwegian tabloid magazine Se & Hør publishes an exclusive interview with a celebrity, they will typically use the expression Bare i Se & Hør! (‘only in Se & Hør!’) to emphasize the uniqueness of the interview, that is, the fact that no other magazine managed to get an interview with this particular celebrity. In this instance, there is nothing downgrading about bare – rather the opposite – and it is from this use that bare’s function as an intensifier has developed. As an intensifier, discourse bare can be used to strengthen the affective commitment of the speaker. It can heighten deontic and evaluative aspects of an utterance, for instance, in connection with imperatives expressing emphatic offers, as in Example (5): (5) Bare ta en kake til! Just take one cake more! Please have another piece of cake! The intensifying function found in emphatic offers, like in (5), may be seen as the result of a politeness strategy where bare underlines the fact that the addressee is not expected to do ‘anything else than’ accepting the offer, that is, an offer given without the call for return services, which, according to Opsahl and Svennevig (2007), reflects the restrictive meaning illustrated in (1). Opsahl and Svennevig (2007) show how the intensifying function associated with emphatic offers has evolved and spread to other deontic utterances and later into other pragmatic domains, such as hyperbolic assessments (cf. Example 6 below) and in dramatic peaks in narratives (cf. Example 7 below). These are the cases where we most typically find the intensifying bare in the data from our young speakers.



Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig

(6) (UNO E-H-f)5 Det er en annen en som heter Petter, han er bare helt vill da, han er gal. There is one other one who is-called Petter, he is just totally wild then, he is crazy. There is this other one who is called Petter, he is just totally wild, he is crazy. In examples such as (6), bare shares certain similarities with the discourse marker just in spoken English, which may also be used to intensify an adjective or a noun (such as in the title of Erman’s 1997 article: “Guy’s just such a dickhead”). In (6), for instance, the discourse marker bare is used to intensify the adjective phrase “helt vill” ‘totally wild’. Bare may also be used with wider scope, intensifying verb phrases, clauses and whole utterances. This function is typically found in narratives, where bare is used as an evaluative marker to highlight the dramatic character of certain events. This can be seen in examples like (7), where we hear from a girl about a group of boys who smashed some equipment belonging to a youth club: (7) (UNO E-H-f) Dem tok gitaren da som tilhørte, (.) var det (klubbnavn), (.) bare knuste-n i huet på hverandre. They took guitar-the then which belonged-to, (.) was it (name of club), (.) just smashed-it in head-the on each-other. They took the guitar which belonged to, was it (name of club), just smashed it onto each others’ head. The ability of bare to display the affective involvement of the speaker, and its more or less explicit connection to evaluation, are associated with functions that are primarily speaker-oriented (cf. 1.2). As we will see below, this is one of the characteristics which separates bare from the two other markers investigated. In the following section we discuss bare’s quotative function in our data. We show that it is closely related to bare’s discourse function as an intensifier. 2.2

Quotative bare

Several instances of bare used as a quotative can be found in the transcription of speech in (8):

(8) (UNO W-S-f)

5. This is the file reference to the UNO corpus, which gives information about the speaker in terms of geography (E = east Oslo, W = west Oslo), age group (S = secondary school ages 13–15, H = high school ages 16–19) and gender (m = male, f = female).



By three means 

(L1) (L2) (L3) (L4) (L5)

Tor kom bort til meg sånn “↓I want to make lo:ve to you”, Tor came over to me such “↓I want to make lo:ve to you” Tor came over to me like this “↓I want to make lo:ve to you” og så tok han og bare dro meg ((med seg)) and then took he and just pulled me with him and then he just pulled me ((along)) så jeg bare “nei (.) du får bare et kyss men da går jeg”, so I just “no (.) you get only one kiss but then go I” so I just “no (.) you can have just one kiss then I’m off ” ((han bare)) “nei få mer” he just “no get more” ((he just)) “no give me more” jeg bare “ha de:t” (latter) beina ut vet du. I just have it (laughter) ran out know you I just “by:e” (laughter) ran off you know.

Example (8) renders a dramatic dialogue between two people via the construction ‘Subject (jeg ‘I’, han ‘he’) + quotative (bare) + quotation proper’. Note that bare is most commonly used in reporting constructions without a quotative verb. In terms of linear order, it always precedes the quotation and introduces it. As can be seen in (8) above, the utterances rendered via bare tend to have a dramatic quality, figuring around the peak of the story or encoding notable events, as in the example above, where the young girl demonstrates how she managed to wriggle herself out of an awkward situation with a clingy suitor. This means that the speakers’ affective involvement that is inherent in discourse marker bare (as discussed above) carries over into its use as a quotative marker. Importantly, quotative bare also occurs without a manifest subject, as in (9): (9) (L1) (L2)

(UNO W-S-f) Det gjør så vondt at jeg holder på å begynne å grine hver gang jeg nyser It does so painful that I held on to begin to cry every time I sneeze It hurts so bad I’m about to burst into tears every time I sneeze for da “atsjoo” og så bare “eh au: uhu” because then “achoo” and then just “boo: hoo” because then “achoo” and then just “boo: hoo”

Here, the first quotation – the sneezing sound – is presented without any quotative marker at all (see Golato this volume; Herrmann and Steinbach; Mathis and Yule 1994 this volume). The second – the crying sound – is introduced by bare, but without any manifest subject. The context makes it clear, however, that both quotations are attributed to the same speaker. This example also shows that the material quoted via bare may be entirely non-verbal.



Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig

Note that bare may co-occur with the other quotatives liksom and sånn. The combination of quotative markers will be explored in more detail in Section 6 below. However, bare rarely co-occurs with reporting verbs. Only a few instances of such combinations can be found in our corpora, such as si ‘say’ + bare in (10), where the speaker quotes from a song and shifts into a singing mode: (10)

(UNO W-H-f) da må du si bare (synger) “tramp tramp tramp på en smurf ” then must you say just (singing) “step step step on a smurf ” then you’ll have to say just (singing)“step step step on a smurf ”

As was pointed out above, in its quotative function bare typically signals heightened speaker involvement. It frames quotations which represent highly emotional or expressive utterances, as in (8), and/or onomatopoetic exclamations, as in (9). Furthermore, bare frequently occurs at or around narrative peaks, where it serves to highlight the dramatic character of quotations. As shown by Fox (this volume), new quotatives may fulfill the pragmatic function of highlighting a particularly dramatic peak in performed narratives, and narrative contexts may in fact be the locus of the development of new quotatives. Indeed, as we pointed out in (6) and (7), discourse marker bare may also serve as an evaluative intensifier, associated with narrative peaks and hyperbolic assessments. Bare thus seems to be associated with cases where the speaker commits fully to the reported utterance and shows heightened involvement. Hence, bare’s role as a quotative seems to be related to its function as a discourse marker with an intensifying function and may also be seen as derived from the restrictive focus adverb illustrated in (1). Note, however, that there is some evidence for semantic bleaching (Hopper and Traugott 2003) amongst the younger speakers in our corpus; there are several occasions where quotative bare seems to have lost some of its pragmatic potential, that is, its intensifying function. As can be seen in (11), bare can be a more or less neutral device for the portrayal of direct speech in discourse: (11) (L1)

(NoTa Oslo 008-W-18-f)6 Vi bare “ja mor vi skal på Carl Berner7 i kveld” We just “yes mum we shall on Carl Berner in night” We just “yes mum we’re going to Carl Berner tonight”

6. This is the file reference to the NoTa corpus. Each speaker is provided with an informant number, such as ‘008’. W-18-f gives information about the speaker in terms of geography (E = east Oslo, W = west Oslo), age and gender (m = male, f = female). 7. Carl Berner is an abbreviation for Carl Berners plass, an area surrounding a square in central Oslo.



By three means 

(L2) hun bare “hva? Skal dere på Carl Berner?” jeg bare “ja” she just “what? Shall you on Carl Berner?” I just “yes” she just “what? Are you going to Carl Berner?” I just “yes” The sequence in (11) does not seem dramatic, unless visiting Carl Berner is an act which involves some sort of danger, which is not implied by the context. Our corpus contains several examples like these, where bare does not involve any heightened involvement but merely signals the introduction of a quote. This use seems to be a semantically bleached version of the use of the quotation marker shown in (8) and (9). The patterns found in our Norwegian data correspond quite well with previous research on the Swedish discourse marker ba(ra) (Eriksson 1992, 1995, 1997; Jönsson 2005; Kotsinas 1994). The quotative function is common to the two languages. It is also found in Danish (cf. Rathje 2009 and Icelandic (Van Alphen, p.c.)). In Swedish, however, the marker is generally realized in the phonetically reduced form [ba]. This kind of articulatory reduction, which goes hand in hand with a loss of semantic content, a stable pragmatic function and strict syntactic placement (i.e. ‘S + bare + quotation proper’), is in line with general characteristics of grammaticalization processes (Eriksson 1992; Hopper and Traugott 2003; see also Coppen and Foolen this volume; pace Vandelanotte this volume). Note that Swedish ba(ra) has in fact turned into a neutral quotation marker (Eriksson 1997). Opsahl and Svennevig (2007) claim that a similar development is in progress in Norwegian. This is not surprising given the fact that – as we show later on – bare is without question the preferred quotation marker among younger speakers. Indeed, amongst this age group, bare has outnumbered the constructions with conventional reporting verbs. This may be taken, albeit somewhat speculatively, as further evidence for the fact that we are dealing with a conventionalized construction resulting from a process of grammaticalization (cf. Güldemann this volume). Importantly, there remains a close developmental functional-pragmatic relation between the discourse marker bare serving an intensifying function and the cases where bare serves as a quotative. The idea that the original semantic and pragmatic properties of the words have persisted during their development into quotatives is not unique. Buchstaller (2004: 131) refers to several studies that provide “cross-linguistic evidence for a close semantic link between the notion of approximation, hedging, focusing, and introducing speech and thought” (cf. also Coppen and Foolen this volume; Güldemann this volume; Meyerhoff 2002). As we have seen above, there is an element of comparison present in some of the nondiscourse uses of the restrictive and downgrading adverb bare. Moreover, in its discourse marker use, bare may serve both as a hedge and as an intensifier. This intensifying function is also visible in many of the cases where bare serves as a quotative.



Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig

The notion of similarity and approximation are, however, more prominent in the case of liksom, to which we now turn. 3. Liksom 3.1

Liksom as a discourse marker

The discourse use of liksom8 originates in its nondiscourse uses as a preposition and conjunction, which have a core meaning of comparison and similarity. (12) is an example of the nondiscourse uses of liksom: (12) Hun er lærer liksom moren. She is teacher like mother-the. She is a teacher like her mother. In (12), the grammatical function of liksom is to connect two elements in a clause. There is a comparison between two entities in the syntactic structure, the two noun phrases lærer ‘teacher’ and moren ‘her mother’. The semantic meaning of liksom can be described as ‘in the same way as’. The nondiscourse use of liksom is archaic and restricted to formal, written language. In present-day spoken Norwegian, expressions such as akkurat som ‘just like’, and slik som or sånn som ‘in the same way as’ are preferred to liksom as a preposition and conjunction. As a discourse marker, however, liksom is frequently used in present-day spoken Norwegian. (13) is one example of this use: (13) Lillesøstera mi har liksom hundre forskjellige kjoler. Little-sister my has like hundred different dresses. My little sister has like a hundred different dresses. Liksom as a discourse marker has the same element of comparison and similarity inherent in nondiscourse liksom. However, while nondiscourse liksom involves a comparison between two recoverable entities in the syntactic structure, discourse liksom may indicate a comparison between entities that are not necessarily on the same syntactic level in the utterance. One of them might not be explicitly expressed but implicitly referred to, for example, ‘what is in the speaker’s mind’ or ‘what is true/correct’. For instance, in (13), liksom is used to indicate a comparison between the number hundre ‘a hundred’ and another number which is, perhaps, more precise or correct but less prototypical (see Berlin 1992; Rosch 1975, 1978).

8. Phonologically, liksom may be realized as either /liksom/ or /lissom/.



By three means 

The relation between the nondiscourse and discourse uses of liksom can be described as the result of a grammaticalization process, whereby the form has developed increased syntactic flexibility and a more generalized meaning (cf. Hasund 2003). 3.2

Quotative liksom

Our data-set contains very few examples of liksom introducing a quotation without a quotative verb. In extract (14), the speaker is telling about how he was not allowed to buy alcohol at the liquor store because he was underage: (14) (L1) (L2)

(UNO E-H-m) (siterer ansatt på polet) “nei du e’ kke gammal nok.” (quoting shop assistant) “no you aren’t old enough” (quoting shop assistant) “No you’re not old enough.” (.) Jeg liksom (..) “Ja. (.) Ja jeg trur jeg tar meg en tur jeg.” (.) I like (..) “Yes. (.) Yes I think I take me a trip I.” (.) I’m like (..) “Right. (.) Yeah I think I’ll just go for a walk.”

Here, we find the construction S (jeg ‘I’) + quotative marker (liksom) + quotation proper: “Ja. Ja jeg trur jeg tar meg en tur jeg”, ‘Right. Yeah I think I’ll just go for a walk’, line 2. In this example, the quotation is probably not a realistic rendering of what the speaker actually said, but rather a dramatized illustration of his emotions or thoughts at the time, in this case his embarrassment at being denied service at the liquor store. In this respect, the construction resembles quotative like in English, which is commonly used with quoted stance, such as attitudes and point of view (cf. Buchstaller 2004; see also Spronck this volume). Since the reporting speaker is not committing himself as to whether the reported speech act is exactly what he said or thought when being refused the alcohol, liksom can be characterized as an epistemic hedge, which reduces the speaker’s commitment to the factuality or exactness of the representation. Note that the core meaning of similarity is clearly discernable, in that liksom indicates an implicit comparison between the quotation and the actual situation in which the quote was realized (verbally or merely in the speaker’s mind – see Andersen 1999; Buchstaller 2004; Hasund 2003). Note also that there are examples of quotative liksom without a manifest subject (see Buchstaller 2004, 2011 for similar results for like; see also Herrmann and Steinbach this volume):



Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig

(15) (L1) (L2)

(UNO W-H-f) Man får vite det en eller annen gang uansett. One gets to-know it one or other time anyway. Sooner or later you get to know anyway. Liksom “nei jeg kom ikke inn på de studiene der”. Like “no I came not in on those studies there”. Like “no I didn’t get into that university.”

Here the speaker first describes a future situation when students get the results of their application to a university. The quotation (line 2) represents the speaker’s potential reaction at being denied entrance and it is not specified whether this reaction is direct speech or inner thought. This quotation is introduced with neither a traditional quotative verb nor with a manifest subject; it is merely prefaced by liksom. As there is no subject, the listener has to infer the source of the quotation from the context. In this case the inference is quite straightforward, the subject of the prior clause (man ‘one’). Note that in some cases, such as (16), line 3, it is less obvious who the source of the quotation is: (16) (L1) (L2) (L3) (L4)

(UNO E-H-m) (Peter is talking about his mother) Hun er alltid sånn pesete med meg. She is always such stressed-out with me. She’s always such a pain. Når jeg har med kamerater og sånn så er a drithyggelig When I have with comrades and such then is she really-nice When I bring my mates home and stuff she’s really nice og liksom, (imiterer kameratene) “Å fy faen å lite kul mor” liksom? and like (mimicking mates) “Oh bloody hell oh little cool mum” like? and like, (mimicking mates) “Wow, what a cool mum!” like? Det er sånn “jo takk du sku kjent a du” It is such “yes thanks you should known her you” It’s like “yeah if you only knew her”

In this case the subject of the previous clause in line 2 is a ‘she’ (the mother). However, the lexical choices (“cool mum”, the use of swearwords) and the prosodic marking of the quotation make it clear that the speaker is quoting his friends rather than his mother. Crucially, in the latter two examples, the quotations represent illustrations, elaborations or exemplifications of some point that the speaker has already made in the previous utterance (see also Golato, this volume): In (15) the speaker illustrates her point that sooner or later one gets to know whether one is accepted at a university by staging her hypothetical reaction when the information is received.



By three means

In (16), the quotation exemplifies the speaker’s point that his mother is cool when around his friends by demonstrating (Clark and Gerrig 1990) how his friends react. In both of these examples, the element of comparison inherent in the meaning of the word liksom is salient. The quotation functions as an illustration of an assertion previously made. It is presented as similar to the point that has been made just before the quotation and thereby exemplifies and dramatizes this point. Note that liksom also occurs with reporting verbs (verba dicendi), such as si ‘say’ (cf. 17), tenke ‘think’ (18): (17) (UNO W-S-f) Da kommer jeg til å si liksom “ja hvor mange har du rota med i kveld” liksom? Then come I to to say like “yes how many have you snogged with to night” like Then I’ll say like “well how many girls have you been snogging tonight” like? (18) (UNO W-H-f) jeg har en halvflaske vodka så jeg tenkte lissom “jeg kan jo ta den i samme slengen?” I have a half-bottle vodka so I thought like “I can yes take that in same toss?” I have a half-bottle of vodka so I thought like “I can drink that too while I’m at it?” In cases when liksom co-occurs with a traditional quotative, it may be debatable whether the form has an independent quotative-introducing function, since it follows a verb with precisely this function. Moreover, while liksom might indeed be used as a quotative introducer in these cases, resulting in double marking, we would like to suggest that the form also functions as an epistemic hedge concerning the faithfulness of the reported material to the original speech act, especially given that (17) is a hypothetical situation and (18) a reporting of inner monologue. Finally, liksom may enter a quotative construction with the existential verb være ‘be’ (especially in the present tense form er ‘is’) in pre-quote position. Consider the examples in (19): (19) (L1) (L2)

(NoTa Oslo 041-E-30-m) De skal jo til Kristiansand første helga i april, They shall yes to Kristiansand first weekend-the in April, They are going to Kristiansand the first weekend of April, og han er liksom “å vi må bestille togbillett” og er helt hyper. and he is like “oh we must order train-ticket” and is totally hyper. and he is like “oh we must order our train tickets” and is totally hyper.





Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig

(L3) jeg bare “ja ja” I just “yes yes” I just “yes yes” In this construction, liksom resembles the ubiquitous quotative construction be + like in English (cf. Andersen 1999; Buchstaller and D’Arcy 2009; Hasund 2003; Tagliamonte and D’Arcy 2007 inter alia). However, the construction være + liksom is not very frequent in Norwegian, especially when compared to the use of bare. Indeed, since, as we mentioned above, liksom rarely introduces a quotation by itself, instead tending to co-occur with verba dicendi or another discourse marker, the net result is that liksom is much less widespread and conventionalized as a quotative marker in our Norwegian data compared to like in English. As the reader will have noticed from Examples (16) and (17), liksom frequently occurs after quotations. In fact, in our data quote-final position is the most common position for this marker with regard to reported speech and thought (cf. Table 3 below). This observation is in line with what has been reported for liksom in Swedish and Finland-Swedish (cf. Eriksson 1997; Kotsinas 1994; Londen 1997). Final liksom can be used in addition to initial quotative verbs or markers, as in (16) and (17), or alone at the end of the quotation, as in (20): (20)9 (UNO E-H-f) (L1) Anita: det er mange som drar til Mekka ikke sant, a lot of people go to Mekka right, de som blir [skikkelig rike og sånn] those who get [really rich and stuff] (L2) Else: [for å få prestisje] [to get prestige] (L3) Anita: drar dit hvert år og: go there every year and: (L4) Tina: ja yes (L5) Anita: og lissom kommer hjem og “jeg har vært i Mekka ti ganger” liksom, and like comes home and “I have been in Mekka ten times” like, and like come home and “I have been to Mekka ten times” like. (L6) Tina: ja yes In this example, Anita produces a quotation which functions as a hypothetical, dramatized illustration of her point about rich Muslims going to Mekka, a quote 9. In longer examples, we have restricted the glossed translation to the utterances containing the discourse marker which is the focus of analysis.



By three means 

that is followed by liksom. We would like to argue that in cases such as (20), quote final liksom should be considered a hearer-oriented, appealing marker, inviting the hearer to visualize or imagine the dramatized situation (see Miller and Weinert 1995 for sentence final like). Indeed, the appealing function is especially conspicuous in cases like this, where the interlocutor responds verbally to the appeal by providing an acknowledgement token: Tina immediately produces a positive minimal response (ja ‘yes’) in line 6 and Anita continues her account. In Scandinavia, research on liksom in spoken language comprises two SwedenSwedish studies (Kotsinas 1994; Eriksson 1997) and one Finland-Swedish study (Londen 1997), showing that liksom in these two varieties of Swedish is used in much the same way as liksom in Norwegian. To our knowledge, there are no previous studies of this marker in Danish. The fact that the Norwegian quotative liksom is more frequently used after a quotation distinguishes it from the English quotative like, which is almost categorically used before a quotation (see Vandelanotte, this volume). In addition, liksom does not enter into collocative quotative constructions to the same extent as like (see Buchstaller et al. 2010; Vandelanotte this volume). As we will see below, one reason for the paucity of liksom + traditional quotative constructions in Norwegian is that the marker sånn is used to a greater extent in clausal quotative constructions. We will now discuss this quotative form in some detail. 4. Sånn 4.1

Sånn as a discourse marker

The discourse marker sånn derives from a deictic demonstrative meaning ‘such’ or ‘like this/that’. Sånn has not been previously studied as a discourse marker in Norwegian, but only as an adnominal demonstrative: “veldig sånn festejente”, literally ‘very such party-girl’, meaning ‘totally like a party girl’ (Lie 2008; cf. also Simonsen and Christensen 1980). A basic meaning of nondiscourse sånn is to point to an entity by comparing it to another one, typically specified in an embedded clause (cf. Ekberg 2007, 2010), as in Example (21): (21) Jeg vil ha en sånn ballong som Jan har. I want to-have a such balloon as Jan has. I want to have a balloon like the one Jan has. The comparison element inherent in sånn makes the form somewhat similar to liksom, and this aspect is traceable also in its quotative use. However, it is the



Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig

deictic function of sånn which is dominant in its development of discourse uses, and by virtue of which sånn can be used to refer to entities available in the speech situation. While pointing to a balloon, the speaker may say: (22) Jeg vil ha en sånn. I want to-have a such. I want one like that. As a discourse marker, sånn may be used as a hedge, particularly in situations when the speaker expresses some degree of uncertainty about whether the phrase used is the correct formulation or whether the facts themselves are correct, as in (23) below. In this use, sånn may, according to Lie (2008) co-occur with other hedges or markers of uncertainty, such as holdt jeg på å si ‘sort of ’ in (24): (23) Det var en sånn ti-femten stykker der. It was a such ten-fifteen people there. There were approximately ten to fifteen people there. (24) Det er en sånn landsby holdt jeg på å si. It is a such village was I about to say. It’s like a village, sort of. The connection between the semantic elements of comparison and the hedging/ approximation function of discourse markers is well-known (see e.g., Buchstaller 2004 and Meyerhoff 2002). Indeed, as we argued above, comparative elements tend to be used with round numbers, or other focal or prototypical elements. For discourse sånn, the hedging function can be glossed as ‘approximately like this’, where ‘approximately like’ contains a comparative element as well as a deictic element. 4.2

Quotative sånn

In contrast to bare and liksom, quotative sånn generally occurs in a clausal structure with a finite verb. Two main types of constructions are found in our Norwegian data: one involving an existential verb, and the other, a reporting verb. The first construction involves existential verbs such as være ‘be’ and bli ‘become/get’. The subject may either be a person (represented by a name or a pronoun), or a formal subject, as in the impersonal construction det er sånn ‘it’s like (this). This construction seems equivalent to the English quotative construction be + like (Andersen 1999, 2000; Hasund 2003). In (25), several variants of this construction can be found. The participants Anne and Jon are talking about which teachers are easy to persuade:



By three means 

(25) (Nota Oslo 029-E-19-m, 030-E-18, f) (L1) Anne: hvis du sier nei og får med deg klassen for å si nei til’n, If you say no and you get the rest of the class with you to, (.) så sier’n nei til slutt [ikke sant.] say no he’ll end up saying no [isn’t it] (L2) Jon: [ja da]blir det noe annet. [yeah then] it’s different. (L3) Anne: ja da blir det sånn “↑neivel greit. hva vil dere ha?” yes then becomes it like-this “↑no OK. what will you have?” yeah then it’s like this “↑no okay. so what do you want?” eller “da tar vi noe annet da.” or “then take we something else then. or “we’ll do something else then.” (L4) Jon: ja vi f-òg får overtalt n til å ha [andre ting.] yeah we also manage to talk him into doing [other things.] (L5) Anne: [ja men sånn] [Yes but like-this] [yeah but like] med de eldre de er sånn at “↑nei vi vil ikke” with older-ones they are like-this that “↑no we will not” with the older ones they’re like this “↑no we don’t want to,” så bare (.) “ja men det gjør vi uansett.” (mimicking voice) then just (.) “yes but that do we anyway.” (mimicking voice) then just (.) “yes but we’ll do it anyway.” (mimicking voice) (L6) Anne: jeg blir sånn “å HERR:e gud.” I become like-this “oh LO:rd-God.” I get like this “JE:sus Christ.” Firstly, we can observe the impersonal construction “da blir det sånn...” ‘then it’s like this...’ (line 3), which does not explicitly attribute the quotation to the teacher, although it is clear from the content of the quote that he is indeed the source of the reported activity.10 The next occurrences are attributed to personal agents, either the various teachers (“de eldre de er sånn at...” ‘the older ones they’re like this’, line 5) or the speaker herself (“jeg blir sånn” ‘I get like this’ line 6). Sånn can be characterized here as a quotative marker with a demonstrative or pointing function. The marker points to the quotation as an embodied reenactment of the dramatized event, and is used with quotations that involve extensive prosodic staging devices. In the example above, the voice quality is shifted in the 10. This construction type can also be observed in (16) above (Det er sånn, jo takk, du sku kjent a du” ‘It’s like, yeah, if you only knew her’).



Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig

quotations, and especially the exclamation “å HERR:e gud” ‘JE:sus Christ’ in line 6 is produced with emphatic prosody. The quotations are thus presented as highly naturalistic and mimetic reenactments (despite the fact that the situation portrayed is presented as a hypothetical or typical situation rather than a specific event). In Example (26), which repeats (8) above, there is also an instance of sånn which involves extensive dramatization of the quote: (26) Tor kom bort til meg sånn “↓I want to make lo:ve to you”, Tor came over to me such “↓I want to make lo:ve to you” Tor came over to me like this “↓I want to make lo:ve to you” In this example, the speaker reenacts the original speaker’s code-switching into English and imitates the boy’s voice quality by shifting into a deeper pitch range (see Clark and Gerrig 1990 for the concept of quotations as demonstrations). As noted above, the semantic core meaning of nondiscourse sånn involves comparison + deixis, with the use of the marker implying a high degree of naturalism in terms of the reenactment of the original quote. As such, it resembles the German quotative so (“und ich so” ‘and I’m like’), which is reported to occur with embodied reenactments (Golato 2000, this volume).11 We therefore interpret sånn as a focus marker, expressing a high degree of epistemic commitment to the exact rendering of the quote. While both discourse sånn and liksom share the same double semantic potential of expressing approximation and exactness, it seems that, as quotatives, they have specialized in separate fields. As we discussed in Section 3.2, liksom before quotations signals approximation whereas sånn foreshadows exactness/precision. Note also that, in constructions with an existential verb, sånn -quotations typically serve as a characterization of the speaker they are attributed to. In the examples above, the quotations are used to portray the teacher as being either stubborn or yielding and the suitor as pushy. In this way, the quotative construction marks reporter’s attitude or stance towards the attributed speaker (cf. Spronck, this volume; see also Bucholtz 2004). Another type of quotative construction involving sånn is with traditional reporting verbs, such as si ‘say’ or tenke ‘think’. In (27), the speaker quotes from an actual speech situation, namely her speech at a New Year’s party (using historical present tense). In (28), the speaker renders her thoughts in a hypothetical situation.

11. For other markers resembling sånn’s combination of comparison and deixis, see Coppen and Foolen (this volume) on the zoiets van-construction in Dutch, as well as Maschler (2001) on Hebrew kaze.



By three means 

(27) (UNO W-H-f) så sier jeg sånn “ja jeg ska- -kke gå nærmere inn på detaljene rundt Helenas kjole” then say I like-this “yes I shall not go closer in to details-the around Helena’s dress” And then I say like-this “well I’m not going into further detail on Helena’s dress” (28) (L1) (L2) (L3)

(Nota Oslo 093, E-18-f) nå er hun brun men hvis jeg hadde sett henne ute, Now she’s really tanned but if I’d seen her outside, så hadde jeg ikke tenkt sånn “↑oi det er et eller annet rart so had I not thought such “↑oh there is something strange I wouldn’t have thought like this “↑oh there’s something strange med henne hun virker ikke helt norsk.” about her she doesn’t seem quite Norwegian.”

In addition, the quotation is presented as a naturalistic reenactment in these types of instances. In (28), the speaker produces a discourse marker indicating surprise (“oi”, line 2) with emphatic prosody and accompanied by a facial expression of surprise (head moved forward and eyes opened wide). Note that, apart from verbs of quotation, sånn can also collocate with verbs of motion or position that refer to accompanying actions or circumstances of the quotation, as in Example (26 = 8). Thus, it might be most fortuitous to analyse sånn not as an independent quotative marker in itself (like liksom), but as constituting part of a complex quotative construction S + være/bli (‘be/become/get’) OR verba dicendi OR motion/position verbs + sånn + quotation proper consider Vandelanotte this volume. Importantly, quotative sånn is often combined with the deictic adverbs her (proximate, ‘here’) or der (distal, ‘there’), most commonly pronounced herre and derre, to form the conventional phrases sånn herre and sånn derre (Lie 2008), as exemplified in (29): (29)

(UNO E-S-f) du så jo Katrine også var litt sånn herre “skal dere ikke gå?” (mimicking) you saw yes Katrine also was a-little like this here “shall you not go? you could see Katrine too she was a bit like this here “Aren’t you leaving?”

Just before the example given in (29), the interlocutors had been portraying Katrine as a jealous person. The quote serves to elaborate this characterization by giving an example of Katrine enacting her jealousy. The use of sånn prepares the hearer for a dramatized reenactment of the speech situation, marking the quote to follow as a demonstration of such a jealous outburst and mimicking Katrine’s voice. Indeed,



Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig

when combined with the deictic adverbs herre and derre, the pointing character of sånn is further accentuated. The quotative use of sånn seems closely related to the Swedish discourse markers sån and såhär (also pronounced in the reduced form sär).12 Kotsinas (1994) notes that såhär may introduce quotations, mainly in combination with quotative verbs. She also mentions that the marker may follow quotations. Similarly, Eriksson (1997) describes the marker’s function in narratives as a boundary marker, introducing, among other things, quotations. However, according to Eriksson, såhär has a mainly demonstrative function, inviting the hearer to get involved in the narrative and to imagine the reported events – an interpretation which is fully compatible with our discussion of sånn above. Öqvist (2000) claims that sär is a rheme marker, introducing new or important information. It also has the pragmatic function of inviting the hearer to cooperate by mobilizing socially shared knowledge in order to arrive at an interpretation of the following constituent (for instance a quotation). Crucially, our analysis of the Norwegian sånn corresponds quite well with those described for sån/sär in Swedish, marking upcoming demonstration, or reported action, especially such that is reenacted with increased speaker involvement, which obviously requires the listener to draw on culturally available knowledge (see Clark and Gerrig 1990). Having described each marker separately, we now turn to a description of their distribution across age groups (Section 5) before we discuss how they combine in quotative constructions (Section 6).

5. Distribution across age groups The design of the NoTa-corpus provides us with the opportunity to compare the distribution of the new quotatives across different generations of speakers. A search for the quotative markers in three age groups revealed large generational differences (see Table 2 and Figure 1): The youngest group of speakers, aged 15–25, used all three markers much more frequently than the older groups. In particular, the use of quotative bare yields an extreme generational variation in that the youngest speakers used the marker 362 times (1.46 per thousand

12. Sär is composed of så(dan) – the Swedish equivalent of sånn, plus här – the equivalent to the proximal deictic adverb her (described above). Sån is used in the south-western parts of Sweden, whereas såhär is used in the other parts.



By three means 

Table 2.  Number of quotative constructions used by different age groups in the NoTa corpus in number of instances and frequency per thousand words13 Age

Bare + quote Sånn + quote Liksom + quote Quote + liksom Si + quote Quote + si

15–25 26–50 51+ Total

362 (1.46)   12 (0.04)     1 (0.003) 375 (0.47)

109 (0.44)   21 (0.08)    6 (0.02) 136 (0.17)

21 (0.08) 11 (0.04) 0 (–) 32 (0.04)

41 (0.17) 12 (0.04)   3 (0.01) 56 (0.07)

26 (0.10) 32 (0.12) 37 (0.14) 95 (0.12)

   5 (0.02)   50 (0.18) 129 (0.47) 184 (0.23)

1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 Bare+Q

Sånn+Q

Liksom+Q Q+liksom 15–25

26–50

Si+Q

Q+si

51+

Figure 1.  Frequency per thousand words of different quotative constructions in relation to age group in the NoTa corpus

words), whereas the speakers in the next age group (26–50) used it 12 times (0.04 per thousand words) and the speakers in the oldest age group (50+) used it only once (0.003). In addition, sånn shows large variation in usage between the generations represented in the NoTa-corpus, with 109 instances produced by the youngest speakers and only 21 and 6 instances in the next age groups (in frequency per thousand 13. The age group of 51+ had, in fact, 11 additional examples of bare + quote, but these were all metalinguistic usages presented to portray how young people used the marker. They were therefore omitted from the score.

 Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig

words 0.44 younger, 0.08 middle, 0.02 older speakers). The differences are slightly smaller for liksom, but even here the youngest speakers use the quote-initial marker twice as often as the middle age group (0.08 versus 0.04) and the quote-final marker four times as often (0.17 versus 0.04). We may conclude that all these quotatives are most common by far among young speakers. Yet, note that they are also all used by adults between 26 and 50. Speakers aged 51 years or more seem to use the markers very sporadically, although some instances of sånn as a quotative are indeed found. What strategies do older speakers use when reporting direct speech? To find out, we counted the frequency of the reporting verb si ‘say’ in quote-initial (Han sa “hei!”, ‘he said “hi!”’) and quote-final position (“Hei!” sa han, ‘“Hi!” he said’). As Figure 1 clearly illustrates, the ‘older’ alternative to using a quotative marker seems to be to quote with si, especially in quote-final position. Note that while the frequency of si in quote-initial position is quite similar among the three age groups (0.10, 0.12, 0.14), in quote-final position the differences are large. The oldest informants produce 129 instances (a frequency of 0.47 per thousand words), the middle group 50 instances (0.18) and the youngest group only 5 instances (0.02) of quotation-final si. An account of the differences between quote-initial and quote-final reporting verbs is beyond the scope of this study, but we hypothesize that the quote-final use is typical of expressive or dramatic contexts, in that it presents the quotation in medias res, without any introduction. The older speakers may be using this construction for dramatic effect as an alternative to the new quotative markers. The results also yield another interesting pattern, namely a difference in frequency among the markers themselves. The total scores reveal that bare is the most common quotative marker with 375 instances in the corpus (0.47 per thousand words), whereas sånn occurs only 136 times (0.17) and liksom is represented with a total of 88 instances (0.11), 32 in quote-initial position (0.04) and 56 in quote-final position (0.07). We may conclude that bare is the most popular, and thus maybe the prototypical, quotative marker in Norwegian, at least among younger speakers in the Oslo region. We also see that liksom is the least used quotative marker, and is most common in quote-final position These results are in line with previous quantitative findings on the marker sånn in both quotative and non-quotative usages in the NoTa corpus (see Lie 2008). The quantitative analysis is also in line with Rathje’s (2011) study of interactional differences between three generations of Danish female speakers. Rathje shows that the youngest speakers use discourse markers like bare and sådan/sån as quotatives considerably more than middle-aged and elderly speakers, who instead prefer to use traditional quotative verbs. Rathje’s youngest speakers are responsible for 81% of the total amount of discourse markers used as quotatives. These findings seem



By three means 

to suggest not only that all three markers – sånn, liksom and bare – are innovations that recently started being used as quotative markers; it thus appears that the recent advent of these quotative markers is a pan-Scandinavian phenomenon. 6. Combinations of the markers We have already pointed out the fact that liksom tends to be post-posed, whereas the other quotative markers occur in pre-posed position. Hence, unsurprisingly, final liksom combines relatively freely with the pre-quote markers, including preposed liksom (cf. Examples 16 and 17 for illustrations of the combination liksom + Q + liksom). The distribution of the combination of the markers present in the NoTa-corpus is given in Table 3. The most frequent combinations in quotative constructions are bare sånn + Q as well as bare + Q + liksom. The next example illustrates an especially rich combination, involving all the three quotative markers: (30)

(UNO E-S-f) Helena: nei. jeg bare sånn “(lagnavn) er jo ikke noe gode” lissom. No. I just like-this “(name of team) are yes not any good” like. No. I just like this “(name of team) they’re no good” you know/like.

The syntactic structure of the reporting utterance is not the one associated with sånn (namely, S + finite verb + sånn) but rather the one associated with bare (S + bare), that is, without a finite verb. Pragmatically, the structure seems to combine the intensifying function associated with bare and the demonstrative function associated with sånn. In addition, we can observe that liksom occurs after the quotation, where it serves a double function of hedging the former quote (hence the translation ‘like’) as well as eliciting hearer response (hence the translation ‘you know’). Table 3.  Distribution of the combination of the markers in in the NoTa corpus Combination of the markers bare + sånn + QUOTATION bare + QUOTATION+ liksom liksom + bare + QUOTATION sånn + QUOTATION+ liksom bare + sånn + QUOTATION+ liksom liksom+ QUOTATION+ liksom liksom + sånn + QUOTATION bare + liksom + QUOTATION

Number of occurrences 19 17  6  5  4  2  2  1



Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig

In these combinations we can ask whether the markers are used with a quotative function or whether one or more of them has another discourse function. We believe that since these markers serve complementary pragmatic functions in the quotative use, they may well be combined as quotatives, but this is a question that deserves a more thorough investigation in future research and on the basis of more data. 7. Conclusions Previous analyses of the English quotative like have identified a range of pragmatic functions, such as epistemic hedge, focus marker and filler, and the same may be said about quotative markers in other languages, as demonstrated by the various contributors to this volume. Some studies of like have claimed that one of these functions is central or primary (Andersen 1999, 2000; Jucker and Smith 1998), whereas others suggest that the whole range of functions may be present in different situations, yielding a pattern of widespread polysemy and multifunctionality (see, for instance, Buchstaller 2004; Hasund 2003; Romaine and Lange 1991). Yet, the meaning of comparison and similarity seems to underlie the various functions and resonate in them. In Norwegian, there is not one, but three different quotative markers with comparative/similative semantics, namely bare, sånn and liksom. The functions associated with like in English are discernable to various degrees in these markers as well, but they seem to be more functionally specialized so that each one of them is less multifunctional than like. Bare derives from an adverb indicating uniqueness and exclusivity, and as such, denies the possibility of alternative versions of the quoted material. This means that the speaker commits fully to the rendering of the reported utterance. Indeed, whereas bare in discourse marker function can serve both affective-intensifying and epistemic-hedging functions, in its quotative uses, it is only the affective-intensifying function which is salient. The speaker stance expressed by bare as a quotative is thus primarily affective, not epistemic (contrary to liksom and sånn).14 From the data we have investigated, it appears that bare may be developing into a more or less neutral quotative, that is, a grammaticalized form that is losing its pragmatic potential as an intensifier, similar to what has happened to the corresponding

14. This intensifying function may be seen as corresponding to the ‘focus’ function of like. Note that the literature on quotative like harbors some terminological confusion regarding pragmatic functions such as ‘focus’ and ‘intensification’, which seem to be used interchangeably (cf. Hasund 2003 for a discussion of this point).



By three means 

Swedish quotative ba(ra). In this respect, it may be seen as resembling the use of like in present-day English as a “wildcard” quotative (Buchstaller 2004: 62). Liksom and sånn both resemble like in that they originate from nondiscourse uses that involve comparison of some sort. They are similar in that both mark the speaker’s epistemic stance, with this stance marking closely connected to the element of comparison inherent in both. These quotatives present the quotation as an illustration or exemplification of a reported state of affairs. However, liksom and sånn express opposite epistemic stances: liksom is an epistemic hedge, expressing approximation and inexactness of the quote to the original speech act, while sånn is an epistemic focus marker, expressing certainty and exactness of the demonstration. As mentioned in 4.1, we use the term focus marker for sånn and intensifier for bare, in order to distinguish the epistemic stance associated with sånn from the affective stance associated with bare. The possibility for sånn and liksom to express opposite degrees of epistemic stance can be seen as related to the element of comparison and similarity inherent in both quotatives. Both liksom and sånn have what we may call a double semantic potential, expressing alternatively ‘approximately similar’ (x is approximately like y) as well as ‘exactly similar’ (x is exactly like y) (see Hasund 2003: 92; cf. also the classification in Quirk et al. 1985: 635 of the English expressions likewise and similarly as ‘equative’ linking adverbials, a term which focuses on the sameness and not the difference of the comparison). Thus, there are two possible paths for the core semantic meanings of nondiscourse liksom and sånn to be redistributed or extended to the discourse level (see Lakoff 1987): one is the extension of the core meaning of approximation, imprecision and ‘not sameness’ to an epistemic hedge (liksom), the other is the extension of the core meaning of exactness, precision and ‘sameness’ to an epistemic focus marker (sånn). In addition to its speaker-oriented epistemic function, liksom in quote-final position also serves a hearer-oriented function as a marker of appeal or hearer involvement, often resulting in double marking. This hearer-oriented function is different from like, which is not regularly used in quote-final position (Hasund 2003). It is also not found for bare. As can be seen, the inventory of quotative constructions in Norwegian has been, and still is, in the process of being reshaped as the new quotative markers replace the traditional constructions with quotative verbs. The availability of three different markers used for introducing quotations has led to a differentiation and specialization among forms involving different associated pragmatic functions. Moreover, among the three markers, bare has commanded an especially salient position as both the most common and the most conventionalized quotative marker.



Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig

References Andersen, Gisle. 1999. Pragmatic Markers and Sociolinguistic Variation: A Corpus-based Study. Dr. art. Thesis. Bergen: University of Bergen. Andersen, Gisle. 2000. “The role of the pragmatic marker like in utterance interpretation.” In Pragmatic Markers and Propositional Attitude, Gisle Andersen and Thorstein Fretheim (eds), 17–38. Amsterdam: John Benjamins. Berlin, Brent. 1992. Ethnobiological Classification. Princetown University Press. Bucholtz, Mary. 2004. “From Stance to Style: Innovative Quotative Markers and Youth Identities in Discourse.” Paper presented at the Sociolinguistic Perspectives on Age, NYU. Buchstaller, Isabelle. 2004. The Sociolinguistic Constraints on the Quotative System – British English and U.S. English Compared. Unpublished doctoral dissertation, University of Edinburgh. Buchstaller, Isabelle. 2006. “Diagnostics of age-graded linguistic behaviour: The case of the quotative system.” Journal of Sociolinguistics 10 (1): 3–30. Buchstaller, Isabelle. 2011. “Quotations across the generations: A multivariate analysis of speech and thought introducers across 4 generations of Tyneside speakers”. Corpus Linguistics and Linguistic Theory 7 (1): 59–92. Buchstaller, Isabelle and D’Arcy, Alexandra. 2009. “Localized globalization: A multi-local, multivariate investigation of quotative be like.” Journal of Sociolinguistics 13: 291–331. Buchstaller, Isabelle, Rickford, John R., Traugott, Elizabeth C., Wasow, Thomas and Zwicky, Arnold. 2010. “The sociolinguistics of a short-lived innovation: Tracing the development of quotative all across spoken and internet newsgroup data.” Language Variation and Change 22: 191–219. Clark, Herbert H. and Gerrig, Richard J. 1990. “Quotations as demonstrations”. Language 66 (4): 764–805. D’Arcy Alexandra. 2007. “Like and language ideology: Disentangling fact from fiction”. American Speech 82 (4): 386–419. Eckert, Penelope. 1997. “Why ethnography?” In Ungdomsspråk i Norden, Ulla-Britt Kotsinas, Anna-Brita Stenström and Anna-MalinKarlsson (eds), 52–62, MINS 43. Stockholm: Stockholms Universitet. Ekberg, Lena. 2007. “‘...sånn svensk å blond å sånt du vet.’ Lexiko-grammatiska drag i Malmöungdomars talspråk.” [Lexico-grammatical features in the speech of Malmø adolescents] In Språket hos ungdomar i en flerspråkig miljö i Malmö, Lena Ekberg (ed.), Nordlund 27. Småskrifter från Nordiska språk vid Lunds universitet, 48–77. Lund: Lunds universitet. Ekberg, Lena. 2010. “Extended use of sån (‘such’) among adolescents in multilingual Malmö, Sweden.” In Multilingual Urban Scandinavia. New Linguistic Practices, Pia Quist and Bente Ailin Svendsen (eds), 17–30. Bristol: Multilingual Matters. Eriksson, Mats. 1992. “Ett fall av grammatikalisering i modern svenska: Ba i ungdomars talspråk.” [A Case of grammaticalization in modern Swedish: Ba in adolescents’ speech] [SoLiD nr. 1 FUMS rapport nr. 166] Uppsala Universitet: Uppsala. Eriksson, Mats. 1995. “A case of grammaticalization in modern Swedish: The use of ba in adolescent speech.” Language Sciences 17: 19–48. Eriksson, Mats. 1997. Ungdomars berättande. En studie i struktur och interaktion. [Storytelling in adolescence. A study of structure and interaction] Skrifter utgivna av institutionen för nordiska språk 43. Uppsala: Uppsala Universitet.



By three means  Erman, Britt. 1997. “‘Guy’s just such a dickhead’: The context and function of just in teenage talk.” In Ungdomsspråk i Norden, Ulla-Britt Kotsinas, Anna-Brita Stenström and AnnaMalin Karlsson (eds), 96–110, MINS 43. Stockholm: Stockholms Universitet. Fretheim, Thorstein. 1980. “Bare – er en enhetlig semantisk beskrivelse mulig?” [Bare – is a unified semantic description possible?] Norskrift 29: 1–26. Fretheim, Thorstein. 1999. “‘Det er bare det at...’” [It is just that ...] Norsk Lingvistisk Tidsskrift 17(1): 155–170. Golato, Andrea. 2000. “An innovative German quotative for reporting on embodied actions: Und ich so/und er so ‘and I’m like/and he’s like’.” Journal of Pragmatics 32: 29–54. Hasund, Ingrid Kristine. 2003. The Discourse Markers ‘like’ in English and ‘liksom’ in Norwegian Teenage Language. A Corpus-Based, Cross-Linguistic Study. Unpublished doctoral dissertation, University of Bergen. Hopper, Paul J. and Traugott, Elizabeth C. 2003. Grammaticalization. Cambridge: Cambridge University Press. Jönsson, Linda. 2005. “Tal i tal. Grammatiska konstruktioner av anföring i tonårsflickors samtal.” [Talk in talk. Grammatical construction of direct speech is teenage girls’ talk] In Samtal och grammatik. Studier i svenskt samtalsspråk, Jan Anward and Bengt Nordberg (eds), 89–108. Lund: Studentlitteratur. Jucker, Andreas H. and Smith, Sara W. 1998. “And people just you know like ‘wow’: Discourse markers as negotiating strategies.” In Discourse Markers: Descriptions and Theory, Andreas H. Jucker and Yael Ziv (eds), 171–201. Amsterdam: John Benjamins. Kärkkäinen, Elise. 2003. Epistemic Stance in English Conversation. A description of its interactional functions, with a focus on I think. Amsterdam: John Benjamins. Kärkkäinen, Elise. 2006. “Stance taking in conversation: From subjectivity to intersubjectivity.” Text & Talk 26: 699–731. Katriel, Tamar and Dascal, Marcelo. 1989. “Speaker’s commitment and involvement in discourse.” In From Sign to Text. A Semiotic View of Communication, Yishai Tobin (ed.), 275–295. Amsterdam, Philadelphia: John Benjamins. Kotsinas, Ulla-Britt. 1994. Ungdomsspråk. [Youth Language] Uppsala: Hallgren and Fallgren. Lakoff, George 1987. Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago, Illinois: University of Chicago Press. Lee, David. 1987. “The pragmatics of just.” Journal of Pragmatics 11: 377–398. Lie, Svein. 1979. “Bare er ikke bare bare og ikke også heller.” [‘Bare’ is not only ‘bare’ an not ‘også’ either] Norskrift 25: 24–45. Lie, Svein. 2008. “Veldig sånn festejente.” [Very such party girl] In Språk i Oslo. Ny forskning omkring talespråk, Janne Bondi Johannessen and Kristin Hagen (eds), 78–95. Oslo: Novus. Lindström, Jan. 2008. Tur och ordning. Introduktion till svensk samtalsgrammatik. [Turn and Order. Introduction to Swedish conversation grammar] Stockholm: Norstedt Akademiska Förlag. Londen, Anne-Marie. 1997. “Då e man ju fri liksom. Om användningen av partikelen liksom i ett finlandssvenskt radiosamtal.” [On the use of the particle liksom in a Finnish-Swedish radio conversation] In Svenskan i Finland 4, Saara Haapamäki (ed.), 101–118. Åbo: Svenska Institutionen. Maschler, Yael. 2001. “veke’ilu haragláyim sh’xa nitka’ot bifním kaze (‘and like your feet get stuck inside like’): Hebrew kaze (‘like’), ke’ilu (‘like’), and the Decline of Israeli dugri (‘direct’) Speech.” Discourse Studies 3(3), 295–326. Mathis, Terrie and Yule, George. 1994. “Zero quotatives.” Discourse Processes 18: 63–76.



Ingrid Kristine Hasund, Toril Opsahl and Jan Svennevig Meyerhoff, Miriam. 2002. “All the same? The emergence of complementisers in Bislama.” In Tom Gueldemann and Manfred von Roncador (eds), Reported Discourse: A meeting ground for different linguistic levels, 341–359. Amsterdam: Benjamins. Miller, Jim and Weinert, Regina. 1995. “The function of like in dialogue.” Journal of Pragmatics 23: 365–393. NoTa: Norsk talespråkskorpus – Oslodelen, [Norwegian spoken language corpus, the Oslopart] Tekstlaboratoriet, ILN, Universitetet i Oslo. (29 September 2010) Ochs, Elinor, Schegloff, Emanuel A. and Thompson, Sandra (eds). 1996. Interaction and Grammar. Cambridge: Cambridge University Press. Opsahl, Toril. 2002. “‘Jeg bare: Hæ?! Hva skjedde her da?’ En studie av de kommunikative funksjonene til diskurspartikkelen bare i et ungdomsspråkmateriale.” [A study of the discourse particle bare’s communticative functions in youth language] Unpublished master thesis, University of Oslo. Opsahl, Toril and Svennevig, Jan. 2007. “Må ha det. Bare må ha det. Bare som pragmatisk partikkel i samtale.” [“Have to have it. Just have to have it” Bare as pragmatic particle in conversation] Norsk Lingvistisk Tidsskrift 25 (1): 29–55. Öqvist, Jenny. 2000. “Jamen förr i tiden så va re ju mera sär “å vicken vacker hatt du har”. Om partikeln ‘sär’s funktioner i samtal.” [But in the old days then it was more like “oh what a beautiful hat you’ve got”] (5 April 2009). Quirk, Randolph, Greenbaum, Sidney, Leech, Geoffrey and Svartvik, Jan. 1985. A Comprehensive Grammar of the English Language. London: Longman. Rathje, Marianne. 2009. Generationssprog i mundtlig interaktion. En sociolingvistisk undersøgelse af generationsspecifikke sproglige og interaktionelle træk i tre generationers talesprog. [Generation language in spoken interaction. Asociolinguistic study of generational differences regarding language and interaction in the speech of three generations] Unpublished doctoral dissertation, Københavns universitet. Rathje, Marianne. 2011. “Quotations and quotatives in the speech of three Danish generations.” In ICLaVe#5 proceedings, Frans Gregersen, Jeffrey Parrott and Pia Quist (eds). Amsterdam: Benjamins. Romaine, Susan and Lange, Deborah. 1991. “The use of like as a marker of reported speech and thought: A case of grammaticalization in progress.” American Speech 66: 227–279. Rosch, Eleanor. 1975. “Family resemblances: Studies in the internal resemblance of categories. Cognitive Psychology 8: 382–439. Rosch, Eleanor. 1978. Principles of Categorization. In Eleanor Rosch and Barbara Lloyd (eds.), Cognition and Categorization, 27–48. NewYork, Hillsdale. Røyneland, Unn. 2009. “Dialects in Norway: Catching up with the rest of Europe?” International Journal of the Sociology of Language 7–31. Schiffrin, Deborah. 1987. Discourse Markers. Cambridge: Cambridge University Press. Selting, Margret and Couper-Kuhlen, Elizabeth (eds). 2001. Studies in Interactional Linguistics. Amsterdam: Benjamins. Simonsen, Hanne Gram and Christensen, Kirsti K. 1980. ”‘Når man byner å gå sånn litt på skolen og sånn’. En femårings bruk av sånn.” [When you start to go such to school and such. A five-year-old’s use of sånn] In The Nordic Languages and Modern Linguistics. Proceedings of the Fourth International Conference of Nordic and General Linguistics in Oslo 1980, Even Hovdhaugen (ed.), 334–344. Oslo, Bergen, Tromsø: Universitetsforlaget.



By three means  Stenström, Anna-Brita, Andersen, Gisle and Hasund, Ingrid Kristine. 2002. Trends in Teenage Talk. Corpus Compilation, Analysis and Findings. Amsterdam: John Benjamins. Tagliamonte, Sali and D’Arcy, Alex. 2007. “Frequency and variation in the community grammar: Tracking a new change through the generations.” Language Variation and Change 19(2): 199–217. Tannen, Deborah. 1984. Conversational Style. Analyzing Talk among Friends. Norwood, NJ: Ablex. Tannen, Deborah. 1989. Talking Voices. Repetition, Dialogue, and Imagery in Conversational Discourse. Cambridge: Cambridge University Press. UNO: Språkkontakt og ungdomsspråk i Norden – Talespråkskorpuset, Oslodelen. [Language Contact and Teenage language in Scandinavia – The Norwegian Corpus of Oslo Teenage Language] English department, University of Bergen. .

Appendix: Transcription Conventions symbol comment , continuing intonation, slightly rising pitch . terminating intonation, falling pitch (statements) ? terminating intonation, rising pitch (statements, questions) : lengthened syllable % incomplete word (.) brief pause (  Dubitative  >  Volitive This tentative implicational scale assumes that unmarked/minimally marked strategies relate to doxic modality and that dubitative and volitive modality are encoded by either similar or more complex strategies. Typologically, it may be expected that languages with current speaker-evaluative meanings have either only doxic, doxic and dubitative or doxic, dubitative and volitive modality but no other combinations. Again, the question of whether this impressionistic hypothesis will survive scrutiny requires a more substantial typological study than I have been able to pursue in this chapter. 4.3

Number and functions of quotative constructions

The number of quotative constructions in the sample languages ranges from as few as just one in, for example, Ngarinyin and Semelai, to four in languages such as Koromfé and Lezgian. A remarkable tendency in the sample is best described with the implicational scale in (28). (28) Direct speech/reportativity > indirect speech > insubordinated speech In words: if a language employs only one type of quotative construction, it is either a bi-clausal direct speech strategy (e.g. a ‘X says’-clause and an appositional message clause) or a grammaticalized (reportative) strategy (i.e. a reportative morpheme). If there are more constructions available, these are most likely indirect speech constructions. If both/all three types are found, the language may also employ an insubordinated strategy as well. This observation is further evidence for rejecting a priori distinctions between grammaticalized and non-grammaticalized strategies when it comes to the encoding of quotation. I will interpret the quotatives as indicated in the second column of Table 2 as paradigmatic oppositions, which prompts the question as to what the semantic oppositions encoded by the different quotatives are. Although we need not expect a one-to-one correlation between the number of quotative constructions and the co-encoding of speaker attitudes, the literature suggests that semantic distinctions in quotatives in languages that have direct and indirect speech can generally be captured with the notion of objectivity (see Coulmas 1986). In the sample, Semelai appears to be the language in which the possibilities for encoding the relation between the reported and the current speakers are maximally restricted. The language has only a single (grammaticalized) quotative construction. It has no framing verbs and in all examples in which modal markers are used, they express an evaluation on behalf of the reported speaker. However, the three languages with the most different quotative constructions – Koromfe,



Stef Spronck

Lezgian and Toqabaqita – are not the languages for which the most fine-grained explicitly encoded speaker attitudes have been reported. It is important to bear in mind, though, that of all numbers in Table 2, the exact number of constructions in the column Q may be the least indicative. This is so because non-(primarily) quotative constructions may be used to encode reports and not everything that looks like a quotative construction is (exclusively) a quotative construction. For example, Martin (1975: 666) describes a process in Japanese where certain nominalizing constructions “[...] may be marked as if quoted”, a process he terms “quotationalization” (see also Oshima and Sano this volume). In Manumbu, quotative constructions may be “[...] employed to express internal speech and thought, desire and intention of third person, reason and purpose and a few other related meanings” (Aikhenvald 2008: 484). A similarly multifunctional construction has been found in Ngarinyin (Rumsey 1982), and is also common in many Papuan languages (Lila San Roque, p.c.). Reported thought and reported speech are also co-encoded in the same constructions in a number of American languages such as Wari (Everett and Kern 1997) and Aguaruna (Larson 1978). Although I have not addressed reported thought in the discussion here, it should be clear that much of the analysis presented in this chapter can also be applied to reported thought with some modifications, as was also indicated by the close relation between the present topic and constructions as the mistaken belief construction in (22). 4.4

Categories in the de dicto domain

Quotative constructions constitute a specific structural and semantic environment, within which particular meanings tend to grammaticalize. Frajzyngier (1991) refers to this grammatical environment as the de dicto domain. Examples of frequently occurring grammaticalization processes in this domain are insubordination, that is, “subordinated” clauses without a “main clause” (Evans 2007) (see example 30), but also some of the modal meanings arising from the appli­ cation of certain tense and mood categories in quotative constructions (see Section 4.5). A number of languages in the sample have grammatical categories that are either exclusively found in quotative constructions, or acquire a specific meaning when appearing in a quotative construction. In the results table, these languages are indicated in column C. One such category is that of logophoricity which in the sample is attested in such genetically diverse languages as Ewe, Lao, Lezgian and West-Greenlandic. The primary function of a logophoric marker in quotative constructions in these languages is to indicate that the referent in the framing clause, the source construction, is the same as the one in the embedded clause, the message construction. While Lao, Lezgian and West-Greenlandic use a



Minds divided 

reflexive construction to encode co-referentiality with the reported speaker in a framed clause, Ewe has a marker that is exclusively used for this purpose. (29) kofi bé ye- a- dzó K. say log irr leave ‘Kofii said that hei would leave’

(Ameka 1991: 62)

Although its main function appears to be deictic/anaphoric, that is, keeping track of discourse participants in a narrated event, a marginal secondary usage of logophoricity has also been reported. In Tuburi (Niger-Congo), logophoric pronouns may sometimes head insubordinated clauses signalling reportativity: see (30). (30) sārā dús sō log disperse thus ‘Theyi [the ancestors] then dispersed/spread out’  (Dimmendaal 2001: 136) A second quotative specific category is the subjunctive in German, which is used in framed clauses in quotative constructions, although Jung (1971) observes that this rule is prescriptive rather than descriptive of spontaneous discourse. A more general and fundamental category that interacts with speaker attitudes in quotative constructions is person. An interesting case is that of languages with grammatical honorific marking (in the sample: Japanese, Korean and Lao). Korean distinguishes six different speech levels: plain, intimate, familiar, blunt, polite, deferential plus a seventh, neutral form (Sohn 1994: 8–11). “[T]he neutral level appears in quotative clauses irrespective of the social status of the addressee” (Sohn 1994: 341), that is, there is no honorific distinction in quotation. However, we cannot approach grammatical person in quotation with a monolithic view. For example, there is a clear asymmetry between the discourse participant (first and second person) and non-participant person (third person) distinctions. For example, a first person present tense framing clause like “I am telling you...” could be interpreted as “I am asserting that p is true” (cf. Huttar and Huttar 1994: 581 on a similar construction in Ndyuka). This effect is not a coincidental, language-specific phenomenon in English but may also be attested in many different languages. For example, Donaldson (1980: 240) reports that the ‘believed true’ marker -ŋadhanga in the Australian language Ngiyambaa (Pama-Nyungan) diachronically derives from the verb SAY inflected for first person. Compare also the Lezgian (31) and Aguaruna (Peru; Jivaroan) (32), which combine first person framing clauses with adverbs meaning ‘true’. či xür-e lišanlu (31) za-qh, Nadja, düz laha-j-t’a, i-poess N right say-aop-cond we:gen village-iness betrothed



Stef Spronck

ruš wa-j-di tir girl be.in-ptp-sbst cop:pst ‘Honestly [litt.: if (I) say (it) rightly], Nadja, I used to have a fiancée in our village’ (Haspelmath 1993: 308) (32)

Nueva Vidak tikima yaigchi, junak dekas N. V.-top very it-is-small that-obj-top truly yaktauchin tajai village-small-obj I-say ‘New Life is really small. I say it is a really small village’ (Larson 1978: 55)

A similar process is also reported by Lee (2007: 379), who cites a description in Martin (1975) of the particle yo, meaning “‘I want you to know’, ‘Believe (you) me ... ’, ‘I tell you’, ‘I’d say’ or ‘Let me tell you’”. ‘As implied in these translations, enhancing one’s role as the deliverer of information indicates the speaker’s desire to be in a superior position over the partner with respect to the utterance contents, implying “I am sure of this more than you are” or “You did not know this, did you” (Lee 2007: 379). Another case in point is the following description of the Cupeño reportative marker ku’ut: “A few sentences are attested where =ku’ut does not seem to have a ‘reportative’ function but instead seems to challenge the validity of a statement by another” (Hill 2005: 292): see (33). (33) yax-qal=et tani-lya’-ika=ku’ut ha$h-ax-qat ivi-y tukmiyat say-pis=2sabs dance-instn-to=rep go-yax-if pdem-o night You say you’re going to go to a dance tonight? (Faye’s translation: You say you want to go to a dance tonight.) (Hill 2005: 292) The meaning ‘contesting the truth’ seems to derive directly from the combination of address (second person) and the quotative construction. In addition, the fact that Hill translates (33) as an interrogative sentence is significant as it underlines the illocutionary function ‘requesting confirmation of the statement’. The indicated asymmetry between first person (and second person in interrogative clauses) and non-first person (second in declarative sentences/third/ hearsay) in quotatives thus seems to interact with particular modal meanings (also cf. Curnow 2002), which is most clearly expressed in languages that have what several researchers have referred to as conjunct/disjunct systems (Bickel and Nichols 2007: 223–224; Hale 1980). In these languages, person agreement is grouped into conjunct forms, generally referring to discourse participants (i.e. first person/second person) and non-participants, that is, third person (or sometimes non-first person), called disjunct. In Tsafiki (Barbacoan), this opposition may also be used to express logophoricity (cf. 34a,b), and even ‘unintentionality’ (cf. 34c).



Minds divided 

(34) a.

ya mantoka jiyoe tie ya man-to=ka ji-yo-e ti-e 3 other-earth=loc go-conj14-decl say-decl (Dickinson 2000: 385) ‘Hei said that hei went to Santo Domingo.’

b.

ya mantoka jie tie ya man-to=ka ji-e ti-e 3 other-earth=loc go-decl say-decl ‘Hei said that hej went to Santo Domingo.’ (ibid.)

c.

la yaka machitechi poreie la ya=ka machite=chi pore-i-e 1masc 3=acc machete=instr cut-disj-decl ‘I cut him (unintentionally) with the machete.’ (Dickinson 2000: 387)

When using the conjunct form (as in 34a), the referent he is understood to be coreferential with the reported speaker. In the absence of the conjunct marker, the referent he is understood to be someone else than the speaker. Similar to Tuburi (30), the pair (34a,b) exemplifies how reference tracking morphology may develop into specific grammatical marking in quotative constructions. The meaning ‘unintentionally’ in (34c) is a secondary meaning of disjunct marker in Tsafiki. Although not exemplified in (Dickinson 2000), it would be interesting to see if this meaning could arise in quotative constructions meaning ‘unintentionally/inadvertently say’. Less grammaticalized person effects may be observed in Manumbu (cf. 35), where the unusual person reference in the framed clause may be assumed to enhance the dramatic effect of the quotation. (35)

l6-k6 mam6k ata 3fsg-link + fsg elder_sibling + link + dat then wa-l6-l a-d6 du [pause] say-3fsg.subj.p-3fsg.obj.p dem.dist-Msg man [pause] wun k6ta an-a:m k6-k6r 1sg:dir.sp.rep now 1du-link + obj:ind.sp.rep eat-des ata wa-na-d thus say-act.foc-3msg.subj.np ‘She said to her elder sister thus: “That mani: ‘Ii want to eat us now’ (hei) said”’ (Aikhenvald 2008: 395)

A final and summarizing observation about the category of person in quotative constructions is that effects observed in the source construction in the preceding

14. Dickinson (2000) refers to this form as “congruent” versus “non-congruent”.

 Stef Spronck

examples, at least in English, also apply to the addressee as encoded in the source construction: compare (36a–h). (36) a. I said I would come b. c. d. e. f. g. h.

> highlighting commitment to truth 15 > request for commitment to You said you would come truth He said he would come > suspension of commitment to truth I said to you/him I would come > inviting inference about truth You said to me/him you would come > inviting inference about truth He said to me he would come > highlighting status of message: report He said to you he would come > inviting inference about truth He said to him he would come > suspension of commitment to truth

First person reported speakers (36a) invite the inference “I am saying it so it is true”, and first person reported addressees invite the inference “as far as I know it is true” (36f). Second person reference in the source construction invites either a request to the reported speaker to recommit to the truth of the message (36b) or to invite some inference about the truth of the message (36d,g), for example, highlighting the status of the reported message as a promise or suggesting that the reported speaker might provide a different message to the reported addressee than to someone else (i.e. contrastive stress). Reported speakers or reported addressees other than current discourse participants (i.e. 36c,h) have no truth enhancing effect, but rather suspend commitment to the truth of the message on behalf of the current speaker. Prosodic prominence of the reported addressee in the form of contrastive stress invites the inference “the reported speaker might have given this message to this reported addressee, but could have said something else to some other addressee”, that is, the message may not be reliable. At least for the languages I have been able to survey here, all these meanings seem to arise primarily from pragmatic inference, which entails that they may be cancelled in non-default contexts. 15. Pragmatic stress on the speech verb might invite the inference of lying a bit too strongly in direct address, possibly making the stress pattern “You said you would come” more natural, but for the sake of exposition and parallelism between (36a–h), I will adhere here to similar patterns for (36a–c) and (36d–h). Politeness likely plays a similar role in what Evans (fc.) calls “second person magnetism”.



Minds divided 

The representation in (37) may be taken as an illustration of the inferential processes reported so far. Inferential processes are always calculated from the perspective of the current speaker, that is, the discourse entity making the report. As she is necessarily present, this layer of evaluation is always present. A second layer of evaluation is that of the reported speaker. The next layer in the representation in (37) is that of the evaluation itself, which may or may not be explicitly expressed. I hypothesize that when the evaluation is not made explicit, an evaluation about the truth of the message is implied. In cases where the two outer layers are discourse participants, the commitment to the truth is strengthened; if the reported speaker is not a discourse participant (i.e. 3rd person), commitment to the truth is suspended. The innermost layer is that of the message itself, the other-discourse element. Constructions highlighting this layer do not primarily evaluate the message but signal lack of responsibility for uttering the information contained in it, although the embedded structure in (37) implies that this cannot be done without some evaluation of the message (e.g. suspension of commitment to the truth).16 The round brackets in (37) should distinguish the elements in this representation from the meaning-form complexes in the representations above: as I have aimed to show, current speaker-evaluative meanings arise in the complex interaction of the semantic elements in (37), although the constructions in (26a–f) do not correspond one-to-one to any of the meanings in (37). (37) (Current speaker (Evaluating discourse entity (Evaluation (Message)))) I am assuming that the meanings ‘suspension of commitment to the truth of the message’ and explicitly doubting the message are different (also compare the inferential scale (27) in Section 4.2). The exact nature of this distinction is yet to be described. 4.5

Evidentiality in quotation

As indicated above, my approach to evidentiality in quotatitive constructions differs slightly from the treatment in, for example, Aikhenvald (2004), where the only evidential category relevant to quotation is a single evidential category reportative. I explicitly assume that quotative constructions do not represent a single evidential value but may express several different evidential types. One way in which this may be done is exemplified in this section: the combination of evidential morphemes with a source construction.

16. Following Tom Güldemann’s definition of reportative discourse as an act of distancing (Güldemann this volume), this distancing may be conceptualized on these two levels.

 Stef Spronck

Ten of the twenty five languages in the sample display grammaticalized reportativity, either with verbal inflection, a verbal suffix or a particle encoding a general type of quotation (mostly ‘hearsay’). This is a considerably higher number than the languages which have fully developed evidential systems and encode other types of sources of information (e.g. visual, non-visual, sensory) as well (5/25). Currently, little is known about the specific combinatory features of reportative markers and evidential markers. Evidence from Tariana suggests that even though non-reportative evidential markers are not generally stacked, reportative markers may combine with other types of evidential markers (Aikhenvald 2003).17 A similar observation can be made for Duna. Most quotative constructions in San Roque (2008) indeed combine a verbum dicendi or reportative marker with an evidential marker. Compare, for example, the Duna quotative constructions (38) and (39), which feature the visual evidential suffix -tia and the sensory evidential suffix -yanua respectively. (38)

Lepani Mospi nga-nda ri-tia-na, k˜ao=koae=pe, psn pln go-int say-pfv.vis.p-spec lie=hyp=q ene=koae, ko waki po=pe true=hyp 2sg hear do.pfv=q ‘Lepani is going to Moresby [people] said, I wonder could it be false or true, have you heard?’ (San Roque 2008: 430)

(39) ko inginiwane akita ri-yanua 2sg offspring how.many say-sns.impl ‘How many children do you have? [she] says.’

(San Roque 2008: 419)

Example (38) appears to illustrate the default evidential marking in quotations in Duna. Although San Roque (2008: 430) suggests that the visual evidential marker might have some modal overtones for some speakers, evidential markers primarily relate to the physical world, in this case the reported utterance as a speech signal. We might interpret the visual evidential marker in (38) as placing the current speaker in the reported speech situation as an observer. The reported speech situation in (39) is slightly different. Example (39) represents a translation, that is, the utterance is relayed to the original addressee of the utterance by a different speaker 17. Lila San Roque points out that combining evidential and modal markers is not an uncommon phenomenon in languages with grammaticalized evidential morphology and rightly stresses that there are (at least) three distinct possible semantic effects for quotative constructions: the evidential is just there to mark the quotative construction (the evidential indicates that there is a report), the evidential takes scope over the report (e.g. I report that someone saw p) or the evidential is ‘inside’ the report (e.g. I saw someone report p) (Lila San Roque, p.c.). The Duna examples in (38) and (39) appear to be of the type ‘evidential inside a quotative construction’, and this seems the most likely type to find current speaker-evaluative meanings in.



Minds divided 

in a different language. The current speaker indicates his role as an interpreter in the reported speech situation by suffixing a sensory evidential marker. For West-Greenlandic, Fortescue (1984: 294) reports that an affix (classified as “authority of assertion” in the Routledge descriptive grammar series) is used “to indicate that the speaker did not witness and event himself but accepts the report as reliable”, but no examples of quotative constructions are given in the grammar. 4.6

Modal categories

As I briefly sketched in Section 3, definitions of modality and related notions are notoriously diverse and, accordingly, descriptions in grammars of the sample languages are exceedingly varied. For example, Lichtenberk (2008: 754) defines mood as “[...] expressions of modality that are not purely lexical”, which would include markers such as the speaker conclusion suffix in Semelai (3), the dubitative marker in Lele (2) and all irrealis inflections. In this study, I interpret mood as the grammatical expression of illocutionary categories such as questions, commands, suggestions, although this definition does not necessarily solve all semantic and morphological ambiguities, either (see also the discussion below). A popular way of understanding the distinction between mood, modality and illocution in functionalist theories of grammar is through a layered clause structure analysis, where modality is predicated at the level of states of affairs (SoA) and mood at the level of utterances (cf. Dik 1997; Van Valin and LaPolla 1997). One clear problem of such an approach, and with a strict structural modality-mood distinction more generally, is that it is contrived for languages that use identical markers both at utterance level and at the level of SoAs, as will be exemplified in Section 4.6. For our present purposes, we may take a relatively naïve approach to modality: modality in quotation is the encoding of an evaluation of a message. This definition fully recognizes that both the message construction (predicating a proposition flagged as a discourse element of some other speaker) and the speaker-evaluative meaning/speaker attitude take on a range of structural guises. Modal markers are notoriously hard to identify, as these may not be straightforwardly distinguished from other grammatical categories. For example, in many languages, there are complex interactions and overlaps between tense forms and modal meanings (cf. Comrie 1985: 23–25, 45) and it is not uncommon that what some authors analyze as modal markers expressing, for example, “shared knowledge” is interpreted as tense forms, such as “general past” by others (cf. San Roque 2008: 350). Ameka (1991: 129) reports that the ingressive aspect marker in Ewe may also be used to express a very similar meaning to the frustrative modal, which is introduced below. There is a more direct link between evaluative meanings and time: the semantics of many modals may be represented on a timeline of two moments of evaluation

 Stef Spronck

such as observed in ‘mistaken belief ’ markers (this is a characteristic modals and quotative constructions share). I will highlight two such categories from the sample in this section: the frustrative and the mirative. The frustrative, which has been identified for Jarawara, Kwaza and Toqabaqita, indicates “[...] that the action was done to no avail – that is, the desired result was not achieved” (Dixon 2000: 293). The category has a double referential relationship between a moment t-1 at which some discourse entity has some intention and a moment t0 at which this intention has failed to materialize. The examples from Kwaza make it especially clear that this category interacts in a striking way with quotative constructions: see (40). (40) eto’hoi kuro-’ra ta-ta-’le-h󰁹-ki child close-imp talk-1o-frust-nom-dec ‘(In vain) the child wanted me to close the door’

(Van der Voort, 2000)

As the translation ‘in vain’ already indicates, the frustrative-plus-quotative combination may either mean that the reported speaker does not succeed in performing the speech act (e.g. because the addressee did not hear him), or that the speech act did not have the intended effect. A category even more closely related to quotative constructions is the mirative: in Turkish, the marker -miş, which has a mirative interpretation, can equally be used to encode hearsay, a pattern also reported for Bulgarian, cf. (41). (41) Kemal gel-miş K come-mirative ‘Kemal came’ (Slobin and Akşu 1982 cited in Delancy 1997: 37) Example (41) may be interpreted as either “They say Kemal came” or (if the context makes it sufficiently clear that every participant in the speech situation is aware that Kemal did come) that the speaker was not expecting the expressed proposition, that is, something like “What do you know, Kemal came!.” In this latter sense, the mirative (see Delancy 1997) is an ‘unprepared mind’ marker, not dissimilar to the ‘mistaken belief ’ marker in (22). It describes an evaluation moment t-1 at which the speaker had some expectation and a moment t0 at which this expectation was disproved. A similar modal category in Korean is termed “appreciative” in Sohn (1994); Kruspe (2004: 289) also indicates that the irrealis in Semelai may be marginally used as a mirative as well. Although the primary meaning of the frustrative (and arguably also the mirative) is not necessarily epistemic, both categories require a factive reading: the evaluation moment t0 has to reflect a fact. Thus, by applying these categories in a quotative construction, the current speaker necessarily expresses that she believes the conveyed message to be true.



Minds divided 

The most widespread modal category in the sample is the irrealis, although the range of its application in the languages varies, as do extended meanings. Ngarinyin appears to be among the most consistent irrealis languages, encoding all non-real world events with an irrealis marker, including all negative clauses (a feature it shares with many Australian languages). In many languages, irrealis markers are said to have overtones of disbelief, but these appear to be inferential rather than semantic and I have found no cases in the sample in which they encode current speaker-evaluative meanings. It seems likely, though, that the irrealis may perform this function in some languages. In addition, however, in sample languages that do contain grammaticalized dubitative markers, it appears that all subjects in complex sentence examples are co-referential with those in the framing clause (and thus in quotative constructions that describe evaluations by the reported speaker), except for the already cited Lele (2). The main conclusion that can be derived from columns MI1 and MI2 appears to be that the number of modal distinctions is not indicative of the presence of current speaker-evaluative meanings. For example, the sample language with the widest range of modal categories, Kolyma Yukaghir, does not seem to be able to encode speaker evaluative meanings using modal markers. On the other hand, languages without modal inflection (i.e. sample languages for which the MI1 and MI2, columns are not filled) may well encode many of the meanings described in this section, including speaker evaluative meanings periphrastically, for example, through complementation strategies. The flexibility of the periphrastic strategy has been most explicitly noted for Japanese: “A sentence can be quoted by adding the particle tó ‘(says/thinks) that’. [...] The sentence quoted can be negative, desiderative etc.; it can be perfect, tentative, etc.; and the quotational verb that follows can undergo all conversions independently of the quoted sentence” (Martin 1975: 996). This is an intriguing observation, of which the full semantic consequences are yet to be systematically described. Yet, with respect to columns MI1 and MI2 it does indicate that focusing solely on languages displaying rich modal inflection and particles when searching for current speaker-evaluative meanings in quotation, as one might be inclined to do guided by languages like Lele, would be myopic. A similar conclusion applies to the remaining columns; each indicates a likely candidate for the expression of speaker-evaluations, but no explicit examples were found in the grammars consulted. I consider whether correlations exist between the number of constructions available for the expression of (epistemic/epistemicdeontic) speaker attitudes in a language and the expressibility of current speakerevaluative meanings in quotation an interesting research question. However, the current survey has not yet unveiled such correlations.

 Stef Spronck

4.7

Illocution and some remaining observations

A category that I did not examine in the survey in Table 2 is illocution, although there are some indications that illocutionary markers may play a role in the encoding of current speaker-evaluative meanings. One reason is that the task of defining illocution separately from mood and modality is complex, and grammars choose different cut-off points for different languages. For example, Jung (1971: 238) and Frajzyngier (2001: 5) define mood (such as declarative, imperative, interrogative etc.) in terms of notions such as “speaker certainty”, or belief in the truth of the encoded proposition (e.g. Frajzyngier 2001: 5; see also Palmer 1986: 9–33 for discussion). These definitions are almost indistinguishable from, for example, John Searle’s definition of “assertives”, one of the five illocutionary acts he distinguishes: “assertives [are] where we tell our hearers (truly or falsely) how things are” (cited from Palmer 1986: 13). However, there are languages with much more fine-grained illocutionary distinctions than English, for example, Lezgian, which has explicitly encoded nondeclarative categories such as imperative, prohibitive, hortative, optative and interrogative (Haspelmath 1993: 149–152). Such specific categories defy straightforward cross-linguistic classifications. More importantly, in a language like Toqabaqita, grammatical markers that in some contexts encode modality, in some other contexts directly seem to specify speech acts, indicating that teasing out these ‘levels’ is slightly artificial in the language, at least on structural grounds. Similar connections between mood and modality might also be seen in modal categories such as “intentive” and “desiderative” in San Roque (2008: 284) and a declarative-assertive marker that Aikhenvald (2003: 398) distinguishes in Tariana. Consider (42a,b). In (42a), the “timitive mood” marker is part of a construction that roughly conveys ‘lest you touch the ants’, a cautionary speech act, whereas in (42b) it designates the speaker’s state of mind/evaluation. (42) a.

lio fas=i faar-a ada ta fuufusi ka look prec=loc underneath tim some ant 3sg.seq too-too qani-a rdp-be.present genp-3sg.obj ‘Look first underneath [the piece of timber], in case there are ants there’ (Lichtenberk 2008: 786)

b. ada dani ka qaru tim rain 3sg.seq fall ‘It might/may rain’

(Lichtenberk 2008: 787)

A similar example can be found in (43a,b), where Lichtenberk glosses the marker qoko as ‘obligation’ in (43a) and ‘I guess’ in (43b).



Minds divided 

(43) a. qoko lae toqo boqo 2sg.seq go oblig ints ‘You should, must go’

(Lichtenberk 2008: 196)

b. too ba-d=i sa-muluqa toqo neri stay lim-3pl.pers=loc adjc-2pl.pers I.guess npast.here ‘I guess [they] are staying with you (are they?)’ (ibid. : 772)  (Lichtenberk 2008: 772) The illocutionary force of (43a) appears to arise from the combination of direct address (second person) and the intensifying/assertive marker, and possibly other factors such as prosody, identifying (43a) as a command (cf. Lichtenberk 2008: 193). This illocutionary force is absent in (43b), and the particle toqo assumes the modal function ‘possibility’. In Kolyma Yukaghir, Malsova observes that questions may be used to “[...] negate the proposition expressed by the clause, or to present it as doubtful” (Maslova 2003: 487) and the optative particle plus nominal resultative inflection expresses “[...] that the situation is strongly desired [by the speaker], yet [she] cannot bring it about” (Maslova 2003: 490). Sohn (1994) similarly indicates that questions are used for a wide range of pragmatic functions (none of which are necessarily modal). What these examples show is that searching for current speaker-evaluative meanings requires a full appreciation of the complex interactions between mood, modality and illocution. Under an analysis in which quotative utterances frame discourse and reported messages are utterances rather than propositions (cf. McGregor 1994), it should not be surprising that illocutionary markers play a role in the encoding of speaker evaluative meanings. An initial impression is that complementizers are a category that is particularly prone to moving between modal and illocutionary in the context of quotative constructions, for example, through grammaticalization processes such as described in Frajzyngier (1995) and Evans (2007). 4.8

Summary of the preceding sections

In the preceding sections the results of the typological survey of the languages in Table 1 were summarized in two ways: by their expression (form/function) in the quotative construction (Section 4.2) and by their grammatical function (Sections 4.3–4.7). The complementizer construction and the message construction were identified as the main constructions expressing speaker-evaluative meanings but are certainly not the only part of the quotative construction encoding speaker attitudes. With respect to the grammatical categories that were shown to play a role in the encoding of speaker attitudes in quotation such as person, evidentiality, modality and mood are several areas requiring further research. In Section 4.4, examples

 Stef Spronck

were introduced showing how evidential categories may be used to refer to the message as forming part of a physical utterance; in Section 4.5, examples of how modal categories may be used to explicitly encode speaker attitudes were introduced. 5. A method for the typological analysis of speaker attitudes in quotative constructions Although understanding and anticipating the perspectives of other discourse participants are essential for acquiring and using a language (cf. Tomasello 2003), the only perspective we have to contribute to the discourse is our own. This is a realization that has to be observed while studying current speaker-evaluative meanings; even ‘objectively’ reported messages reflect how the current speaker perceives the reported message. A quotative construction does not directly reflect the mind of the reported speaker; it represents the mind of the reported speaker through that of the current speaker. As a result, current speaker evaluations are an inalienable part of the semantics of a quotative construction and we may call this the inherent modality of quotative constructions. Objectivity versus subjectivity in quotation lies at the heart of the traditional opposition between de dicto and de re interpretations: consider (44). (44) Oedipus said that his mother was beautiful Ever since Quine (1956), it has been observed that an indirect speech construction, as in (44), may encode either a faithful reflection of the reported speaker’s words (a de dicto interpretation) or a rephrasing of the contents of those words in such a way as to preserve their referential value (a de re interpretation). Given our knowledge of Greek mythology, we will find it unlikely that Oedipus phrased the message as in the embedded clause in (44), leading to interpreting (44) as a de re rendition of an original utterance such as “Jocasta is beautiful”. In more recent studies of quotation, the notions de dicto and de re generally have received less attention, presumably due to the clear non-universality of indirect speech (also pointed out in Coulmas 1986) and the finding that quotatives generally do not represent literal repeats of the reported speaker’s words, so that even purportedly de dicto utterances are not a faithful rendition of the original speaker’s words. Yet, should this mean the de dicto/de re opposition has lost relevance? My answer would be no, although I would have to redefine the distinction in order to make it compatible with the analysis pursued here. In my present proposal, the key notion in this redefinition is evidentiality. Although to date, relatively little research has addressed the topic of evidentiality in quotative constructions, the original speech moment has always been a



Minds divided 

central concern in accounts of quotation. In order to interpret a reported message, the addressee needs to have some indication about the relation of the current speaker to the discourse context she is reporting about; this is exactly the communicative aspect that has grammaticalized in the category of evidentiality. The behavior of this category in quotative constructions may give us a more meaningful view on the semantics of quotative constructions than classical direct/indirect speech oppositions or traditional interpretations of de dicto and de re, and examples like in Duna, (38) and (39), are a case in point. What has lead to the often reported intuition that some quotative constructions are a more ‘accurate’ reflection of a reported speech situation than others is, I claim, due to the evidential value of the quotative construction. Some evidential values may reflect that the current speaker was a discourse participant in the reported speech situation (prototypically the addressee) and fully represents herself as such at the speech moment t0. For lack of a better term, we may define the evidential value as the degree of mental contact the current speaker has with the reported speech situation. If this mental contact is direct, the current speaker places herself fully in the reported discourse situation and we may expect that the quotative construction reflects the reported message as accurately as possible. If mental contact is less direct, the addressee may infer that the current speaker was a less prototypical discourse participant in the reported speech situation (e.g. she overheard the message) and/or that the reported message is more likely to have been rephrased by the current speaker. However, no matter how accurately and objectively the current speaker aims to convey the reported message, she is the one speaking and may be assumed to have evaluated the message. This observation leads to inherent modality, but inherent modality may be overridden or specified with an encoded modal value through one of the strategies as surveyed in Section 4.2. We may assume, as does Frajzyngier (1995, 2001), that if a language allows several strategies, the unmarked strategy will generally signal agreement with the truth of the reported message: compare the implicational scale in (27). The main definitions of values and constructions that are reflected in quotative constructions are summarized in (45). (45) Evidential value: Indicates the degree of mental contact the current speaker has with the reported speech situation at speech moment t0. Modal value: Evaluation of the current speaker of the relationship between the reported speaker and some proposition p (= 19). Source construction: Represents the reported speaker in the reported speech situation. Reported message construction: A proposition the current speaker presents as a discourse element of some other speaker.

 Stef Spronck

Two questions remain: How can evidential and modal values be determined and how do the elements in (45) relate to one another? As mentioned in Section 1, the term construction as used here needs to be understood as in constructionist approaches to grammar, that is, as a conventionalized pairing of meaning and form (Croft 2001; Goldberg 1995, 2006; Tomasello 2003; see also Vandelanotte this volume). The semantic representation of a quotative construction necessarily requires the elements in (45):, see (46): (46) Quotative construction [[SOURCE construction] (modal value) [MESSAGE construction]]evidential value In words: a quotative construction combines a source construction, a message construction and a (semantic) modal value in a semantic-structural complex which expresses an evidential value. The semantic structure in (46) is non-hierarchical, but it represents the analysis that when all elements between the square brackets are present with the values as conventionalized in a particular language, they are part of a quotative construction, which means that they represent a particular evidential value. Quotative constructions may differ in the way in which either of these constructions and values is represented, that is, when there exists an opposition in a language between quotative constructions, we may expect that languages make distinctions either in the representation of the source, in the representation of the message, the modal value or the evidential value. In other words, the different structures in which modal and evidential values grammaticalize shape the different quotative constructions in a particular language, although there may be some discourse functional motivations for representing the source or reported message constructions differently. For example, Frajzyngier (2001: 375) observes that in Lele, “[t]he choice between direct speech and indirect speech is an issue not merely of style but rather of resolving the potential ambiguity between the participants of the reported speech and the participants of the ongoing speech.” Discourse functional motivations may require particular quotative constructions, which means that a different representation of the source or reported message constructions does not necessarily have to indicate different modal or evidential values, although this might generally be expected. Some of the structures to which this semantic representation may be paired were considered in Section 4. When applying this analysis in a particular language, the first step is to determine the number of quotative constructions (i.e. the number of semantic and structural oppositions between constructions encoding quotation), and secondly, the way in which the modal and evidential values and the representation of the source and message in the construction differ. Let us assume that (47a–e) (= 16a–e) are the quotative constructions in English. The primary semantic opposition



Minds divided 

between (47a), (47b) and (47d) lies in their evidential value, from more to less direct, where (47a) purportedly expresses the message ‘as said’, (47b) may still specify the way in which the message was conveyed ‘originally’ but is paraphrased, and (47d) may not even specify the way in which the current speaker learned about the message (e.g. the current speaker may just infer that the reported speaker was thinking the reported message). (47)

a. b. c. d. e.

He said: “I am here” He said (that) he was there He maintained (that) he was there He went like, “I am here” According to him he was there

Examples (47a,b,c) seem to mainly contrast with respect to their modal value. Where (47a) excludes the possibility of co-encoding a current speaker attitude (e.g. prosodic stress on the framing verb does not result in a modal meaning), in (47b), there appears to be room for a modal interpretation, as evidenced by (47c), seemingly the same construction with a different speech verb, which implies noncommitment to the truth of the message. The source construction in example (47e) contrasts with those in (47a–d), which might be motivated by referential or other discourse factors. Although the modal and evidential values are independent in the representation, it may be assumed that a general tendency in languages is that if current speakers express a stronger evaluation of a message (i.e. the construction has a marked modal value), the message is more likely to be less represented as if the current speaker is directly relaying the message from the perspective of the reported speech situation (i.e. the construction less likely expresses an evidential value that suggests direct mental contact with the reported speech situation). However, this is by no means always the case, nor is it a necessary conceptual correlation. For example, a construction encoding a meaning such as “Can you believe s/he told this blatantly obvious lie?” will most probably present the reported message as a directly relayed utterance. 6. Some closing remarks: Semantics and pragmatics In this chapter I have aimed to show that the analysis of quotation necessarily includes meanings of source information and evaluation, the grammaticalized counterparts of which are the categories of evidentiality and modality. A cross-linguistic account of quotative constructions is well served by examining the behavior of these categories in quotation in order to discover and reinterpret periphrastic

 Stef Spronck

strategies in other languages that otherwise may have gone unnoticed, and also by eventually predicting grammaticalization paths. The present account has focused on constructions, that is, grammatical realizations of meaning. It remains to be demonstrated how modal and evidential values in quotative constructions are employed in quotative utterances, that is, how quotative constructions correlate with the usage of quotation in discourse (cf. Larson 1978; Vincent and Perrin 1999). In a fine-grained account of Aguaruna (Peru; Jivaroan) texts, Larson (1978) distinguishes seven usage categories with up to seven different subtypes; Vincent and Perrin (1999) identify the following four functions in their French corpus: Narrative (advancing the story), Appreciative (evaluations), Authority (arguments of authority) and Support (arguments of exemplification). The last three, they maintain, are either motivated on an emotional or rhetorical basis (Vincent and Perrin 1999: 293). I would argue, however, that a discourse account of quotation crucially relies on the semantics of the constructions employed, and that relevant correlations between discourse function and types of construction can only be properly determined if the meaning of the latter is firmly established. Also, by studying grammatical realizations of pragmatic functions in quotation, we may shed light on the cross-linguistic pragmatics of quotative constructions under the assumption that pragmatics precedes semantics in grammaticalization. This is not to deny that the occurrence of certain quotative constructions is highly reliant on such notions as discourse context or hearer expectancy, that is, the speaker’s ability to anticipate what the addressee will understand as a ‘normal’ continuation of the discourse. Given a particular context, certain constructions may gain, for example, a certain modal meaning, while others are perceived as neutral contributions. This is especially true for current speaker-evaluative meanings. If a speaker is asking for a fact and her discourse participant provides her with a third person quotative construction, she may conclude that the speaker is uncertain about the information. Similarly, if a speaker uses a quotative construction with a less direct evidential value than might be expected, pragmatic effects might be triggered. However, just as we cannot recognize these contexts without studying the psycho-social construction of discourse, we may also not recognize the constructional semantics that are enabling these discourse meanings if we do not have a proper understanding of the semantics of quotative constructions. This was the topic I aimed to address in the present chapter. Examining speaker attitudes in quotative constructions remains a largely unexplored area in descriptive and theoretical linguistics, despite a number of pioneering studies that the present account is heavily indebted to. What I hope to have demonstrated, however, is that further research into this topic will not only contribute to a better understanding of such categories as modality and evidentiality,



Minds divided 

but also of such core concepts of linguistics as clause-hood (in examining the scope and syntactic status of current speaker framed evaluative meanings) and referentiality (by studying who is the evaluating discourse participant and how that meaning is established). I hope to have presented a viable framework for analyzing the respective parameters that need to be taken into account for studying this fascinating topic. References Aikhenvald, Alexandra Y. 2003. Tariana. Oxford: Oxford University Press. Aikhenvald, Alexandra Y. 2004. Evidentiality. Oxford: Oxford University Press. Aikhenvald, Alexandra Y. 2008. Manumbu. Oxford: Oxford University Press. Ameka, Felix K. 1991. Ewe: Its Grammatical Constructions and Illocutionary Devices. Unpublished PhD thesis, The Australian National University. Besnier, Niko. 1993. “Reported speech and affect on Nukulaelae Atoll.” In Responsibility and Evidence in Oral Discourse, Jane Hill and Judith. T. Irvine (eds), 161–181. Cambridge: Cambridge University Press. Besnier, Niko. 2000. Tuvaluan: A Polynesian Language of the Central Pacific. London: Routledge. Bickel, Balthasar and Nichols, Joanne. 2007. “Inflectional morphology.” In Language Typology and Syntactic Description, Vol. 3, 2nd ed., Timothy Shopen. (ed), 169–240. Cambridge: Cambridge University Press. Boye, Kasper. 2010. “Semantic maps and the identification of cross-linguistic generic categories: Evidentiality and its relation to epistemic modality.” In Linguistic Discovery 8–1 (online: http://journals.dartmouth.edu/cgi-bin/WebObjects/Journals.woa/2/xmlpage/1/issue/34, accessed 9 April 2011). Boye, Kasper and Harder, Peter. 2009. “Evidentiality: Linguistic categories and grammaticalization.” Functions of Language 16(1): 9–43. Bybee, Joan, Perkins, Revere and Pagliuca, William. 1994. The Evolution of Grammar: Tense, Aspect and Modality in the Languages of the World. Chicago: University of Chicago Press. Bybee, Joan and Fleischman, Suzanne (eds). 1995. Modality in Grammar and Discourse. Amsterdam/Philadelphia: John Benjamins. Chafe, Wallace and Nichols, Johanna (eds). 1986. Evidentiality: The Linguistic Coding of Epistemology, Advances in Discourse Processes, Vol. XX. Norwood: Ablex. Clark, Herbert and Gerrig, R.J. 1990. “Quotations as demonstrations.” Language 66: 764–805. Clift, Rebecca. 2006. “Indexing stance: Reported speech as an interactional evidential.” Journal of Sociolinguistics 10(5): 569–595. Coate, Howard and Oates, Linette. 1970. A Grammar of Ngarinjin, Western Australia. Canberra: Australian Institute of Aboriginal Studies. Collins, Daniel E. 2001. Reanimated Voices: Speech Reporting in a Historical-Pragmatic Perspective. Pragmatics and Beyond. New Series, 85. Amsterdam/Philadelphia: John Benjamins. Comrie, Bernard. 1985. Tense. Cambridge: Cambridge University Press. Coulmas, Florian. 1986. “Reported speech: Some general issues.” In Direct and Indirect Speech, F. Coulmas (ed.), 1–28. Berlin/New York/Amsterdam: Mouton de Gruyter.

 Stef Spronck Curnow, Timothy Jowan. 2002. “Types of interaction between evidentials and first person subjects.” Anthropological Linguistics 44: 178–196. Croft, William. 2001. Radical Construction Grammar. Syntactic Theory in Typological Perspective. Oxford: Oxford University Press. De Brabanter, Philippe and Dendale, Patrick (eds). 2008. “Commitment: The term and the notions.”, Special issue of Belgian Journal of Linguistics 22 (1): 1–14. Delancy, Scott. 1997. “Mirativity: The grammatical marking of unexpected information.” Linguistic Typology 1(1): 33–52. Dench, Alan. 1995. Martuthunira: A Language of the Pilbara Region of Western Australia, Canberra: Pacific Linguistics. Dendale, Partick and Tasmowski, Liliane. 2001. “Introduction: Evidentiality and related notions.” Journal of Pragmatics 33: 339–348. Dickinson, Connie. 2000. “Mirativity in Tsafiki.” Studies in Language 24: 379–421. Dik, Simon C. 1997. The Theory of Functional Grammar. Part 2: Complex and Derived Constructions. Kees Hengeveld (ed.), Berlin/New York: Monton de Gruyter. Dimmendaal, Gerrit J. 2001. “Logophoric Marking and Represented Speech in African Languages as Evidential Hedging Strategies.” Australian Journal of Linguistics 21(1): 131–157. Dixon, Robert. 2000. Jarawara. Oxford: Oxford University Press. Donaldson, Tamsin.1980. Ngiyambaa, the Language of the Wangaaybuwan. Cambridge: Cambridge University Press. Enfield, Nick. 2003. Lao. Berlin: Mouton de Gruyter. Evans, Nicholas. 1995. A Grammar of Kayardild. Berlin: Mouton de Gruyter. Evans, Nicholas. 2003. Bininj Gun-wok. 2 vol. Canberra: Pacific Linguistics. Evans, Nicholas. 2006. “A view with a view: Towards a typology of multiple perspective constructions” Proceedings of the Thirty-First Annual Meeting of the Berkeley Linguistics Society: 93–120. Evans, Nicholas. 2007. “Insubordination and its uses.” In Finiteness: Theoretical and Empirical Foundations, Irina Nikolaeva (ed.,), 366–431. Oxford: Oxford University Press. Evans, Nicholas. forthcoming. “Some problems in the typology of quotation: A canonical approach.” Evans, Nicholas D. and Wilkins, David P. 2000. “In the mind’s ear: The semantic extensions of perception verbs in Australian languages.” Language 76(1): 546–592. Everett, Daniel and Kern, Barbara. 1997. Wari’. London: Routledge. Feuillet, Jack. 1996. ‘Réflexions sur les valeurs du médiatif ’, In L’ énonciation médiatisée, Zlatka Guentchéva (ed.), 71–86. Louvain/Paris: Éditions Peeters. Fortescue, Michael. 1984. West-Greenlandic. London: Routledge. Frajzyngier, Zygmunt. 1991. “The de dicto domain in language.” In Approaches to Grammaticalization, Volume 1: Focus on Theoretical and Methodological Issues, Elisabeth Closs Traugott and Bernd Heine (eds), 219–251. Amsterdam/Philadelphia: John Benjamins. Frajzyngier, Zygmunt. 1995. “A functional theory of complementizers.” In Modality in Grammar and Discourse, Joan Bybee and Suzanne Fleischman (eds), 453–502. Amsterdam/Philadelphia: John Benjamins. Frajzyngier, Zygmunt. 2001. A Grammar of Lele. Stanford: CSLI. Goffman, Erving. 1979. “Footing.” Semiotica 25(1–2): 1–30. Goldberg, Adele E. 1995. Constructions. A Construction Grammar Approach to Argument Structure. Chicago/London: The University of Chicago Press. Goldberg, Adele E. 2006. Constructions at Work. Cambridge: Cambridge University Press.



Minds divided  Grice, Paul. 1989. Studies in the Way of Words. Cambridge/London: Harvard University Press. Guentchéva, Zlatka (ed.). 1996. L’ énonciation médiatisée, Louvain/Paris: Éditions Peeters. Güldemann, Tom. 2008. Quotative Indexes in African Languages: a Synchronic and Diachronic Survey, Berlin: Mouton de Gruyter. Hale, A. 1980. “Person markers: Finite conjunct and disjunct forms in Newari.” In Papers in Southeast Asian Linguistics 7, R. Trail (ed.), 95–106. Canberra: Pacific Linguistics. Haspelmath, Martin. 1993. Lezgian. Berlin: Mouton de Gruyter Haßler, Gerda. 2002. “Evidentiality and reported speech in Romance languages”. In Reported Discourse, A Meeting Ground for Different Linguistic Domains, Tom Güldeman and Manfred von Roncador (eds), 143–172. Amsterdam/Philadelphia: John Benjamins. Hill, Jane H. 2005. A Grammar of Cupeño, Berkely/Los Angeles: University of California Press. Hill, Jane H. and Judith T. Irvine 1993. (eds) Responsibility and Evidence in Oral Discourse. Cambridge: Cambridge University Press. Hopper, Paul J. and Traugott, Elisabeth Closs. 1993. Grammaticalization. Cambridge: Cambridge University Press. Huttar, George L. and Huttar, Mary L. 1994. Ndyuka: A Descriptive Grammar. London: Routledge. Jakobson, Roman. 1957. Shifters, Verbal Categories and the Russian Verb, Russian Language Project, Department of Slavic Languages and Literatures. Cambridge, MA: Harvard University. Jung, Walter. 1971. Grammatik der Deutschen Sprache, 4. ed. Leipzig: VEB Bibliographisches Institut. Kimball, Geoffrey. 1991. Koasati Grammar. Lincoln: University of Nebraska Press. Kockelman, Paul. 2004. “Stance and subjectivity.” Journal of Linguistic Anthropology 14(2): 127–150. Kornfilt, Jaklin. 1997. Turkish. London: Routledge. Kruspe, Nichole. 2004. A Grammar of Semelai. Cambridge: Cambridge University Press. Larson, Mildred L. 1978. The Functions of Reported Speech in Discourse. Dallas: SIL. Lee, Duck-Young. 2007. “Involvement and the Japanese interactive particles ne and yo.” Journal of Pragmatics 39(2): 363–388. Levinson, Stephen. 1988. “Putting linguistics on a proper footing: Explorations in Goffman’s participation framework.” In Goffman: Exploring the Interaction Order, P. Drew and A. Wootton (eds), 161–227. Oxford: Polity Press. Lewis, Geoffrey. 2000.Turkish Grammar. Second edition. Oxford: Oxford University Press. Lichtenberk, Frantisek. 1995. “Apprehensional epistemics.” In Modality in Grammar and Discourse, Joan Bybee and Suzanne Fleischman (eds), 293–327. Amsterdam/Philadelphia: John Benjamins. Lichtenberk, Frantisek. 2008. Toqabaqita. 2 vol. Berlin: Mouton de Gruyter. Martin, Samuel E. 1975. A Reference Grammar of Japanese. Honolulu: University of Hawai’i Press. Maslova, Elena. 2003. A Grammar of Kolyma Yukaghir. Berlin: Mouton de Gruyter. McGregor, William. 1994. “The grammar of reported speech and thought in Gooniyandi.” Australian Journal of Linguistics 14(1): 63–92. Mushin, Ilana. 2001. Evidentiality and Epistemological Stance: Narrative Retelling. Pragmatics and Beyond New Series, 87. Amsterdam/Philadelphia: John Benjamins. Nedjalkov, Igor. 1997. Evenki. London: Routledge. Palmer, Bill. 1986. Mood and Modality. Cambridge: Cambridge University Press. Quine, W.V. 1956. “Quantifiers and propositional attitudes.” Journal of Philosophy 53: 177–187. Rennison, John R. 1997. Koromfé. London: Routledge Rumsey, Alan. 1982. An Intra-Sentence Grammar of Ungarinjin. Canberra: Pacific Linguistics.

 Stef Spronck Rumsey, Alan. 1990. “Wording, meaning, and linguistic ideology”. American Anthropologist 92: 346–361. Russ, Charles V.J. 1994. The German Language Today: A Linguistic Introduction. London: Routledge. San Roque, Lila. 2008. Duna. Unpublished PhD thesis, Australian National University. Scatton, Ernest A. 1984. A Reference Grammar of Modern Bulgarian, Columbus: Slavica Publishers. Sicoli, Mark A. 2007. Tono: A Linguistic Ethnography of Tone and Voice in a Zapotec Region. Unpublished PhD thesis, University of Michigan. Slobin, Dan I. and Ayhan A. Akşu. 1982. “Tense, aspect and modality in the use of the Turkish evidential”. In Tense-aspect: Between Semantics & Pragmatics. Paul J. Hopper (ed.), 185– 200. Amsterdam/Philadelphia: John Benjamins. Sohn, Ho-Min. 1994. Korean. London: Routledge. Tomasello, Michael. 2003. Constructing a Language. A Usage-Based Theory of Language Acquisition. Cambridge, MA: Harvard University Press. Vandelanotte, Lieven. 2005. “Tense in indirect speech or thought: Some proposed modifications.” In Tense, Aspect and Modality, Bart Hollebrandse, Angeliek van Hout and Co Vet (eds), 61–75. Amsterdam/Philadelphia: Rodopi. Van der Auwera, Johan and Plungian, Vladimir. 1994. “Modality’s semantic map.” Linguistic Typology 2(1): 79–124. Van der Voort, Hein. 2000. Kwaza. Berlin: Mouton de Gruyter. Van Valin, Robert and LaPolla, Randy. 1997. Syntax: Structure, Meaning and Function. Cambridge: Cambridge University Press. Verstraete, Jean-Christophe. 2005. “The semantics and pragmatics of composite mood marking: The non-Pama-Nyungan languages of northern Australia.” Linguistic Typology 9: 223–268. Vincent, D. and Perrin, L. 1999. “On the narrative vs. non-narrative functions of reported speech: A socio-pragmatic study.” Journal of Sociolinguistics 3(3): 291–313. Vološinov, V.N. 1973. Marxism and the Philosophy of Language. Matejka, Ladislav and I.R. Titunik (translators), Studies in Language 1. New York/London: Seminar Press. von Roncador, Manfred. 1988. Zwischen Direkter und Indirekter Rede, Nichwörtliche Direkte Rede, Erlebte Rede, Logophorische Konstruktionen und Verwandtes. Linguistische Arbeiten 192. Tübingen: Max Niemeyer. Wierzbicka, Anna. 1974. “The function of direct and indirect discourse.” Papers in Linguistics 7(3):267–307.

Thetic speaker-instantiating quotative indexes as a cross-linguistic type Tom Güldemann

Humboldt Universität zu Berlin and MPI-EVA Leipzig Quotative indexes often do not simply encode a speech event but turn out to be grammatical constructions with the dual function of referring to a state of affairs and of orienting the audience to the presence of reported discourse. A still poorly-analyzed quotative strategy without overt reference to a speech event is the use of thetic non-verbal identificational and presentational clauses that focus on the identity of the speaker as the source of the reported discourse. With data that are diverse in terms of geography, genealogical affiliation and diachronic dimension, the article will on the one hand demonstrate that this pattern is cross-linguistically recurrent, and thus robust as a type, and on the other hand aims to explore the special morphosyntactic and functional nature of this subtype of quotative index. Keywords: direct reported discourse, identificational clause, mimesis, presentational clause, quotative index, speaker (representation)

1. Quotative indexes from a typological perspective 1.1

Functional properties

The linguistic structures which accompany reported discourse in spoken language have often been viewed simply as sentential expressions of a state of affairs referring to speech or cognition. In this approach, an English phrase like he said (to me) would be a ‘canonical’ structure containing the following semanto-syntactic units as apparently essential components: a verbal predicate encoding a speech event, a nominal referring to the speaker, and, less required, a nominal referring to the addressee.1 This expectation is in stark contrast to the fact that quotes are quite often 1. It goes without saying that, as a reviewer remarked, the situation in written genres can be quite different from conversation. The following discussion concerns primarily spoken language.

 Tom Güldemann

not accompanied by all these elements. Accordingly, it seems necessary to approach these constructions under a broader perspective. In Güldemann (2008), I carried out an investigation of such structures in 39 African languages, based on their grammatical descriptions as well as the analysis of limited text corpora. There, I have proposed the term “quotative index” and defined it as follows: A quotative index is a segmentally discrete linguistic expression which is used by the reporter for the orientation of the audience to signal in his/her discourse the occurrence of an adjacent representation of reported discourse.

One major result of the formal analysis of more than 3200 text tokens of quotative indexes with direct quotes is that the expression type cannot be reduced conceptually to a sentence that focuses on propositional content by merely conveying a state of affairs; instead, it often turns out to represent a routinized construction with a potentially considerable degree of grammaticalization (see Güldemann 2008: Section 2.5 for more details). In functional terms it is clear then that there must be another function of quotative indexes beyond plain event representation. I propose to derive this function from the very nature of reported discourse itself, which I define as follows (note that this definition refers to the entire categorial scale between extreme direct and extreme indirect discourse and includes internal cognition and perception): Reported discourse is the representation of a spoken or mental text from which the reporter distances him-/herself by indicating that it is produced by a source of consciousness in a pragmatic and deictic setting that is different from that of the immediate discourse.

That is, the reporter also needs to orient her/his audience towards the presence of the quote, because this ‘alien’ text is associated with a different pragmatic and deictic interpretation. This reorientation is in fact often achieved by markedly pointing to the quote, either by means of demonstratives and other deictic elements that also have uses outside reported discourse, or by means of more grammaticalized quotatives, which are dedicated to signaling the presence of a quote. Depending on the discourse context, this grammatical function of quote orientation often predominates over the representation of the state of affairs, which in turn correlates with the grammatical(ized) nature of quotative indexes. 1.2

Morphosyntactic typology

Associated with the above general observation is the possibility to classify quotative indexes from a cross-linguistic perspective into basic types which differ in their morphosyntactic structure and the semantic focus they convey. I have derived



Thetic speaker-instantiating quotative indexes as a cross-linguistic type 

the following typology from both the African sample data as well as a survey of the relevant literature on other African and non-African languages. A first basic distinction should be made between quotative indexes which largely conform to the structure of normal verbal clauses, as the above he said (to me), and those which do not. For the second type, consider (1) from Kera (Chadic, Afroasiatic):2 this structure only consists of a noun phrase for speaker reference and a following grammaticalized quotative particle. Since it strongly deviates from a normal Kera clause based on a verbal predicate, it is called non-clausal. (1) k6mar kaalay ma {...} young men q Die Jungen riefen nun [lit.: the young men Q]: “...”

(Ebert 1975: 106)

Such non-clausal quotative indexes can be subdivided into a class which – in line with the above central function – involves quote orientation, as (1) from Kera with a dedicated quotative particle, and another class which can be called participantoriented in encoding exclusively the speaker and/or the addressee of the quote. These participant-oriented quotative indexes will be discussed in more detail from Section 1.3 onwards with one special type being the major topic of this article. Clausal quotative indexes can also be distinguished according to whether or not they highlight quote orientation. Structures like he said (to me), which are simple propositions that merely represent a state of affairs like other canonical clauses of a language, are viewed as highlighting the event rather than the quote and are called here monoclausal event-oriented; these have often but inadequately been considered in previous typological studies to be the default pattern of quotative indexes. The quote-oriented types of clausal quotative indexes come in three subtypes. The first one, called monoclausal quote-oriented, is superficially comparable with the previous monoclausal event-oriented one in that its nucleus is also a simple verbal clause. Its major difference is that the verb used does not refer to speech etc. outside the construction at issue. Although such an element is often translated in a simplified manner as ‘say’, its language internal properties qualify it as a dedicated quotative verb. This can be conceptualized as a grammaticalized quotative marker in verbal disguise, as cèe in (2) from Hausa (Chadic, Afroasiatic). (2) yaa cèe masà {...} 3m.s:perf qv dat:3s il lui a dit [he has “said” to him]: “...”

(Gouffé 1970/1: 80)

2. The glosses of examples have been unified for the purpose of this chapter and thus may differ from the original source. I give, however, always the original translation. Since this does not always reflect the actual morphosyntactic structure, a more literal translation in English may be provided in square brackets.

 Tom Güldemann

± Clausehood

–

± Quote-oriented

–

+ +

–

± Quote orienter

+ –

± Clausehood of quote orienter

+ –

Non-clausal Non-clausal Monoclausal Monoclausal Monoclausal quoteeventparticipantquotebipartite oriented oriented oriented oriented see below

see (1) of Kera

he said (to me)

see (2) of Hausa

see (3b)

+ Biclausal bipartite see (3a)

Figure 1.  Basic semanto-syntactic types of quotative indexes (Güldemann 2008: 516)

The two other clausal quote-oriented types display quote-orienting elements separate from and in addition to a verbal predicate so that they are syntactically bipartite. Depending on whether this quote-orienter is clause-like or not, one can further differentiate here between biclausal bipartite and monoclausal bipartite. This distinction can be illustrated by the two colloquial English phrases in (3), where the quote-orienter he says in (3a.) after then Peter tells him is a clause while like in (3b.) is not. (3) a. then Peter tells him, he says {...} b. then Peter tells him like {...} The basic semanto-syntactic typology of quotative indexes briefly represented above is summarized in Figure 1, including the associated examples given above (it goes without saying that more research might bring additional types of quotative index to light). 1.3

Non-clausal speaker representation

While from a cross-linguistic perspective the representation of the speech event is frequently omitted, there is one semantically expected element of quotative indexes that is far more central: namely the reference to the speaker. Thus, my typological investigation on the form and meaning of more than 3200 tokens of quotative indexes has established the following hierarchy of statistical occurrence (3225 = 100%) of the four relevant semanto-syntactic components (Güldemann 2008: 142–6):



Thetic speaker-instantiating quotative indexes as a cross-linguistic type 

Speaker (92%) > Quote orientation (71%) > Event (50%) > Addressee (31%) Note the very regular speaker representation against the low 50%-appearance of a verb expressing speech or cognition. Overall, speaker representation turns out to be the most central element of quotative indexes. One reflex of this general tendency is the recurrent existence of quotative indexes which consist of nothing but some kind of participant encoding. These would normally count as non-clausal participant-oriented in the above classification (see Figure 1). Depending on the language, they can be an occasional or a more regular phenomenon. While this strategy can also concern the addressee, the more frequent case is bare speaker representation, like a possible English phrase then Peter. Non-clausal participant-oriented quotative indexes have been reported, for example, in Turkish (Tietze 1959: 99–100), Modern Hebrew (Zuckermann 2006: 477–8), spoken Puerto Rican Spanish (Cameron 1998), and written English (Ware 1993: 163) and are thus far from unfamiliar or ‘exotic’. The following example is from the Australian language Warrwa, where the speaker nominal is marked by ergative case3 (W. McGregor p.c.). (4)

kinya-rnirl-ma {marlu jana-n-ngurndany ingan, this-p-erg {no where-loc-idef 3min.nom:be:prs ingan nyunu marduwarra} 3min.nom:be:prs there river} They said [lit.: those], “He might be somewhere, perhaps down on the river.”

Note that it is irrelevant for a cross-linguistically oriented evaluation of such participant-oriented quotative indexes whether they are rare and arguably cases of verb ellipsis or whether they represent a recurrent grammatical routine. In both cases they comply with the above definition of a quotative index and as a type require a comparison with other types of quotative index. In my cross-African study, such quotative indexes were encountered in analyzed texts of 13 out of 39 sample languages, normally without any recognition of this strategy in the grammatical description consulted. Consider the following examples from Kunama (Isolated, “Nilo-Saharan”) and Lamang (Chadic, Afroasiatic), respectively, where the structure was an occasional phenomenon. (5) báddi ína nonyna gàmba-sī {...} then dem frog:det lizard-obj Nun sprach der Frosch zur Eidechse [lit.: then the frog to lizard]: “...”  (Reinisch 1881–90, 1: 172) 3. A similar situation is found in Miriwoong and possibly Kija, where the speaker is marked by an ablative suffix.

 Tom Güldemann

(6) na sLm-à-tàŋ gùléŋ {...} then all-poss-3p again All of them said [lit.: then all of them again], “...”

(Wolff 1994: 333)

A more regular use of such structures is evident in languages of the Saharan family (“Nilo-Saharan”). I observed in Güldemann (2008: 116) that in a text corpus of Kanuri, 8% of about 200 quotative index tokens with direct reported discourse only displayed a reference to the speaker and/or the addressee before the quote. The morphological marking of the two nominals was as described by Hutchison (1981: 215–6) in his discussion of the so-called “agent postposition” yè (which is normally prominent in passive-like constructions): In making direct quotation, the agent/source normally occurs initially marked by the agent postposition. It may be followed by the listener marked by the indirect postposition +rò [cf. (8) where this addressee phrase occurs on its own]. The relevant form of the verb ngìn say, or whatever quotation verb is being used, may then occur either before or after the direct quote.

Here are two examples from my text corpus without any verb: (7) mai-ye {...} king-agt The king asked [lit.: the king] “...”

(Geider n.d.)

(8) bultu-ro {...} hyena-obl And she told the hyena [lit.: to the hyena] “...”

(Geider n.d.)

While I did not view such quotative indexes in Kanuri as establishing a genuine construction type, Lukas’ (1953: 177–8) description of reported discourse in closely related Tubu suggests that at least there it is a quite salient strategy: Sehr oft wird in der Tubusprache die direkte Rede unmittelbar, d.h. ohne ein Verb des Sagens, eingeführt. Diese Art der Einführung der direkten Rede hat etwas Lebendiges und zieht die Aufmerksamkeit auf das, was unmittelbar folgt. [Direct speech in Tubu is very often introduced immediately, that is, without a speech verb. This way of introducing direct speech has something lively to it and draws the attention to what immediately follows.]

(9) mflffúr {yf} hyena {Yes} die Hyäne sprach [lit.: the hyena]: ‘Ja’ (10) adéma yi anyíma du {túsfp!} woman agt man obl {leave me!} die Frau sprach zu dem Mann [lit.: the woman to the man]: ‘entlasse mich!’



Thetic speaker-instantiating quotative indexes as a cross-linguistic type 

(11) maná du {yír ginẹs!} gopher obl {come and divide!} er sprach zum Erdhörnchen [lit.: to the gopher]: ‘komm und teile!’ (12) ál jirkannu Dázza yi {...} this because.of pn agt deshalb sagen die Dazza [lit.: therefore, the Dazza]: “...” Lukas’ examples show that the marking of the participants in Tubu is the same as that in Kanuri, in that the addressee is marked by the oblique postposition du ~ ru (a cognate of Kanuri ro), as in (10) and (11), and that the speaker is mostly encoded with the so-called subject postposition (y)i (which, like the Kanuri cognate ye, is better analysed as marking non-topical agents), as in (10) and (12) (cf. also Lukas 1953: 164–5, 158–61). Non-clausal participant-oriented quotative indexes can become highly grammaticalized in the sense that they achieve the status of a fixed construction, which was the case in 2 of the 13 relevant African languages. One such language is Tikar (Benue-Congo, Niger-Congo), where the essential element of one type of quotative index is a special pronoun called “anaphoric particle” (Stanley 1982: 32), which cross-references 3rd-person speakers.4 Indeed, the quotative index in (13) lacks other possible elements entirely; it consists only of what I would call a quotative pronoun – another type of quotative marker, this time in pronominal disguise.5 (13) nũ {nũ tšiâ byεbi} s/he:q {s/he did badly} He (said) he has acted badly.

(Jackson 1987: 105)

In the remainder of this article, I will deal with a type of, mostly non-clausal, participant-oriented quotative index which is formally more circumscribed: the speaker is encoded by way of a construction which in the given language is canonically used for identifying or presenting nominal referents. It can be translated into English most closely as something like ‘it/this/that/(t)here is X’. In the sense that such a structure merely asserts the identity and/or presence of a referent, it is a non-verbal clause (Note that one can speak here of an assertive “clause”, even if

4. See Van de Velde (2008: 142–4) for a similar case in Eton (Bantu, Benue-Congo, NigerCongo). 5. The quotative pronoun is recruited from a functionally versatile paradigm series which also serves to express a logophoric subject in the quote. This is the reason why Example (13) displays a sequence of the same pronoun form.

 Tom Güldemann

the structure lacks a verb like English be).6 I will, accordingly, call the type henceforth ‘speaker-instantiating’ quotative index. 2. Speaker-instantiating quotative indexes: The data 2.1

Historical cases

Speaker-instantiating quotative indexes have so far been largely neglected in the literature. Ironically, in my own cross-linguistic study of quotative indexes (Güldemann 2008) and subsequent research, they first called my attention from a diachronic rather than synchronic perspective. Accordingly, I will start with these historical cases. Güldemann (forthcoming) tackles the history of the quotative marker j(n) in Egyptian (Afroasiatic), which is attested in some form throughout most of its history, undergoing drastic changes in the course of time. In its earliest stage, j(n) is best characterized as a defective quotative verb stem which prototypically precedes a full noun phrase referring to the speaker, as shown in the following example. c3-t (14) {m twt nf} jn psd-(tj) wr-t {who is like him?} q:pst pn-f.d be.great:pap-f be.great:pap-f “Who is like him?” said the Two Great and Powerful Enneads.  (Kammerzell and Peust 2002: 302)

J(n) must be characterized as a defective verb because it does not conjugate regularly; it is part of a paradigm of three suppletive stems which all serve as default bases of quotative indexes and are commonly distinguished according to different values for tense-aspect-modality. Table 1 shows these stems together with their basic semantic and syntactic characteristics; the last element, stative j, is commonly thought to be related to the form jn but occurs only later in history – hence the notation j(n) in the present article. Table 1.  The quotative verb paradigm in Earlier Egyptian Form

TAM value

Syntactic distribution

k3󰁪r(-) jn(-) j-

Future Present Preterite Stative

before “suffixal” pronoun before noun and “suffixal” pronoun before noun and plural “suffixal” pronoun before singular “suffixal” pronoun

6. Such a usage leads to a double sense of (non)-clausal, namely, referring either to the assertive force of an expression or to its morphosyntactic character of involving a verb. Nothing presented in this chapter hinges on this potentially complicating terminological detail. As such, I will continue to use non-clausal in the sense of non-verbal, as introduced in Section 1.2.



Thetic speaker-instantiating quotative indexes as a cross-linguistic type 

The frequency of the simple quotative pattern in (14) decreases in later stages of Egyptian, being replaced by what I have introduced above as a bipartite quotative index; this employs [j(n) SPEAKER] after the quote and a phrase with a verb, notably dd ‘say’, before the quote. Consider (15), in which the first part of the quote is preceded by nw rf ddnk Rc(w) ‘this is just what you, Re, have said’ and followed by jt tw Rc(w) – an instance of the j(n)-construction at issue. Rc(w) {hwj z3(j)} (15) wjj Rc(w) nw rf dd-n-k excl pn dem ana say:rel-pst-2m.s pn {be it that my son} O Re, this is just what you, Re, have said: “Be it that my son”, j-t tw Rc(w) {b3j s󰁸mj w3šj} q:stat-2m.s 2m.s pn {is besouled, is mighty, is strong} so you, Re, say, “is besouled, is mighty, and is strong!”  (Kammerzell and Peust 2002: 302) Moreover, in later stages j(n) is more often used preceding a pronominal element rather than a lexical noun, as in (15). Since in ancient Egyptian normal conjugated verbs are also followed by a pronoun, j(n) looks then more verb-like than in the other context. The major argument developed in Güldemann (forthcoming) is that the properties of j(n) as a verb lexeme are, however, only apparent and/or acquired secondarily. The reasoning behind this idea is that Egyptian possesses another element jn, which can be shown to originally be a non-verbal element that establishes an identificational/presentational clause of the type ‘it is X’. This also became grammaticalized within more complex syntactic structures, for example, in cleft sentences, as illustrated in (16). (16) jn-m j.jr n-k id-who act:pap for-2m.s Who is the one who acts for you?

(Kammerzell and Peust 2002: 303)

The conclusion from this is that the identificational function of jn can be assumed to also underlie the earliest quotative pattern [jn SPEAKER] – a case of the nonverbal speaker-instantiating quotative index at issue here. This account is largely compatible with the empirical facts observed for the diachronic development of jn as a quotative marker. The earliest stage [jn NOUN] would be a plain speakeroriented quotative index, quite parallel to those exemplified in Section 1.3; the only difference is that the speaker nominal is explicitly foregrounded by the element jn. As long as no other material (e.g. a true speech predicate) elaborates the nuclear jn-constituent, even tokens deviating from the original [jn NOUN] are still comparable to purely participant-oriented examples presented in Section 1.3.

 Tom Güldemann

Thus, (17) from Later Egyptian, with a pronominal speaker representation after jn, could be interpreted as ‘(it’s) me to her’ rather than ‘I said to her’ and, disregarding differences in constituent order with respect to the quote, would thus turn out to be essentially the same as (5) from Kunama and (10) from Tubu. (17) {...} jn-j n-s q-1m.s obl-3f.s {...} ich zu ihr [I to her]

(Jordan 2009: 13)

The interpretation of jn as a verb would have occurred over time by analogy to other truly clausal quotative indexes. In Egyptian, this ‘verbification’ was facilitated by the fact that a simple constituent [VERB.STEM-(n)-(PRO)NOUN] is part of the overall conjugation pattern, called ‘sdm-(n)-f form’ in Egyptian philology. Since the presence versus absence of n with canonical verbs conveys a tense-aspect difference, the original quotative pattern in jn could have been interpreted as an n-preterite of a verb j, triggering the later emergence of a stative quotative counterpart j without final n. The entire historical change can be sketched as follows: jn ‘it is ...’ > Quotative marker jn ... > Quotative verb j(n)Egyptologists hardly distinguish j(n) in its latest stage from a normal speech-verb lexeme, hence its common philological analysis as simply ‘say’.7 At this point, the most salient indication of a quite unusual history is its suppletive relation to other equally defective verbs in the quotative paradigm. Akkadian (Semitic, Afroasiatic), geographically and historically close to the earlier varieties of Egyptian, displays a quotative index that also seems to have started out as a speaker-instantiating structure. In its earliest stage, Early Old Babylonian, the relevant pattern was a syntactically self-contained structure [enma SPEAKER-(ma) (ana ADRESSEE)], used with direct reported discourse, as in (18): enma anāku-ma {...} (18) enma pn1 ana pn2 {ašmama ahtadu} q pn to pn {I heard it and was happy} q i-emph This is what pn1 says to pn2: I heard (his letter) and was happy (about it). (And now) this is what I say (i.e. answer): “...” [more narrow translation: (Deutscher 2000: 69) pn1 (says) to pn2 {...} I (say) {...}] The crucial question here concerns the source of the grammaticalized quotative marker enma before the speaker reference. Since there is no good language-internal structural evidence for relating it to a verb or noun referring semantically to 7. See, however, Chetveruchin (1988) for a diverting hypothesis that on an abstract level proposes a directionality which is similar to that proposed here.



Thetic speaker-instantiating quotative indexes as a cross-linguistic type 

speech (cf. Deutscher’s 2000: 69–70 extensive discussion in Footnote 25), the original pattern in (18) is not too different from non-clausal participant-oriented quotative indexes like (5) from Kunama, (10) from Tubu and (17) form Egyptian. Apart from the particle’s change in form to umma, the quotative index as a whole underwent considerable restructuring in later stages of the language, as outlined in detail by Deutscher (2000: 66–87). First of all, from Early Old Babylonian on but fully established in Later Old Babylonian, the nuclear constituent follows a speech verb, for example, ‘say’ in (19). This example qualifies as a monoclausal bipartite quotative index according to the above typology, in which, however, the non-verbal quote orienter umma šunuma is complex and co-varies somewhat tautologically with the speaker noun, here nappāhū ‘smiths’. (19) pīqat nappāhū iqabbû-kum umma šunu-ma {...} perhaps smiths they.say-to.you q they-emph perhaps the smiths might say to you “..”. (Deutscher 2000: 77) The latest stage is characterized by further grammaticalization of this innovative pattern; it encroaches on contexts of indirect reported discourse, umma itself developing to a dependent function word. One formal aspect of this process is that the originally obligatory speaker reference after umma is no longer necessary, as in (20) from Neo-Babylonian. (20) ašāl-šu umma {...} I.asked-him q I asked him “...”

(Deutscher 2000: 83)

Even more clearly than in Egyptian j(n) one can diagnose a historical trajectory from a non-clausal quotative index towards a monoclausal bipartite one by means of increasing co-occurrence with a preceding speech predicate. Regarding the etymological origin of Akkadian enma/umma, the two hypotheses that it is related either to the Hebrew noun n’um ‘speech’ (Baumgärtner 1974) or to some unidentified speech verb (Deutscher 2000: 69–70, Footnote 25) appear to be motivated crucially by the default assumption that semantically opaque quotatives should originate in elements that display utterance semantics – an idea which, as mentioned in Section 1, is actually not borne out by cross-linguistic data. However, as opposed to most of the previous analyses of Egyptian j(n), von Soden (1965–81, vol.1: 218) has actually entertained a historical origin which is of the type at issue here. In Güldemann (2008: 364–5, forthcoming), I argue in favor of this older hypothesis according to which [enma SPEAKER] would turn out to be a presentational structure ‘there/it is X’. That is, the original enma is proposed to be bimorphemic whereby en is a ‘presentative’ focus marker and ma is an emphatic enclitic which can also occur after the speaker nominal itself, as in (18) and

 Tom Güldemann

(19) above.8 Compared to the other two hypotheses, von Soden’s reconstruction is more viable for two reasons: First, it ties in with the typological data presented here; second, and more importantly, it has robust family-internal support in that a number of other Semitic languages have likely cognates of en which are also used as both presentational markers and quotative-complementizers. This in turn means that early Semitic languages as a whole are implicated in the phenomenon described here. A third historical case for a speaker-instantiating quotative index is Tonga-Inhambane (Benue-Congo, Niger-Congo) (see Güldemann 2008: 67–8, 365–7; forthcoming). Here the possible reconstruction arises first of all out of internal evidence (supplemented by some comparative data) in the sense that the different evolutionary stages outlined above for Egyptian j(n) are encountered in the synchronic structure of the language. Consider in this respect the following examples. (21) {nyinguhongola} kha Rasi {I am going} q pn “I am going” says Rasi

(Lanham 1955: 140)

(22) kh-iso {khumani ahipalago} q-8pro {who is it that defeats us?} They [animals] said, “Who is it that defeats us ...”

(Lanham 1955: 140)

(23) si-rengo si-ngu-kh-iso {...} 8-animal 8-prs-qv-8pro the animals say, “...”

(Lanham 1955: 139)

Example (21) is a case of a non-clausal quotative index with a particle kha preceding the speaker noun; khiso in (22) arguably involves the same quotative element kh- (< *kha) which has fused with a pronominal speaker representation (here of agreement class 8) and thereby changed its a-vowel through assimilation and analogy; (23) displays the same form khiso but now with canonical verb prefixes, making it a prototypical instance of a morphologically irregular quotative verb (hence glossed as QV) because normal verbs have entirely different suffix morphology. I argue that these different types of quotative indexes should be related to each other in historical terms as follows: Quotative marker kha >Speaker-inflected quotative kh- >Quotative verb khThe first two non-verbal patterns in this development can be plausibly derived historically from an identificational structure [kha (PRO)NOUN] ‘it is X’. Apart from supporting comparative data in other Bantu languages (see Section 3), this can still be discerned in Tonga-Inhambane itself from the fact that the person-inflected 8. The change from enma to later umma remains unexplained by all hypotheses.



Thetic speaker-instantiating quotative indexes as a cross-linguistic type 

quotative paradigm is the same as that for identificational pronominals; for example, khiso of (22) alone means ‘it is them’ referring to a class-8 noun.9 There are several other similarities of Tonga-Inhambane (-)khV- with Egyptian j(n) and Akkadian enma/umma, supporting the hypothesis about a similar history. Firstly, the speaker-inflected quotative kh- illustrated in (22) has developed into a general quotative~complementizer which is also used as a quote orienter in bipartite quotative indexes, as in (24). This can be recruited for both direct and indirect reported discourse. (24) nyamayi adi-wujisa kh-uye {...} woman.1 1:pst-ask q-1pro The woman asked saying, “...”

(Lanham 1955: 140)

Secondly, Tonga-Inhambane *khV has developed a polyfunctionality pattern that is similar to Egyptian jn in yet another respect: it not only involves functions in non-verbal clauses and reported discourse, but also serves as a marker of syntactically peripheral participants, among them the agent in passives (cf. Lanham 1955: 141 Footnote 1, 212–3). In Güldemann (2008: 368–9), I have proposed another potential case of a reconstructable speaker-instantiating pattern, namely the Hausa (Chadic, Afroasiatic) quotative index based on cee. Synchronically, cee is a typical case of a quotative verb, in the sense that it is dedicated to reported discourse and displays various conjugational irregularities. The normal pattern of a quotative index with cee has already been illustrated in (2) above. As in the previous cases, the verb (in its nominalized form ceèwaa) has developed into a complementizer in a bipartite structure that is virtually restricted to indirect reported discourse. (25) sai ya tàmbàyee ni ceèwaa {...} then 3m.s:pfv ask 1s comp dann fragte er mich [then he asked me]: “...”

(Wolff 1993: 516)

The main reason for the idea that the verb might have a similar origin to the previous quotatives is the observation that cee and its dialectal variants tane, cane, cene show a clear formal relation to some gender-sensitive identificational and presentational markers of Hausa. For example, cee is segmentally identical to the feminine form of the paradigm of postposed identificational markers: nee (M.S), cee (F.S), nee (P) (cf. Wolff 1993: 494–5; Newman 2000: 161, 545–7); thus consider (26). 9. The situation, especially with nouns, is more complex, though, because the reconstructed vowel *a changed in the pronominal paradigm to u or i, which arguably was transferred to contexts with lexical nouns in the more frequent identificational but not in the quotative structure; hence quotative [kha NOUN] versus identificational [khu NOUN].

 Tom Güldemann

(26) a. mootàa cee car.f.s f.s.id es ist ein Auto [it is a car] b. wannàn mootàa cee this.f.s car.f.s f.s.id dies ist ein Auto [this is a car]

(Wolff 1993: 494)

The present hypothesis implies several changes in the morphosyntax of the element, such as gender neutralization towards the feminine form and increasing ‘verbification’ to a quotative verb, as it no longer immediately follows the speaker nominal but, instead, an inflectional complex typical for verbal clauses. Whether this scenario is feasible needs to be clarified by more detailed historical research, though. Summarizing the above information, there is robust evidence from several languages for the hypothesis that identificational and/or presentational structures with scope over a nominal referring to the source of a reported text have been recruited as a regular quotative index for direct reported discourse. Since these cases have heretofore evaded recognition, certain properties of the nucleus of a quotative index should be considered as important heuristics for the detection of such an unusual history, namely, its non-verbal behavior – or at least defective verbal character – and/or its marked morphosyntax, notably possible pronominal co-variation with the speaker referent other than canonical subject-verb agreement. In those cases where the acquisition of verbal features by the earlier nonverbal element can be diagnosed – namely, in Egyptian, Tonga-Inhambane, and probably Hausa – one can speak of lexicalization, or more specifically, what I have called above ‘verbification’; that is, the element in question, assumed to originate in a grammatical particle, acquires morphosyntactic and semantic properties which are typical of an open part-of-speech class, namely, verb lexemes. This ‘becoming-(like)-a-verb’ is associated with the gradual category shift of the construction as a whole, namely from a non-clausal to a monoclausal quotative index. Another common denominator of these cases emerges from their history of grammaticalization in the sense that the structures expand formally and functionally. That is, they come to be employed in a bipartite quotative index in conjunction with verbal predicates and they are extended to contexts of indirect reported discourse. This development is sketched in Table 2 by four constructional stages (only Akkadian and Hausa have reached the last stage). Lexicalization and grammaticalization, although in principle different and independent, can have in these cases one important point in common: an earlier non-clausal quotative index of the speaker-instantiating type becomes disguised



Thetic speaker-instantiating quotative indexes as a cross-linguistic type 

Table 2.  Simplified historical development from non-clausal to monoclausal bipartite quotative indexes across the historical cases 1. 2. 3. 4.

[ [[SPEAKER + SPEECH.CLAUSE] [[SPEAKER + SPEECH.CLAUSE] [[SPEAKER + SPEECH.CLAUSE]

(QUOTE) (QUOTE)

[PRES/ID SPEAKER] [PRES/ID SPEAKER] [Q~COMP SPEAKER] Q~COMP

(QUOTE)] (QUOTE)] QUOTE] QUOTE]

as a canonical verbal form, either as the nucleus of a monoclausal structure or as the verb-like quote orienter of a biclausal bipartite structure. 2.2

Modern cases

The historical scenario reconstructed above receives support from the fact that the assumed original structure of a non-verbal clause of identification or presentation is also attested as a quotative index in modern languages. The cases that have come to my attention will be briefly presented below. One such case, the colloquial English construction [this is SPEAKER] observed in the language of adolescents in London (Cheshire and Fox 2007), is treated by Fox in this volume, to which the reader is referred for more details. Here it suffices to highlight the points important for the present discussion. As far as the use of this type of quotative index is concerned, it is reported to be restricted to direct reported discourse. Moreover, it is strongly preferred by (predominantly female) speakers in conversational self-reporting narratives in the historic present at particularly dramatic and performative stretches of the discourse. With respect to formal characteristics, consider first the following examples (see Fox this volume, examples (9) and (10), also in Cheshire and Fox 2007). (27) This is my mum {what are you doing? I was in the queue before you} (28) This is them {what area are you from. what part?} This is me {I’m from (inner London)} The speaker nominal after the identificational copula predicate can be recruited from a wide range of types but pronouns seem to occur more frequently. While this nominal is called by the authors subject, it is clear from word order and case marking (cf. (28) with two non-subject pronouns) that this term is not appropriate as a syntactic characterization (see Section 3 below for a more detailed discussion). In view of my previously used term non-clausal in the sense of verbal and the cases discussed in Section 2.1, this English pattern is different, because it involves – besides the demonstrative identifier this – the equational verb be (I will also postpone the discussion of this point to Section 3). Another interesting fact from

 Tom Güldemann

a more general perspective is the apparent existence of another just slightly different English quotative structure already documented by Milroy and Milroy (1977: 54–6, so cited by Cheshire and Fox 2007), namely [here is/was SPEAKER], based on the presentational deictic adverb here. This implies that one and the same language can potentially recruit clauses of both the identificational and the presentational type. According to Dejan Matic (p.c.), to whom I owe all the following information, in colloquial varieties of Bosnian-Croatian-Serbian, a presentational clause can also be employed as a consistently preposed construction with direct reported discourse, as illustrated in (29). (29) i evo (ti) njega {...} and here (2s:dat) 3m.s:gen And there he says [lit.: and here (for you) of him], “...” Example (30) demonstrates that this quotative index is identical with a canonical presentational construction. (30) evo (ti) Petr-a   (or eto (ti) Petra with the distal deictic ‘there’) here (2s:dat) pn-gen (T)here is Peter (for you)! As in previous cases, the speaker referent is not encoded in the typical form for subjects; here it is in the genitive case. Another specialty of this pattern is the optional occurrence of the clitic form ti of the so-called ethical dative, which in general expresses a special interest felt by the person indicated in this form, which in (29) is the audience to which the reporter addresses the reported discourse construction (the non-clitic alternative tebi or even a form other than 2nd-person singular is not possible). Finally, quotative indexes, as in (29), are also repeatedly preceded by a particle kad, which normally means ‘when’ but apparently does not perform this function in the context at issue. The discourse features of this quotative pattern are also comparable to those of the English case in that its typical genre is spoken colloquial narratives. It also occurs especially in contexts where the reporter wants to vividly perform a dynamic verbal interaction between narrative characters, in particular, involving quote content that is unexpected and remarkable before the background of the previously available information. I assume that more in-depth studies in other languages will reveal cases in addition to English and Bosnian-Croatian-Serbian. Just to give a further example, the following structure is, although arguably an occasional ad-hoc formation, a possible quotative index in colloquial Portuguese; it is based on the deictic adverb aí ‘there’ followed by the speaker nominal:



Thetic speaker-instantiating quotative indexes as a cross-linguistic type 

(31) aí os gajos {...} there the guys the guys were/said like [lit.: there the guys], “... While the previous examples are all from European languages, it can be expected that similar structures are also employed outside Europe. A case in point is the Polynesian language Tongan. According to Broschart (1994: 64), the following reported discourse construction in which the quote precedes the quotative index is attested in the spoken language. (32) {alu atu!} ko Sione {go away!} q pn “Go away!” said John. [lit.: it is John]

(Jürgen Broschart p.c.)

The central element ko of this quotative index is a highly versatile element which has received quite some attention in grammatical descriptions of the language, although its quotative use is not described in the sources available to me except for Broschart’s short mention. While ko is commonly treated as a polyfunctional preposition, it emerges from studies such as Churchward (1953: 100–5), Broschart (1994: 49–50, 57–61, 66–70, 74–94), and Custis (2004: 18–47, 124–63) that its overall functional profile is in fact quite comparable to that of Egyptian jn and Tonga-Inhambane kha~khV-. One use of ko can be characterized as the syntactic nucleus in non-verbal clauses. This function, which I assume to be the original one, can motivate its occurrence in complex bipartite cleft-like structures (among them those with focus function), as a marker of peripheral participant roles (e.g. essive ‘as/being X’) and also, as opposed to the elements of Egyptian and Tonga-Inhambane, as an essential element in equational clauses. In (33), ko is illustrated in the context most relevant for the present discussion, namely as the nucleus of a clause which according to the translation can be both identificational and presentational. (33) ko e tamasi’i id/pres det boy es/da ist ein Junge [it/there is a boy]

(Broschart 1994: 58)

Summarizing the findings regarding the modern cases of speaker-instantiating quotative indexes, the following points can be made. Firstly, they are all restricted to direct reported discourse and pertain to the domain of colloquial and spontaneous oral language use, which presumably is the reason why usually they are not addressed in grammatical descriptions. From those cases which have received more detailed linguistic reflection, it can also be concluded that they are pragmatically marked, having a particularly vivid and highlighting flavor, and are partly in competition with other similarly marked strategies.

 Tom Güldemann

3. Speaker-instantiating quotative indexes as a cross-linguistic type In the following section, I try to relate the historical cases discussed in Section 2.1 with the modern cases in Section 2.2. In spite of the very different status of these two groups in the grammar of the relevant languages, I would venture that a quite robust common pattern emerges. After outlining this common ground I conclude that one can speak of a unitary cross-linguistic type of what I call speaker-instantiating quotative index. One first shared point concerns in fact their linguistic treatment in the past in the sense that, until recently, they have been quite elusive. For the historical cases, the reconstructed scenario proposed here had not been entertained before (TongaInhambane, Hausa) or had not found its way into the typological discussion on quotative indexes (Egyptian, Akkadian). Regarding the modern cases, the research dedicated to them is very limited or even non-existing; moreover, the first closer attention to them came out of sociolinguistic and discourse-oriented rather than morphosyntactic research on the relevant languages. The net result is that grammatical descriptions would mostly not reveal the very existence of these structures. Although the elusiveness of the historical and modern cases is different in nature, the reason for it is the same: previous research, particularly with a typological orientation, had a strong tendency to start from the simplistic assumption that a monoclausal event-oriented pattern of the type [SPEAKER SPEECH.VERB] epitomizes a quotative index, thereby overlooking structures which deviate from it. That this is a misleading approach to the formal design of reported discourse is, however, increasingly recognized – the collection of articles in the present volume is a good example of this scientific reorientation (cf. also several contributions in Güldemann and von Roncador 2002). The structures presented above also seem to share considerable formal similarities, even though the empirical data for both the modern and historical cases are still partly unclear or simply insufficient. Most importantly, all the structures presented above share that they are non-verbal clauses (or copula clauses without a canonical subject topic) with a foregrounding function with scope over a nominal that refers to the source, aka speaker, of an associated direct quote; hence the term speaker-instantiating introduced here. Recall from Section 1.2 that there exist other non-clausal quotative indexes with a similar foregrounding function, namely those which are semanto-syntactically more oriented to the quote itself. Consider, for example, such foregrounding and often non-clausal structures in European languages, as with English like, German so ‘like this, so’ (Golato 2000, this volume), Swedish ba ‘just’ (Eriksson 1995), and the like, which partly have been intensively studied by now and are also treated again in this volume.



Thetic speaker-instantiating quotative indexes as a cross-linguistic type 

It is, of course, necessary to determine whether a foregrounding device in a quotative index is oriented to the speaker or to the quote itself. Depending on the overall structure of an individual language, this is not always a trivial issue. A good diagnostic is at least the presence versus absence of co-variation with the speaker category. Such a distinction, if subtle, can be shown to be relevant in some languages. Shona (Benue-Congo, Niger-Congo), for example, uses two relevant markers, presentational ha- and identificational ndi-, in a context that is intimately related to direct reported discourse, namely in the introduction of ideophones (see Güldemann (2008) and further below regarding ‘mimesis’). Consider the following two examples: (34)

imbwá héyo {pikú nyáma mu-mbá 9.dog pres:9.dem {ideo:snatch meat ine-house washu toro} ideo:run.off ideo:disappear There is the dog taking the meat from the house, running off and disappearing (Fortune 1971: 250)

(35) imbwá ndi-ye {nyama pikú mu-mbá so-muridzi} 9.dog id-pro {meat ideo:snatch ine-house ess-owner As for the dog, it is a taking of the meat from the house as its owner; viz. The dog took the meat from inside the house as if it were its owner  (Fortune 1971: 250) While in both sentences the initial agent noun imbwá ‘dog’ is linked to the eventencoding ideophone phrase based on pikú ‘snap’ by means of a non-verbal element, only the presentational ha-form in (34) agrees in person, gender, and number with its preceding noun. The pronominal ye suffixed to identificational ndi- in (35) is invariable, which strongly suggests that this construction rather foregrounds the following ideophone phrase (or alternatively its initial noun nyama?). Only this non-agreeing form ndiye is mentioned by Fortune to be also used with direct reported discourse, as in (36). (36) ndi-ye {sáráí} id-pro {stay! = goodbye!} and he said, Goodbye; [contextual meaning:] and he died  (Fortune 1971: 252) Therefore, I conclude for Shona that the ndiye construction is evidence for a nonclausal identificational quotative index but that it does not refer to a specific speaker referent and is thus not of the speaker-instantiating type. Note, however, that the Shona presentational marker ha- is a likely cognate of the identificational-quotative

 Tom Güldemann

marker kha~khV- of Tonga-Inhambane discussed in Section 2.1 (cf. Güldemann 1996: 202–7), so that a more abstract functional pattern emerges here. From the comparison of the clausal bases of the above quotative indexes one must diagnose a certain amount of diversity. As pointed out, normally the structure displays a non-verbal element. This, however, varies in being either presentational (English – here-type, Bosnian-Croatian-Serbian, Portuguese) or identificational (English – this-type). In the historical cases, its specific nature remains unclear because the earliest meaning or function of a given element which was the input for the quotative use need not be the same as that inferred from its other/later functions. Note in this respect that all four cases, Egyptian, Akkadian, Tonga-Inhambane, and Hausa, involve comparative evidence according to which the element in question can be argued to display both presentational and identificational uses within the respective language family. Answering this question is made even more difficult by the fact that the distinction between these two functions may be difficult to make even synchronically in individual languages, as in the above case of Tongan ko in (33) with both an identificational and presentational reading. As is clear from the existence of the English pattern [this/here is SPEAKER] based on be, non-verbal elements are not the only option in speaker-instantiating quotative indexes, implying that they can come close to canonical verbal clauses. In spite of all these morphosyntactic differences regarding clause syntax, I would, nevertheless, identify a common structural denominator. In an unmarked event-oriented monoclausal quotative index ‘X said,’ the speaker is normally encoded as the topical subject about which the verb predicates the state of affairs ‘say/ speak’ etc.; it is thus a categorical, as opposed to thetic, statement in the sense of Sasse (1987). This is clearly opposed to the speaker-instantiating structures at issue here, in which the speaker nominal is in pragmatic and syntactic terms not treated as a topical participant about which further information is presented; hence it is not a subject in any canonical sense of this term.10 It is this property which renders the quotative structure to be thetic. The marked detopicalization of the speaker~agent in turn is related to the phenomenon observed by Matras and Sasse (1995) and Güldemann (2008: 63–4) that some languages display thetic devices even in quotative indexes which display a verbal clause. Another conclusion regarding formal properties of speaker-instantiating quotative indexes is that they are in principle free with respect to their order vis-à-vis 10. The status of the speaker as a non-topical agent can even hold for non-grammaticalized cases of participant-oriented quotative indexes, as shown in Section 1.3 by the Saharan languages, which encode the speaker nominal by the marker yè that has precisely this detopicalizing function.



Thetic speaker-instantiating quotative indexes as a cross-linguistic type 

the quote. While most cases are preposed to the quote, Egyptian and Tongan (also) display postposed and/or intraposed order patterns. It is an issue of future research to investigate whether it is coincidence that exactly these two languages are the ones with basic VSO clause order (the other languages are all SVO except for Akkadian, which is SOV, presumably innovated due to language contact). Note in this respect that I have so far not come across postposed quotative indexes with exclusive participant reference, whether grammaticalized or not, in head-final SOV languages (cf. in Section 1.3, the examples (5) from Kunama and (7)–(12) from two Saharan languages as relevant non-grammaticalized cases). Due to the limited data and the fact that the order of quotative index and quote is a far more complex matter (see Güldemann 2008: Section 3.2 for more extensive discussion), typological generalizations regarding non-clausal participant-oriented quotative indexes in general and speaker-instantiating ones in particular must await further research on a wider data base. Despite the limited amount of data at hand, one can discern properties which the eight cases also share in functional terms. One first point is that, in spite of their quite unusual form, they are functionally ‘canonical’ with respect to my typological generalization (see Section 1.3) in that the most central ingredient of quotative indexes is speaker representation. Moreover, this requirement is not just ‘fulfilled’ by the structure at issue; speaker representation is in fact its only semantic content and in pragmatic terms is even made more salient. This informationstructural status of the speaker as a non-topical agent is clearly reflected in its marked syntactic exposition and, if relevant, the accompanying morphology. In this sense speaker-instantiating quotative indexes are actually good representatives of their expression type. They should not be treated as exceptional, but rather as representing a specific variant of a more universal theme: like a number of other structural types without semantically specific verbs, they invoke the implied state of affairs (speech, cognition, etc.) only indirectly and thus background it with respect to other components of the reported discourse construction. Another functional commonality of the above structures is that they are originally restricted to direct reported discourse. This is particularly clear in the modern cases but also holds for the reconstructed historical cases of Egyptian and Akkadian in their earliest attestation. The expansion into indirect reported discourse holding for the historical cases is clearly associated with increasing grammaticalization – a phenomenon which applies also to other morphosyntactic types of quotative indexes with a regular initial bias to direct quotes (Güldemann 2008: 161–171). Moreover, the cases which are better characterized in terms of their discourse properties, namely English and Bosnian-Croatian-Serbian, seem to indicate that the respective quote representation has a particularly high degree of pragmatic

 Tom Güldemann

salience and stylistic expressivity. As one should expect, this property is also observed in more elaborate descriptions of non-grammaticalized cases of participant-oriented quotative indexes (cf. the quote by Lukas 1953: 177 on the relevant strategy in the Saharan language Tubu in Section 1.3). Admittedly, the data presented here are still limited. I would dare to venture, however, that the functional properties identified here are systematic rather than coincidental. Under this hypothesis, the place of these structures in the overall domain of reported discourse can be determined more precisely. This is related to a more comprehensive, though less common, approach to reported discourse as a whole proposed by Güldemann (2008). One important aspect of this approach (developed in detail, e.g. by von Roncador 1988) is that reported discourse is viewed as a scalar domain between the opposite poles of maximally direct and maximally indirect reported discourse; the different focal categories on this scale are determined by a specific kind and degree of interference by the reporter perspective. The direct end of the categorial scale, which is relevant here, can be characterized by the least amount of reporter interference – relevant expressions pretend to reproduce/perform/enact the reported part as close to the invoked original as possible, irrespective of whether this original is real or only constructed/ hypothesized. By now, this approach to direct reported discourse has a long tradition, especially in studies focusing on discourse analysis and pragmatics (see references in Güldemann 2008: 291–2), and is also well represented in this volume. Accordingly, I have proposed in Güldemann (2002, 2008: Section 4.4), to which the reader is referred for more details, that direct reported discourse, so viewed primarily as the enactment of linguistic behaviour, pertains to a wider domain in language that also includes expressions such as iconic representational gestures, non-linguistic sound imitations (mostly by means of the speech organs), as well as ideophones and related lexical signs. While these are often equated with or subsumed under reported discourse, I consider them in principle as semantically and structurally independent. This wider domain, subsuming all four expression types on an equal footing, I call mimesis and define it with respect to its relevance in modern human languages as follows: Mimesis in language is a second mode of representing states of affairs in addition to the descriptive mode with canonical linguistic signs; it evokes for both speaker and hearer a sensory perception, rather than a mental representation, this by expressive, iconic means like sound symbolism, suprasegmental speech modulation and representational gestures.

Mimesis, as such, is not a purely linguistic phenomenon but a far more general cognitive capacity of humans that has been argued to have preceded language itself. Donald (1998: 49) writes:



Thetic speaker-instantiating quotative indexes as a cross-linguistic type 

Mimesis is a non-verbal representational skill rooted in kinematic imagination – that is, in an ability to model the whole body, including all its voluntary actionsystems, in three-dimensional space. This ability underlies a variety of distinctively human capabilities, including imitation, pantomime, iconic gesture, imaginative play and the rehearsal of skills. My hypothesis is that mimesis led to the first fully intentional representations early in hominid evolution, and set the stage for the later evolution of language.

The specific nature of mimesis in language is defined by its relation to unmarked linguistic expressions which convey propositional content by means of lexical items – the ‘normal’ signs of language. The performance character of mimesis is not only associated with a distinct kind of ‘enacted’ representation of semantic content but also tends to have typical discourse properties, notably an increased involvement and empathy on the part of the speaker within the communication process as well as an association with foregrounded information. This is clearly related to the previous characterization of the opposition between (extreme) direct and indirect reported discourse. Direct reported discourse can thus be modeled as a true intersection of the two domains, whereby each has some categories not shared with the other, that is, non-mimetic (e.g. extreme indirect) reported discourse on the one hand and non-reporting mimesis, like iconic gesture, ideophones etc, on the other hand. This is schematized in Figure 2. The mimetic character of direct reported discourse can be expected to entail certain choices regarding its formal expression, one aspect being the structure of the quotative index. In particular, I present evidence in Güldemann (2008) that structures which are propositionally truncated and involve a foregrounding of the quote and its source are overall typical for the more mimetic types of reported discourse. The strong pragmatic sensitivity of mimesis also leads me to assume that reported discourse structures closely associated with this domain are more often subject to linguistic innovation and tend to feed historical change in the overall encoding system. This is also demonstrated in this volume by holistic and fine-grained studies of entire quotative inventories of colloquial speech in which a strong mimetic character of the reported discourse construction correlates with the choice of such specific quotative indexes. The speaker-instantiating quotative

Reported discourse

Direct reported discourse

Mimesis

Figure 2.  Direct reported discourse as an intersection of two domains

 Tom Güldemann

indexes at issue here seem to fall clearly in this more mimetic range, both in terms of the type of quote representation (direct aka mimetic) and, as far as this can be discerned, its discourse properties (discourse-salient information). Indeed, these properties are reflected in the form of the quotative index: the linguistic material involved is semantically reduced to the speaker reference, but at the same time conveys the pragmatic function of foregrounding. The possible expectation that such thetic nominal-oriented clauses should be suitable for mimetic expressions in general is indeed borne out by some empirical data in that they are also employed for mimesis other than direct reported discourse; one such example has been briefly presented above with the case of Shona, exemplified in (34) and (35), in which both presentational and identificational clauses, which in other languages index direct quotes, are also recruited for the introduction of ideophones. In view of my hypothesis that speaker-instantiating quotative indexes are designed for mimetic direct reported discourse, an important problem arises, however, regarding the data at hand. Apart from their overall limited amount, they clearly involve a kind of ‘empirical gap’ (provided, of course, that my reconstructions in Section 2.1 are adequate). On the one hand, there are the very young colloquial instances which clearly pertain to the mimetic domain and are here apparently in competition with equally innovative alternative strategies. On the other hand, there are the historical cases, which hardly reflect these supposedly inherent pragmatic features and whose grammaticalization history implies a high degree of previous routinization. Obviously, languages differ greatly as to what type of linguistic expression is subject to increasing conventionalization; what in one language becomes a default structure remains in another language idiosyncratic, unstable and/or nonstandard. Nevertheless, one would wish to find cases which are in-between the two extremes presented in this article. Data of this kind could throw light on the question under what structural and sociolinguistic conditions such a pattern develops towards a default quotative index, even to the extent that, under my analysis in Section 2.1, it becomes structurally “streamlined” according to the form of a normal event-referring verbal clause. Obviously, the future study of speaker-instantiating quotative indexes promises to be an interesting field for both morphosyntactic and sociolinguistic research on reported discourse and beyond. References Baumgärtner, Walter. 1974. Hebräisches und aramäisches Lexicon zum Alten Testament. Leiden: E. J. Brill.



Thetic speaker-instantiating quotative indexes as a cross-linguistic type  Broschart, Jürgen. 1994. Präpositionen im Tonganischen: Zur Varianz und Invarianz des Adpositionsbegriffs. Bochum-Essener Beiträge zur Sprachwandelforschung 26. Bochum: Universitätsverlag Dr. N. Brockmeyer. Cameron, Richard. 1998. “A variable syntax of speech, gesture, and sound effect: direct quotations in Spanish.” Language Variation and Change 10: 43–83. Cheshire, Jenny and Fox, Sue. 2007. “‘This is me, this is him’: Quotative use among adolescents in London.” Paper presented at the 2nd International Conference on the Linguistics of Contemporary English, Toulouse, July, 2007. Conference handout on the web: http://www. lancs.ac.uk/fss/projects/linguistics/innovators/documents/ICLCE2_FoxCheshire_000.pdf (accessed 21/04/09) Chetveruchin, Alexander S. 1988. “Unexpected linguistic interpretation of jn ‘say(s), said.” Göttinger Miszellen, Beiträge zur ägyptologischen Diskussion 104: 75–88. Churchward, C. Maxwell. 1953. Tongan Grammar. London: Oxford University Press. Custis, Tonya. 2004. Word Order Variation in Tongan: A Syntactic Analysis. Unpublished Ph.D. thesis, University of Minnesota. Deutscher, Guy. 2000. Syntactic Change in Akkadian: The Evolution of Sentential Complementation. Oxford: Oxford University Press. Donald, Merlin. 1998. “Mimesis and the executive suite: Missing links in language evolution.” In Approaches to the Evolution of Language: Social and Cognitive Bases, James R. Hurford, Michael Studdert-Kennedy and Chris Knight (eds), 44–67. Cambridge: Cambridge University Press. Ebert, Karen H. 1975. Sprache und Tradition der Kera (Tschad) Teil I: Texte. Marburger Studien zur Afrika- und Asienkunde A6. Berlin: Dietrich Reimer. Eriksson, Mats. 1995. “A case of grammaticalization in Modern Swedish: The use of ba in adolescent speech.” Language Sciences 17(1): 19–48. Fortune, George. 1971. “Some notes on ideophones and ideophonic constructions in Shona”. African Studies 30(3/4): 237–257. Geider, Thomas. n.d. “Kanuri folktale collection.” Unpublished ms. Golato, Andrea. 2000. “An innovative German quotative for reporting on embodied actions: Und ich so/und er so ‘and I’m like/and he’s like.’’ Journal of Pragmatics 32: 29–54. Gouffé, Claude. 1970/1. “Sur les emplois grammaticalisés du verbe «dire» en haoussa.” Comptes Rendus du Groupe Linguistique d’Études Chamito-Semitiques 15: 77–88. Güldemann, Tom. 1996. Verbalmorphologie und Nebenprädikationen im Bantu: Eine Studie zur funktional motivierten Genese eines konjugationalen Subsystems. Bochum-Essener Beiträge zur Sprachwandelforschung 27. Bochum: Universitätsverlag Dr. N. Brockmeyer. Güldemann, Tom. 2002. “When ‘say’ is not say: The functional versatility of the Bantu quotative marker ti with special reference to Shona.” In Reported Discourse: A Meeting Ground for Different Linguistic Domains, Tom Güldemann and Manfred von Roncador (eds), 253–287. Amsterdam: John Benjamins Güldemann, Tom. 2008. Quotative indexes in African languages: A Synchronic and Diachronic Survey. Empirical Approaches to Language Typology 34. Berlin: Mouton de Gruyter. Güldemann, Tom. forthcoming. “How typology can inform philology: Quotative j(n) in Earlier Egyptian.” In Egyptian-Coptic Linguistics in Typological Perspective, Martin Haspelmath, Sebastian Richter and Eitan Grossman (eds), Amsterdam: John Benjamins. Güldemann, Tom and von Roncador, Manfred (eds). 2002. Reported Discourse: A Meeting Ground for Different Linguistic Domains. Typological Studies in Language 52. Amsterdam: John Benjamins.

 Tom Güldemann Hutchison, John P. 1981. The Kanuri Language – A Reference Grammar. Madison: African Studies Program, University of Wisconsin. Jackson, Ellen M. 1987. “Direct and indirect speech in Tikar.” Journal of West African Languages 17 (1): 98–109. Jordan, Birgit. 2009. Der Sprechakt des Zitats in ausgewählten mittel- und neuägyptischen Texten. M.A. thesis: Johannes-Gutenberg-Universität Mainz. Kammerzell, Frank and Peust, Carsten. 2002. “Reported speech in Egyptian: Forms, types, and history.” In In Reported Discourse: A Meeting Ground for Different Linguistic Domains, Tom Güldemann and Manfred von Roncador (eds), 289–322. Amsterdam: John Benjamins Lanham, Leonard W. 1955. A Study of Gitonga of Inhambane. Bantu Linguistic Studies 1. Johannesburg: Witwatersrand University Press. Lukas, Johannes. 1953. Die Sprache der Tubu in der zentralen Sahara. Veröffentlichungen des Instituts für Orientforschung der Deutschen Akademie der Wissenschaften zu Berlin 14. Berlin: Akademie-Verlag. Matras, Yaron and Sasse, Hans-Jürgen (eds). 1995. Verb-subject order and theticity in European languages. Berlin: Akademie-Verlag Milroy James and Lesley, Milroy. 1977. “Speech and context in an urban setting.” Belfast Working Papers in Language and Linguistics 2: 1–85. Newman, Paul. 2000. The Hausa Language: An Encyclopedic Reference Grammar. New Haven/ London: Yale University Press. Reinisch, Leo. 1881–90. Die Kunama-Sprache in Nordost-Afrika I-IV. Sitzungsberichte der philosophisch-historischen Classe der kaiserlichen Akademie der Wissenschaften 98,(1): 87–174/ 119,5/ 122,5/ 123,1. Sasse, Hans-Jürgen. 1987. “The thetic/categorical distinction revisited.” Linguistics 25: 511–580. Stanley, Carol. 1982. “Direct and reported speech in Tikar narrative texts.” Studies in African Linguistics 13(1): 31–52. Tietze, Andreas. 1959. “Die Eingliederung der wörtlichen Rede im Türkischen.” Wiener Zeitschrift für die Kunde des Morgenlandes 55: 89–121. Van de Velde, Mark L. O. 2008. A Grammar of Eton. Mouton Grammar Library 46. Berlin/ New York: Mouton de Gruyter. von Roncador, Manfred. 1988. Zwischen direkter und indirekter Rede: Nichtwörtliche direkte Rede, erlebte Rede, logophorische Konstruktionen und Verwandtes. Linguistische Arbeiten 192. Tübingen: Max Niemeyer. von Soden, Wolfram. 1965–81. Akkadisches Handwörterbuch, 3 vols. Wiesbaden: Otto Harrassowitz. Ware, Jan. 1993. “Quote formulae in The Final Diagnosis.” Journal of Translation and Text Linguistics 6(2): 161–178. Wolff, H. Ekkehard. 1993. Referenzgrammatik des Hausa: Zur Begleitung des Fremdsprachenunterrichts und zur Einführung in das Selbststudium. Hamburger Beiträge zur Afrikanistik 2. Münster/ Hamburg: LIT. Wolff, H. Ekkehard. 1994. Our People’s Own (Ina Lamaŋ): Traditions and Specimens of Oral Literature from Gwa1 Lamaŋ Speaking Peoples in the Southern Lake Chad Basin in Central Africa. Afrikanistische Forschungen 11. Hamburg: Research and Progress Verlagsgesellschaft. Zuckermann, Ghil‘ad. 2006. “Direct and indirect speech in straight-talking Israeli.” Acta Linguistica Hungarica 53(4): 467–481.

part iii

Functional and formal perspectives

On the characteristics of Japanese reported discourse A study with special reference to elliptic quotation David Y. Oshima and Shin-ichiro Sano

Nagoya University and International Christian University This chapter aims to deepen the understanding of Japanese reported discourse by directing particular attention to the phenomenon in which the predicate of communication or attitude is elided (quotative predicate ellipsis, ‘QPE’). We will taxonomize and examine varieties of the QPE, and demonstrate that its proper understanding leads to straightforward accounts of two issues that have been taken to evidence the ‘idiosyncracy’ of Japanese reported discourse: namely (i) why a quotative phrase may co-occur with a nominal direct object under the same predicate, and (ii) why a quotative phrase may occur under a predicate that is not a predicate of communication or attitude. Keywords: quotative predicate ellipsis, direct/indirect reported discourse, grammatical status of quotes, idiomatic extension of quotative constructions

1. Introduction Reported discourse in Japanese has been subject to extensive studies in the past several decades, and emphasis has often been put on its ‘idiosyncracy’ in comparison with its counterparts in European languages. This chapter aims to deepen the understanding of Japanese reported discourse, directing particular attention to the phenomenon called quotative predicate ellipsis (QPE), whereby the predicate of communication or attitude in reported discourse is elided. Whereas the presence of the QPE has been recognized as a major characteristic of Japanese reported discourse, the fact that the QPE manifests itself in a variety of different forms, carrying different semantic and discourse functions, has been largely overlooked. The current work taxonomizes and examines varieties of the QPE, and demonstrates that a proper understanding of the QPE leads to straightforward accounts of two puzzling issues regarding Japanese reported discourse,

 David Y. Oshima and Shin-ichiro Sano

namely (i) why a quotative phrase may co-occur with a nominal direct object under the same predicate, and (ii) why a quotative phrase may occur under a predicate that is not a predicate of communication or attitude. 2. Characteristics of reported discourse in Japanese Reported discourse in Japanese has attracted a great deal of scholarly attention (Kato 2010; Maier 2009; Fujita 2000a; Kamada 2000; Hirose 1995; Kuno 1988; Coulmas 1986). The most extensively discussed type is marked with the so-called quotative marker/complementizer to or its stylistic variant tte, where the latter is more colloquial.1 Reported discourse with to/tte can either be in the direct or indirect mode, as illustrated below (the sentence forms in (2) also allow for a direct interpretation).2 (1) direct a. Yumi-wa “Zenbu watashi-no sei-yo” {to/tte} Y.-top all I-gen responsibility-sfp to/tte it-te-i-ru. say-ger-ipfv-prs ‘Yumi says “It’s all my fault”.’ b. Yumi-wa “Anata-ga suki-des-u ” {to/tte} it-ta. Y.-top you-nom fond-be.polite-prs to/tte say-pst ‘Yumi said “(I) like you”.’ (2) indirect a. Yumi-wa [zenbu watashi-no sei-da] Y.-top all I-gen responsibility-be.prs {to/tte} it-te-i-ru. to/tte say-ger-ipfv-prs ‘Yumi says that it is all my fault.’

1.

Tte alternates with te when it immediately follows a nasal, as exemplified in the following:

(i) “Otoosan” te it-ta-no? father te say-pst-q ‘Did he say “Dad”?’ 2. Throughout the chapter, direct quotes in examples are put between double quotation marks and italicized; indirect quotes are put between square brackets and not italicized. Also, where naturally occurring data are cited, the source is specified with a code referring to the list provided as Appendix. Example sentences without a source specification were constructed by the authors.



On the characteristics of Japanese reported discourse 

b. Yumi-wa [anata-ga suki-da] {to/tte} it-ta. Y.-top you-nom fond-be.prs to/tte say-pst ‘Yumi said that (she) liked you.’ A reported discourse construction with to/tte may be headed by a wide variety of communication (speech) and attitude (epistemic) predicates, including iu ‘say’, tsugeru ‘announce, tell’, hanasu ‘speak, tell’, tazuneru ‘inquire’, and omou ‘believe, think’ (Teramura 1982: 173–176). In the following, we will refer to those communication/attitude predicates that select for a to/tte-marked quotative phrase as ‘quotative predicates’. Certain communication/attitude predicates, including so-called factive predicates (i.e. those predicates that presuppose the truth of the content of its complement clause, e.g. English know, remember), select for a nominal clause headed by the nominalizer koto and followed by a case particle like o. Quotative constructions with koto invariably are in the indirect mode.3 (3)

Yumi-wa [watashi-ga mukashi Amerika-ni Y.-top I-nom a.long.time.ago America-dat sun-de-i-ta] koto-o shit-te-i-ru. live-ger-ipfv-pst koto-acc know-ger-ipfv-prs ‘Yumi knows that I lived in the U.S.A. a long time ago.’

Besides to and koto, the particles ka, kadooka, and yoo(ni) are commonly regarded as markers of reported discourse, which are respectively used to report (i) a (wh- or yes-no) question, (ii) a yes-no question, and (iii) a request, invitation, etc. (Searle’s “directives”), and which are only compatible with the indirect mode. There is no broad consensus among scholars as to what else, if any, should be added to the list of reported discourse markers (see Kamada 2000: 14–21). The discussion in the current chapter concentrates on reported discourse with to/tte, which is the only marker that is compatible with both the direct and indirect modes. In this section, we will illustrate some intriguing properties of Japanese reported discourse (with to/tte) and critically review previous discussions of them. 2.1

Alleged continuity between direct and indirect reported discourse

As was mentioned above, the same particle to/tte can be used to introduce either direct or indirect reported discourse. Furthermore, unlike languages such as 3. Some communication/attitude predicates are compatible both with to/tte and koto. In such cases, the choice between to/tte and koto tends to cause a subtle semantic/pragmatic difference (Maynard 1986: 192–193).

 David Y. Oshima and Shin-ichiro Sano

English, Japanese indirect quotes do not undergo a change in tense, mood, or word order, which would formally differentiate them from independent clauses and direct quotes. As Coulmas (1986: 172) points out, these conditions contribute to generating “relatively many Japanese sentences that could be given a direct or indirect reading”. The sentence forms in (2), for example, are ambiguous, also allowing the direct interpretation; that is, (2a) can be interpreted as ‘Yumi said “It’s all my fault”’, and (2b) as ‘Yumi said “I like you”’. The direct interpretation is sometimes singled out, however, by features that can be present only in a direct quote or in an independent clause, such as the occurrence of a sentence-final particle – for example, yo in (1a) – or a predicate in its polite form – for example, suki-des-u in (1b). Based on such observations, authors like Kuno (1988) and Kamada (2000) argue that in Japanese there is no clear distinction between direct and indirect reported discourse, and that there exist cases of ‘medium’ (or ‘semi-direct’) reported discourse (see also Maier 2009; McClain 1981; Toyama 1981). This claim, which we call the ‘continuum hypothesis’, is at odds with the widely accepted view that the direct and indirect modes of reported discourse differ in kind, the former being about the relation between an individual and a linguistic object and the latter being about the relation between an individual and a semantic object (proposition) (see Clark and Gerrig 1990; Partee 1973, inter alia). We will demonstrate that the continuum hypothesis lacks sufficient empirical support. Two types of data have been claimed to evidence the continuum hypothesis: (i) imperative morphology in otherwise indirect reports, and (ii) partial deictic shift. The first is exemplified in (4). (4)

Hiroshii-wa boku-ni [yatsui-no otooto-o H.-top I-dat he-gen younger.brother-acc tetsuda-e] to it-ta. help-imp to say-pst ‘Hiroshi requested me to help out his younger brother.’

In (4), the third person pronoun yatsu, which refers to the secondary speaker – that is, the individual whose speech is reported – indicates that the report is in the indirect mode. The imperative ending, on the other hand, would indicate that it must be in the direct mode, under the assumption that an imperative cannot occur in an indirect quote, as is the case in languages such as English. Kuno (1988: 76) introduces the term ‘blended discourse’ to refer to such putative mixed reported discourse. However, the assumption that universally imperatives cannot be reported in the indirect mode seems dubious. It has been reported that in languages like Korean and Amharic, imperative forms can be used in indirect quotes (Pak, Portner, and Zanuttini 2004: 359; Schlenker 2003: 69). In addition, there seems to be no a priori semantic reason that indirect reports of imperatives are impossible. Portner (2005)



On the characteristics of Japanese reported discourse 

speculates that, among the variety of clause types, only (i) declaratives, (ii) interrogatives, and (iii) imperatives are universally attested, and proposes that the first denote propositions, the second denote sets of propositions, and the third denote properties. If we adopt the view that imperatives denote a property, rather than some illocutionary-force bearing entity, it would be intriguing why they cannot be reported in the indirect mode, while the other two universal clause types, declaratives and interrogatives, can. One can even make a case that sentences like the following are ‘indirect imperatives’, although they do not involve imperative morphology.

(5) The man {told/asked} me to leave.

Considering these points, it seems reasonable to suppose that Japanese allows an imperative form to occur in an indirect quote, while certain languages (including English) do not (see Schwager 2006: 204–221 for relevant discussions and additional references). Consequently, imperative morphology in otherwise indirect quotes is not an unequivocal sign of medium reported discourse. Let us now turn to the second type of supposed evidence for the continuum hypothesis: partial deictic shift. (6) a.

Hiroshi-wa [kinoo boku-no otooto-ga H.-top yesterday I-gen younger.brother-nom kare-no uchi-ni it-ta] to it-te-i-ru. he-gen home-dat go-pst to say-ger-ipfv-prs ‘Hiroshi is saying that my younger brother went to his place yesterday.’

b.

Hiroshi-wa [kinoo boku-no otooto-ga H.-top yesterday I-gen younger.brother-nom kare-no uchi-ni ki-ta] to it-te-i-ru. he-gen home-dat come-pst to say-ger-ipfv-prs ‘Hiroshi is saying that my younger brother came to his place yesterday.’

Sentences (6a,b) contain three types of deictic, or viewpoint-sensitive, expressions in the quote: (i) the temporal adverbial kinoo ‘yesterday’, (ii) the personal pronouns boku ‘I’ and kare ‘he’, and (iii) the deictic motion verb iku ‘go’ or kuru ‘come’. In (6a), all three are anchored to the external speaker’s perspective, so that the quote is canonically indirect. In contrast, in (6b) the deictic predicate ‘remains’ anchored to the secondary speaker’s (Hiroshi’s) perspective, while the other deictic elements are interpreted relative to the external speaker’s perspective. Kamada (2000) argues that, in such a sentence, the ‘degree of indirectness’ is smaller in the sense that the secondary speaker’s perspective is partially maintained (see also Kuno 1988). It is worth noting that the same phenomenon is attested in other languages including English, as illustrated in (7) (Oshima 2006a, 2006b; Hockett 1990):

(7) John claims that I {went/came} to his place yesterday.

 David Y. Oshima and Shin-ichiro Sano

If we adopt the idea detailed in Oshima (2006a, 2006b) that certain viewpointsensitive predicates (e.g., English go and come) can be used as secondary indexicals in the sense of Schlenker (2003), sentences like (6b) can be accounted for without compromising the view that direct and indirect modes of reported discourse are discrete.4 In (6b), the verb kuru ‘come’ is used as a secondary indexical and is anchored to the context associated with the reported utterance, while the adverb kinoo and the pronouns boku and kare, which are lexically specified to be primary indexicals, are anchored to the external context.5 As such, all involved deictic expressions (the deictic motion verb, the pronouns, and the adverb) denote entities/ events in the outside world rather than linguistic objects – or in other words, they are used linguistically, rather than meta-linguistically – so that a sentence like (6b) can be treated as a purely (rather than partially) indirect report. Sentences like (6b), therefore, do not serve as valid evidence for the continuum hypothesis either. In this section, we have argued that the claim made in the literature that Japanese lacks a clear distinction between direct and indirect reported discourse does not hold up to scrutiny. We will turn next to another putative characteristic of Japanese reported discourse: the adverbial-like nature of the quotative phrase. 2.2

Syntactic realization of a quote: Complement or modifier?

The syntactic status of the quotative phrase marked by to/tte has been a matter of debate. It has often been treated as a complement of the communication/attitude predicate (Moriyama 1990; Nitta 1982; Okutsu 1974; Teramura 1982 among others), partly because it typically corresponds to a complement clause in a different language such as English, and also because its omission may lead to the unacceptability of the sentence, except where the quoted material is contextually recoverable (# in the examples to follow indicates that the sentence form requires appropriate contextual information to be acceptable).6 (8) a. Yumi-ga #(“Watashi-ga yari-mas-u” to) it-ta. Y.-nom I-nom do-polite-prs to say-pst ‘Yumi said “I will do it”.’ 4. Indexicals are those expressions whose contribution to propositional content depends on a context of utterance or attitude. Primary indexicals (e.g. I, here, now) are those indexicals that are interpreted relative to the context of the external utterance. Secondary indexicals are those indexicals that are interpreted relative to the context of a reported utterance or attitude. 5. The same argument applies to a similar phenomenon that involves ‘empathy-loaded’ predicates like yaru and kureru, which both translate as ‘give’ but reflect different viewpoints, to some extent analogous to go and come (Oshima 2006a). 6. In Japanese, predicate complements in general, whether being nominal or clausal, may be omitted when they are contextually recoverable. For example, (8a) without the quotative phrase would be acceptable if it is uttered in reply to the question “Did anyone say ‘I will do it’?”.



On the characteristics of Japanese reported discourse 

b.

Hiroshi-wa #([watashi-ga kaisha-o yame-ru] to) H.-top I-nom company-acc quit-prs to omot-te-i-ru. believe-ger-ipfv-prs ‘Hiroshi believes that I am going to quit my job.’

However, a contrary position is held by authors like Fujita (2000a, 2000b, 1999, 1988), Niwa (1994), Sunakawa (1989, 1988), and Shibatani (1978), who advocate the view that the to/tte-marked quotative phrase is better understood as an adverbial modifier. An alleged piece of evidence is the ability of a to/tte-marked (direct or indirect) quotative phrase to co-occur with a nominal direct object referring to what is said or thought of under the same quotative predicate; this is illustrated in Examples (9) to (11). (9) a.

Sono sakka-to at-ta toki, watashi-wa, “Anata-no that author-with meet-pst when I-top you-gen sakuhin-wa issatsu-mo yon-da koto-ga ari-mas-en” work-top one.clf-any read-pst matter-nom exist-polite-neg to shoojiki-ni hontoo-no koto-o it-ta. to honestly true matter-acc say-pst ‘When I met that author, I honestly told him the truth, saying “I have not read any of your books”.’

b.

Sono sakka-to at-ta toki, watashi-wa, that author-with meet-pst when I-top [kare-no sakuhin-wa issatsu-mo yon-da koto-ga he-gen work-top one.clf-any read-pst matter-nom na-i] to shoojiki-ni hontoo-no koto-o it-ta. not.exist-prs to honestly true matter-acc say-pst ‘When I met that author, I honestly told him the truth, saying that I had not read any of his books.’

(10) Kare-wa “Soo shi-mas-u” to jikkoo-o yakusoku-shi-ta. he-top that.way do-polite-prs to execution-acc promise-do-pst ‘He promised to carry it out, saying “I will do it”.’  (adapted from Fujita 2000a: 80) (11)

Hiroshi-wa [watashi-no otooto-o otoshiire-ta H.-top I-gen younger.brother-acc ensnare-pst no-wa jibun-da] to tsumi-o kokuhaku-shi-ta. pro-top self-be.prs to sin-acc confess-do-pst ‘Hiroshi confessed his sin, saying that it was he who ensnared my younger brother.’

 David Y. Oshima and Shin-ichiro Sano

This contrasts, for instance, with the situation in English, where a communication or attitude predicate cannot co-occur both with a complement clause and a nominal object referring to what is said or thought of within a single clause.7 (12) a. He told me the good news. b. He told me “The job is yours”. c. *He told me a good news “The job is yours”. In sentences (9) to (11), one may omit either the quotative phrase or the direct object, but not both. Examples (13) to (15) illustrate this point: (13) Watashi-wa #(hontoo-no koto-o) it-ta. I-top true matter-acc say-pst ‘I told (him) the truth.’ (14) Kare-wa #(jikkoo-o) yakusoku-shi-ta. he-top execution-acc promise-do-pst ‘He promised to carry it out.’ (15) Hiroshi-wa #(tsumi-o) kokuhaku-shi-ta. H.-top sin-acc confess-do-pst ‘Hiroshi confessed his sin.’ In sum, the to/tte-marked quotative phrase behaves like an adverbial modifier in that it may co-occur with a nominal direct object referring to the content of speech or thought while also having a characteristic of an argument in that its absence may lead to unacceptability. This issue will be revisited in Section 3.1. 2.3

Elliptic quotation

In Japanese (as well as many other languages; see below), quotative constructions are attested where the quotative predicate is not explicitly present (elliptic quotation). Fujita (1988 and subsequent works) extensively discusses a particular type of elliptic quotation, which is exemplified in (16) (see also Maynard 1986: 186–7; Shibatani 1978: 83). 7. One may suspect that the quotative phrases in sentences (9)–(11) modify the object nominal as an appositive, analogous to the that-clause in the English sentence below.

(i) John told me the fact that Bob resigned as chairman.

There is, however, no reasonable basis to believe that the quotative phrases in question form a constituent with the object nominal co-occurring with them. For example, the sequence of a to/ tte-marked quotative phrase followed by a nominal cannot serve as a clause subject as can a nominal followed by an appositive clause, as in (ii).

(ii) The fact that Bob resigned as chairman surprised me.



On the characteristics of Japanese reported discourse 

(16) a.

Makoto-ga “Ohayoo” to hait-te-ki-ta. M.-nom good.morning to enter-ger-come-pst ‘Makoto came in, saying “Good morning”.’ (lit. ‘Makoto came in, “Good morning”.’)

b. Emiko-wa, “Masaka” to furikaet-ta. E.-top it.can’t.be to look.back-pst ‘Emiko looked back, thinking to herself “It can’t be true”.’ (lit. ‘Emiko looked back, “It can’t be true”.’)  (adapted from Fujita 2000a: 74) Fujita (2000a) argues that sentences like (16a,b), which lack an element like ‘saying’ or ‘thinking (to oneself)’, are instances of one of the two major types of Japanese quotative constructions, which he labels as ‘Type II’. (17) Type I: The matrix predicate describes an action or state that matches the speech or thought described by the quotative phrase. (e.g. (1), (2), (9)) Type II: The matrix predicate describes an action or state that accompanies the speech or thought described by the quotative phrase. (e.g. (16)) Fujita (2000a) claims that it is not appropriate to analyze the Type II quotative construction as involving ellipsis of a quotative predicate, on the grounds that, in Japanese, ellipsis of a predicate is rather rare and tends to result in a sentence that sounds ‘defective’.8 This argument is sensible, but it overlooks one crucial point: unlike ellipsis of other kinds of predicates, ellipsis of a quotative predicate (QPE) is rather common and attested in a variety of environments. The following sentence exemplifies a different type of the QPE, where a quotative predicate in the matrix environment is (or at least appears to be) elided. (18) A: Hiroshi-mo ku-ru-no? H.-also come-prs-q ‘Will Hiroshi come too?’ B: Iya, [kyoo-wa isogashi-i] tte. no today-top busy-prs tte ‘No, he said that he would be busy today.’ It is also worth noting that ellipsis of a communication or attitude predicate is attested in other languages such as Korean (Baek 2006: 193–202; H. Sohn 1999: 325–326) and African languages, including Ewe (Kwa, Niger-Congo) and Kera 8. Fujita also notes that if a constituent is elided (or ‘missing’) in a Type II construction, there must be a unique way to syntactically recover it, whilst also noting that there seems to be no such way. See Section 3.1 for discussion on this point.

 David Y. Oshima and Shin-ichiro Sano

(Chadic, Afro-Asiatic) (Güldemann 2008: 160–161 et passim, this volume; Dimmendaal 2001: 132; Frajzyngier 1996: 125–141), and less prominently in European languages like Spanish and English (Buchstaller 2011; Cameron 1998).9 Considering these points, it seems sensible to hypothesize that Fujita’s Type II construction does indeed involve ellipsis of the quotative predicate. In the next section, we will examine varieties of the QPE phenomenon, which is to be regarded as a major characteristic of Japanese reported discourse, and argue that the Type II construction is a kind of elliptic quotation. 3. Varieties of elliptic quotation This section has two aims: Firstly, it will argue that QPE is a key to solving the two puzzles regarding Japanese reported discourse, namely (i) why a quotative phrase may co-occur with a nominal direct object under the same predicate, and (ii) why a quotative phrase may occur under a predicate that is not a predicate of communication or attitude. Secondly, this section will offer detailed descriptions of a variety of QPE phenomena. We will examine the forms, functions, and licensing conditions of five types of the QPE: (i) suspensive QPE, (ii) sentence-final QPE, (iii) QPE in noun-modifying constructions, (iv) QPE in topic phrases, and (v) QPE in concessive constructions. The first two types involve ellipsis of a quotative predicate in its ordinary, literal use, while the next two involve ellipsis of iu ‘say’ occurring as part of an idiomatic phrase. The fifth type involves ellipsis of iu both in the literal use and as part of an idiomatic phrase. 3.1

Suspensive QPE

We propose that the type of QPE exemplified in (16) above, which we term the suspensive QPE, is generated through ellipsis of iu ‘say’ or omou ‘believe, think’ in their te-form, or gerund form in Bloch’s (1946) terminology, heading a dependent clause. A dependent clause headed by a te-form (gerund clause) may engage in a variety of semantic relations with the main clause (Makino and Tsutsui 1986: 464–467), among which the most canonical are: (i) simultaneous occurrence, (ii) sequential occurrence, (iii) manner, and (iv) cause.10 9. The two types of QPE illustrated in (16) and (18), which are to be labeled below as the suspensive QPE and the sentence-final QPE, may be regarded as instances of ‘non-clausal, quoteoriented quotative index’ in Güldemann’s (2008: 516, this volume) typology of quotative indexes (his term referring to expressions that indicate reported discourse). 10. One may find it more appropriate to group the first two uses under the label of ‘simultaneous or sequential occurrence’.



On the characteristics of Japanese reported discourse 

(19) (simultaneous occurrence) a. Migite-de ten-o sashi-te, hidarite-de right.hand-by sky-acc point.at-ger left.hand-by chi-o sashi-ta. ground-acc point.at-pst ‘He pointed at the sky with his right hand, and at the ground with his left hand.’ b. Hiroshi-ga chizu-o mi-te, Ken-ga unten-shi-ta. H.-nom map-acc look-ger K.-nom drive-do-pst ‘Ken drove while Hiroshi looked at the map.’ (20) (sequential occurrence) a. Ha-o migai-te, hige-o sot-ta. tooth-acc brush-ger beard-acc shave-pst ‘He brushed his teeth and (then) shaved.’ b. Hiroshi-ga booru-o nage-te, Ken-ga uketome-ta. H.-nom ball-acc throw-ger K.-nom receive-pst ‘Hiroshi threw the ball and Ken caught it.’ (21)

(manner) Basu-ni not-te, kaisha-ni it-ta. bus-dat ride-ger company-dat go-pst ‘He went to work by bus.’

(22)

(cause) Ishi-ni tsumazui-te, koron-da. stone-dat stumble-ger fall-pst ‘He stumbled on a stone and fell.’

A suspensive QPE construction may carry out any of these four functions, as shown in (23) to (26); this agrees well with our claim that this construction is generated through ellipsis of a quotative predicate in its te-form.11

11. As sentences like (19b) and (20b) illustrate, the subject of a gerund clause does not need to match that of the superordinate clause. The same observation holds for a suspensive QPE construction, as illustrated by Example (i) (note that the predicate kireru ‘hang up’ is intransitive and selects for an inanimate subject); see Fujita (2000a: 77–79) for similar remarks. (i) “Urusee, bakazaru-ni-wa kankei nee” tte kire-ta. annoying stupid.ape-dat-top relation not.exist.prs tte hang.up(intr.)-pst ‘She said “It’s none of your business, you stupid ape”, and the phone hung up.’  [Ikebukuro: 134]

 David Y. Oshima and Shin-ichiro Sano

(23)

(simultaneous occurrence) “Dareka i-mas-en-ka?” to (it-te) doa-o tatai-ta. anybody be-polite-neg-q to say-ger door-acc knock-pst ‘He knocked on the door, saying “Is anybody here?”.’

(24)

(sequential occurrence) “Soredewa mata ashita” to (it-te) heya-o de-te-it-ta. then again tomorrow to say-ger room-acc exit-ger-go-pst ‘He said “Okay, then I will see you tomorrow”, and left the room.’

(25) (manner) a. “Omae-ga ik-e” to (it-te) Hiroshi-ni shorui-o you-nom go-imp to say-ger H.-dat document-acc tori-ni ik-ase-ta. get-dat go-caus-pst ‘He made Hiroshi fetch the document, saying “You go”.’ b. (26)

“Dare-ni-demo machigai-wa ar-u” to (omot-te) jibun-o who-dat-any mistake-top exist-prs to think-ger self-acc nagusame-ta. console-pst ‘He consoled himself, thinking “Anyone can make a mistake”.’

(cause) “Moo doose maniaw-ana-i” to (omot-te) hashir-u no-o already anyway be.on.time-neg-prs to think-ger run-prs pro-acc yame-ta. stop-pst ‘He stopped running, thinking “I won’t make it anyway”.’

Interestingly, it appears that ellipsis of it-te ‘say’ is only compatible with the first three interpretations of the te-form (‘simultaneous occurrence’, ‘sequential occurrence’, and ‘manner’), and ellipsis of omot-te ‘think’ is only compatible with the ‘manner’ and ‘cause’ interpretations. It is impossible, for example, to elide it-te on the ‘cause’ interpretation in (27), and omot-te on the ‘simultaneous occurrence’ interpretation in (28). (27)

“Ore-igai-wa minna munoo-da” to *(it-te) hankan-o I-except-top all incompetent-be.prs to say-ger antipathy-acc kat-ta. buy-pst ‘He incurred hostility, saying “Everyone except me is incompetent”.’



On the characteristics of Japanese reported discourse 

(28)

“Kitto taikutsu-daroo” to *(omot-te) probably boring-probably to think-ger yomi-hajime-ta-ga, igai-to omoshirokat-ta. read-begin-pst-although unexpectedly interesting-pst ‘I started to read it, thinking “Probably it will be boring”, but it turned out to be more interesting than I expected.’

Another interesting restriction on the suspensive QPE construction (for which we cannot offer an explanation) is that it cannot be accompanied by an adverbial modifier (Fujita 2000a: 75). To put it simply, a quotative predicate in its gerund form cannot be elided when it is modified by an adverbial. (29)

Oogoe-de “Dareka i-mas-en-ka?” to *(it-te) doa-o loud.voice-by anybody exist-polite-neg-q to say-ger door-acc tatai-ta. knock-pst ‘He knocked on the door, saying “Is anybody here?” in a loud voice.’

As is the case for the ordinary quotative construction, to in the suspensive quotative construction may alternate with its stylistic variant tte. (30)

“Shachoo, okuruma dashi-mash-oo-ka ...” tte Ueda-ga president car take.out-polite-vol-q tte U.-nom makuramoto-ni tats-u kamoshiremasen-na. bedside-dat stand-prs it.may.be.the.case.polite-sfp ‘[The ghost of the deceased chauffeur] Ueda may appear by your bedside, saying “Do you need a ride, boss?”.’ [naname: 137]

The suspensive QPE construction is typically in the direct mode, but is compatible with the indirect mode, too (Fujita 2000a: 108–109, 2000b: 99). (31)

[Zettai-ni kimi-ni make-taku-na-i] to, certainly you-dat lose-want-neg-prs to moorensyuu-shi-te-i-ru-yo. practice.hard-do-ger-ipfv-prs-sfp ‘He has been practicing hard, saying that he definitely does not want to lose to you.’

Considering the discussion so far, we can now give a straightforward solution to the previously puzzling syntactic status of a to/tte quotative phrase (illustrated in Section 2.2). We propose that quotative constructions like (9a,b), (10) and (11)

 David Y. Oshima and Shin-ichiro Sano

involve the suspensive QPE. (10) (repeated below with the elidable predicate it-te), for example, can be analyzed as having the structure in (33).12 (32)

Kare-wa “Soo shi-mas-u” to (it-te) jikkoo-o he-top that.way do-polite-prs to say-ger execution-acc yakusoku-shi-ta. promise-do-pst ‘He promised to carry it out, saying “I will do it”.’

(33) [S Kare-wa [S “Soo shi-mas-u” to (it-te)] jikkoo-o yakusoku-shi-ta] This analysis implies that a to/tte quotative phrase and a nominal direct object do not co-occur under a single quotative predicate, and thus a to/tte-quotative phrase may well be treated as a complement.13 When the main predicate is iu ‘say’, as in (9) – repeated below as (34) – the non-elliptic version of a suspensive quotative construction sounds awkward or marginal: (34) a.

Sono sakka-to at-ta toki, watashi-wa, that author-with meet-pst when I-top “Anata-no sakuhin-wa issatsu-mo yon-da you-gen work-top one.clf-any read-pst koto-ga ari-mas-en” to (?it-te) matter-nom exist-polite-neg to say-ger

12. Note that Fujita (2000a) considers sentences like (9a,b), (10) and (11) as instances of the Type I quotative construction. That is, what we call the suspensive QPE construction does not exactly correspond to but subsumes Fujita’s type II construction. 13. Under a predicate describing a gesture or facial expression, a to/tte quotative phrase representing the conveyed message may co-occur with a direct object, without implying that there actually was a verbal utterance. (i)

Kare-wa “Acchi-ni ik-e” to te-o fut-ta. he-top there-dat go-imp to hand-acc wave-pst a. ‘He waved his hand, saying “Go away”.’ b. ‘He waved his hand, to convey the message: “Go away”.’’

This type of sentence (e.g. (i) on interpretation (b)) cannot be paraphrased with the version with it-te, which would imply that the gesture was actually accompanied by an utterance, or with the version with omot-te (Kare-wa ... to {it-te/omot-te} te-o fut-ta). The reason that a pseudo-quotative construction like this is possible may have to do with the fact that the direct object in it is not a canonical object in that it denotes a body part of the referent of the subject, so that the combination of the predicate and the direct object is construed as a single unit (which then can be combined with a to/tte quotative phrase).



On the characteristics of Japanese reported discourse 



shoojiki-ni hontoo-no koto-o it-ta. honestly true matter-acc say-pst ‘When I met that author, I honestly told him the truth, saying “I have not read any of your books”.’

b.

Sono sakka-to at-ta toki, watashi-wa, that author-with meet-pst when i-top [kare-no sakuhin-wa issatsu-mo yon-da he-gen work-top one.clf-any read-pst koto-ga na-i] to (?it-te) shoojiki-ni matter-nom not.exist-prs to say-ger honestly hontoo-no koto-o it-ta. true matter-acc say-pst ‘When I met that author, I honestly told him the truth, saying that I had not read any of his books.’

We believe, however, that this does not necessarily constitute a counter-argument against our analysis, as syntactic or morphological ellipsis is often not merely an optional process, but a pragmatically obligatory one (McShane 2005: 24–25), and is required for the sentence to be felicitous and natural. The following pair of English sentences illustrates this point; (35b) is arguably the non-elliptic version of (35a), and yet it sounds less natural. (35) a. John has been feeling depressed, but he doesn’t know why. b. ?John has been feeling depressed, but he doesn’t know why he has been feeling depressed. The awkwardness of (34a,b) without ellipsis may be analogized with that of (35b). We will now turn to another type of QPE, where a quotative predicate in the sentence-final position is elided. 3.2

Sentence-final QPE

As illustrated in (18) above, a quotative predicate heading a main clause can also be elided. This phenomenon, which we call the sentence-final QPE, has been widely considered as one of the peculiar characteristics of the Japanese quotation, evidenced by the fact that it is often mentioned in textbooks and reference books for learners of Japanese (e.g. Banno et al. 1999: 97; Sunakawa et al. 1998: 232; Makino and Tsutsui 1986: 510–511). In the sentence-final QPE, typically, the elided predicate corresponds to some form of iu ‘say’, such as it-ta (perfective past), it-te-i-ru (imperfective present), and

 David Y. Oshima and Shin-ichiro Sano

it-te-i-ta (imperfective past).14 Unlike in the case of the suspensive QPE construction, the sentence-final QPE construction may be accompanied by an adverbial modifier. (36)

Yamada-san-wa hakkiri “Watashi-wa iki-mas-en” tte y.-suffix-top clearly I-top go-polite-neg tte (it-te-i-ru/ it-ta/ ...). say-ger-ipfv-prs say-pst ‘Yamada {says/said} “I won’t go” in a clear tone.’

An interesting constraint on the sentence-final QPE is that a communication predicate referring to an eventuality in the future cannot be elided; for example, (36) cannot be interpreted as ‘Yamada will say “I won’t go” in a clear tone’. The quote in a sentence-final QPE construction is typically a direct quote, but can also be indirect. (37) [Omae-ga ik-u-nara ik-ana-i] tte. you-nom go-prs-cond go-neg-prs tte ‘He says he won’t go if you are going.’ The sentence-final QPE typically involves the form tte, rather than to (most examples of the sentence-final QPE in the aforementioned references are with tte). A sentence-final QPE construction with to, however, is not impossible. (38)

Watashi, Takigawa-ni it-te izoku-no I T.-dat go-ger surviving.family-gen kata-to omenikakat-te ohanashishi-ta ori-ni, person-with meet.hon-ger talk.hon-pst time-dat “Jisatsu-nasar-u zenya soshite toojitsu, suicide-do.hon-prs previous.night then same.day donna goyoosu-deshi-ta-ka” tte what.kind.of appearance-be.polite-pst-q tte ukagat-tara, “Futsuu-to kawari-mas-en-deshi-ta” to. ask.hon-after usual-with differ-polite-neg-polite-pst to ‘When I went to Takigawa and talked to a family member of the high school student who committed suicide, I asked him “How was he on the night before and on the day when he committed suicide?”, and then he said “He looked just the same as usual”.’ [Diet: 165th session]

14. The sentence-final QPE is also possible in speech in the polite style (characterized by the use of polite predicate forms); this means that iu in polite forms, e.g. ii-mashi-ta (polite perfective past), can be a target of ellipsis.



On the characteristics of Japanese reported discourse 

In addition, there are cases where the elided predicate corresponds to some form of omou ‘believe, think’. Ellipsis of omou heading a matrix clause is more restricted, however, in that (i) the matrix subject must be first person, and (ii) the quote cannot represent just any mental process, but must represent either an inference or an intention. (39) exemplifies ellipsis of omou with quote decribing the speaker’s inference, and (40) with quote describing the speaker’s intention. (39) [Mitarai-san-nara kyoomi-o omochininar-u nja-na-i-ka] to. M.-suffix-cond interest-acc have.hon-prs aux-neg-prs-q to ‘I thought you might be interested, Mr. Mitarai.’  [dance: 14, modified] (40) (In reply to “Oh, weren’t you asleep?”) [Kore-kara nemur-oo-ka] to, nanika? this-from sleep-vol-q to something ‘I was just thinking of getting some sleep. Any news?’  [dance: 246, modified] 3.3

Extensions of sentence-final QPE

The forms to and tte are often used as a sentence-final particle or part of an idiomatic phrase occurring at the end of a sentence. It is natural to hypothesize that such uses of to/tte emerged from the sentence-final QPE construction, through reanalysis. However, it is beyond the scope of the current work to fully examine all such extended uses of to/tte; here, we will just offer three representative examples (see Kato 2010; Sunakawa et al. 1998: 197, 232–233; Okamoto 1995 for relevant discussions). Firstly, tte (but not to) may be used as a sentence-final particle that indicates the speaker’s eagerness to convince the hearer to accept or comply with the content of the utterance.15 (41) a. Zettai muri-da-tte. certainly impossible-be.prs-tte ‘It’s definitely impossible, I’m telling you.’ b. Yame-ro-tte. stop-imp-tte ‘Stop it!’ It is quite natural to hypothesize that tte in this usage emerged from the idiomatic construction ‘P to/tte iu n(o)da’ (exemplified in (42); P is a placeholder for a clause 15. Tte as a sentence-final particle, as well as the ‘P to/tte iu n(o)da’ construction to be mentioned below, may be regarded as an instance of the cross-linguistic tendency for quotative markers/constructions to acquire the function of ‘illocution reinforcement’ (Güldemann 2008: 411–417 and references therein).

 David Y. Oshima and Shin-ichiro Sano

and n(o)da is an auxiliary verb with varied functions including emphasis), through QPE and reanalysis.16 (42) [Yame-ro] tte i-u nda. stop-imp tte say-prs aux.prs ‘I’m telling you to stop!’ Secondly, the sequence of (some form of) the copula da and to/tte functions to indicate the speaker’s incredulity or amazement. The expression preceding dato/ datte is typically a repetition of (part of) the immediately preceding utterance by a different interlocutor, and does not need to be a complete clause. (43) a. Yuurei datte? ghost datte ‘Did you say “ghost”? (I can’t believe my ears.)’ b. Boku-ga uso-o tsui-te-i-ta datte? I-nom lie-acc tell-ger-ipfv-pst datte ‘Did you say that I was lying to you? (I can’t believe my ears.)’ Thirdly, the sequence of (some form of) the auxiliary n(o)da and to/tte can be used as a hearsay expression, which can roughly be translated as ‘I hear that’ or ‘they say that’.17 (44) Kono yama-ni-wa kitsune-ga i-ru ndato. this mountain-dat-top fox-nom be-prs ndato ‘I hear that there are foxes on this mountain.’ The expression preceding n(o)dato/n(o)datte must be a clause, and cannot be a nominal or some other kind of utterance fragment. The ‘P n(o)dato/n(o)datte’ construction needs to be distinguished from the regular sentence-final QPE construction. This is evident in contexts where the segment n(o)da is turned into the polite form n(o)desu, as in (45); the construction is thus interpreted as indicating deference of the (external) speaker towards the (external) 16. It is also possible to elide the segement iu in the ‘P to/tte iu n(o)da’ construction and leave the rest as it is. (i) [Yame-ro] tte nda. stop-imp tte aux.prs ‘Stop it!’ 17. Functional extension of a quotative marker/construction into a hearsay/evidential expression is attested across languages, including Tibeto-Burman (Saxena 1988), Papuan (de Vries 1990: 294), and a number of African languages (Güldemann 2008: 406–410). See also Spronck (this volume) and Buchstaller (2011) for relevant discussions.



On the characteristics of Japanese reported discourse 

hearer, rather than of the ‘original’ speaker (i.e., the person from whom the speaker of (45) heard that Suzuki would meet Tanaka) towards the reporter-speaker. (45) Suzuki-san-ga ashita Tanaka-san-ni a-u ndesutte. S.-suffix-nom tomorrow T.-suffix-dat meet-prs ndesutte ‘I hear that Suzuki will meet Tanaka tomorrow.’ In other words, (45) conveys that the speaker is being polite to the hearer, rather than some other person being polite to the speaker.18 This implies that the part preceding to/tte is not a quote. (The same point holds for the aforementioned ‘X dato/datte’ construction.) 3.4

QPE in a noun-modifying construction

The sequence of the quotative particles to/tte and iu ‘say’ has an idiomatic use as an adposition occurring in the noun-modifying construction: ‘X toiu/tteiu N’. The exact function and distribution of this construction have been a matter of extensive debate (Matsumoto 1998; Terakura 1983 inter alia). In most typical cases, the relation between the two constituents linked by toiu/tteiu is [X: name, N: entity], [X: content, N: information-bearing entity], or [X: characterization of form, N: linguistic or auditory entity].19 (46)

[X: name, N: entity] Ozu Yasujiro {toiu/tteiu} hito O. Y. toiu/tteiu person ‘the person called Yasujiro Ozu’

18. (45) can be contrasted with a direct report like (i), where the polite form in the quote indicates that the original speaker (= Suzuki) spoke in a polite manner. (i) Suzuki-san-wa “Watashi-wa ashita Tanaka-san-ni ai-mas-u” to it-ta. S.-suffix-top I-top tomorrow T.-suffix-dat meet-polite-prs to say-pst ‘Suzuki said “I will meet Tanaka tomorrow”.’ 19. Of these three relations indicated by toiu, the [X: name, N: entity] is probably the most basic; this can be inferred from the fact that iu as a main predicate may represent this relation, but not the other two. (i) Kono hito-wa Ozu Yasujiro {to/tte} ii-mas-u. this person-top O. Y. to/tte say-polite-prs ‘This person is called Yasujiro Ozu.’ It is also worth noting that cross-linguistically it is quite common for a quotative marker/construction to be also used in a ‘naming construction’ (Güldemann 2008: 398–406 and references therein).

 David Y. Oshima and Shin-ichiro Sano

(47)

[X: content, N: information-bearing entity] raion-ga nige-ta {toiu/tteiu} nyuusu lion-nom escape-pst toiu/tteiu news ‘the news that a lion escaped’

(48)

[X: characterization of form, N: linguistic or auditory entity] “Tetsu-wa atsu-i uchi-ni ut-e” {toiu/tteiu} kotowaza iron-top hot-prs while-dat strike-imp toiu/tteiu proverb ‘the proverb “Strike while the iron is hot”’

Tteiu occurring in this noun-modifying construction can be shortened into tte (National Institute for Japanese Language 2006: 65; Sunakawa et al. 1998: 231). This process may be regarded as a variant of the QPE. (49) a. Ozu Yasujiro tte hito b. raion-ga nige-ta tte nyuusu c. “Tetsu-wa atsu-i uchi-ni ut-e” tte kotowaza Unlike tteiu, toiu cannot undergo this reduction process (recall that to is a more formal variant of tte; as quotative markers, their difference is purely stylistic). Thus, the segment tte in (49a–c) cannot be replaced with to. 3.5

QPE in topic phrases

The topic markers towa and tte, which are illustrated in (50), can be analyzed as having emerged through the etymological processes shown in (51) and (52), which involve ellipsis of iu ‘say’ (among other things). (50) a.

Guusuu-towa ni-de warikir-e-ru even.number-towa two-by divide.without.remainder-pot-prs kazu-no koto-des-u. number-gen matter-be.polite-prs ‘Even numbers are those numbers that can be divided by 2 without a remainder.’

b. Yamada-san-tte shinsetsu-des-u-ne. Y.-suffix-tte kind-be.polite-prs-sfp ‘Yamada is nice, isn’t he?’ (51) [X toiu no] wa > X towa X toiu pro top X towa (52) [X tte(iu) no] wa > X tte X tte(iu) pro top X tte



On the characteristics of Japanese reported discourse 

We hypothesize that the topic marker towa emerged from ‘X toiu no wa’, where toiu is the adposition mentioned in Section 3.4, no is a pronoun that roughly translates as ‘(the) one’, and wa is a so-called topic marker. While the sequence ‘X toiu no wa’ is interpreted in the expected way from the functions/meanings of its subparts, the phrase ‘X towa’ has a more specialized function; typically, it is used as a part of a sentence that explains the meaning/identity of a word or name (Iori et al. 2001: 333–334; Sunakawa et al. 1998: 354–355). Similarly, the topic marker tte can be regarded as having emerged from ‘X tte(iu) no wa’, which consists of a noun modifier ‘X tte(iu)’, a pronoun, and a topicmarker. The exact function of the topicalization construction with tte is hard to pin down, but it can be characterized approximately as ‘to take up an entity or concept, and make an emotive judgment or evaluation on it’ (Makino and Tsutsui 1986: 509–510; Sunakawa et al. 1998: 231–232). 3.6

QPE in concessive constructions

The concessive adverbial clauses in the forms of ‘P kara {to/tte} itte’ and ‘P {to/tte} ittatte’, which are exemplified in (53) and (54) respectively, may undergo shortening that involves ellipsis of (a form of) iu ‘say’. (53) a.

Fuhei-o iw-ana-i kara {to/tte} it-te complaint-acc say-neg-prs because to/tte say-ger manzoku-shi-te-i-ru wakede-wa-na-i. satisfaction-do-ger-ipfv-prs aux-top-neg-prs ‘He is not satisfied, even though he does not complain.’

b.

Kodomo-da kara {to/tte} it-te yurus-u koto-wa child-be.prs because to/tte say-ger forgive-prs matter-top deki-na-i. able-neg-prs ‘I cannot forgive him, even though he is a child.’

(54) a. [Ku-ru-na] {to/tte} it-tatte, ku-ru-daroo. come-prs-neg.imp to/tte say-even.if come-prs-probably ‘He would come even if you tell him not to.’ b.

[Isog-e] {to/tte} it-tatte, kore-ijoo-wa hurry-imp to/tte say-even.if this-more.than-top muri-da-yo. impossible-be.prs-sfp ‘You tell me to hurry up, but I can’t go any faster.’

 David Y. Oshima and Shin-ichiro Sano

c. [Ryokoo-ni ik-u] {to/tte} it-tatte, trip-dat go-prs to/tte say-even.if mikkakan-dake-des-u. for.three.days-only-be.polite-prs ‘I’m indeed going to make a trip, but (contrary to what you might think) it’s only for three days.’ ‘P kara {to/tte} itte’ is an idiomatic construction that literally consists of a clause, kara ‘because’, to/tte, and the gerund form of iu. It must be subordinate to a clause in the negative polarity and roughly translates as ‘even though P’ (Makino and Tsutsui 1995: 103–105). The segment ‘kara {to/tte} itte’ may be shortened into ‘karatte’, as illustrated in (55), without a discernable difference in the meaning. (55) a. Fuhei-o iw-ana-i karatte manzoku-shi-te-i-ru wake-de-wa-na-i. (≈(a)) b. Kodomo-da karatte yurus-u koto-wa deki-na-i. (≈(b)) The presence of the variant form karattutte (or karattatte), which is used in a very colloquial style, suggests a reduction process along the lines of (56) (phonological descriptions are presented between slashes, where Q represents a special phoneme that turns the following consonant into a geminate). (56) kara tte itte /karaQteiQte/ > karattutte /karaQtuQte/ > karatte /karaQte/ (The segment /Qtei/ changes into /Qtu/, and then the segment /Qtu/ drops.) This analysis implies that the formation of the form karatte involves not only ellipsis of the communication predicate, but also ellipsis of the quotative marker. Let us next turn to ‘P {to/tte} ittatte’; this adverbial construction consists of a clause P, to/tte, and the infinitive form of iu followed by the particle tatte ‘even if, even though’.20 As a whole it roughly translates as ‘even if X says P’ (as in (54a)), ‘even though X says P’ (as in (54b)), or, idiomatically, ‘contrary to what you might expect from (X’s saying) P’ (as in (54c)). (Note that the third use does not necessarily involve actual saying, while the first and second use do.)

20. It is interesting to note that similar ambiguity between ‘even if ’ (concessive conditional) and ‘even though’ (concessive) is exhibited by the English (even) if construction. Sweetser (1990) presents examples like the following, referring to them as ‘speech-act concessive conditionals’. (i) a. (Even) if there are ten beers in the fridge, we gotta work. b. (Even) if he’s a stuffed shirt, he’s not a fool.

(Sweetser 1990: 138)



On the characteristics of Japanese reported discourse 

Analogous to ‘P kara {to/tte} itte’, the ‘P {to/tte} ittatte’ construction may be shortened into (‘P ttuttatte’ and) ‘P ttatte’, as illustrated in (57) (Makino and Tsutsui 2008: 597–99). (57) a. [Ku-ru-na] ttatte, ku-ru-daroo. (≈ (54a)) b. [Isog-e] ttatte, kore-ijoo-wa muri-da-yo. (≈ (54b)) c. [Ryokoo-ni ik-u] ttatte, mikkakan-dake-des-u. (≈ (54c)) The formation of the form ttatte is illustrated in (58); as is the case of karatte, it involves not only ellipsis of iu, but also ellipsis of the quotative marker tte. (58) tte it tatte /QteiQtaQte/ > ttuttatte /QtuQtaQte/ > ttatte /QtaQe/ (The segment /Qtei/ changes into /Qtu/, and then the segment /Qtu/ drops.) 4. Conclusion It was demonstrated that in Japanese the basic communication predicate iu (either in its literal use or as part of idiomatic constructions) may be elided in a variety of configurations through similar but distinct processes. In addition, it was shown that ellipsis of the basic attitude predicate omou is also possible, though under more limited circumstances. Considering that quotative predicate ellipsis (QPE) is a rather common phenomenon in Japanese, it appears reasonable to conclude, contra Fujita, that the so-called Type II quotative construction (the construction where a quotative phrase appears to occur under a predicate that is not a predicate of communication or attitude) is generated through QPE. The same type of QPE also accounts for the interesting observation that a to/tte quotative phrase may co-occur with a nominal direct object under the same predicate; it only appears so because a subordinate predicate that selects for a quotative phrase (namely the gerund form of iu ‘say’ or omou ‘think, believe’) may be elided. In sum, the apparent ability of the to/tte quotative phrase to co-occur with a nominal direct object and to occur under a predicate that is not a predicate of communication or attitude – this having been regarded as evidence of the alleged idiosyncrasy of reported discourse in Japanese – can be understood as an effect resulting from the elidability of a quotative predicate. We conclude by mentioning two issues that are in need of future research. First, the relations and connections between varieties of the QPE are yet to be clarified. Diachronic investigation is called for, e.g. to determine whether the suspensive QPE historically preceded the sentence-final QPE or vice versa, and

 David Y. Oshima and Shin-ichiro Sano

whether one of them was etymologically derived from the other. Second, it would be interesting to investigate other languages in which the QPE is attested, and more generally, how conditions differ across languages under which the QPE is possible. In Korean, which is known to have a great deal of grammatical similarities with Japanese, varieties of ellipsis phenomena are attested that are analogous to the suspensive QPE, the sentence-final QPE, and other forms of QPE in Japanese (Baek 2006: 193–202; H. Sohn 1999: 325–326; S. Sohn 1994: 396–401; Martin 1992: 331–333). It has also been reported that a communication predicate in a main clause is often dropped in some African languages (Güldemann 2008: 160–161, this volume; Dimmendaal 2001: 132; Frajzyngier 1996: 125–141), and that a similar phenomenon is observed in European languages such as English and Spanish (Buchstaller 2011; Cameron 1998). Cross-linguistic research is called for to determine how elliptic quotative constructions in Japanese and other languages such as Korean contrast with each other, and to identify universals and the range of variation regarding elliptic quotation. Acknowledgments We are grateful to the editors Ingrid Van Alphen and Isabelle Buchstaller and two anonymous reviewers for valuable comments and discussions which helped us improve this work in a number of ways. Thanks also to Chiemi Fujiwara for valuable comments on a previous version of this work. All remaining errors are our own. References Baek, Bongja. 2006. Oegugeoroseoui hangugeo munbeop sajeon [A reference grammar of Korean as a foreign language]. Seoul: Hau, 2nd edn. Banno, Eri, Ohno, Yutaka, Sakane, Yoko, Shinagawa, Chikako and Tokashiki, Kyoko. 1999. Genki: An Integrated Course in Elementary Japanese, vol. 2. Tokyo: The Japan Times. Bloch, Bernard. 1946. “Studies in colloquial Japanese I: Inflection.” Journal of the American Oriental Society 66: 97–130. Buchstaller, Isabelle. 2011. “Quotations across the generations: A multivariate analysis of speech and thought introducers across 5 generations of Tyneside speakers”. Corpus Linguistics and Linguistic Theory 7: 59–92. Cameron, Richard. 1998. “A variable syntax of speech, gesture, and sound effect: Direct quotations in Spanish”. Language Variation and Change 10: 43–83. Clark, Herbert H., and Gerrig, Richard J. 1990. “Quotations as demonstrations.” Language 66 (4): 764–805. Coulmas, Florian. 1986. “Direct and indirect speech in Japanese.” In Direct and Indirect Speech, F. Coulmas (ed.), 161–178. Berlin: Mouton de Gruyter.



On the characteristics of Japanese reported discourse  Dimmendaal, Gerrit J. 2001. “Logophoric marking and represented speech in African languages as evidential hedging strategies.” Australian Journal of Linguistics 21(1): 131–157. de Vries, Lourens. 1990. “Some remarks on direct quotation in Kombai.” In Unity in Diversity: Papers Presented to Simon C. Dik on his 50th Birthday, Harm Pinkster and Inge Genee (eds), 291–308. Dordrecht: Foris Frajzyngier, Zygmunt. 1996. Grammaticalization of the Complex Sentence: A Case Study in Chadic. Amsterdam: John Benjamins. Fujita, Yasuyuki. 1988. “Inyooron no shikai [The scope of research on quotation].” Nihongogaku 7(9): 30–45. Fujita, Yasuyuki. 1999. “Inyoo koobun no koozoo [The quotative structure in Japanese].” Kokugogaku 198: 1–15. Fujita, Yasuyuki. 2000a. Kokugo inyoo koobun no kenkyuu [A Study of the Quotative Structure in Japanese]. Osaka: Izumi Shoin. Fujita, Yasuyuki. 2000b. “Bunpooron to shite no nihongo inyoo hyoogen no kenkyuu no tame ni: Futatabi Kamada Osamu no shoron ni tsuite [A note for the syntactic analysis of quotative expressions in Japanese: A second criticism on Kamada Osamu’s theory].” Memoirs of the Faculty of Education, Shiga University: Humanities and Social Science 50: 85–104. Güldemann, Tom. 2008. Quotative Indexes in African Languages: A Synchronic and Diachroinc Survey. Berlin: Mouton de Gruyter. Hirose, Yukio. 1995. “Direct and indirect speech as quotations of public and private expression.” Lingua 95(4): 223–238. Hockett, Charles F. 1990. “bring, take, come, and go.” Journal of English Linguistics 23(1/2): 239–244. Iori, Isao, Matsuoka, Hiroshi, Nakanishi, Kumiko and Yamada, Toshihiro. 2001. Chuujookyuu o oshieru hito no tame no nihongo bunpoo handobukku [The Handbook of the Japanese Grammar for Teachers of Intermediate/Advanced Classes]. Tokyo: 3A Corporation. Kamada, Osamu. 2000. Nihongo no inyoo [Quotation in Japanese]. Tokyo: Hituzi Syobo. Kato, Yoko. 2010. Hanashi kotobo ni okeru inyoo hyoogen: Inyoo hyooshiki ni chuumoku shite [Quotative Expressions in Spoken Language: With Special Attention to Quotative Markers]. Tokyo: Kurosio Publishers. Kuno, Susumu. 1988. “Blended quasi-direct discourse in Japanese.” In Papers from the Second International Workshop on Japanese Syntax, W.J. Poser (ed.), 75–102. Stanford: CSLI Publications. Maier, Emar. 2009. “Japanese reported speech: Against a direct-indirect distinction.” In New Frontiers in Artificial Intelligence: JSAI 2008 Conference and Workshops, Asahikawa, Japan, June 11–13, 2008, Revised Selected Papers, vol. 5447 of Lecture Notes in Computer Science, H. Hattori, T. Kawamura, T. Ide, M. Yokoo, and Y. Murakami (eds), 187–200. Berlin: Springer. Makino, Seiichi and Tsutsui, Michio. 1986. A Dictionary of Basic Japanese Grammar. Tokyo: The Japan Times. Makino, Seiichi and Tsutsui, Michio. 1995. A Dictionary of Intermediate Japanese Grammar. Tokyo: The Japan Times. Makino, Seiichi and Tsutsui, Michio. 2008. A Dictionary of Advanced Japanese Grammar. Tokyo: The Japan Times. Martin, Samuel E. 1992. A Reference Grammar of Korean: A Complete Guide to the Grammar and History of the Korean Language. Tokyo: Tuttle.

 David Y. Oshima and Shin-ichiro Sano Matsumoto, Yoshiko. 1998. “The complementizer toyuu in Japanese.” In Japanese/Korean Linguistics, vol. 7, N. Akatsuka, H. Hoji, S. Iwasaki, S.O. Sohn, and S. Strauss (eds), 243–255. Stanford: CSLI Publications. Maynard, Senko K. 1986. “The particle -o and content-oriented indirect speech in Japanese written discourse.” In Direct and Indirect Speech, F. Coulmas (ed), 179–200. Berlin: Mouton de Gruyter. McClain, Yoko M. 1981. Handbook of Modern Japanese Grammar. Tokyo: Hokuseido Press. McShane, Marjorie J. 2005. A Theory of Ellipsis. Oxford: Oxford University Press. Moriyama, Takuro. 1990. Nihongo dooshi jutsugo bun no kenkyuu [A Study of Verb Constructions in Japanese]. Tokyo: Meiji Shoin. National Institute for Japanese Language. 2006. Nihongo hanashi kotoba koopasu no koochikuhoo [Construction of the Corpus of Spontaneous Japanese], no. 124 of NIJLA Reports. Nitta, Yoshio. 1982. “Kaku no hyoogen keishiki: Nihongo [Case-marking expressions: Japanese].” In Gaikokugo to no taishoo: I [Comparison with Foreign Languages: I], vol. 10 of Kooza nihongogaku, Y. Miyaji, H. Teramura, and Y. Kawabata (eds), 118–138. Tokyo: Meiji syoin. Niwa, Junko. 1994. “Fukushiteki shuushoku seibun “to” o seiri suru: Gutaika no “to”, inyoo no “to”, jootai fukushi no “to” o megutte [An analysis of adverbial element “TO”: on TO of specification, of quotation, and of manner adverbs].” In Tsukuba daigaku ryuugakusei sentaa nihongo kyooiku ronshuu, vol. 9: 19–28. Okamoto, Shigeko. 1995. “Pragmaticization of meaning in some sentence-final particles in Japanese.” In Essays in Semantics and Pragmatics. Festschrift for Charles Fillmore, M. Shibatani and S.A. Thompson (eds), 219–246. Amsterdam: John Benjamins. Okutsu, Keiichiro. 1974. Seisei nihon bunpooron [Generative Japanese Syntax]. Tokyo: Taisyukan. Oshima, David Y. 2006a. Perspectives in Reported Discourse. Unpublished Ph.D. thesis, Stanford University. Oshima, David Y. 2006b. “Motion deixis, indexicality, and presupposition.” In Proceedings of SALT, vol. 16, Masayuki Gibson and Jonathan Howell (eds), 172–189. Ithaka: CLC Publications. Pak, Miok D., Portner, Paul and Zanuttini, Raffaella. 2004. “Deriving clause types: Focusing on Korean.” In Proceedings of the Linguistics Society of Korea 2004, 359–368. Seoul: Hanshin Publishing. Partee, Barbara. 1973. “The syntax and semantics of quotation.” In A Festschrift for Morris Halle, Stephen R. Anderson and Paul Kiparsky (eds), 410–418. New York: Holt, Reinhart and Winston. Portner, Paul. 2005. “The semantics of imperatives within a theory of clause types.” In Proceedings of SALT, vol. 14, Kazuha Watanabe and Robert B. Young (eds), 235–252. Ithaka: CLC Publications. Saxena, Anju. 1988. “On syntactic convergence: The case of the verb ‘say’ in Tibeto-Burman.” In Proceedings of BLS, vol. 14, Shelley Axmaker, Annie Jaisser and Helen Singmaster (eds), 375–388. Berkeley: Berkeley Linguistics Society. Schlenker, Philippe. 2003. “A plea for monsters.” Linguistics and Philosophy 26(1): 29–120. Schwager, Johanna M. 2006. Interpreting Imperatives. Unpublished Ph.D. thesis, University of Frankfurt. Shibatani, Masayoshi. 1978. Nihongo no bunseki [Analyses of the Japanese Language]. Tokyo: Taisyukan. Sohn, Ho-Min. 1999. The Korean Language. Cambridge: Cambridge University Press.



On the characteristics of Japanese reported discourse  Sohn, Sung-Ock S. 1994. “Phonological reduction and reanalysis in modern Korean: A case of incipient grammaticalizations.” In Proceedings of CLS, vol. 30(1), K. Beals, J. Denton, R. Knippen, L. Melnar, and H. Suzuki (eds), 391–404. Chicago: Chicago Linguistic Society. Sunakawa, Yuriko. 1988. “Inyoo koobun ni okeru ba no nizyuusei ni tsuite [On the duality of the field in quotative constructions].” Nihongogaku 7(9): 14–29. Sunakawa, Yuriko. 1989. “Inyoo to wahoo [Quotation and reported discoruse].” In Nihongo no bunpo/buntai: Joo [Syntax and Style of Japanese: Part 1], vol. 10 of Kooza nihongo to nihongo kyooiku, Y. Kitahara (ed.), 355–387. Tokyo: Meiji Shoin. Sunakawa, Yuriko, Komada, Satoshi, Shimoda, Mitsuko, Suzuki, Mutsumi, Tsutsui, Sayo, Hasunuma, Akiko, Bekes, Andrej and Morimoto, Junko (eds). 1998. Kyooshi to gakushuusha no tame no nihongo bunkei jiten [A Dictionary of Japanese Phrase Patterns for Teachers and Learners]. Tokyo: Kurosio Publishers. Sweetser, Eve E. 1990. From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. Terakura, Hiroko. 1983. “Noun modification and the use of to yuu.” The Journal of the Association of Teachers of Japanese 18(1): 23–55. Teramura, Hideo. 1982. Nihongo no shintakusu to imi [Syntax and Semantics of Japanese], vol. 1. Tokyo: Kurosio Publishers. Toyama, Shigehiko. 1981. Nihongo no sugao [The True Face of Japanese]. Tokyo: Chuo Koronsha.

Appendix: Sources of data [dance]: Mitarai Kiyoshi no dansu (a novel; author: Shimada Soji, publisher: Kodansha, year of publication: 1993) [Diet]: The on-line full-text database of the Minutes of the National Diet of Japan (administered by National Diet Library; http://kokkai.ndl.go.jp/) [Ikebukuro]: Kudoo Kankuroo kyakuhon: Ikebukuro uesuto geetopaaku (A collection of scripts for a TV series; author: Kudo Kankuro, publisher: Kadokawa Shoten, year of publication: 2005) [naname]: Naname yashiki no hanzai (a novel; author: Shimada Soji, publisher: Kodansha, year of publication: 1992)

Quotative go and be like Grammar and grammaticalization Lieven Vandelanotte

University of Namur and University of Leuven This chapter addresses the question how semantically non-reportative and grammatically intransitive verbs such as be (like) and go could come to be used in English quotative constructions. It rejects analyses which evoke the notion of ‘reporting verb’ or, for like, of complementizer, and argues instead for an interclausal analysis in which clauses such as I’m like or he went as a whole are analysed as conceptually dependent on a complement clause. This analysis of the combinatorics involved in these constructions helps to explain their emergence as an analogical process in which ‘imitation clauses’ are apprehended as ‘reporting clauses’, and invites a reassessment of the extent to which this initial innovation and its further developments constitute a case of ‘grammaticalization’. Keywords: be like, go, constructions, interclausal complementation, grammaticalization vs. constructionalization

1. Introduction1 The past decades have witnessed a spate of publications, mostly from a variationist and interactional sociolinguistic perspective, devoted to the use of go and, in particular, to be like and its variants (such as be all) as quotatives in English. The focus of these articles was mainly to tease out the position of these variants within the quotative system as a whole across different age and sex groups in different English-speaking countries (see e.g. Barbieri 2005, 2007, 2009; Blyth et al. 1990; Buchstaller 2004, 2006, 2008; Buchstaller and D’Arcy 2009; Cukor-Avila 2002; Dailey-O’Cain 2000; D’Arcy 2004; Ferrara and Bell 1995; Hansen-Thomas 2008; Macaulay 2001; Rickford et al. 2007; Romaine and Lange 1991; Tagliamonte and D’Arcy 2004; Tagliamonte and Hudson 1999; Winter 2002). Comparatively less 1. I am grateful to two anonymous referees and both editors of this volume for many helpful suggestions on the level of content as well as expression.

 Lieven Vandelanotte

attention has been devoted to trying to capture the way in which the innovations might initially have arisen. While trying to pinpoint the initial impetus for language change is not easy, this chapter seeks to draw up a plausible model of the grammatical and semantic conditions under which the first novel uses could have arisen from a construction grammatical perspective. The argument I make to this effect will turn out to be closely linked to a specific analysis of the syntagmatic structure of reportative go and be like constructions. The kinds of constructions this paper examines are illustrated in (1) below. In this example, go is used to present that which people told someone who had found a lost kitten (everyone was going), and be like is used for the speaker’s self-report:

(1) And everyone was going “Oh you’ve got no chance of getting it out of there may as well just leave it” and I was like “I cannot leave it there I cannot leave it ’cos it was just crying all the time it was so sad.”  (Collins Wordbanks, spoken British component 1995)2 In focusing on go and be like, I do not intend to claim that both construction types are identical in all respects; rather, the point will be made that they can be grouped together at a schematic level as ‘Innovative Intransitive Quotative’ constructions (Vandelanotte and Davidse 2009) involving a similar syntagmatic structure. Butters (1980) was the first to propose as the origin of the quotative use of go an earlier onomatopoeic use, recorded in the Oxford English Dictionary (OED), as in to go bang or to go clatter. In a similar vein, Meehan (1991) used OED data to trace different shades of meaning in the history of like, such as imitation, approximation and focus, and she related both the quotative and the focus use to the core meaning of similarity. Romaine and Lange (1991) proposed a specific grammaticalization path for like in the vein of Traugott’s (1982) model, arguing that it has acquired a quotative complementizer use as an offshoot of its textual use as a conjunction, which is itself a grammaticalization of like’s original prepositional use. This grammaticalization account of like has taken root and has remained virtually unchallenged, with the exception of Buchstaller (2002, 2004), who questions the linearity of the proposed path from a synchronic point of view, favouring instead a radial network approach (in the sense of Lakoff 1987) to the meaning aspects of like playing a role in its quotative use.

2. Data in this paper are drawn from the Collins Wordbanks online corpus (reprinted with the kind permission of HarperCollins publishers), the Corpus of Contemporary American English (abbreviated COCA; Davies 2008), the Internet (in which case the URL is given in a footnote), and from published articles. Regarding data from the Collins Wordbanks and COCA corpora, an indication is provided of the specific component of the corpus as well as the year from which the example is drawn.

Quotative go and be like 



Important elements identified in the received accounts of go and like just outlined have undeniably played a role in engendering their quotative uses; foremost among these are imitative meanings (see Section 4.1), backed up by cross-linguistic parallels even in genetically unrelated languages (for overviews see Foolen 2001, 2007; Güldemann 2002, 2008; Meyerhoff 2002; Schourup 1983; Buchstaller and Van Alphen, this volume; for discussion of specific examples see also Coppen and Foolen, this volume; Hasund et al., this volume). However, an essential point on which I believe the standard account needs reconsidering is its focus on the lexical items go and like in isolation. As argued among others by Traugott (2003) and the contributors to Bergs and Diewald (2008), linguistic change tends to take place “in highly constrained pragmatic and morphosyntactic contexts” (Traugott 2003: 645), affecting not just lexemes but, in most cases, morphosyntactic strings and thus larger constructional units. In Section 2, I briefly state the problems resulting from treating go and like as lexemes undergoing (or having undergone) change in isolation. In Section 3, I summarize the model of syntagmatic structure proposed in Vandelanotte (2009: Ch. 2) for direct speech and thought constructions in English and extend it to quotative go and be like constructions. Section 4 shows how this model affords a better insight into the precise locus of change – which is the clause (e.g., everyone was going or I was like in 1) rather than the lexeme (go or like) – and spells out a construction grammar interpretation of the initial linguistic innovation. Section 5 revisits the question of grammaticalization (cf. Noël 2007) – understood variably as the creation of new grammatical constructions or the assigning of “increasingly grammatical, operator-like function” (Traugott 2003: 645) – and Section 6 briefly considers current and future developments in the usage patterns associated with quotative go and be like. Section 7 summarizes the argument. 2. Problems with item-based accounts Early reports on the use of go to introduce direct speech (Butters 1980; Schourup 1982) comment on the ungrammaticality of (2):

(2) *What did he go?

(Butters 1980: 305; Schourup 1982: 149)

Butters saw this as an extension which had not happened yet but “[might] come in due time” (1980: 305), clearly on the basis of an expectation that go was acquiring the meaning of the verb say, which does allow this type of usage (What did he say?). While Schourup (1982: 149) adopts Butters’ analysis of go as having acquired a ‘say’ sense, he is more circumspect and attributes the ungrammaticality of (2) to the fact that in it, go does not introduce a direct quote. This comment is on the

 Lieven Vandelanotte

right track, but in my view the fundamental point is that the locus of change is not a verb but a clause. The question is not so much whether the verb go can acquire a ‘say’ sense, but rather whether clauses such as I went as a whole can acquire the meaning of quoting clauses such as I said. I will return to this point below. Similar comments apply to Meehan’s (1991) and Romaine and Lange’s (1991) analysis of quotative like as an example of grammaticalization. Meehan (1991) traces the history of like in OED attestations, charting its meaning developments from a core sense of similarity into focuser and quotative uses, with intermediate steps. As direct precursor to quotative uses, she sees the conjunction sense ‘as if ’ in an example such as Like you could afford the house I want. Romaine and Lange’s (1991: 257–265) analysis applies Traugott’s (1982) grammaticalization path from propositional to textual to expressive (‘interpersonal’) meaning to the case of like, with preposition like taking nominal complements as the propositional starting point, conjunction like taking sentential complements as the textual extension thereof (as in Winston tastes good like a cigarette should), and discourse marker uses as the expressive, syntactically free endpoint. Quotative like is seen as a “further specialization within the textual component” (Romaine and Lange 1991: 261), from conjunction to quotative complementizer, with “a dummy verb be” to complete the syntactic frame. While the developments in the history of like of course deserve attention as they are instrumental in the change from similarity to quotative uses, the analysis of like as a complementizer, as advocated by Romaine and Lange (1991), is problematic in a number of respects. Firstly, while Old English allowed the sporadic use of complementizer Þæt even for direct speech or thought (Rissanen 1991: 274), present-day English normally restricts the use of complementizers to indirect speech or thought.3 Indeed, the deictic and expressive features of typical quotative go and be like constructions show all the hallmarks of direct speech or thought,4 as shown in Example (1) by the use of the discourse marker Oh and the second person pronoun you (instead of the first person pronoun one would have expected to find in indirect speech). A second, related problem concerns the English indirect speech complementizers that, if and whether, all of which involve restrictions as to the type of complement that can follow. For instance, that does not easily tolerate discourse markers and interjections (3b) (cf. Wilkins 1992) which be like has no 3. Particularly in journalism, mixed forms can be found where partial direct quotes occur in what otherwise looks like indirect speech, as in Schuelke’s (1958: 90) newspaper example Dr. Loper said last night he wished ‘to express my gratitude to all the Stanford students for all the time spent’ in the search for his small son. For discussion and references, see Vandelanotte (2009: 6). 4. Recent extensions into the indirect realm cannot be excluded, as will be illustrated in Section 6 below.

Quotative go and be like 



problems with (3a), and whether does not license interrogative complements (4b), again as opposed to be like (4a):5 (3) a. And then I got hit in the foot and I was like “wow, this is real”.  (COCA, spoken component 1999) ? b. ...and I exclaimed that wow, this is real. (4) a. She was like “Did you tell him that?” (Internet)6 b. *... She asked whether did you tell him that? A further related observation is that, unlike established complementizers, like, together with its putative complement, does not form a recognizable unit in the syntagmatic structure of the utterance. Thus, whereas a that-complement, for instance, can be topicalized (5a) or form the focus of a question (6a), the same is not true of a putative like-complement (5b) (6b): (5) a. That I cannot leave it there, that’s what I said. b. *Like “I cannot leave it there”, that’s what I was. (6) a. What did you say? – That I cannot leave it there. b. *What were you? – Like “I cannot leave it there”. If like lacks all these properties which established complementizers possess, its status as a complementizer is difficult to accept. Indeed, it will be argued in the next section that instead of linking two clauses, like forms part of the be like clause in the structural assembly of the whole construction. Just as it was argued above that go on its own is not the relevant locus of change, so I claim that like should not be considered as such a locus in isolation. In other words, go in its quotative uses does not mean say, and like in quotative be like constructions is not a complementizer. 3. Solving the puzzle: An interclausal type of complementation In this section, I summarize the general model of the syntagmatic structure of reported speech constructions in present-day English proposed in Vandelanotte (2009: Ch. 2) and show how this model can be applied to quotative go and be like constructions (cf. Vandelanotte and Davidse 2009). The starting point for this 5. As a reviewer correctly points out, the real constraint is that the complementizers do not license direct speech (notwithstanding the exception alluded to in note 3). The point of Examples (3–4) is merely to suggest that Romaine and Lange’s complementizer analysis of like creates more problems than it solves: if like were a complementizer, then why does it not behave like bona fide complementizers such as that or whether? 6. http://movies.about.com/od/breach/a/breach020207_3.htm

 Lieven Vandelanotte

alternative proposal is the observation that a traditional verbal complementation analysis which most analysts assume (e.g. Cappelen and Lepore 2007: 131; Clark and Gerrig 1990: 771; Noonan 2007; Quirk et al. 1985; Ransom 1986) runs into serious problems on closer inspection. 3.1

Problems in the verbal complementation analysis

According to the traditional analysis, the quoted material forms the direct object of the reporting verb, and the reporting clause together with its complement is analysed as transitive. However, as Hopper and Thompson (1980) have argued, we should allow for different degrees of transitivity. Reported clauses, then, do not appear as very strong examples of direct objecthood. For instance, whereas prototypically direct objects are affected by the process in some way, a reported clause is not an affected ‘undergoer’ of a verbal process. This feature can be illustrated by means of passivization, which is unmarked for strongly transitive clauses (e.g., he hit her/she was hit by him) but very marked for reporting constructions (e.g., he said “I love you”/ ?“I love you” was said by him). Indeed, with a weakly transitive predicate like say, making the quoted material the subject of a passive is only possible in highly constrained pragmatic contexts where the correct or contested attribution or specific wording of quotes is at issue, as in (7). With the non-transitive verbs in the quotative go and be like structures in (1), passivization looks highly unsuccessful (8).

(7) “Play it again Sam” was actually said by Bobby Fischer when he beat Reshevsky in a Sicilian Najdorf. (Internet)7 (8) *“Oh you’ve got no chance of getting it out of there may as well just leave it” was being gone by everyone and “I cannot leave it there I cannot leave it” was been like by me ‘cos it was crying all the time it was so sad.

Further confirmation of the deviant behaviour of reported clauses compared to bona fide direct objects comes from word order patterning. Whereas true direct objects can only occur before subject and verb in very marked contexts, for instance with contrastive stress (e.g., all this I can still remember), sentence-initial quotes are not very marked (in the written form) and are not associated with contrastive stress (e.g., “The war has educated many of us,” he reflected, Cobuild). What is more, inversion is possible following sentence-initial direct quotes, even if this is mainly restricted to literary contexts (see Collins and Branigan 1997; Green 1980: 590–594). Most transitive clauses, however, do not allow a clause-initial

7.

http://www.chessgames.com/perl/chessgame?gid=1233918

Quotative go and be like 



direct object. Munro (1982: 312) illustrated this difference with the examples “What’s up?” said John as against *Linguistics likes Mary. Whereas a weakly transitive verb like say can, again in restricted contexts, take a highly schematic nominal complement (he said it, they said a few words), perhaps the most serious objection against the verbal complementation analysis is that indisputably intransitive verbs can also be used in quotative constructions. Verbs such as insist or reflect are cases in point, as are many ‘behavioural’ verbs (Halliday 1994: 139–140) used in literary discourse as more colourful devices for introducing speech or thought representations (e.g. nod, smile, smirk, sigh, yawn, etc.); compare *He sighed it to “No,” he sighed. The verbs centrally involved in the more recent quotatives discussed in this paper are further cases in point, with go being an intransitive motion verb and be a copular verb (compare *He went it, *I was like a few words). I conclude from this brief discussion that an analysis of English quotative constructions in terms of verbal complementation, which imputes transitivity onto the verbs used in them and treats their reported clauses as direct objects, is problematic.8 I agree with Munro (1982) that, while the degree of transitivity of the verbs used in quotation may vary, quoting clauses in the syntax of quotation are intransitive. The next subsection proposes an alternative line of analysis which circumvents the problems surveyed above. 3.2

An alternative proposal: Interclausal complementation

The alternative which I propose here is to adopt an interclausal analysis, variants of which have been put forward by Halliday (1994: Ch. 7) and McGregor (1997: Ch. 6), both of whom argue that the relevant syntagmatic relation holds between the reporting clause (as a whole) and the reported clause, rather than between the so-called reporting verb and the reported clause. That the fundamental syntagmatic split in quoting constructions is between reporting and reported clause is suggested by various observations. Typographical conventions such as commas, colons or semicolons separating the two, or quotation marks singling out the reported clause, have gestural and prosodic counterparts in speech, namely quoting gestures and a prosodically less prominent pronunciation of reporting clauses (e.g., Halliday 1994: 250; Klewitz and CouperKuhlen 1999). Likewise, the very fact that reporting clauses as a whole can ‘move 8. This contrasts in interesting ways with so-called factive constructions (featuring predicates such as accept or forget), which can be analysed as transitive clauses with direct objects. For discussion, see Davidse (1994), Vandelanotte (2009: 26–31) and Vandelanotte and Davidse (2009: 783–784).

 Lieven Vandelanotte

around’ in the encompassing quoting construction, adopting initial, medial or final position in the sentence, suggests their unithood (cf. Longacre 2007: 387–388). One might add to this the fact that reporting clauses, as a whole, can ‘alternate’ with zero, resulting in the ‘unframed’ zero-quotative (Mathis and Yule 1994), providing that sufficient contextual cues as to the reportative nature of the reported clause are present (as when a description of a mental state or of behaviour accompanying speech is followed by an ‘unframed’ quote: He was furious. “I’m going to get all of you fired!”). Crucially, quoting constructions involve two deictic centres, the current speaker’s and the represented speaker’s, and the former is construed by the reporting clause as a whole, and not just by the reporting verb. Indeed, the reporting clause as a whole describes the speech act, the content of which is given in the reported clause. By the same token, the whole reporting clause tends to construe the deictic ‘frame’ (McGregor 1997: 254–255) containing the spatial, temporal and personal reference points with regard to which the quote is interpreted. The way in which I conceive of the syntagmatic relationship between reporting and reported clause differs from that proposed by Halliday (1994), who analyses direct speech and thought in terms of parataxis, the linking of clauses of equal status which can potentially stand on their own.9 In his review of Halliday (1985), Hudson (1986: 797) already pointed out that there is an undeniable asymmetry between reporting and reported clause: while the former may interrupt the latter (e.g., “I’m not afraid of you”, he said, “and I will kick your ass.”), the reverse is not true (e.g., *He, “I’m not afraid of you and I will kick your ass,” said). More fundamentally, the requirement that both components in a paratactic complex can stand on their own cannot be applied successfully to direct speech and thought constructions. While the reported clause can occur on its own given sufficient contextual cues (see above), the reporting clause is grammatically incomplete without its quote, as in *Brutus said (Hudson 1986: 797), *everybody was going, or *I was like.10 In order to incorporate this asymmetry resulting from the reporting clause’s ‘incompleteness’, I propose to adopt Langacker’s (1987: Ch. 8) model of complementation relations which starts precisely from the observation that when component structures combine to form composite constructions they typically manifest “substantial asymmetry” (1987: 300), with one component being conceptually 9. It also differs from McGregor’s (1990, 1994, 1997: Ch. 6) approach in which no relation of either constituency or dependency is argued to hold between reporting and reported clause but rather a ‘conjugational’ or ‘interpersonal’ relation of framing or perspective-taking. Here I will argue for a specific type of dependency relation. See Vandelanotte (2009: 40) for discussion. 10. Clauses featuring intransitive verbs such as He reflected or She insisted can occur on their own but are not pragmatically equivalent to reporting clauses; remove the quote and you remove the quotative interpretation.



Quotative go and be like 

more dependent and the other more autonomous. Technically, the more dependent substructure is said to have a salient “elaboration site” which is elaborated by the more autonomous substructure. For instance, in an ordinary transitive clause like he hit her, the verb hit has two ‘elaboration sites’, because the predicate makes salient schematic reference to a person hitting and a person (or thing) being hit. The primary elaboration site (referred to as ‘trajector’ in Cognitive Grammar) is elaborated by he; the secondary elaboration site (the ‘landmark’) by her. Applying this to quoting constructions, I submit that the reporting clause is conceptually dependent on the reported clause, which the reporting clause needs for its completion. Reported clauses are relatively more autonomous, as they can appear in the right context on their own, or take part in other constructions, for instance as the subject of a copular clause (e.g., “I love you” is such a simple yet difficult thing to say). Apart from conceptual dependence versus autonomy, complementation (and modification) relations are characterized by Langacker (1987: Ch. 8) in terms of a second dimension, that of ‘profile determinacy’, or in more traditional terms, headhood. The head of a composite construction is defined as that component structure which designates the same kind of thing or event as the composite construction (cf. Van Langendonck 1994). In quoting constructions, this is the reporting clause; an example such as He said, “I love you” designates first and foremost a saying event, not a loving event.11 Uniting the two parameters of conceptual dependence and headhood leads us to the conclusion that the reporting clause is a conceptually dependent head, whereas the reported clause is a conceptually autonomous complement. What is unusual about this analysis is not the application of Langacker’s model per se, but rather its application to an interclausal relation (cf. Vandelanotte and Davidse 2009: 790–791), thus deviating from Langacker’s (1997) own assumption of verbal rather than clausal complementation for quoting constructions. The analysis can be extended to quotative go and be like clauses, and has the advantage of neither imputing transitivity nor a reporting verb sense onto these verbs. What is elaborated by the complement is not an “elaboration site” of the verb (nor of like, as per Section 2 above), but rather of the clause (everybody was going, I was like, etc.). Adopting Croft’s (2001: 274 Fig. 7.1) simplified representation of conceptual dependence relations, we can represent the resulting analysis for the examples with go and be like in (1) as in Figures (1) and (2) respectively. In these figures ‘e-site’ is the abbreviation for ‘elaboration site’ and the arrow indicates the relation of elaboration. 11. This situation is different with ‘subjectified’ uses of reported speech constructions, as with hedging uses of I think or I guess; for an analysis, see Vandelanotte (2006, 2009: Ch. 8).

 Lieven Vandelanotte

Everybody was going

e-site

"Oh you've got no chance …"

Head

Complement

Figure 1.  Conceptual dependence analysis of a quotative go construction

I was like Head

e-site

"I cannot leave it there" Complement

Figure 2.  Conceptual dependence analysis of a quotative be like construction

In the next section, I show how this interclausal model of the syntagmatic structure of quotative go and be like constructions helps to better understand the initial innovation which, as argued in Section 2, took place not so much at the level of the isolated lexemes go and like, but at the level of the go- and be like-clauses. 4. Relocating the initial innovation: ‘Imitation clauses’ constructionally apprehended as ‘reporting clauses’ 4.1

The imitative meaning of be like and go clauses

A first step in understanding the analogy allowing go and be like clauses to be understood as reporting clauses consists of recognizing a link between ‘imitation’ and ‘reporting.’ This step is a precondition for a clause such as he was like to be understood, very roughly, as (I will now imitate what) he said. Meanings such as similarity, approximation, comparison, exemplification and imprecision are all highlighted in Meehan’s (1991) diachronic overview of like’s semantic development, reinterpreted synchronically by Buchstaller (2002, 2004) in terms of a network with prototypical similarity/comparison/approximation meanings extending towards quotative, focus, epistemic/pragmatic hedging and filler meanings. With regard to go, Butters (1980) was the first to suggest that the onomatopoeic (and hence imitative) use of go in go bang, clatter, cluck, etc. is at the origin of its quotative use. This onomatopoeic use, in which non-humanly produced sounds are performed, dates back at least to the eighteenth century (the oldest example recorded in the OED dating from 1791) and is itself linked metonymically to the use of go with bells, clocks, watches and the like, which produce a sound when they move (or ‘go’). The chronology of the OED examples suggests that the strictly onomatopoeic use (go bang, clatter and the like) is followed first by examples in which people utter sounds such as ‘Yo-yo-yo-yo-yoe’, ‘chip, chip, chip’



Quotative go and be like 

or ‘lah-de-ah-de-ay’,12 and only later by examples in which whole clauses are quoted with a go-clause introducing the quote. In addition to the metonymic links between bells ‘going’, imitative go and, eventually, fully quotative go, a wide-ranging conceptual metaphor (Lakoff and Johnson 1980) which views a conversation as a journey in which people go from one place to another (e.g., we covered a lot of ground, let’s go back to what you were saying earlier, don’t even go there) may have played a facilitating role in the analogization process from imitation to reporting. Hence, whereas for be like the imitation sense derives fairly transparently from the semantics of like, for go it arises only through association with the onomatopoeic uses illustrated above, and not from some imitative meaning origin for go in isolation. The result is broadly similar, however, with speakers using clauses such as everybody was going or I was like to announce that they are about to give a partial or ‘approximative’ imitation of thought, emotion states or words. This kind of meaning ties in well with influential descriptions of direct quotations, including those of Wierzbicka (1974) and Clark and Gerrig (1990). Wierzbicka (1974) analysed direct quotation as an imaginative act in which the current speaker poses briefly for the represented speaker and ‘dramatically’ re-enacts the former’s speech or thought. Clark and Gerrig (1990: 766–768) view direct quotations as “nonserious actions” in which another’s words are imitated or “depicted” rather than described. Such depictions are necessarily selective in that only non-incidental aspects effectively get to be acted out (e.g., in English newspaper reports of what the Russian president has said, the language used will typically be considered incidental and therefore not be depicted). This view implies that the selective depiction thus given is typically not a ‘literal’, word for word repetition, all the more so since material is often presented as speech or thought without having actually been uttered (as is the case with negated or hypothetical reported discourse, did not say, might have said; see Golato, this volume; Vandelanotte 2009: 118–130). In spite of its obvious appeal, however, Clark and Gerrig’s approach is not without its problems. McGregor (1994: 77–81) has cogently argued that the ‘demonstration-description’ distinction cannot be calqued straightforwardly onto the direct-indirect speech distinction, since both types, so also indirect speech, “demonstrate their referents, the words uttered by speakers, albeit from different 12. Buchstaller (2004: 154–156) specifically proposes that the onomatopoeic words following go in early imitative uses are treated as manner adverbs (following a suggestion by Clark and Gerrig 1990: 772) and that routinization of this go + manner adverbial combination led to entrenchment of this usage. The plausibility of this seems to depend on the frequency with which non-onomatopoeic, ‘prototypical’ manner adverbials (such as slowly) tended to co-occur with go. Another (perhaps more plausible) analog might be formed by uses of go as a copular verb (like be in be like), which, according to the OED, date back to the sixteenth century (go mad, thick, cold, etc.).

 Lieven Vandelanotte

viewpoints” (1994: 81). In other words, there is a fundamental sense in which both direct and indirect speech indicate that the reported material is not to be taken at face value as a direct description of entities and relations or events they are involved in, but rather as a ‘second order’ representation of someone else’s description of ‘first order’ entities and events. Conversely, not only indirect reports but also direct quotes “refer to, and describe some state of affairs, or some world” (McGregor 1994: 81), as suggested by Partee’s (1973: 417) examples of anaphora and ellipsis between direct quotes and the surrounding discourse – as in the case of pronominal anaphora in The sign says, “George Washington slept here,” but I don’t believe he really did (Partee 1973: 417). Similar problems appear when the notion of ‘non-seriousness’, which Clark and Gerrig use in their definition of demonstration, is scrutinized, as argued by both McGregor (1994: 70, 90) and Noh (2000: 56–57). Indirect speech is also a way of saying something without actually asserting it oneself, and so in this sense, is as ‘non-serious’ as direct speech. Conversely, there are various perfectly ‘serious’ uses of direct speech, as in warnings (I’m telling you “Piss off!”: McGregor 1994: 70), denials (I say, “I didn’t do that”: Noh 2000: 57) and pieces of advice (My advice is “Buy now”: Noh 2000: 57). All of this suggests that the demonstration-description divide is too dichotomous (see also Vandelanotte 2009: 128–139); the fundamental point that can and should be upheld is that demonstration is involved in speech and thought representation in general. Returning to the problem of be like and go clauses, the relevant point is that a meaning shift from ‘imitation clause’ to ‘reporting clause’ is plausible thanks to a metonymic link between a more general human skill, namely imitation (or ‘demonstration’) by any means (as in Clark and Gerrig’s 1990: 765 example of imitating someone’s limp or tennis serve or the swinging of a pendulum), and a more specific subtype of that skill, namely ‘imitating’ someone’s speech or thought. Crucially, by locating the change at the clausal level, we have avoided conferring an implausible verb sense of speech or thought onto verbs such as go or complex predicates such as be like. 4.2

A constructional interpretation of the imitation-reporting analogy

The avoidance of implausible verb senses is among the reasons discussed by Goldberg (1995: 9–10) for favouring a construction grammar approach to language (e.g., Croft 2001; Fillmore 1988; Goldberg 1995, 2006; Östman and Fried 2005; see also Spronck, this volume). While there are different flavours of construction grammar – some more “radical” and others more “vanilla” (Croft 2005) – they all agree that constructions should be recognized at all levels of grammatical analysis (from concrete to abstract, from small to large), and should be conceived of as conventional form-meaning pairings.



Quotative go and be like 

While hardly new, the idea that constructions themselves carry meaning (e.g., Bolinger 1968; Davies 1979; Haas 1954; Halliday 1973) provides an interesting backdrop against which to interpret linguistic innovation. With respect to one much-discussed example, Sam sneezed the napkin off the table, Goldberg’s (1995) account holds that the creative occurrence of a non-motion verb sneeze in the caused motion construction is made possible precisely because the construction itself carries meaning. While Langacker (2005) agrees to an extent, he justifiedly cautions against overstating the case: by locating too much meaning in the caused motion construction and too little in the verb, there is a real danger of losing the sense of unexpected, novel, creative usage instantiated in the sneeze example. Langacker refines Goldberg’s account by stressing that assembling components into a larger complex structure always requires an active integration of their meaning through perceived correspondences. In the sneeze example, the innovative use of sneeze is possible only because (and to the extent that) sneeze can be brought into correspondence with the constructionally required sense of causing motion. As sneezing involves a forward expulsion of air, such correspondence can be found; one can think of other verbs for which such correspondence is hard or impossible to retrieve, which makes their attempted use in a caused motion construction infelicitous (e.g. see, read, sleep). This line of reasoning can also be applied to quotative go and be like constructions. Just as a novel and very locally perceived correspondence enabled the use of sneeze in the caused motion construction, so could ‘imitation clauses’ involving go and be like be brought into correspondence with ‘reporting clauses’ since the two construe the same types of event but at different levels of generality. The findings in the literature – which suggest a chronology where the innovation starts with non-lexicalized sounds and extends to emotion states and thoughts and eventually also to speech (e.g., Ferrara and Bell 1995: 279; Tagliamonte and Hudson 1999: 152) – support the idea that the innovation first caught on in a bridging context (Evans and Wilkins 2000) of ‘performing’ non-lexicalized sounds, allowing both a straightforward ‘imitation’ interpretation (‘I am acting out what someone/something did’) and a ‘report’ interpretation (‘I am repeating what someone said’). The sneeze example is different from the case of quotative go and be like in at least two respects. Firstly, the correspondence or analogy obtains between the predicate sneeze and caused motion predicates, rather than between whole clauses, as with go and be like quotatives. Secondly, there is a different degree of entrenchment. Clearly sneeze has not become frequently and robustly associated with the caused motion construction, and as such, its use in examples like Sam sneezed the napkin off the table will typically be perceived as remarkable and unusual by the average non-linguist. In contrast, go and be like clauses have become firmly

 Lieven Vandelanotte

entrenched and, at least within the social groups who have fully adopted the constructions, their use is not remarkable. The fact that construction grammar does not recognize the kind of lexiconsyntax dichotomy which traditional approaches assume is an essential part of the account described above. Under a strict lexicon-syntax partitioning in which lexical items are slotted into abstract rule patterns leaving both unaffected, the lexical items used would have to match the requirements stipulated in the rule. For the innovation discussed here to be possible, this would mean that go and be like would need to have the required speech or thought sense, which brings us back to the problems discussed in Section 2. Construction grammar, along with other broadly ‘functional’ approaches to language, rejects this dichotomy and, instead, works with a continuum of constructions, where more complex constructions involve an active integration or mutual ‘accommodation’ (Langacker 1987: 75–76) of component structures. As Langacker (1987: 76) points out, “[a]ccommodation is one of the primary forces driving the growth of schematic networks”. This can be illustrated by even a simple case, such as the meaning of the verb run, which is adjusted at least in part when construed with a human subject as opposed to a fourlegged animal subject, since the two cases involve slightly different types of movement.

4.3

I’m like, he went, this is me in a network of constructions

As was stressed in the introduction, the treatment of quotative go and be like constructions together should not be taken to imply that they are in all respects similar. Construction grammars conceive of the inventory of constructions in a language as a taxonomy going from concrete usage events all the way up to abstract constructions, with intermediate steps. If we put the level of individual usage events to one side, part of the relevant taxonomy would see the quotative go construction and the quotative be like construction as separate ‘micro’-constructions at one level, yet belonging together to what Vandelanotte and Davidse (2009) have called the “Innovative Intransitive Quotative” (IIQ) ‘meso’-construction at a higher level. This IIQ construction, together with ‘canonical’ direct speech and thought constructions (involving verbs of speech and thought in their reporting clauses), would then form a broader category of the direct speech/thought ‘macro’-construction at a higher level again.13 13. The terminology to refer to subsequent levels of constructional schemas is borrowed from Traugott (2007: 525), who adds “constructs” as the lowest level, namely that of concrete instances of micro-constructions.



Quotative go and be like 

The more recent innovation of this is + speaker – present in inner-city London and described by Fox (this volume) – is not covered by the analysis presented here and so is not part of the same ‘IIQ’ meso-construction. As one reviewer correctly suggested, this recent phenomenon is reminiscent of the historical emergence of canonical quotation constructions, where a cataphorically referring demonstrative pronoun that ‘pointed forward’ to the complement clause in structures like He geseah þæt. Hit wæs god ‘He saw that. It was good’ (Mitchell 1985: 14) or þa cwæð se Hælend þæt ðu segst ‘And Jesus said unto him, Thou sayest’ (Matt(WSCp) 27: 11 cited in Mitchell 1985: 7; the gloss given is the King James Bible version).14 This deictic demonstrative þæt, which is generally agreed to have given rise to presentday complementizer that (e.g. Gorrell 1895; Harris and Campbell 1995: 287–289; Hopper and Traugott 2003: 190–194; Mitchell 1985: 13–14), bears a clear resemblance to the ‘new’ use of cataphoric this in the newcomer This is me. Both the historical demonstrative origin of complementizer that and this recent use of this have cross-linguistic parallels, as deictic functions have been identified in diachronic pathways leading to quotative complementizers (e.g., Güldemann 2002 on Shona ti; Meyerhoff 2002 on Bislama olsem). In spite of these links, however, if complementizer that occurs in canonical present-day English reported-speech constructions, it is a complementizer introducing the reported clause complement and no longer refers deictically, whereas in the new this is + pronoun constructions it does so cataphorically. In the broadly functional approach adopted here (e.g., Halliday 1994 [1985], McGregor 1997), phoric meanings are not captured in terms of constituency and dependency relations, but are viewed instead as textual relations weaving constituents and dependents into coherent and cohesive text. For this reason I do not consider the this is + pronoun construction as part of the IIQ construction as described above. Instead, it could be viewed as a separate meso-construction, alongside the Innovative Intransitive Quotative and canonical (he said, she thought-type) direct speech/thought constructions (DST), as represented in Figure 3. Needless to say, this figure is incomplete in different ways. It does not show, for instance, different subconstructions within the go and be like families (such as go like, be all, be all like, etc.); nor does it show the relations these constructions maintain with, for example, non-quotative go, be like or this is + pronoun... constructions. Much more is involved in this rich corner of English grammar than is represented in Figure 3, but it does at least give an idea of the kind of network relations assumed here.

14. As the second person pronoun shows, the example from Matthew 27 involves direct speech, which could also be introduced by þæt in Old English.

 Lieven Vandelanotte Macro-construction

DST

Meso-constructions Micro-constructions Constructs

DSTverbal/cognition This is PRON

IIQ GoQuot

Be LikeQuot

DSTverbal

DSTcognition

concrete usage events

Figure 3.  Partial network of English direct speech and thought constructions

5. Revisiting the question of grammaticalization In this section, I first consider the question to what extent, or in what sense, the initial innovation – using go and be like clauses as intransitive quotative constructions – can be captured in terms of ‘grammaticalization’, as has usually been done (Section 5.1). In Section 5.2, this discussion is extended to further developments such as be all or go kinda. 5.1

Quotative be like and go: Grammaticalization or constructionalization?

The alternative account of the emergence of quotative go and be like constructions sketched here raises the question to what extent grammaticalization is involved on this account. A commonly used definition of grammaticalization is provided by Hopper and Traugott (2003: 18), who view it as “the change whereby lexical items and constructions come in certain linguistic contexts to serve grammatical functions, and once grammaticalized, continue to develop new grammatical functions”. In Romaine and Lange’s (1991) model of the evolution of like, several clear symptoms of grammaticalization can be discerned: like’s grammatical category changes from preposition to conjunction (decategorialization and recategorialization; cf. Hopper 1991) and its initial ‘propositional’ meaning weakens and gives way to a more grammatical meaning (semantic ‘attrition’ or desemanticization; cf. Lehmann 1985, 1995), which is also more subjective on the cline from nonsubjective (propositional) to subjective (speaker-centred) and intersubjective (addressee-centred; cf. Traugott 2010; Traugott and Dasher 2002). While subjectification is “typical of semantic change in general” and thus “not limited to grammaticalization” (Traugott and Dasher 2002: 89–90), decategorialization and some kind of shift from lexical to grammatical, or from grammatical to ‘more’ grammatical, as in the case of increased pragmatic meaning (Traugott 1988), are



Quotative go and be like 

typically treated at least implicitly as key features of grammaticalization, with more advanced cases in morphosyntax sometimes additionally displaying such characteristics as phonetic reduction and cliticization (see e.g. Lehmann 1985). Thus, for instance, intensifying uses of pure (Macaulay 2006; Vandewinkel 2010; Vandewinkel and Davidse 2008) illustrate such features as decategorialization (from adjective to adverb) and a shift from lexical (‘unmixed’) to more pragmatic or grammatical meanings (pure functioning as intensifier, indexing speaker involvement, or as submodifier), with attendant increase in the syntactic contexts in which pure is used. In the realm of speech and thought reporting, grammaticalized and subjectified uses of such clauses as I think, I guess, I mean and I say have also been diagnosed in terms of specific grammaticalization features such as decategorialization, desemanticization, ossification, pragmatic strengthening and the like (see e.g. Brinton 2005: 291–293; Vandelanotte 2009: 283–294). Compared to such treatments, then, the use of the term grammaticalization in much of the literature on be like, go, and also this is + speaker quotatives seems to suggest the implicit adoption of the Romaine and Lange (1991) view, which I argued against above. Indeed, the account that was given in the preceding sections would mean that the change does not really involve grammaticalization in a strict sense. While one could maintain that a meaning shift from ‘imitation clauses’ to ‘reporting clauses’ constitutes a move towards more ‘grammatical’ meaning, it is not clear what the foundation for such an interpretation would ultimately be. More importantly perhaps, the meaning shift does not involve decategorialization and recategorialization in any real sense. In fact, the clausal status attributed to clauses such as everybody was going or I was like constitutes one of the hallmarks of the analysis defended here, and is essential to allow for the spread, particularly of the quotative be like construction, from initial uses in first person present tense to different person and tense forms. This broadening of the grammatical paradigm also shows that any treatment which might seek to view quotative go and be like constructions as ossifying into formulaic phrases (in the way that I think or I guess have in most of their uses) would be misguided. Instead I interpret this broadening, along with the simultaneous extension from reports of sounds, inner states and emotions to full-fledged speech reports (Ferrara and Bell 1995: 279; Tagliamonte and Hudson 1999: 152), as symptomatic of the constructions’ ongoing analogical integration into the ‘canonical’ direct speech and thought construction. Such integration (or ‘accommodation’ in Langacker’s 2005 terms) constitutes a process which need not proceed at the same rate for quotative go and be like, nor go on to completion, as I will suggest briefly in Section 7 below. What seems to confuse the issue of the potential involvement of grammaticalization is the different uses to which the term has been put, as argued in detail by

 Lieven Vandelanotte

Noël (2007). In his view, the confusion has come about as a result of the increased interest in constructions within grammaticalization, which has led some to define grammaticalization as the creation of new constructions. Noël (2007) goes on to argue cogently that the two developments should be kept apart. On the one hand there is the primary emergence of constructions, dubbed ‘constructionalization’ or ‘schematization’, and on the other the change of a construction’s meaning towards the grammatical end of the lexical-grammatical cline, dubbed (for want of a better term) ‘grammaticalization’.15 While grammaticalization implies constructionalization, the reverse is not the case: once a new construction has been formed, it need not be put onto a route towards increased grammatical meanings. Constructionalization in this sense of the initial formation of a construction is obviously involved in the coming into being of the quotative go and be like constructions, also in my interpretation of their emergence. Through a process of analogical extension,16 imitation clauses are apprehended as reporting clauses and thus come to be added to the inventory of reporting clauses used in the direct speech or thought construction in English, fitting in to a broader taxonomy. Once formed, the constructions have undergone a broadening of the grammatical paradigm in terms of person, tense forms and type of reported material (sounds, emotion states, thought, and speech), and have spread to different English-speaking regions around the globe. While these developments effectively increase the conformity of the new constructions to the ‘prototype’ or ‘macro-construction’ (cf. Traugott 2007) of English direct speech and thought, it is not clear how this increases the grammatical meaning of the constructions. Rather, they can be viewed as a logical (partial) fulfilment of the analogical trajectory on which the quotative go and be like constructions were put.

5.2

Extensions of be like and go quotatives

Different variants of quotative be like and (to a lesser extent) go constructions likewise need not be analysed in terms of grammaticalization in its strict sense. Use of bare be as in (9) below, for instance – reported sporadically in very small quantities in Tagliamonte and Hudson (1999), Macaulay (2001) and Tagliamonte and D’Arcy (2004) – can be viewed as a clipped form of be like in which like has been dropped

15. Strictly speaking, since lexical items are also ‘constructions’ in construction grammar, they are also ‘grammatical’. 16. Note in this connection that Fischer (2010) has recently argued that, while analogy is a cause as well as a mechanism of change, grammaticalization is only an epiphenomenon.



Quotative go and be like 

abruptly (as argued by Macaulay 2001: 16 for the Scottish case),17 and not as the result of phonological reduction (or ‘attrition’; Lehmann 1985) of the kind typically found in more advanced stages of grammaticalization, which would involve a gradual weakening from /bi˜ laIk/ to /bi˜/ via intermediate stages such as /bi˜ laI/ and /bi˜ l/.18 Here again, analogy may play a role, in the sense that bare be quotatives may have taken bare go quotative clauses as a model, an analogization process made plausible by the copular function shared by be and go in some of its uses (e.g., go mad, go sour, etc.). (9) Even in class he’d be “Come here girlfriend.”  (Tagliamonte and D’Arcy 2004: 500) Forms such as be just (Macaulay 2001; Tagliamonte and D’Arcy 2004; Tagliamonte and Hudson 1999), be all (Buchstaller et al. 2010; Rickford et al. 2007; Waksler 2001), and be pure (reported sporadically in Glasgow by Macaulay 2006), illustrated in (10) to (12) below, can be seen as involving additional hedging and boosting when compared to bare be, due to the focusing sense of just and the amplifying or intensifying meaning of all (Buchstaller and Traugott 2006) and pure (Macaulay 2006). (10) And Angela’s just “Did you do anything last night?”  (Tagliamonte and Hudson 1999: 155) (11) I said something funny, and he’s all “Write that down!”  (Waksler 2001: 132) (12) and she’s pure “You got it wrong” (Macaulay 2006: 275) Particularly in combinations with sorta/sort of and kinda/kind of, as in (13) (cf. Aijmer 2002: 186; De Smedt, Brems and Davidse 2007; Margerie 2010), and also, perhaps, in the combination with just and all as in (10) and (11), an analogical extension from like to other particles with hedging functions seems to have 17. Interestingly, Buchstaller (p.c.) reports occurrences of bare be with quotative function in data from the 1960s, which might suggest that quotative bare be at some point constituted an independent development, without any involvement of like and perhaps more akin to the emergence of quotative go-clauses (the link then being that both be and, in some of its uses, go are copular verbs). If borne out, however, this hypothesis need not exclude the possibility that synchronically bare be (re-)emerges as “like-less be like”. 18. A reviewer pointed to Drager’s (2006, 2009) work suggesting that like demonstrates phonological attrition (see also Hasund, Opsahl and Svennevig, this volume). The picture emerging from her data from a girls’ high school in New Zealand is fairly complex, but involves variable realizations of the vowel in like as diphthong or monophthong and of realization or dropping of the final /k/. The data so far do not seem to point to a clear directional change, but rather to variation on this point between different social groups among the girls. In any case, variation of this kind does not explain the occurrence of bare be as in Example (9).

 Lieven Vandelanotte

occurred. Note that this is not a case of paradigmaticization (Lehmann 1985), which is commonly understood in grammaticalization as the process by which a newly grammaticalized form enters into an existing paradigm (as with grammaticalized I think entering, in its epistemic uses, the paradigm of modal adverbs like probably or possibly, and in its textual, discourse-organizational uses that of discourse markers you know or well; see Vandelanotte 2009: 285–294). Rather, here we have a case of a new construction which, once established, extends its paradigm by analogically recruiting suitable substructures (like, kinda, sorta, etc.). Concurrently doubled forms emerge, as noted for example by Macaulay (2001) and Rickford et al. (2007) and illustrated in (14) to (16): (13) When he got hurt before the playoffs, everyone in the room was kinda, ‘O.K., what’s Savvy going to do? Is he going to sit out or come back?’   (COCA, magazine component 2009) (14) I mean, I was sort of like, “Wow, that’s it?”  (COCA, spoken component 1999) (15) “I think she was kinda like, ‘Put it in the closet and don’t even pay attention to it,’ ” says a friend. (COCA, magazine component 2004) (16) And she’s all like, “Well you HAVE to. Are you allergic to them?” And I’m like, “No. They just make me nauseous.” (Rickford et al. 2007: 21) Quotative uses of go combining with like have also been observed (e.g. Buchstaller 2004; Macaulay 2001; Winters 2002), as in (17) below, with the same kinds of extensions with other hedges also being attested, such as kinda in (18). Some kind of blend between be like and go seems to have taken place, perhaps facilitated further by a perceived structural parallelism between copular be + like and copular go + like. (17) I closed the door. I locked the door. At some point I remember peeking out the window in the living room. And it was just a lot of back and forth because I went like “close the door, go back to Brian, peek out the window, go back to Brian”. (COCA, spoken component 2008) (18) Then I read the chapter of Naruto where cripple Nagato was shown. I went kinda “WHAT THE FUCK THAT WAS MY IDEA ...” (Internet)19 While the kinds of extensions exemplified in (9) to (18) are, to date, not very frequent, they do attest to the rude health of the family of quotative be like and go constructions. Not only have the ‘innovative intransitive quotatives’ as a whole become robustly entrenched in the grammar of English quotation, but variations on them have been emerging, producing sometimes very local variants as with 19. http://www.fandom.com/forums/showthread.php?p=1891915



Quotative go and be like 

Glaswegian be pure (Macaulay 2006) and sometimes short-lived fads as in the case of be all, for which Rickford et al. (2007) and Buchstaller et al. (2010) demonstrate a quick rise-fall pattern. The case of quotation becoming a prominent site for innovation and renewal among adolescents leading the changes invites comparison with intensifiers going in and out of fashion in fairly quick succession (e.g. Barnfield and Buchstaller 2010; Macaulay 2006; Paradis 2000). The point made by Paradis (2000: 147) that “teenagers tend to exaggerate rather than modulate” helps explain why intensifiers, with their lexical-emotional content, constitute a prominent pool of resources for expressing identity. The case of quotatives is presumably more diffuse, as it is only in broader discourse functions, such as the highlighting of dramatic peaks in performance discussed by Fox (this volume), that expressive and emotive meanings come to the fore (see also e.g. Mayes 1990; Shuman 1993).20 In the next section, I briefly consider some grammatical aspects of the current state and future developments of be like and go quotatives. 6. Where do we go from here? Current and future developments While the extent to which quotative be like and go constructions have become available to speakers in different regions, age brackets and social groups is a question for sociolinguistic study (see Section 1), for those speakers who do use the forms a further question pertains to the extent to which the innovation behaves like ‘canonical’ direct speech or thought constructions. There are several things to note in this regard. First, be like is used more frequently with thoughts and inner monologue compared to go, which still prefers to combine with speech (Buchstaller 2008: 24–25).21 While this mirrors differences to some extent among ordinary reporting clauses (e.g. she said versus she thought), a second difference is perhaps more telling, as it relates to a more general property of canonical direct speech or thought. Whereas the quotative go construction, which arguably has a longer 20. On an anecdotal level, the so-called ‘Valley Girl’ sketches in the first series (2004) of The Catherine Tate Show, a popular BBC television sketch show, suggest that a wider, non-linguistic audience is sensitive to the (over-)use of be like to focus on and dramatize emotions. An example of such a sketch was made available by the BBC Worldwide YouTube channel at http://www. youtube.com/watch?v=-fGZtrBeDcQ. 21. In assessing this, it is often hard to tell speech and thought apart. Who is to say, for instance, that Buchstaller’s (2008: 25) example for reported thought I went “oh shit” does not involve speech? Dancygier (2012: Chapter 7) offers a fascinating, subtle analysis of speech versus thought reports in narratives suggesting that the distinction is rarely salient and often misleading. In spite of these reservations, a broad difference in typical usage between be like and go can probably be upheld.

 Lieven Vandelanotte

pedigree (Section 4.1), clearly allows sentence-medial or sentence-final position of the go-clause, as in (19) and (20), this is less easily tolerated with be like (21b). Admittedly, interposed or postposed quotative go clauses are not frequent,22 but the acceptability contrast suggested in the constructed Example (21), while not absolute, does suggest a somewhat higher degree of integration of quotative go compared to quotative be like into the direct speech construction. (19) The green frog blinked and turned to hop off. “Hiccup’” he went, “hiccup, hiccup.” “Try a drink of water, green frog,” Harry called after it.  (COCA, magazine component 2006) (20) “Fuck, fuck, fuck,” he went. Creaking back and forth. (Internet)23 (21) a. “We’ll see about that,” he went. b. ??“We’ll see about that,” he was like. Another feature setting apart quotative be like from canonical reporting clauses was remarked on by Singler (2005), who argued that negated and interrogative say clauses are much more likely to occur than negated or interrogative quotative be like clauses, but who seems confident that the use of the like variant will also spread to these contexts. Using the Corpus of Contemporary American English (Davies 2008-) to find relevant examples (not just with be like but also with go) suggests that the occurrence of the innovative quotatives in contexts of sentential negation is slightly easier to come by, though very infrequent: (22) and (23); for interrogatives I had to turn to a general Internet search to find relevant examples: see (24) and (25): (22) There are times when I have seen things that I’ve done that I did laugh at. But I don’t go, “Oh, my god, is that fabulous?” I, I don’t do that, no.  (COCA, spoken component 2007) (23) “I’m not like, ‘I’m the man and I have to do everything,’” Donovan said, “but I have to be at my best.” (COCA, newspaper component 2006) (24) Did she intentionally pick the most incongruous goddamn shoes in the universe? Did she go, “wow, what would look absolutely worst in the fucking world with my giant huggable shit-sack... oh, how about these patentleather Nazi gardening boots!” (Internet)24

22. In written genres, this may be due in part to some competition from the established form go on ‘continue’, as in “It’s the mention of Simpson that’s intriguing,” he went on (Collins Wordbanks, written British component: fiction, 1990). 23. http://siblingshot.blogspot.com/2009/08/sno-doz.html 24. http://www.somethingawful.com/d/fashion-swat/wool-swat-german.php

Quotative go and be like 



(25) One time my grandma was flipping through the channels and stopped on a music video for that “song” My Lip Gloss is Poppin. She watched the whole thing then kept flipping like nothing had happened. I want to know what was going through her mind. Was she like What the HELL are people watching these days? or was she like Oh neat. I don’t know. Too bad I can’t ask her. (Internet)25 While Singler’s point is well-taken, it should be borne in mind that negated and interrogative say clauses are in themselves very infrequent and require a particular pragmatic point for them to be used at all. With first and second person subjects, negated and interrogative say reporting clauses respectively are typically used to clarify or ask clarifications about one’s precise intentions (e.g. I’m not saying... ‘I don’t mean...’, are you saying... ‘do you mean...’). With third person subjects, negated or questioned say is used to verify what the precise wording was, or to check the facts when it comes to contested speech (He didn’t say X (but Y), Did he say X?). Considering the approximative meaning origin particularly of be like, which might initially have made be like less suited to insist on or question the precise wordings, the fact that be like quotative constructions occur in negated (23) and interrogative (25) uses is remarkable, and supports the idea that they already behave as more ‘general purpose’ choices than perhaps they are given credit for. Very many verbs which can be used in ‘canonical’ reporting clauses, especially of the more ‘literary’ or ‘descriptive’ kind (ponder, shout, smile, to name but a few) do not normally have these pragmatically specialized uses (consider, e.g. the questionable constructed examples *I did not smile ‘yes’, I smiled ‘no’! and *Did you ponder “That’ll be the life!”?). A final development worth noting concerns direct to indirect quotation. The quotative uses of be like and go have until now been associated almost exclusively with direct quotation, despite Romaine and Lange (1991) arguing that pragmatically be like blurred the boundaries. However, some infrequent inroads into the indirect domain have been reported. In her British English data from around 1994–1995, Buchstaller (2004: 186–188) notes a very low incidence (1.37%) of indirect speech reports with go, absent from her American data. (26) is such a rare example, in which she in she goes and she heard a noise both refer to the participant referred to as Angela. Similarly, in his collection of internet data on be all, Van Elslander (2007) reports a single example of indirect quotation (27), following two short direct quotes. In his example, the context with the preceding direct quote how are you? indicates that the two occurrences of he are coreferential, and thus constitute an indirect rendering of a direct I’m okay.

25. http://oneword.com/2008/12/gloss/full.html

 Lieven Vandelanotte

(26)

and the first thing that went through me mind was, I was telling Mandy was ehhhmmm, when Angela stayed that night, and she goes .hhh she heard a noise downstairs, someone speaking. and I thought, “flipping hell, who is that?” (British English data c. 1994–1995; Buchstaller 2004: 186)

(27) Then he called me tonite, and I’m all: “fine” and “how are you?” and he’s all he’s okay, ... (Internet,26 cited in Van Elslander 2007: 14) It remains to be seen whether the type of usage illustrated in (26) and (27) takes root and expands beyond relatively rare occurrences. 7. Conclusion This paper has reconsidered the locus of change leading to (initially) innovative be like and go quotatives. Rather than proposing a ‘say’ sense for go (Butters 1980), or a quotative complementizer function for like (Romaine and Lange 1991), I have argued that the unit which undergoes change is the ‘imitation’ clause as a whole – everybody was going or I was like in (1) – which comes to be apprehended as a reporting clause. This proposal meshes well with a general model of the syntagmatic structure of the English direct speech or thought construction that views the primary relationship as one of interclausal complementation with the reporting clause as conceptually dependent head vis-à-vis a more autonomous complement. This model was argued to avoid fundamental problems in the standard verbal complementation model, whilst also doing justice to the syntagmatic divide between reporting and reported clause. Whereas particularly Romaine and Lange’s (1991) analysis of like showed clear signs of grammaticalization proper, the alternative analysis proposed here does not invite a grammaticalization interpretation in its strict sense, i.e. as involving a development towards the grammatical end of the lexical-grammatical cline involving core features such as decategorialization. The initial emergence, along with its further developments, of quotative be like and go constructions can, on the other hand, be clearly upheld as an example of constructionalization or ‘schematization’ in the sense of Noël (2007). Whether the English innovative intransitive quotative constructions will go on to become used more naturally and frequently

26. http://msshad.typepad.com/ordinaryms/2007/01/i_made_a_phone_.html

Quotative go and be like 



in indirect speech or thought constructions is one of the main questions to pose in future data collections and analysis. References Aijmer, Karin. 2002. English Discourse Particles: Evidence From a Corpus [Studies in Corpus Linguistics 10]. Amsterdam/Philadelphia: John Benjamins. Barbieri, Federica. 2005. “Quotative use in American English: A corpus-based, cross-register comparison.” Journal of English Linguistics 33: 222–256. Barbieri, Federica. 2007. “Older men and younger women: A corpus-based study of quotative use in American English.” English World-Wide 28: 23–45. Barbieri, Federica. 2009. “Quotative be like in American English: Ephemeral or here to stay?” English World-Wide 30: 68–90. Barnfield, Kate and Buchstaller, Isabelle. 2010. “Intensifiers on Tyneside: Longitudinal developments and new trends.” English World-Wide 31: 252–287. Bergs, Alexander and Diewald, Gabriele. 2008. Constructions and Language Change [Trends in Linguistics. Studies and Monographs 194]. Berlin/New York: Mouton de Gruyter. Blyth, Carl, Recktenwald, Sigrid and Wang, Jenny. 1990. “I’m like, ‘Say what?!’: A new quotative in American oral narrative.” American Speech 65: 215–227. Bolinger, Dwight D. 1968. Aspects of Language. New York: Harcourt, Brace and World. Brinton, Laurel J. 2005. “The development of I (say): Grammaticalization, pragmaticalization, or lexicalization.” In Opening Windows on Texts and Discourses of the Past [Pragmatics and Beyond New Series 134], Janne Skaffari, Matti Peikola, Ruth Carroll, Risto Hiltunen and Brita Wårvik (eds), 279–299. Amsterdam/Philadelphia: John Benjamins. Buchstaller, Isabelle. 2002. “He goes and I’m like: The new quotatives re-visited.” Paper presented at the University of Edinburgh Postgraduate Conference. www.ling.ed.ac.uk/~pgc/ archive/2002/proc02/buchstaller02.pdf. Buchstaller, Isabelle. 2004. The sociolinguistic constraints on the quotative system – British English and US English compared. Unpublished PhD thesis, University of Edinburgh. Buchstaller, Isabelle. 2006. “Diagnostics of age-graded linguistic behaviour: The case of the quotative system.” Journal of Sociolinguistics 10: 3–30. Buchstaller, Isabelle. 2008. “The localization of global linguistic variants.” English World-Wide 29: 15–44. Buchstaller, Isabelle and D’Arcy, Alexandra. 2009. “Localized globalization: A multi-local, multivariate investigation of quotative be like.” Journal of Sociolinguistics 13: 291–331. Buchstaller, Isabelle and Traugott, Elizabeth Closs. 2006. “The lady was al demonyak: historical aspects of adverb all.” English Language and Linguistics 10: 345–370. Buchstaller, Isabelle, Rickford, John R., Traugott, Elizabeth Closs, Wasow, Thomas and Zwicky, Arnold. 2010. “The sociolinguistics of a short-lived innovation: Tracing the development of quotative all across spoken and internet newsgroup data.” Language Variation and Change 22: 191–219. Butters, Ronald R. 1980. “Narrative go ‘say’.” American Speech 55: 304–307. Cappelen, Herman and Lepore, Ernie. 2007. Language Turned On Itself: The Semantics and Pragmatics of Metalinguistic Discourse. Oxford: Oxford University Press.

 Lieven Vandelanotte Clark, Herbert H. and Gerrig, Richard J. 1990. “Quotations as demonstrations.” Language 66: 764–805. Collins, Chris and Branigan, Philip. 1997. “Quotative inversion.” Natural Language and Linguistic Theory 15: 1–41. Collins WordbanksOnline corpus. [http://www.collinslanguage.com/wordbanks/default.aspx]. London: Collins Publishers. Croft, William. 2001. Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford: Oxford University Press. Croft, William. 2005. “Logical and typological arguments for Radical Construction Grammar.” In Construction Grammars: Cognitive Grounding and Theoretical Extensions [Constructional Approaches to Language 3], Jan-Ola Östman and Mirjam Fried (eds), 273–314. Amsterdam/Philadelphia: Benjamins. Cukor-Avila, Patricia. 2002. “She say, she go, she be like: Verbs of quotation over time in African American Vernacular English.” American Speech 77: 3–31. Dailey-O’Cain, Jennifer. 2000. “The sociolinguistic distribution of and attitudes toward focuser like and quotative like.” Journal of Sociolinguistics 4: 60–80. Dancygier, Barbara. 2012. The Language of Stories: A Cognitive Approach. Cambridge: Cambridge University Press. D’Arcy, Alex. 2004. “Contextualizing St. John’s Youth English within the Canadian quotative system.” Journal of English Linguistics 32: 323–345. Davidse, Kristin. 1994. “Fact projection.” In Perspectives on English. Studies in Honour of Professor Emma Vorlat [Orbis Supplementa 2], Keith Carlon, Kristin Davidse, and Brygida Rudzka-Ostyn (eds), 257–284. Leuven: Peeters. Davies, Eirian C. 1979. On the Semantics of Syntax: Mood and Condition in English. London: Croom Helm. Davies, Mark. 2008. “The Corpus of Contemporary American English (COCA): 410+ million words, 1990-present”, (23 September 2009). De Smedt, Liesbeth, Brems, Lieselotte and Davidse, Kristin. 2007. “NP-internal functions and extended uses of the ‘type’ nouns kind, sort and type: Towards a comprehensive, corpus-based description.” In Corpus Linguistics 25 Years On [Language and Computers – Studies in Practical Linguistics 62], Roberta Facchinetti (ed.), 227–257. Amsterdam/New York: Rodopi. Drager, Katie. 2006. “Social categories, grammatical categories, and the likelihood of ‘like’ monophthongisation.” In Proceedings of the 11th Australasian International Conference on Speech Science and Technology, Paul Warren and Catherine I. Watson (eds), 384–387. Auckland: University of Auckland. Drager, Katie. 2009. “Language, stance, and identity at Selwyn Girls’ High.” In Proceedings of the 5th Biennial International Gender and Language Association Conference (IAGALA 5, Wellington 2008), Julia de Bres, Janet Holmes and Meredith Marra (eds), 419–433. Wellington: Victoria University of Wellington. Evans, Nicholas and Wilkins, David. 2000. “In the mind’s ear: The semantic extensions of perception verbs in Australian languages.” Language 76: 546–592. Ferrara, Kathleen and Bell, Barbara. 1995. “Sociolinguistic variation and discourse function of constructed dialogue introducers: The case of be + like.” American Speech 70: 265–290. Fillmore, Charles J. 1988. “The mechanisms of ‘Construction Grammar.’” Berkeley Linguistics Society 14: 35–55. Fischer, Olga. 2010. “An analogical approach to grammaticalization.” In Grammaticalization: Current Views and Issues [Studies in Language Companion Series 119], Katerina Stathi,



Quotative go and be like  Elke Gehweiler, and Ekkehard König (eds), 181–219. Amsterdam/Philadelphia: John Benjamins. Foolen, Ad. 2001. “Marking voices in discourse. Quotation markers in English and other languages.” Paper presented at the 7th International Cognitive Linguistics Conference, UCSB, Santa Barbara. Foolen, Ad. 2007. “New quotative markers and commitment.” Paper presented at the conference ‘The notion of commitment in linguistics’, University of Antwerp, 11–13 January 2007. Goldberg, Adele. 1995. Constructions: A Construction Grammar Approach to Argument Structure [Cognitive Theory of Language and Culture]. Chicago: University of Chicago Press. Goldberg, Adele. 2006. Constructions At Work: The Nature of Generalization in Language. Oxford: Oxford University Press. Gorrell, J. Hendren. 1895. “Indirect discourse in Anglo-Saxon.” Publications of the Modern Language Association of America 10: 342–485. Green, Georgia M. 1980. “Some wherefores of English inversions.” Language 56: 582–601. Güldemann, Tom. 2002. “When ‘say’ is not say: The functional versatility of the Bantu quotative marker ti with special reference to Shona.” In Reported Discourse. A Meeting Ground for Different Linguistic Domains [Typological studies in language 52], Tom Güldemann and Manfred von Roncador (eds), 253–287. Amsterdam/Philadelphia: John Benjamins. Güldemann, Tom. 2008. Quotative Indexes in African Languages: A Synchronic and Diachronic Survey [Empirical Approaches to Language Typology 34]. Berlin/New York: Mouton de Gruyter. Haas, William. 1954. “On defining linguistic units.” Transactions of the Philological Society 53: 54–84. Halliday, M.A.K. 1973. Explorations in the Functions of Language. London: Arnold. Halliday, M.A.K. 1994. An Introduction to Functional Grammar. 2nd ed. [1st ed., 1985]. London: Arnold. Hansen-Thomas, Holly. 2008. “An investigation of innovative quotatives in adolescent Chicana English in Texas.” Intercultural Pragmatics 5: 19–39. Harris, Alice C. and Campbell, Lyle. 1995. Historical Syntax in Cross-Linguistic Perspective. Cambridge: Cambridge University Press. Hopper, Paul J. 1991. “On some principles of grammaticization.” In Approaches to Grammaticalization. Volume I: Focus on Theoretical and Methodological Issues, Elizabeth Closs Traugott and Bernd Heine (eds), 17–35. Amsterdam/Philadelphia: Benjamins. Hopper, Paul J. and Thompson, Sandra A. 1980. “Transitivity in grammar and discourse.” Language 56: 251–299. Hopper, Paul J. and Traugott, Elizabeth Closs. 2003. Grammaticalization [Cambridge Textbooks in Linguistics]. Cambridge: Cambridge University Press. Hudson, Richard. 1986. “Systemic grammar.” Linguistics 24: 791–815. Klewitz, Gabriele and Couper-Kuhlen, Elizabeth. 1999. “Quote-unquote: The role of prosody in the contextualization of reported speech sequences.” Pragmatics 9: 459–485. Lakoff, George. 1987. Women, Fire and Dangerous Things: What Categories Reveal about the Mind. Chicago: University of Chicago Press. Lakoff, George and Johnson, Mark. 1980. Metaphors We Live By. Chicago: University of Chicago Press. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar. Volume I: Theoretical Prerequisites. Stanford: Stanford University Press.

 Lieven Vandelanotte Langacker, Ronald W. 1997. “Constituency, dependency, and conceptual grouping.” Cognitive Linguistics 8: 1–32. Langacker, Ronald W. 2005. “Construction grammars: Cognitive, radical, and less so.” In Cognitive Linguistics: Internal Dynamics and Interdisciplinary Interaction [Cognitive Linguistics Research 32], Francisco J. Ruiz de Mendoza Ibáñez and M. Sandra Peña Cervel (eds), 101–159. Berlin/New York: Mouton de Gruyter. Lehmann, Christian. 1985. “Grammaticalization: Synchronic variation and diachronic change.” Lingua e Stile XX: 303–318. Lehmann, Christian. 1995. Thoughts on Grammaticalization [LINCOM Studies in Theoretical Linguistics 1]. München/Newcastle: LINCOM EUROPA. Longacre, Robert E. 2007. “Sentences as combinations of clauses.” In Language Typology and Syntactic Description. Vol. 2: Complex Constructions, 2nd ed., Timothy Shopen (ed.), 372–420. Cambridge: Cambridge University Press. Macaulay, Ronald. 2001. “You’re like ‘why not?’ The quotative expressions of Glasgow adolescents.” Journal of Sociolinguistics 5: 3–21. Macaulay, Ronald. 2006. “Pure grammaticalization: The development of a teenage intensifier.” Language Variation and Change 18: 267–283. Margerie, Hélène. 2010. “On the rise of (inter)subjective meaning in the grammaticalization of kind of/kinda.” In Subjectification, Intersubjectification and Grammaticalization [Topics in English Linguistics 66], Kristin Davidse, Lieven Vandelanotte and Hubert Cuyckens (eds), 315–346. Berlin/New York: Mouton de Gruyter. Mathis, Terrie and Yule, George. 1994. “Zero quotatives.” Discourse Processes 18: 63–76. Mayes, Patricia. 1990. “Quotation in spoken English.” Studies in Language 14: 325–363. McGregor, William B. 1990. “The metafunctional hypothesis and syntagmatic relations.” Occasional Papers in Systemic Linguistics 4: 5–50. McGregor, William B. 1994. “The grammar of reported speech and thought in Gooniyandi.” Australian Journal of Linguistics 14: 63–92. McGregor, William B. 1997. Semiotic Grammar. Oxford: Clarendon Press. Meehan, Teresa. 1991. “It’s like, ‘What’s happening in the evolution of like?’: A theory of grammaticalization.” Kansas Working Papers in Linguistics 16: 37–51. Meyerhoff, Miriam. 2002. “All the same? The emergence of complementizers in Bislama.” In Reported Discourse. A Meeting Ground for Different Linguistic Domains [Typological studies in language 52], Tom Güldemann and Manfred von Roncador (eds), 341–359. Amsterdam/ Philadelphia: John Benjamins. Mitchell, Bruce. 1985. Old English Syntax. Volume II: Subordination, Independent Elements, and Element Order. Oxford: Clarendon Press. Munro, Pamela. 1982. “On the transitivity of ‘say’ verbs.” In Studies in Transitivity [Syntax and Semantics 15], Paul Hopper and Sandra Thompson (eds), 301–318. New York: Academic Press. Noël, Dirk. 2007. “Diachronic construction grammar and grammaticalization theory.” Functions of Language 14: 177–202. Noh, Eun-Ju. 2000. Metarepresentation: A Relevance-Theory Approach [Pragmatics and Beyond New Series 69]. Amsterdam/Philadelphia: Benjamins. Noonan, Michael. 2007. “Complementation.” In Language Typology and Syntactic Description. Vol. 2: Complex Constructions, 2nd ed. Timothy Shopen (ed.), 52–150. Cambridge: Cambridge University Press.



Quotative go and be like  OED. 2008. Oxford English Dictionary online, . Oxford: Oxford University Press. Östman, Jan-Ola and Fried, Mirjam (eds). 2005. Construction Grammars: Cognitive Grounding and Theoretical Extensions [Constructional Approaches to Language 3]. Amsterdam/ Philadephia: Benjamins. Paradis, Carita. 2000. “It’s well weird. Degree modifiers of adjectives revisited: The nineties.” In Corpora Galore: Analyses and Techniques in Describing English, John M. Kirk (ed.), 147–160. Amsterdam: Rodopi. Partee, Barbara H. 1973. “The syntax and semantics of quotation.” In A Festschrift for Morris Halle, Stephen R. Anderson and Paul Kiparsky (eds), 410–418. New York: Holt. Quirk, Randolph, Greenbaum, Sidney, Leech, Geoffrey and Svartvik, Jan. 1985. A Comprehensive Grammar of the English Language. London: Longman. Ransom, Evelyn N. 1986. Complementation: Its Meanings and Forms [Typological Studies in Language 10]. Amsterdam/Philadelphia: John Benjamins. Rickford, John R., Wasow, Thomas, Zwicky, Arnold and Buchstaller, Isabelle. 2007. “Intensive and quotative all: Something old, something new.” American Speech 82: 3–31. Rissanen, Matti. 1991. “On the history of that/zero as object clause links in English.” In English Corpus Linguistics: Studies in Honour of Jan Svartvik, Karin Aijmer and Bengt Altenberg (eds), 272–289. London: Longman. Romaine, Suzanne and Lange, Deborah. 1991. “The use of like as a marker of reported speech and thought: A case of grammaticalization in progress.” American Speech 66: 227–279. Schourup, Lawrence. 1982. “Quoting with go ‘say’.” American Speech 57: 148–149. Schourup, Lawrence. 1983. Common Discourse Particles in English Conversation [Working Papers in Linguistics 28]. Columbus, Ohio: The Ohio State University. Schuelke, Gertrude L. 1958. “‘Slipping’ in indirect discourse.” American Speech 33: 90–98. Shuman, Amy. 1993. “‘Get outa my face’: Entitlement and authoritative discourse.” In Responsibility and Evidence in Oral Discourse [Studies in the Social and Cultural Foundations of Language 15], Jane H. Hill and Judith T. Irvine (eds), 135–160. Cambridge: Cambridge University Press. Singler, John. 2005. “Like, quote me. Like quotative – a part of English grammar now?” PBS ‘Do you speak American?’ website, ‘Sez who?’ section, 2005, (23 September 2009). Tagliamonte, Sali and D’Arcy, Alex. 2004. “He’s like, she’s like: The quotative system in Canadian youth.” Journal of Sociolinguistics 8: 493–514. Tagliamonte, Sali and Hudson, Rachel. 1999. “Be like et al. beyond America: The quotative system in British and Canadian youth.” Journal of Sociolinguistics 3: 147–172. Traugott, Elizabeth Closs. 1982. “From propositional to textual and expressive meanings: Some semantic-pragmatic aspects of grammaticalization.” In Perspectives on Historical Linguistics [Current Issues in Linguistic Theory 24], Winfred P. Lehmann and Yakov Malkiel (eds), 245–267. Amsterdam/Philadelphia: John Benjamins. Traugott, Elizabeth Closs. 1988. “Pragmatic strengthening and grammaticalization.” Berkeley Linguistics Society 14: 406–416. Traugott, Elizabeth Closs. 2003. “Constructions in grammaticalization.” In The Handbook of Historical Linguistics [Blackwell Handbooks in Linguistics], Brian D. Joseph and Richard D. Janda (eds), 624–647. Oxford: Blackwell. Traugott, Elizabeth Closs. 2007. “The concepts of constructional mismatch and type-shifting from the perspective of grammaticalization.” Cognitive Linguistics 18: 523–557.

 Lieven Vandelanotte Traugott, Elizabeth Closs. 2010. “(Inter)subjectivity and (inter)subjectification: A reassessment.” In Subjectification, Intersubjectification and Grammaticalization [Topics in English Linguistics 66], Kristin Davidse, Lieven Vandelanotte and Hubert Cuyckens (eds), 29–71. Berlin/ New York: Mouton de Gruyter. Traugott, Elizabeth Closs and Dasher, Richard B. 2002. Regularity in Semantic Change [Cambridge Studies in Linguistics 97]. Cambridge: Cambridge University Press. Vandelanotte, Lieven. 2006. “Speech or thought representation and subjectification, or on the need to think twice.” Belgian Journal of Linguistics 20: 137–168. Vandelanotte, Lieven. 2009. Speech and Thought Representation in English: A Cognitive-Functional Approach [Topics in English Linguistics 65]. Berlin/New York: Mouton de Gruyter. Vandelanotte, Lieven and Davidse, Kristin. 2009. “The emergence and structure of be like and related quotatives: A constructional account.” Cognitive Linguistics 20: 777–807. Vandewinkel, Sigi. 2010. “Strengthening uses of pure/puur in English and Dutch.” English Text Construction 3: 44–73. Vandewinkel, Sigi and Davidse, Kristin. 2008. “The interlocking paths of development to emphasizer adjective pure.” Journal of Historical Pragmatics 9: 255–287. Van Elslander, Thibaud. 2007. “I was all: ‘Well, go get another one.’ The innovative uses of all.” Unpublished term paper, University of Namur. Van Langendonck, Willy. 1994. “Determiners as heads?” Cognitive Linguistics 5: 243–259. Waksler, Rachelle. 2001. “A new all in conversation.” American Speech 76: 128–138. Wierzbicka, Anna. 1974. “The semantics of direct and indirect discourse.” Papers in Linguistics 7: 267–307. Wilkins, David P. 1992. “Interjections as deictics.” Journal of Pragmatics 18: 119–158. Winter, Joanne. 2002. “Discourse quotatives in Australian English: Adolescents performing voices.” Australian Journal of Linguistics 22: 5–21.

Quotation in sign languages A visible context shift Annika Herrmann and Markus Steinbach University of Göttingen

Assuming that quoting utterances and thoughts is a universal property of natural languages, sign languages are also expected to have various linguistic means to mark quotation. Like spoken languages, sign languages have regular forms of indirect reported speech. However, sign languages mostly draw on a specific grammatical means for quotation called role shift. Role shift is a particularly interesting form of reported speech that combines properties of both direct and indirect speech. Based on a pilot corpus study, we discuss the formal and functional properties of role shift in German Sign Language (DGS). We argue that role shift can be analyzed as a nonmanual agreement operator that overtly agrees with the signer and the addressee of the reported utterance and triggers a context shift from the actual context to the context of the reported utterance. Keywords: sign languages, German Sign Language (DGS), role shift, context shift, nonmanuals, agreement

1. Introduction1 Sign languages are natural languages with complex grammars that use the threedimensional signing space and various articulators to simultaneously express grammatical features. As opposed to the oral-auditory modality of spoken languages, sign languages are visual-manual languages that are perceived visually and 1. We would like to express our gratitude to our deaf informants, who offer us an invaluable insight into German Sign Language and the Deaf culture. We are also very much indebted to Elisabeth Engberg-Pedersen, Diane Lillo-Martin, Josep Quer, Roland Pfau, Wendy Sandler, and Bencie Woll for their constructive comments and many illuminating perspectives from other sign languages. Furthermore, we thank the audience of the CILS conference (Colloque International sur les Langues de Signes) in Namur, Belgium, and the participants of the sign language workshop at the SLE 42nd Annual Meeting in Lisboa, Portugal, for their helpful feedback.

 Annika Herrmann and Markus Steinbach

produced manually. Apart from the hands – that is, the manual articulators – additional articulatory channels such as the upper part of the body, the head and the face – that is, the nonmanual articulators – may carry complex linguistic information and are thus used to encode various aspects of sign language grammar. Examples of nonmanual features on the face are eyebrow movement, eye gaze patterns, eye blinks, mouth gestures, and facial expressions. In general, nonmanuals accompany manual signs and may fulfill lexical, morphological, syntactic, semantic, and prosodic functions, some of which will be discussed in the next section. Nonmanual articulators also play an essential role with respect to the marking of quotation in sign languages. The visual-manual modality of sign languages offers a specific means to quote utterances, thoughts and actions. By using movements of the upper part of the body, head turns and eye gaze shifts, a signer may take over the perspective of other signers to express the quoted material. This possibly modality-specific form of quotation is called role shift or shifted reference. We will argue below that the sideward movement of the upper part of the body indicates the quoted signer, whereas a shift of the midsagittal plane of the body and the head and a break in eye gaze contact away from the actual addressee indicate the addressee of the quoted utterance. These nonmanual means marking role shift occur simultaneously to the manual signing stream and have scope over the whole quoted material. Since role shift has some very specific properties, a thorough investigation of this visual form of quotation is important to fully understand the grammatical and pragmatic properties of different forms of reported speech and to develop a comprehensive linguistic theory of quotation. In this paper, we discuss the formal and functional properties of role shift as a means of reported speech in German Sign Languages (DGS) on the basis of elicited data.2 Section 2 provides some introductory remarks on the relevant aspects of sign language grammar that relate directly to quotation in DGS and other sign languages. In Section 3, we briefly summarize the basic findings about the phenomenon of verb agreement in sign languages, as it is essential for the understanding of the use of role shift markers. Section 4 focuses on the formal aspects of role shift and presents the results of a pilot corpus study of DGS data. The findings show that the nonmanual markers encoding role shift are hierarchically structured and interrelated, thus indicating a minimal and maximal marking of quotation. In Section 5, we elaborate on the functional properties of role shift. Presenting the similarities and differences between role shift and verb agreement, we present an 2. As in spoken languages, role shift can also be used for hypothetical reported discourse in sign languages (for hypothetical discourse in spoken German, cf. Golato this volume; Haberland 1986). Concerning the possibility of reporting thoughts with role shift, further investigations are necessary. In this paper, we focus on prototypical examples of reported speech.



Quotation in sign languages 

analysis of role shift as a nonmanual agreement phenomenon. Taking a crossmodal perspective, sign language role shift is then compared to its counterparts in spoken languages in Section 6. In Section 7, we briefly summarize the main findings and discuss some open issues for further research. 2. Some aspects of sign language grammar From a universal perspective, sign languages are linguistically quite similar to spoken languages. As the natural and native languages of deaf people around the world, they exhibit a fascinating complexity on all levels of the grammar. Despite some general similarities across sign languages, they also show an enormous amount of typological variation and dialectal differences within one single sign language are no exception (cf. Johnston and Schembri, to appear; Perniss et al. 2007). Recent linguistic as well as psycho- and neurolinguistic research has proven that both spoken and sign languages use basically the same grammatical system and have astonishingly similar neuro-cognitive foundations (cf. Emmorey 2003). Sign languages have the advantage of simultaneously expressing grammatical aspects within one sign or utterance by layering different articulators such as the above-mentioned manual (i.e. hands) and nonmanual (i.e. upper part of the body, head and face) means. This is the reason why the relatively slow production time of sign units compared to the relatively fast production time of word units is compensated for at the clausal level (cf. Meier 2002). Each individual sign language is subject to specific constraints on all levels of grammar, such as syntactic restrictions on word order, or phonological constraints on syllable structure. DGS, for instance, has an underlying SOV sentence structure in both main and embedded clauses. Furthermore, matrix clauses typically precede their subordinate clauses (cf. Happ and Vorköper 2006; Steinbach 2007). Embedding is rarely marked by subjunctions, but more often expressed by prosodic pauses and nonmanual markers. This becomes relevant for quotation in DGS as discussed below. The functions of nonmanuals can either be affective (i.e. paralinguistic) or grammatical. Grammatical uses of nonmanuals may affect all levels of grammar. Some lexical signs, such as the DGS adjective thin, for instance, inherently include nonmanual features (in this case ‘sucked in cheeks’). Moreover, in most sign languages, many adverbials are expressed only by nonmanual means. Likewise, many morphosyntactic and prosodic features and various syntactic constructions are marked by nonmanuals, as is the case with negation (i.e. headshake), sentence types (i.e. raised or lowered eyebrows, forward head tilt), subordination (i.e. brow raise, head nod), and topicalization (i.e. brow raise, backward head tilt) (cf. Happ

 Annika Herrmann and Markus Steinbach

and Vorköper 2006; Pfau and Quer 2010; Pfau and Steinbach 2006; Steinbach 2007; Wilbur 2003). Despite the fact that affective and grammatical nonmanuals are performed via the same articulatory channel, there are clear criteria to tell these two functions apart. Most importantly, they differ in scope and timing. As opposed to the more gradual and inconsistent affective means, grammatical nonmanual markers show a clear onset and offset and systematically align with and spread over specific constituents (cf. Emmorey 1999; Wilbur 2003). Nonmanuals that are an integral part of grammar usually comprise of only a few restricted articulators, whereas different muscles are used for affective purposes (cf. Corina et al. 1999; Reilly Mcintyre and Bellugi 1990). Signers are very consistent in the use of linguistic nonmanual features but vary with respect to the production of affective expressions. Below we show that nonmanual means used for sign language quotation are equally subject to the clear structural constraints mentioned above for grammatical markers, thus suggesting a grammatical analysis of nonmanuals in role shift. Note finally that sign languages do not have a standardized writing system, so most research on sign languages looks at ‘oral’ language use such as discourse from video elicitation and spontaneous data recording. Thus, we cannot directly investigate equivalents of the graphematic device of quotation marks in sign languages. The quotation mark gesture (i.e. so-called air quotes) that is often used in hearing conversations to express a scare quote meaning that signals the speakers’ attitude towards the quoted material is less productive in sign language dialogues (for scare quotes see Gutzmann and Stei 2011; Klockow 1980; Predelli 2003). In sign languages, the quotation mark gesture is used, for example, to quote titles of movies or books. In DGS, it is often glossed as title and furthermore acts as an indicator for a direct translation from the surrounding spoken language. Only in some rare cases may it be used as a kind of scare quote expressing the signer’s distance to what is said (cf. Boyes Braem 1992; Herrmann and Steinbach 2007; Werner 2007). 3. Verb agreement in sign languages The three-dimensional signing space in front of the signer is first of all used to express lexical signs, but also to encode topographical spatial relations as well as abstract grammatical relations such as pronominalization and agreement (for pronouns, cf. Lillo-Martin and Klima 1990; McBurney 2002; Meier 1990; for verb agreement, cf. Mathur and Rathmann 2010; Meir 2002; Steinbach and Pfau 2007). Pronouns correspond to loci in the signing space, they are introduced by the pointing sign index (ix). If the referent is present, the location in the signing space corresponds to the actual position of the referent. In case of non-present referents, the index-sign locates the abstract referent at a distinctive locus in the signing



Quotation in sign languages 

2 3

2nd person 3

1 3rd person

1st person

Figure 1.  Signing space with loci for pronominal reference

space typically used for anaphoric pronouns (usually to the right or left of the signer).3 Subsequent reference to these loci is then interpreted as pronominal reference to the previously established referent (see Figure 1). Sign languages use the same means as spoken languages for pronominalization and verb agreement (cf. Pfau and Steinbach 2006 for similar grammaticalization paths of agreement in spoken and sign languages). However, unlike spoken languages, sign languages use spatial localizations in the signing space, thereby expressing referential indices directly on the verb by modulating its phonological form. Sign languages do not use agreement affixes expressing morphosyntactic categories like, for example, person (cf. Aronoff et al. 2005). Moreover, in sign languages, agreement can only be expressed on a subset of verbs. Sign languages distinguish between three different verb classes: plain verbs, spatial verbs and agreement verbs (cf. Padden 1986). Plain verbs are lexically specified for their beginning and endpoints. Therefore, plain verbs cannot agree with one or two of their arguments. Spatial verbs express topographic relations in the signing space by modification of the path movement. Hence, the beginning and endpoint of the path movement are determined by topographic locations. Agreement verbs, on the other hand, do not agree with spatial referents, but with one or two of their arguments, which are pronominally linked to the signing space. Agreement thereby involves a path movement of the dominant hand from the position of the subject to the position of the object, indicated by the referential indices on the verb give in Example (1). Furthermore, many agreement verbs show a particular facing of the palm of the hand towards the object to indicate agreement. Thus, verb agreement in sign languages is locus agreement since agreement verbs agree with

3. It is still under debate whether there is evidence for a threefold distinction of personal pronouns, including a separate second person pronoun for the addressee (as indicated in Figure 1), or whether pronouns should be analyzed by differentiating only between first and non-first pronouns (cf. Berenz 2002; Cormier 2007; Engberg-Pedersen 1993; McBurney 2002; Meier 1990 for a thorough discussion). It would also be worth investigating the difference between the reference to present referents (deictic reference) and to non-present referents (anaphoric reference).

 Annika Herrmann and Markus Steinbach

the local indices of the subject and/or the object.4 In the following sign language examples, signs are always glossed in small caps. Subscripts indicate points in the signing space. For a comprehensive list of all notational conventions used in the examples, see the appendix. (1) yesterday girl ix3 book 3give1 ‘Yesterday, the girl gave me a book.’

(DGS)

Interestingly, some agreement verbs in sign languages are purely object-agreement verbs, such as thank and trust in DGS. However, no verbs exist in sign languages that agree with the subject only, as is the case with verb agreement in many spoken languages. Hence, object agreement is more frequent and less marked than subject agreement in most investigated sign languages. In addition to manual path movement, eye gaze in American Sign Language (ASL) can be analyzed as an instance of nonmanual object agreement in case of agreement verbs (cf. Thompson Emmorey and Kluender 2006; for similar results in DGS see Hosemann 2011).5 Eye gaze will turn out to be an integral part of role shift in DGS. In Section 4.3, we argue that in role shift, eye gaze functions as a nonmanual agreement marker. 4. Formal aspects of role shift in sign languages Since quotation of utterances and thoughts is assumed to be a universal phenomenon (cf. Coulmas 1986), it is of particular interest to investigate how it is expressed in the visual-manual modality. Quotation in sign languages has been addressed in various linguistic studies and the phenomenon of role shift as an important sign-language-specific means of quotation has been noted for a number of sign languages of the world. What we call role shift in this paper has also been discussed under the notions of shifted reference, referential shift, role playing, role

4. The nature of the agreement controllers is still under discussion. In this paper, we use the notions of subject and object, but our analysis of role shift does not depend on a specific analysis of verb agreement. For a detailed discussion of the nature of verb agreement in sign languages see Meir (1998, 2002), Liddell (2000), Neidle, Kegl, MacLaughlin, Bahan and Lee (2000), Rathmann and Mathur (2002), Mathur and Rathmann (2010), and Steinbach and Pfau (2007). 5. Neidle et al. (2000) analyze eye gaze as an independent agreement mechanism, which also occurs on plain verbs. However, this was challenged by the findings of Thompson et al. (2006) eye tracking study, so that eye gaze is assumed to be a syntactic agreement marker that only accompanies agreement verbs in ASL and DGS.



Quotation in sign languages 

taking, constructed dialogue, body shift and perspective shift.6 Particularly interesting is the discussion about similarities and differences between role shift and constructed action (for analyses of constructed action in sign languages, cf. Fischer and Kollien 2010; Lillo-Martin, to appear; Metzger 1995; Quinto-Pozos 2007; Winston 1991). A typical example of constructed action involves gestural and affective imitations of other persons, as is illustrated in the DGS examples in (2). In (2a), the signer introduces the referent boy and then shifts into his perspective, copying the actions of the shepherd’s boy while herding his sheep. Similarly, in (2b), constructed action (ca) is used for the same purpose, exemplifying the boring repetition of the boy’s actions with the additional affective facial expression of being bored.7 (2) Examples from the fable ‘The shepherd’s boy and the wolf ’ (DGS) a. boy castands holding a stick take-care castands holding a stick, looking around ‘The (shepherd’s) boy stood there with a stick in his hand, herded (the sheep) and looked around.’ b. nice everywhere ixa but boring ixa same++ castands holding a stick,

looking around bored

‘It was nice everywhere, but very boring and always the same, like standing with a stick in ones hand, looking around bored.’

Keeping constructed action in mind, in the following we focus on role shift as reported dialogue between two other signers. Nevertheless, the blurred boundaries between role shift and constructed action as well as the constant interaction, including blendings and overlaps of the two structures suggest that the picture is much more complex. It is most probably the case that role shift can be analyzed as a specific form of constructed action. Focusing on role shift as reported discourse, we will leave the discussion about the relation between role shift and constructed action, and the clarification concerning terminology for future research (cf. also Section 6 for a brief discussion of role shift in spoken languages). 6. Role shift seems to be subject to similar constraints across a range of sign languages. Thorough investigations have been presented for American Sign Language (ASL), as for instance in Padden (1986); Meier (1990); Poulin (1994); Lillo-Martin (1995); Metzger (1995); Lee et al. (1997); Emmorey and Reilly (1998); Janzen (2004); Lillo-Martin and Müller de Quadros (2011) and LilloMartin (to appear). For research concerning the form and function of role shift in other sign languages see: Happ and Vorköper (2006); Herrmann and Steinbach (2007) and Werner (2007) for DGS; Boyes Braem (1992,1995) for German-Swiss Sign Language (DSGS); Engberg-Pedersen (1993, 1995) for Danish Sign Language (DSL); Zucchi (2004) for Italian Sign Language (LIS); Quer (2005, 2011) for Catalan Sign Language (LSC); Pyers and Senghas (2007) for Nicaraguan Sign Language (NSL); Poulin and Miller (1995) for Quebec Sign Language (LSQ) and Ahlgren (1990) for Swedish Sign Language (SSL). 7. The examples are taken from a DGS data set of different fables (in this case “The shepherd’s boy and the wolf ”) that were elicited on the basis of the ECHO fables (cf. Crasborn et al. 2007; for more information see http://www.let.ru.nl/sign-lang/echo/).

 Annika Herrmann and Markus Steinbach

Based on a pilot corpus study of DGS (4.1), the following subsections illustrate how DGS expresses reported speech and why role shift is a modality specific means of quotation (4.2). We will then discuss the various nonmanual markers used to express role shift (4.3) and present statistical findings for the interaction of these nonmanual markers (4.4). 4.1

Methodology and data

The data that are used for the present study on role shift in DGS are taken from a corpus of data that were elicited within the framework of a dissertation project investigating modal meaning and focus particles (cf. Herrmann 2009). Eight native DGS signers participated in the study. While a signing interviewer was leading the participants through different tasks, two camcorders captured the signer’s torso and face. The video-recorded data were synchronized and thoroughly annotated using the annotation software ELAN.8 In this corpus, consisting of a total of 750 ELAN-files, we found 171 instances of role shift. The ELAN-files that include one or more role shifts comprise either a signed picture story or a short elicited reported dialogue of DGS. One file consists of approximately 3–6 sentences. In this chapter, we evaluate various nonmanual markers and their distribution within our data set. As described below in more detail, we analyze the coded material with respect to body lean, head position, eye gaze, and facial expressions and evaluate their occurrences in role shift, their mutual interaction, and their dependency on a matrix clause. 4.2

Role shift: Direct or indirect speech?

Sign languages, like spoken languages, exhibit regular modality-independent forms of indirect speech. Example (3) shows that, in DGS, the sentence-final complement clause is not marked by a lexical complementizer but usually by prosodic means such as a short pause ‘:’. Moreover, certain nonmanual markers may indicate of the embedded clause into the main clause. (3) lena3 say: ix3 like book++ read (DGS) ‘Lena3 says that she3 likes reading books.’ In indirect speech in sign languages, we find a regular interpretation of indexicals and pronouns, which have to be adapted to the actual context of the utterance. 8. ELAN (European Distributed Corpora Project Linguistic Annotator) is an annotation tool developed at the Max Planck Institute for Psycholinguistics in Nijmegen. The program can be downloaded from www.lat-mpi.eu.



Quotation in sign languages 

In Example (3), Lena is thus referred to by a third person pronoun (ix3). Due to the specific visual-manual modality of sign languages, however, role shift is a frequently used additional modality-dependent means of reported speech. Role shift is a systematic strategy of shifting the signers’ actual perspective to quote utterances. In sign languages, it is frequently used to report utterances or to recount conversations between two participants (cf. Padden 1986). Interestingly, role shift has some advantages over indirect speech. As opposed to indirect speech, role shift makes use of the signing space and allows for the simultaneous expression of grammatical and affective features related to the reported utterance (which allows, for example, the omission of matrix clauses). Moreover, since it is not possible to have embedded interrogatives in indirect speech in sign languages, embedded interrogatives can only be reported by role shift. The reason for this restriction is the fact that clauses with sentential nonmanual features cannot be embedded. Like embedded declarative clauses, embedded interrogative clauses do not differ syntactically from their non-embedded counterparts. Hence, in cases of indirect speech, the accompanied nonmanual features specific to interrogatives (i.e., raised or furrowed eyebrows and head tilts in DGS) are automatically assigned to the actual signer instead of the quoted signer. This phenomenon is quite similar to the restriction in many spoken languages that imperatives cannot be embedded in indirect speech (cf. d’Avis 2007, but see Oshima and Sano this volume for Japanese). Since sign languages do not have a specific morphosyntactic and prosodic marking of embedded interrogatives, the nonmanual question markers cannot be interpreted in the embedded context of indirect speech. Consequently, embedded interrogatives can only be expressed by role shift. This is illustrated by Example (4a). Role shift can, of course, also be used to quote regular declaratives like in Example (4b), and other sentence types. In the glosses, ‘< >’ stands for the nonmanual markers indicating role shift. Again, the colon represents the prosodic pause and ‘y/n’ stands for the nonmanual markers of yes/no-interrogatives in DGS, that is, eyebrow raise and a forward head tilt. The lines above the glosses indicate the scope of the nonmanual markers. 3a< >3b y/n (4) a. tim ix3a anna ix3b ask: ix2 sad ix2 ‘Tim asked Anna whether she is sad.’ 3a< >3b b. tim ix3a anna ix3b say: tomorrow 1help2 ‘Tim said to Anna that he will help her tomorrow.’

(DGS)

 Annika Herrmann and Markus Steinbach

The shifted context is indicated by the local indices ‘3a’ and ‘3b’; that is, both examples in (4) trigger a context shift leading to the interpretation that the embedded sentence has been uttered in a context different from the actual context, with Tim (i.e. ‘3a’) being the signer and Anna (i.e. ‘3b’) being the addressee of the utterance. Consequently, 1st and 2nd person personal pronouns such as ix2 ‘you’ in (4a) and agreement verbs such as help in (4b) are interpreted within the scope of the nonmanuals marking role shift. Role shift triggers a context shift. The first person pronoun ‘I’ does not refer to the actual signer but to the reported signer, even though the index-sign ix1 is pointing towards the actual signers’ chest. In contrast to indexical personal pronouns and agreement verbs, temporal and local indexicals such as tomorrow in (4b) are, to a certain degree, ambiguous; in other words, they need not be interpreted in the shifted context but can also be interpreted in the actual context, as can be seen in Example (5) (cf. Herrmann and Steinbach 2007: 168; Quer 2005). 3a< >3b (5) a. yesterday peter ix3a say: tomorrow ix1 arrive (DGS) ‘Yesterday Peter said that he will arrive tomorrow.’ b. Gestern (Mi) hat Peter gesagt, dass er morgen (Do/Fr) kommt. ‘Yesterday (Wed) Peter said that he will arrive tomorrow (Thur/Fr.).’ c. Gestern (Mi) hat Peter gesagt: „Ich komme morgen (Do).“ ‘Yesterday (Wed) Peter said: “I will arrive tomorrow (Thur).” ’

Role shift thus combines properties of both indirect and direct reported speech. As in direct speech, 1st and 2nd person pronouns and agreement verbs have to be interpreted in the original context of the quoted utterance marked by role shift.9 Moreover, as in direct speech in spoken languages, all kinds of speech acts can be reported in role shift. Furthermore, facial expressions are usually associated with the reported signer and are often used as supportive means to mark the quoted utterance. This can be compared to intonational variations of the quoted material to indicate the reported speaker’s voice or his/her way of speaking in direct quotations in spoken language (cf. Section 6). Yet, role shift in sign languages, like indirect speech in most spoken languages, has a specific morphosyntactic marking. In sign languages, specific 9. Similar examples can also be found in reported speech in spoken languages like Slave (Northern Athapaskan, a language spoken in the Northwest of Canada), where 1st and 2nd person indexicals in the complement clauses of verbs like ‘to say’ need not be interpreted in the context of the actual utterance but can be bound by the context of the reported utterance (for discussions and examples cf. Herrmann and Steinbach 2007; Quer 2005, 2011; Rice 1986; Schlenker 2003; Zucchi 2004).



Quotation in sign languages 

nonmanual means that will be discussed in the next subsection in more detail are used to indicate role shift. In addition, the interpretation of temporal and local indexicals in role shift resembles their interpretation in indirect speech. Recall from Example (5) that both kinds of indexicals can but need not be interpreted in the scope of the nonmanual markers indicating role shift. A similar ambiguity can be found with temporal and local indexicals in indirect speech in spoken languages, as can be seen in the German Example (5b).10 Another similarity between indirect speech and role shift is that both need not be verbatim.11 This mixture of properties makes it difficult to categorize role shift clearly as indirect or direct speech. Role shift seems to be part of a continuum between indirect and direct speech, most probably closer to direct speech.12 In the next two sections, we will describe the nonmanual markers used for role shift in DGS and their interaction. 4.3

Nonmanual markers for role shift

When quoting utterances, signers change their position and shift into the perspective of the reported signer by adjusting their body and head position as well as their eye gaze and facial expressions. In all investigated sign languages, role shift is generally marked by these nonmanual indicators which have scope over the whole quoted material, that is, the embedded clause. The following list summarizes the relevant markers of role shift in DGS: i. Eye gaze change (eg) towards the locus of the addressee of the quoted utterance ii. Change of head position (hp) towards the locus of the addressee of the quoted utterance

10. German and DGS seem to have different preferences for the interpretation of temporal and local indexicals in indirect speech and role shift. In indirect speech in German, there is a preference for the interpretation of temporal and local indexicals in the actual context of utterance. By contrast, in role shift in DGS, there is a preference for the opposite interpretation. However, both interpretations of temporal and local indexicals are only preferences, since there is always a certain degree of ambiguity. This clearly differs from direct speech in German, where temporal and local indexicals are not ambiguous. They are always interpreted in the reported context, as can be seen in Example (5c). 11. Note, however, that in spoken language the original utterance is also not always reproduced verbatim in direct quotations (cf. Brendel et al. 2011; Clark and Gerrig 1990). 12. In Section 6, we briefly return to this issue. There, we compare role shift in spoken and sign languages and discuss whether role shift is a modality-specific grammatical form of reported speech.

 Annika Herrmann and Markus Steinbach

iii. Body lean (bl) including a sideward movement of the upper part of the body towards the locus of the quoted signer and a midsagittal body shift towards the locus of the addressee of the quoted utterance iv. Facial expression (fe) associated with the quoted signer A bl is defined as a sideward body movement towards the location of the reported signer and/or a slight shift of the upper part of the body towards the addressee of the reported utterance. The feature hp was coded when the head turned towards the right or left in order to mark the addressee of the reported utterance. hp usually co-occurs with an eye-gaze change towards the addressee of the quoted utterance. If the eye gaze of the signer was directed from the actual addressee (usually the interviewer or the camcorder) towards a location in the signing space that coincided with the addressee of the reported utterance, this was annotated as eg. In addition to the linguistic facial expressions that accompany the quoted material, the signers also very often adopt affective facial expressions of the reported person. In order to tell whether these two kinds of facial expressions are used differently in role shift, the category of fe would need further detailed analysis to clearly distinguish between affective and grammatical facial expressions in the coding. Since all facial expressions are usually interpreted as associated with the reported signer anyway, we decided not to tease apart the grammatical and affective functions of facial expressions and only coded changes of facial expressions in general. The spreading of these four nonmanual features indicates the beginning and the end of the quotated material.13 A prototypical role shift involves all four nonmanual markers, but depending on the context, role shift can also be marked by only one or two of these means. In the following, we focus on the first three markers, since eye gaze, head position and body lean are related to loci in the signing space. Facial expressions have a different function in role shift. Unlike the other nonmanuals marking role shift, facial expressions are not linked to loci in signing space. Moreover, since they express characteristic features of the signer and the original utterance (for the body as subject in sign languages, cf. Meir et al. 2007), they are a genuine part of the reported utterance. Hence, facial expressions do not directly mark role shift. They can, however, be used as an implicit role shift marker since they are directly linked to the quoted signer. A significant change of facial expressions may therefore presuppose reported speech. A classical example of role shift is illustrated in Example (6) from the DGS data corpus. In this example, the signer retells a dialogue between a mother and a child (‘g-pu’ stands for the palm-up gesture). 13. The quotation ends when the signer returns to a neutral position or shifts into another perspective.



Quotation in sign languages 

3b< >3a fe, eg, hp, bl neg neg (6) a. ix3b e. ix3b mother ix3a: e. ix3b hey ix1 like ixa stay play wish  (DGS) ‘Emma to her mother: “Hey, I don’t want to stay here and play.” ’ >3b 3a< fe, eg, hp, bl b. ixdual always ixa play++ g-pu ‘“But, the two of us always play here, what’s up?”’ The signer uses all of the markers for role shift in both (6a) and (6b). He breaks the eye contact (eg) with the actual addressee and looks to the location associated with the addressee of the quoted utterance. Furthermore, head position (hp) and body posture (bl) adjust as described in the list above. Interestingly, the body

EMMA

MOTHER

IX3b

STAY

IXA

3b
3a

g-pu

>3b

Figure 2.  Classical role shift marking (6a)

3a
3b eg

(7) a. g-fw e.m.m.a. go mum tell: ix1wish not here long play++(DGS) ‘Emma goes to her mother says: “Mum, I don’t want to play here any longer.” ’ 3b< >3a b. fe, eg, hp, bl ix3b mother look: g-pu ixdual always go++ g-pu: pf g-pu ‘The mother replies: “What’s up, we always come here? I am surprised.”’

Examples (6) and (7) show that the overt realization of the nonmanual markers depends on the loci assigned to the signer and the addressee of the reported utterance. As a consequence, the matrix clause specifying signer and addressee, as well as the reported speech act, can be omitted (for quotative predicate ellipsis in Japanese, cf. Oshima and Sano this volume). In the case of a signer reporting a conversation between two or more participants, the signer may shift between the perspectives without repeating their names or the definite descriptions if the referents are previously and saliently established in the signing space. Moreover, since the reported utterance contains additional nonmanual features marking sentence type as well as expressive meaning, matrix clauses containing speech-act predicates or adverbial modifications expressing the attitude of the reported signer are also not necessary. Matrix clauses are only indispensable at the beginning of a speech report when the context does not clearly identify the signer and the addressee of the quoted utterances. As mentioned above, the usage of all nonmanual markers signals a prototypical and maximal marking of role shift. In the following section, we will present



Quotation in sign languages 

EMMA

MUM


3b

Figure 4.  Role shift with minimal marking (7a)

MOTHER

3b
3a

Figure 5.  Role shift with maximal marking (7b)

results of a statistical evaluation of the various nonmanual markers for role shift in DGS, their interaction and their frequency of occurrence. 4.4

Interaction of nonmanuals for role shift in DGS

Out of 171 role shifts found in the data, 168 (98%) are marked by facial expressions. 147 (86%) show an eye gaze change towards the addressee of the reported situation. In 131 (77%) of the cases, we observe head movement towards the addressee of the reported utterance. Surprisingly, only 82 (48%) of the role shifts involve a kind of body lean. This difference between eg and hp on the one hand and bl on the other seems to result from the fact that bl is more salient and more costly to produce since the signer has to move the whole upper part of the body. In contrast to this, a change in eye gaze and head position is less salient and less costly. The different behavior of the nonmanual markers thus supports the observation that eg is the least marked agreement marker of role shift. All role shifts in our corpus are marked by at least one of the four nonmanual features discussed in the previous section, but usually comprise of a combination of these means. Hence, body leans are the least frequent means of marking role shift in DGS. By contrast, facial expressions and eye gaze changes are the most frequent features to mark role shift, followed by head movements. Interestingly, we find some rare instances of the use of facial expressions without eye gaze change on the one hand,

 Annika Herrmann and Markus Steinbach

and a break of the gaze direction towards the addressee of the actual utterance without accompanying facial expressions on the other hand. The degree of role shift marking seems to depend on certain contextual restrictions such as the presence of a matrix clause (cf. below), the kind of reported utterance, the text type and the register. It might seem odd that we do not find a 100% correlation between role shift and a single marker such as eye-gaze change. Indeed, while the latter is the most important agreement marker of role shift in DGS, even eye-gaze change does not appear consistently throughout the data. It is important to mention that due to the simultaneous availability of various articulators, sign languages generally show a relatively flexible variation in the use of nonmanuals and their combination to express specific grammatical functions. Sign languages generally seem to permit the omission of grammatical markers in certain contexts (cf. subject agreement marking in subject-object-agreement verbs). Thus, we would not expect clear-cut results with respect to a single agreement marker. Even if a perspective shift is exclusively indicated by the facial expressions in certain rare cases (i.e. without an agreement marker), this seems to be sufficient to interpret a sentence as a quoted utterance (see Section 4.3). Nevertheless, the dependencies between the four nonmanual markers of role shift become manifest in the data. This is illustrated in Table 1, which gives the percentage of the examples in which the nonmanual marker listed in the left column is accompanied by the corresponding nonmanual marker listed in the top line (i.e. 91% of all role shifts marked by bl are also marked by eg). On the one hand, Table 1 shows that all three agreement markers are usually accompanied by a facial expression; on the other hand, it also shows that there is a clear dependency between the agreement markers. Body leans are very likely to be accompanied by other nonmanuals (range: 100% – 82%).14 By contrast, only 51% of all examples marked by eye gaze or head positions are articulated with a body lean. The reason for this might be a certain physical interdependency between the Table 1.  Relation between nonmanual markers of role shift If a then b a

bl hp eg

FE

b EG

HP

BL

100%   98%   99%

91% 96% –

82% – 86%

– 51% 51%

(eg = eye gaze, hp = head position, bl = body lean, fe = facial expressions)

14. The few instances where bl is either the only marker of role shift or lacks an accompanying head movement are most probably due to the situation of video recording, where the signers could have tried to look at the interviewer or the camcorder while shifting the upper part of the body.



Quotation in sign languages 

articulators: eg depends on hp, and hp on bl. If a signer moves the upper part of the body, s/he usually also moves the head, and if s/he moves the head, the eye gaze usually also changes. bl without hp and hp without eg requires additional articulatory effort since the signer has to move two related articulators in different directions. Therefore, eg is more frequent and articulatorily less marked than hp, which, in turn, is more frequent and articulatorily less marked than bl. Another interesting aspect is the interaction between nonmanuals and the matrix clause. In our corpus, 79 examples – that is, 46% of all role shifts – are explicitly introduced by a matrix clause. An obvious hypothesis with respect to the behavior of nonmanuals in role shift is that when a matrix clause is present, all three nonmanuals can be more easily omitted, whereas they occur more frequently in role shifts without a matrix clause. This strategy is clearly observable for the three nonmanual features as demonstrated in Table 2. Table 2 shows that bl, hp and eg become more frequent when the matrix clause is missing. Interestingly, the table also illustrates that all three markers are equally sensitive to the presence of a matrix clause. They show the same tendency to be omitted more frequently in role shifts with a matrix clause than in role shifts without a matrix clause. In sum, hp and eg occur much more frequently than bl (86% and 77% as opposed to 48%). The nonmanual markers are more often omitted in roles shifts introduced by an overt matrix clause, and they interact on a specific hierarchy that indicates a maximal and a minimal marking of role shift, illustrated in Figure 6. 5. Functional aspects of role shift in sign languages One of the most interesting aspects of quotation in sign languages is the fact that signers systematically use the signing space to quote utterances of other signers. Table 2.  Relation between nonmanual markers of role shift and the occurrence of matrix clauses

a b a b a b

If mc then bl If no mc then bl If mc then hp If no mc then hp If mc then eg If no mc then eg

N

%

a/b

34 48 53 78 64 83

43% 52% 67% 85% 81% 90%

0.71

(mc = matrix clause, eg = eye gaze, hp = head position, bl = body lean)

0.68 0.77

 Annika Herrmann and Markus Steinbach Minimal marking

EG EG, HP EG, HP, BL Maximal marking

Figure 6.  Minimal and maximal marking of role shift (eg = eye gaze, hp = head position, bl = body lean)

The originally gestural nonmanual means for role shift, such as body and head shifts and the redirection of eye gaze, are grammaticalized nonmanuals with a specific function, namely, to indicate a context shift, thereby referring to signers and addressees of a different utterance (for the grammaticalization of gestures in sign languages, cf. Section 6; Pfau and Steinbach 2006; Wilcox 2004). Syntactically, role shift can be analyzed as an indexical expression whose nonmanual features overtly agree with the signer and the addressee of the shifted context. bl overtly agrees with the signer and the addressee, whereas hp and eg only agree with the addressee. Hence, morphosyntactically, role shift corresponds to verb agreement. Recall from Section 3 that agreement verbs express agreement with the subject and the object by modifying phonological features such as path movement and/or hand orientation. Since the subject and the object of a sentence, as the signer and the addressee of a reported utterance, can be linked to referential loci in the signing space, sign languages can use the spatial means of verb agreement also for role shift. Role shift can thus be analyzed as a nonmanual agreement operator. As mentioned above, the loci of the signer and the addressee of the reported utterance act as anchors for the respective perspective shifts during quotation. As the signers’ eye gaze and the head are turned towards the addressee of the quoted situation, eg and hp overtly agree with the addressee of the reported utterance. By contrast, bl primarily agrees with the signer of the reported utterance because the signers’ body leans towards the locus of the quoted signer. In sum, the nonmanual morphosyntactic realizations of role shift overtly agree with the loci introduced for the signer and the addressee of the shifted context. Since eg (found in 85% of our examples) and hp (found in 76% of our examples) are the most frequent means



Quotation in sign languages 

of marking role shift (bl only occurs in 47% of all utterances), context shift is mainly indicated by a change of addressee. In this respect, role shift equals manual verb agreement as described in Section 3. For both manual and nonmanual agreement, object or addressee agreement is marked more frequently than subject or signer agreement. Besides these similarities, there are also some differences between verb agreement and role shift. The two main differences are the overt realization of agreement and the agreement controller. Role shift uses nonmanual means to express agreement, whereas agreement verbs use manual means such as path movement and hand orientation. Concerning the controller, agreement verbs agree with two of their arguments (i.e. the subject and the object). By contrast, role shift does not agree with syntactic arguments but with higher-level discourse-semantic entities, namely, the signer and the addressee of a reported utterance. Hence, both signer and addressee agreement is triggered by contextual features. A third difference between verb agreement and role shift becomes obvious when considering the scope of the agreement marking. Nonmanual signer and addressee agreement affects the whole clause, while manual subject and object agreement only affects the verb. Despite these minor differences, the intriguing similarities between role shift and verb agreement suggest that role shift can be analyzed as a nonmanual agreement operator. Semantically, the nonmanual agreement operator triggers a context shift from the actual context to a reported context. This context shift is marked by overt agreement, with the loci assigned to the signer and the addressee of the reported utterance. Lillo-Martin (1995), Zucchi (2004) and Quer (2005, 2011) argue for American Sign Language (ASL), Italian Sign Language (LIS) and Catalan Sign Language (LSC) that role shift can be analyzed either as a point of view predicate or a point of view operator. Similar to Quer’s (2011) point of view operator, we assume that the nonmanual agreement operator of our analysis occupies a very high position in the functional domain of the clause and has scope over the entire embedded clause. Recall that role shift is compatible with all sentence types and that the nonmanual features marking role shift spread across the whole c-command domain of the operator (for a syntactic analysis of nonmanual features, see Wilbur and Patschke 1999). The operator binds first and second person indexicals, since signer and addressee are strictly tied to the utterance context. As a consequence of the overt nonmanual marking (i.e. the signer/addressee agreement), the interpretation of indexical personal pronouns and agreement verbs is obligatorily shifted to the context of the reported utterance, as has been shown in Example (4) above. By contrast, temporal and local indexicals are less strictly tied to the utterance context. Therefore, they can, but need not, be interpreted in the shifted context (cf. Example 5).

 Annika Herrmann and Markus Steinbach

Note finally that sentences introduced by a nonmanual agreement operator can be embedded under the corresponding performative verbs (i.e. say, ask, command, ...). In this case, the embedded operator is co-indexed with the matrix verb, and the subject and the object of the matrix clause are linked to the signer and the addressee of the shifted context (for a similar idea, see Lillo-Martin 1995). Recall from the previous section that the onset of the nonmanual features marking role shift can be aligned with the manual sign of the matrix verb, that is, role shift may already start with the sentence-final matrix verb co-indexed with the nonmanual role shift operator. 6. Reported speech in spoken and sign languages As pointed out in Section 2, sign languages do not have a standardized writing system. Therefore, role shift in sign languages cannot be directly compared to (the highly standardized forms of) quotation in the written modality of spoken languages; yet, they should be compared to the more colloquial forms of quotation in spoken languages, which are, for example, often introduced by lexical markers such as ‘I’m like/she’s like’ or the German equivalent ‘und ich so/und sie so’ (cf. Brendel et al. 2011; Golato 2000; Streeck 2002; see also Fox and others in this volume). It has been observed that speakers also productively use various gestures in spoken narration to indicate a context shift. In addition, speakers tend to change the voice quality to shift into the perspective of another character. Hence, role shift is used in both the visual-manual and the oral-auditory modality to indicate direct reported speech (cf. Günthner 1999; Lillo-Martin, to appear; Tannen 1989). However, only in sign languages has role shift become a genuine part of the grammatical system, because the visual-manual modality, unlike the oral-auditory modality, offers the unique property of grammaticalizing manual and nonmanual gestures. Since gestures use the same articulatory channel that is also active in the production of signs, it is not uncommon for manual and nonmanual gestures to become grammaticalized in sign languages (cf. Pfau and Steinbach 2006; Wilcox 2004). By contrast, spoken languages use a completely different articulatory and perceptual system. Therefore, gestures and role shifts used in direct reported speech cannot become an integral part of the grammatical system of a spoken language, but are only used as co-speech gestures, supporting the shift of perspectives in spoken narration. It has been claimed that spoken languages can only integrate gestures into the grammatical system that belong to the same articulatory and perceptual domain, such as intonational patters used to mark focus and sentence types (cf. Gussenhoven 2004). Systematic changes in prosody and voice quality can, in principle, develop into role shift markers and thus become part of the



Quotation in sign languages 

grammatical system of a spoken language. By contrast, co-speech gestures such as eye gaze change, head movement and body leans cannot be grammaticalized functional elements of spoken language grammars. Comparing role shift in spoken and sign languages, we can conclude that the fundamental ability to shift into another character or perspective and to use role shift as a means of marking quotations in discourse and narration seems to be modality-independent. However, the development of role shift into a nonmanual grammatical device systematically marking quotations seems to be a modalityspecific characteristic of sign languages, which have the unique property of grammaticalizing manual and nonmanual gestures. 7. Conclusion In this paper, we have argued that role shift is a grammaticalized nonmanual gesture used to mark reported speech in sign languages. Although the ability to shift into another character or to take over the perspective of another person can also be observed in spoken language, the specific use of nonmanual markers in role shift seems to be a modality-specific grammatical device to express reported speech. Role shift is not merely a co-speech gesture, since in sign languages, unlike in spoken languages, manual and nonmanual gestures can develop into grammaticalized expressions and can thus be integrated into the grammatical system. We have shown that in DGS, role shift combines properties of both indirect and direct quotation. Role shift is expressed by three nonmanual agreement markers: eye gaze, head position and body lean. In addition, the quoted material is often accompanied by facial expressions associated with the original signer of the quoted utterance. All four nonmanual markers can be used simultaneously, but role shift can also be minimally marked, that is, by a change of eye gaze or facial expression only. For the analysis of role shift, we propose a nonmanual agreement operator with maximal scope over the embedded clause. The operator agrees with the signer (sideward body movement) and the addressee (eye gaze, head position and body shift) of the reported utterance. Signer and addressee agreement, like subject and object verb agreement, is locus agreement. Semantically, the nonmanual agreement operator triggers a context shift from the actual context to a reported context. As a consequence, personal indexicals and agreement verbs are interpreted in the context of the reported utterance. More research is needed in order to arrive at a comprehensive understanding of the phenomenon of role shift, including the following: studies on the interaction of nonmanuals marking role shift with other affective and linguistic nonmanuals related to the quoted material, studies on the relation between role shift and

 Annika Herrmann and Markus Steinbach

constructed action, on the acquisition and typological variation of role shift (cf. Lillo-Martin and Müller de Quadros 2011; Morgan and Woll 2003; Reilly 2000), on the realization of reported thought in sign languages, on the expression of speaker’s attitude in role shift (cf. Herrmann and Steinbach 2007; Spronck this volume), on role shift in different kinds of discourse, on similarities and differences between spoken and sign languages with respect to perspective shift and constructed action, and on the continuum between different kinds of perspective shifts in both modalities. References Ahlgren, Inger. 1990. “Deictic pronouns in Swedish and Swedish Sign Language.” In Theoretical Issues in Sign Language Research, Vol. 1: Linguistics, Susan D. Fischer and Patricia Siple (eds), 167–174. Chicago: Chicago University Press. Aronoff, Mark, Meir, Irit and Sandler, Wendy. 2005. “The paradox of sign language morphology.” Language 81: 301–344. Berenz, Norine. 2002. “Insights into person deixis.” Sign Language & Linguistics 5: 203–227. Boyes Braem, Penny. 1992. “Zitat und Zitieren in der Gebärdensprache der Gehörlosen.” Zeitschrift für Semiotik 44: 79–110. Boyes Braem, Penny. 1995. Einführung in die Gebärdensprache und ihre Erforschung. Hamburg: Signum. Brendel, Elke, Meibauer, Jörg and Steinbach, Markus. 2011. “Exploring the meaning of quotation.” In Understanding Quotation, Elke Brendel, Jörg Meibauer and Markus Steinbach (eds), 1–33. Berlin: Mouton de Gruyter. Clark, Herbert and Gerrig, Richard. 1990. “Quotations as demonstratives.” Language 66: 764–805. Corina, David, Bellugi, Ursula, and Reilly, Judy. 1999. “Neuropsychological studies of linguistic and affective facial expressions in deaf signers.” Language and Speech 42: 307–331. Cormier, Kearsy. 2007. “Do all pronouns point? Indexicality of first person plural pronouns in BSL and ASL.” In Visible Variation: Comparative Studies on Sign Language Structure, Pamela Perniss, Roland Pfau and Markus Steinbach (eds), 63–101. Berlin: Mouton de Gruyter. Coulmas, Florian. 1986. “Reported speech: Some general issues.” In Direct and Indirect Speech, Florian Coulmas (ed.), 1–28. Berlin: Mouton de Gruyter. Crasborn, Onno, Mesch, Johanna, Waters, Dafydd, Nonhebel, Annika, van der Kooij, Els, Woll, Bencie and Bergmann, Brita. 2007. “Sharing sign language data online: Experiences from the ECHO project.” International Journal of Corpus Linguistics 12(4): 535–562. d’Avis, Franz Josef. 2007. “Zitat und Sprecherbezug.” In Zitat und Bedeutung. Linguistische Berichte Sonderheft, Elke Brendel, Jörg Meibauer and Markus Steinbach (eds), 67–88. Hamburg: Buske Verlag. Emmorey, Karen. 1999. “Do signers gesture?” In Gesture, Speech, and Sign, Lynn S. Messing and Ruth Campbell (eds), 133–159. Oxford: Oxford University Press. Emmorey, Karen. 2003. “The neural systems underlying sign languages.” In The Oxford Handbook of Deaf Studies, Language, and Education, Marc Marschark and Patricia E. Spencer (eds), 361–376. Oxford: Oxford University Press.



Quotation in sign languages  Emmorey, Karen and Reilly, Judy. 1998. “The development of quotation and reported action: Conveying perspective in ASL.” In Proceedings of the Twenty-Ninth Annual Stanford Child Language Research Forum, Eve Clark (ed.), 81–90. Stanford, California: CSLI Publications. Engberg-Pedersen, Elisabeth. 1993. Space in Danish Sign Language. The Semantics and Morphosyntax of the Use of Space in a Visual Language. Hamburg: Signum. Engberg-Pedersen, Elisabeth. 1995. “Point of view expressed through shifters.” In Language, Gesture, and Space, Karen Emmorey and Judy Reilly (eds), 133–154. Hillsdale, New Jersey: Lawrence Erlbaum. Fischer, Renate and Kollien, Simon. 2010. “Gibt es constructed action in Deutscher Gebärdensprache und in Deutsch (in der Textsorte Bedeutungserklärung)?” DAS ZEICHEN 86: 502–510. Golato, Andrea. 2000. “An innovative German quotative for reporting on embodied actions: Und ich so/und er so ‘and I’m like/and he’s like’.” Journal of Pragmatics 32: 29–54. Günthner, Susanne. 1999. “Polyphony and the ‘layering of voices’ in reported dialogues: An analysis of the use of prosodic devices in everyday reported speech.” Journal of Pragmatics 31: 685–708. Gussenhoven, Carlos. 2004. The Phonology of Tone and Intonation. Cambridge: Cambridge University Press. Gutzmann, Daniel and Stei, Erik. 2011. “Quotation marks and kinds of meaning. Arguments in favor of a pragmatic account.” In Understanding Quotation, Elke Brendel, Jörg Meibauer and Markus Steinbach (eds), 161–193. Berlin: Mouton de Gruyter. Haberland, Hartmut. 1986. “Reported speech in Danish.” In Direct and Indirect Speech, Florian Coulmas (ed.), 219–253. Berlin: Mouton de Gruyter. Happ, Daniela and Vorköper, Marc-Oliver. 2006. Deutsche Gebärdensprache. Ein Lehr- und Arbeitsbuch. Frankfurt am Main: Fachhochschulverlag. Herrmann, Annika. 2009. Modal Particles and Focus Particles in Sign Languages. A Cross-linguistic Study on DGS, NGT and ISL. Unpublished doctoral dissertation, Universität Frankfurt am Main. Herrmann, Annika and Steinbach, Markus. 2007. “Wenn ‘ich’ nicht ich ist. Zitieren in Gebärdensprachen.” In Zitat und Bedeutung. Linguistische Berichte Sonderheft, Elke Brendel, Jörg Meibauer and Markus Steinbach (eds), 153–179. Hamburg: Buske. Hosemann, Jana. 2011. “Eye gaze and verb agreement in German Sign Language: A first glance.” Sign Language & Linguistics. Special Issue 14(1): 76–93. Janzen, Terry. 2004. “Space rotation, perspective shift, and verb morphology in ASL.” Cognitive Linguistics 15(2): 149–174. Johnston, Trevor and Schembri, Adam. to appear. “Sociolinguistic variation.” In Sign Language. An International Handbook, Roland Pfau, Markus Steinbach and Bencie Woll (eds). Berlin: Mouton de Gruyter. Klockow, Reinhard. 1980. Linguistik der Gänsefüßchen. Untersuchungen zum Gebrauch der Anführungszeichen im gegenwärtigen Deutsch. Frankfurt am Main: Haag und Herchen. Lee, Robert G., Neidle, Carol, MacLaughlin, Dawn, Bahan, Benjamin and Kegl, Judy. 1997. “Role shift in ASL: A syntactic look at direct speech.” In Syntactic Structure and Discourse Function: An Examination of two Constructions in American Sign Language (American Sign Language Linguistic Research Project Report 4), Carol Neidle, Dawn MacLaughlin and Robert G. Lee (eds), 24–45. Boston: Boston University Press. Liddell, Scott K. 2000. “Indicating verbs and pronouns: Pointing away from agreement.” In The Signs of Language Revisited: An Anthology to Honor Ursula Bellugi and Edward Klima,

 Annika Herrmann and Markus Steinbach Karen Emmorey and Harlan Lane (eds), 303–320. Mahwah, New Jersey: Lawrence Erlbaum Associates. Lillo-Martin, Diane. 1995. “The point of view predicate in American Sign Language.” In Language, Gesture, and Space, Karen Emmorey and Judy Reilly (eds), 155–170. Hillsdale, New Jersey: Erlbaum. Lillo-Martin, Diane. to appear. “Utterance reports and constructed action.” In Sign language. An International Handbook, Roland Pfau, Markus Steinbach and Bencie Woll (eds). Berlin: Mouton de Gruyter. Lillo-Martin, Diane and Klima, Edward S. 1990. “Pointing out differences: ASL pronouns in syntactic theory.” In Theoretical Issues in Sign Language Research, Vol. 1: Linguistics, Susan D. Fischer and Patricia Siple (eds), 191–210. Chicago: University of Chicago Press. Lillo-Martin, Diane and Müller de Quadros, Ronice. 2011. “Acquisition of the syntax-discourse interface: The expression of point of view.” Lingua 121: 623–636. Mathur, Gaurav and Rathmann, Christian. 2010. “Verb agreement in sign language morphology.” In Sign Languages, Diane Brentari (ed.), 173–196. Cambridge: Cambridge University Press. McBurney, Susan. 2002. “Pronominal reference in signed and spoken language: Are grammatical categories modality-dependent?” In Modality and Structure in Signed and Spoken Languages, Richard P. Meier, Kearsy Cormier and David Quinto-Pozos (eds), 329–369. Cambridge: Cambridge University Press. Meier, Richard P. 1990. “Person deixis in American Sign Language.” In Theoretical Issues in Sign Language Research, Vol. 1: Linguistics, Susan D. Fischer and Patricia Siple (eds), 175–190. Chicago: Chicago University Press. Meier, Richard P. 2002. “Why different why the same? Explaining effects and non-effects of modality upon linguistic structure in sign and speech.” In Modality and Structure in Signed and Spoken Languages, Richard P. Meier, Kearsy Cormier and David Quinto-Pozos (eds), 1–25. Cambridge: Cambridge University Press. Meir, Irit. 1998. “Syntactic-semantic interaction in Israeli Sign Language verbs: The case of backward verbs.” Sign Language & Linguistics 1: 3–33. Meir, Irit. 2002. “A cross-modality perspective on verb agreement.” Natural Language & Linguistic Theory 20: 413–450. Meir, Irit, Padden, Carol A., Aronoff, Mark, and Sandler, Wendy. 2007. “Body as subject.” Journal of Linguistics 43: 531–563. Metzger, Melanie. 1995. “Constructed dialogue and constructed action in ASL.” In Sociolinguistics in Deaf Communities, Ceil Lucas (ed.), 255–271. Washington, DC: Gallaudet University Press. Morgan, Gary and Woll, Bencie. 2003. “The development of reference switching encoded through body classifiers in British Sign Language.” In Perspectives on Classifier Constructions in Sign Language, Karen Emmorey (ed.), 297–310. Hillsdale, New Jersey: Lawrence Erlbaum Associates. Neidle, Carol, Kegl, Judy, MacLaughlin, Dawn, Bahan, Benjamin and Lee, Robert G. 2000. The Syntax of American Sign Language. Functional Categories and Hierarchical Structure. Cambridge, Massachusetts: MIT Press. Padden, Carol A. 1986. “Verbs and role-shifting in American Sign Language.” In Proceedings of the Fourth National Symposium on Sign Language Research and Teaching, Carol A. Padden (ed.), 44–57. Silver Spring, Maryland: National Association of the Deaf.



Quotation in sign languages  Perniss, Pamela, Pfau, Roland and Steinbach, Markus. 2007. “Can’t you see the difference? Sources of variation in sign language structure.” In Visible Variation: Comparative Studies on Sign Language Structure, Pamela Perniss, Roland Pfau and Markus Steinbach (eds), 1–34. Berlin: Mouton de Gruyter. Pfau, Roland and Quer, Josep. 2010. “Nonmanuals: their grammatical and prosodic roles.” In Sign Languages, Diane Brentari (ed.), 381–402. Cambridge: Cambridge University Press. Pfau, Roland and Steinbach, Markus. 2006. “Modality-independent and modality-specific aspects of grammaticalization in sign languages.” Linguistics in Potsdam 24: 5–94. Poulin, Christine. 1994. “Null arguments and referential shift in American Sign Language.” MIT Working Papers in Linguistics 23: 267–281. Poulin, Christine and Miller, Christopher. 1995. “On narrative discourse and point of view in Quebec Sign Language.” In Language, Gesture, and Space, Karen Emmorey and Judy Reilly (eds), 117–131. Hillsdale, New Jersey: Lawrence Erlbaum. Predelli, Stefano. 2003. “Scare quotes and their relation to other semantic issues.” Linguistics and Philosophy 26: 1–28. Pyers, Jennie and Senghas, Ann. 2007. “Reported action in Nicaraguan and American Sign Languages: Emerging versus established systems.” In Visible Variation. Comparative Studies on Sign Language Structure, Pamela Perniss, Roland Pfau and Markus Steinbach (eds), 279–302. Berlin: Mouton de Gruyter. Quer, Josep. 2005. “Context shift and indexical variables in sign languages.” In Proceedings of Semantics and Linguistics Theory 15, Effi Georgala and Jonathan Howell (eds), 152–168. Ithaca, New York: CLC Publications. Quer, Josep. 2011. “Reporting and quoting in signed discourse.” In Understanding Quotation, Elke Brendel, Jörg Meibauer and Markus Steinbach (eds), 277–302. Berlin: Mouton de Gruyter. Quinto-Pozos, David 2007. “Can constructed action be considered obligatory?” Lingua 117(7): 1285–1314. Rathmann, Christian and Mathur, Gaurav. 2002. “Is verb agreement the same crossmodally?” In Modality and Structure in Signed and Spoken Languages, Richard P. Meier, Kearsy Cormier and David Quinto-Pozos (eds), 370–404. Cambridge: Cambridge University Press. Reilly, Judy. 2000. “Bringing affective expression into the service of language: Acquiring perspective marking in narratives.” In The Signs of Language Revisited: An Anthology to Honor Ursula Bellugi and Edward Klima, Karen Emmorey and Harlan Lane (eds), 415–433. Mahwah, New Jersey: Lawrence Erlbaum Associates. Reilly, Judy, McIntire, Marina, and Bellugi, Ursula. 1990. “The acquisition of conditionals in American Sign Language development: Grammatical facial expressions.” Applied Psycholinguistics 11: 369–392. Rice, Keren D. 1986. “Some remarks on direct and indirect speech in Slave (Northern Athapaskan).” In Direct and Indirect Speech, Florian Coulmas (ed.), 47–76. Berlin: Mouton. Schlenker, Philippe. 2003. “A plea for monsters.” Linguistics & Philosophy 26: 29–120. Steinbach, Markus. 2007. “Gebärdensprache.” In Schnittstellen der germanistischen Linguistik, Markus Steinbach, Ruth Albert, Heiko Girnth, Annette Hohenberger, Bettina Kümmerling-Meibauer, Jörg Meibauer, Monika Rothweiler and Monika Schwarz-Friesel (eds), 137–185. Stuttgart: Metzler. Steinbach, Markus and Pfau, Roland. 2007. “Grammaticalization of auxiliaries in sign languages.” In Visible Variation. Comparative Studies on Sign Language Structure, Pamela Perniss, Roland Pfau and Markus Steinbach (eds), 301–339. Berlin: Mouton.

 Annika Herrmann and Markus Steinbach Streeck, Jürgen. 2002. “Grammars, words, and embodied meanings: On the uses and evolution of so and like.” Journal of Communication 52: 581–596. Tannen, Deborah. 1989. Talking Voices: Repetition, Dialogue, and Imagery in Conversational Discourse. Cambridge: Cambridge University Press. Thompson, Robin, Emmorey, Karen and Kluender, Robert. 2006. “The relationship between eye gaze and verb agreement in American Sign Language. An eye-tracking study.” Natural Language and Linguistic Theory 24: 571–604. Werner, Judith. 2007. Direkte und indirekte Rede in der Deutschen Gebärdensprache (DGS). Unpublished Master’s Thesis, Universität Mainz. Wilbur, Ronnie B. 2003. “Modality and the structure of language.” In Oxford Handbook of Deaf Studies, Language, and Education, Marc Marschark, Patricia E. Spencer and Peter E. Nathan (eds), 332–346. Oxford: Oxford University Press. Wilbur, Ronnie B. and Patschke, Cynthia 1999. “Syntactic correlates of brow raise in ASL.” Sign Language & Linguistics 2: 3–41. Wilcox, Sherman. 2004. “Gesture and language; cross-linguistic and historical data from signed languages.” Gesture 4: 43–73. Winston, Elizabeth A. 1991. “Spatial referencing and cohesion in an American Sign Language text.” Sign Language Studies 73: 397–410. Zucchi, Sandro. 2004. Monsters in the Visual Mode? Unpublished Manuscript, Università degli Studi di Milano.

part iv

Language variation and change

Performed narrative The pragmatic function of this is + speaker and other quotatives in London adolescent speech* Sue Fox

Queen Mary, University of London This chapter is an investigation of an innovation in the quotative system in London and it addresses the question of how and why such innovations emerge. The paper compares and contrasts the (socio-)linguistic profile and discourse functions of be like and the even newer this is + speaker using a combination of variationist quantitative methods and discourse analytic qualitative methods. It is argued that the alternation of different quotatives can be seen as a performance feature and that new quotatives fulfil the pragmatic function of highlighting a particularly dramatic peak in performed narratives. In London be like is behaving more like lexical verbs such as say, leaving a gap in the system for the emergence of the innovative this is + speaker. In many ways, this new competitor parallels be like when it was first reported, strengthening the argument that new quotatives have a specific role in the discourse. Keywords: Quotative constructions, pragmatic functions, performance features, performed narratives, this is + speaker

1. Introduction The emergence and rapid rise of be like in many English varieties has been extensively documented in recent years (Blyth, Recktenwald and Wang 1990; Buchstaller 2004, 2006a; Ferrara and Bell 1995; Macaulay 2001; Romaine and Lange 1991; Tagliamonte and D’Arcy 2004, 2007; Tagliamonte and Hudson 1999 and many others). As the present volume demonstrates, this relatively new quotative is not solely restricted to English varieties, with van in Dutch (Coppen and Foolen this volume; * My thanks to Jenny Cheshire for invaluable help and support in writing this chapter. I am also grateful to the editors and to an anonymous reviewer for helpful comments on an earlier draft of the chapter.

 Sue Fox

Foolen 2008), bare and liksom in Norwegian (Opsahl et al. this volume) and und ich so in German (Golato 2000, this volume) showing similarities in use to their English counterpart. Many of these studies have generally focused on the distribution of the quotatives and the internal and external factors constraining their use. What has been less studied is how and why such innovations emerge. An investigation of London English (Cheshire and Fox 2007) demonstrated that the new quotative be like has spread to the speech of adolescents but that there is also a new competitor within the quotative system, this is + speaker, as shown in Examples (1) and (2).

(1) this is me “what..what’s your..what’s your problem?”1 (Zack_1, 1:27:05)2 (2) this is them “what area are you from . what part?” (Alex_1, 20:15)

Catching a newcomer to the system at such an early stage of its development provides an ideal opportunity to examine the role that it plays when it is first introduced; this in turn may shed light on why and how such innovations emerge. I report here on the findings from a corpus of speech from nineteen adolescents and six elderly speakers from an inner-city borough of London, an area of multicultural and social diversity characteristic of many large metropolises. I will first present the distribution of the whole range of quotatives, but will move on to focus on be like and this is + speaker as used by the adolescents. I then widen the analysis further to consider the role that new quotatives may play. I will discuss whether new quotatives take over or fulfil a particular function within the discourse and whether different quotatives have different pragmatic functions, perhaps in relation to the type of personal experience being narrated. 2. The study The corpus I analyze comes from a subsample of speakers from the project Linguistic Innovators: The English of Adolescents in London.3 This project investigated language use in two locations in London (see Figure 1), one inner and one outer borough, but here the focus is only on language use in the inner area of London, since it is only here where, among other innovations, the emergence of the newest quotative this is + speaker is found. 1.

For transcription conventions see Appendix 1

2. All examples (unless otherwise stated) come from the Linguistic Innovators: the English of Adolescents in London corpus (see Footnote 3). The coding sequence indicates the pseudonym of the speaker, the interview number and the counter number. 3. Linguistic innovators: the English of adolescents in London 2004–7, funded by the Economic and Social Research Council, Principal Investigator Paul Kerswill, Co-investigator Jenny Cheshire, Research Associates Susan Fox and Eivind Torgersen (ref. RES 000-23-0680).



Enfield Barnet BUCKS

BERKSHIRE

Harrow

Haringey

Waltham Redbridge Forest Brent Isling- HackHillingdon Camden ton ney Barking Tower Newham H Westa Ealing City Hamlets m m Ken.Minister e & rs Chel. Southmi wark Greenwich th Hounslow Lambath Bexley Wandsworth Richmond Lewisham

ESSEX Havering

Merton Kingston Sutton SURREY

KENT

Bromley Croydon

SCALE Miles

0 0

Km

10 10

Performed narrative 

Figure 1.  Inner London (Hackney) and outer London (Havering) research locations

= LONDON BOROUGHS

HERTFORDSHIRE

 Sue Fox

The nineteen adolescents in the sub-sample, aged between sixteen and nineteen years, are comprised of eight female and eleven male speakers. The participants in this sub-sample are all from a working class background, they were born in the local community (with the exception of two adolescents who were born elsewhere but came to the area before the age of two years old) and all went to local primary and secondary schools. While all of the participants are native speakers of English, some of them have additional languages which they may use or have access to and which reflect their ethnic backgrounds as well as their friendship networks. For example, four of the participants are of Bangladeshi origin and speak Sylheti with some members of their families, one participant is Columbian whose home language is Spanish and many others have access to English (African and Caribbean) creoles through family or friendship networks. Some of the participants are also monolingual ‘Anglos’, meaning speakers of British origin whose families have resided in the area for two or more generations and who correspond roughly to ‘white British’, the term used as an ethnicity classification in the 2001 UK Census and other official documents. As a baseline for comparison, and to consider the possibility of language change due to socio-demographic differences, I analyze speech from six white monolingual Anglo speakers aged over seventy years old, three male and three female, who were also born in the local community and who have lived all of their lives there. In keeping with the apparent time model, it is assumed that the speech of the elderly participants is representative of the speech of that area of London from approximately fifty years ago when that generation acquired the language (Bailey et al. 1991), and when the population was predominantly white working class.4 The fieldwork for the study was carried out during 2005 and 2006. The fieldworker conducted informal interviews with the elderly participants mainly in their homes. For the adolescents, however, the recordings were made in a local college which they all attended and where they were enrolled on vocational courses such as painting and decorating, car mechanics and nursery nursing. The fieldworker conducted a period of participant observation and gradually got to know the students during their leisure time at college. They were then recorded in their self-selected friendship pairs or in small friendship groups in a room attached to the student common room, away from the more formal classroom context. The adolescents were encouraged to talk about topics which they would normally talk about amongst themselves and were also encouraged to lead the discussion. The recordings are all highly informal, sometimes with little input from the fieldworker. 4. The use of apparent time differences to study language change in progress has been a tenet of sociolinguistics since Labov (1963) used it in his work on Martha’s Vineyard. The underlying assumption is that when all other factors (such as social class) are held constant, then differences among generations of speakers reflect actual diachronic developments in a language.



Performed narrative 

Every instance of a verb or quotative expression introducing constructed dialogue (Tannen 1986), reported thought and non-lexicalized sounds/gestures was extracted for analysis. Also included were instances of a zero quotative (Matthis and Yule 1994), where it was clear from the context that direct reported speech was introduced to the discourse with no quotative marker. This resulted in a total of 1652 tokens for this subsample of speakers: 370 for the older speakers and 1282 for the adolescent speakers. 3. The Results Table 1 shows the overall distribution of the quotatives. Among the elderly speakers, say represents almost 71% (N = 261) of the total number of quotatives used, with the zero form being the second most used at almost 19% percent (N = 70). The complete range of quotatives employed by the elderly speakers is largely restricted to the most established quotation introducers of say, think, go and zero with other forms accounting for less than two percent. It is perhaps not surprising that there is no use of be like among this group, since its use is reported to have emerged in British English among young people only in the early 1990s (Buchstaller 2006a). In contrast to this, the adolescents have a much wider range of quotatives: all of those used by the elderly speakers in addition to be like, this is + speaker and a small proportion of tell (Examples 3–5).

(3) the police will come round like and they’re like “I want to search all of youse” (Jess, 18:55) (4) this is him “I’ve never seen a girl like it” (Laura_2, 16:20)

Table 1.  Distribution of quotatives among inner London speakers

Say Think Go Zero Be Like This is + speaker Tell Others Total N

Older speakers

Adolescents

70.8% (261) 4.1% (15) 4.6% (17) 18.9% (70)

27.4% (351) 12.8% (164) 11.7% (150) 15.1% (193) 24.4% (313) 4.8% (61) 1.9% (24) 2% (26) 1282

1.6% (6) 370

 Sue Fox



(5) but I told her “I’ve grown and I’ve learnt from my mistakes”(Adam, 10:10) (6) he goes “why do you think that?” (Tina_2, 16:40) (7) so I thought “ah I’ve had enough of this now” (Mark_2, 29:15)

The adolescents also have higher frequency rates of go and think (Examples 6–7). There may be several reasons for this. First, Buchstaller (2006a) suggests that the use of go is cyclical, so go may be undergoing a resurgence among the adolescents compared to the elderly speakers. Secondly, the higher frequency rate of think could be due to the fact that young people quote more of their inner thought processes in their narratives (Tagliamonte and D’Arcy 2007: 19). Indeed, in the data reported on here, 16% (N = 204/1282) of the total of quotatives used by the adolescents are used to report internal dialogue compared to just 4.6% (N = 17/370) among the elderly speakers. The quotative types used to report internal dialogue are predominantly be like and think for the adolescents with think being the most frequent form at 12.8% (N = 164/1282). For the elderly speakers, however, think accounts for almost all of the quotatives used to report internal thought at 4.1% (N = 15/370) overall, with zero used for the remaining 0.5% (N = 2/370). The higher use of think among adolescents could therefore be attributed to age-grading effects, inasmuch as young people may tell narratives in different ways to their older counterparts – including more of their thoughts and inner states (but see Buchstaller 2011). However, we must also take into account methodological differences in data collection which could contribute to the more varied alternation of quotative forms due to the different types of narratives elicited during the recordings. The majority of the older speakers were recorded individually in more traditional oneto-one sociolinguistic interviews where the fieldworker played a traditional interviewer role. In contrast, the majority of the adolescents were recorded with one, two or occasionally three friends, with the fieldworker sometimes playing a minimal role. Although the recordings from the elderly speakers included many narratives, the context in which the adolescents were recorded, that is to say with their friends, would seem to have given them more scope for lively, spontaneous narratives which were performed (Wolfson 1978) rather than a simple report of past events. In other words, the older speakers were more concerned with transmitting explicit information about a past event, whereas the storytelling of the adolescents occurred more in the context of social interaction and dramatic performance (Goffman 1986: 508) and could be viewed as an opportunity to ‘show off ’ in front of their friends. In this situation, it is perhaps unsurprising that these young people express their attitudes, emotions and opinions in the telling of their narratives; the quotative think might therefore be seen as a stance marker (see Buchstaller 2011). It captures the essence of evaluating something/someone



Performed narrative 

and expresses convergent or divergent alignment to that thing or person (Du Bois 2007). These observations, then, could go some way to explaining the high frequency of the think quotative among young people: notwithstanding the methodological differences, the high frequency of think and the wider range of quotative expressions used generally amongst the younger speakers could indeed reflect a difference in narrative style. I will return to the point that certain types of narratives lend themselves to a more varied alternation of quotative types in a later section. For now, though, I will focus specifically on the uses of be like and this is + speaker. 4. Be like Perhaps the most striking difference between the two age groups is the considerably reduced amount of say usage among the adolescents, with this quotative verb accounting for around 27% of the total number of quotatives – less than half the frequency of the older elderly speakers. However, as elsewhere among many young people in English speaking communities, there is substantial use of be like, representing around 24% of the total number of quotatives and seemingly having replaced say to a certain extent. It is clear from this distribution that be like is well established among London adolescents, even though the figure is not as high as some other communities that have been researched, for example in Canada, where be like “represents a full 58 percent of the total number of all quotatives” among Toronto youth (Tagliamonte and D’Arcy 2004: 501) and the USA, where be like “has established itself as the primary quotative of much of America’s vernacular English,” particularly in younger generations (Singler 2001: 276). In the UK, be like appears to have emerged a little later, seemingly in several locations simultaneously, and continues to be used with increasing frequency. In Glasgow, for example, be like accounted for 14% of quotative usage by university students in 1997 (Macaulay 2001). Similarly, in York, be like was used 18% of the time by university students in 1996 (Tagliamonte and Hudson 1999), but by 2006 this figure had increased to 60% (Baker, Cockeram, Danks, Durham, Haddican and Tyler 2006). On Tyneside, be like was also detected in the early 1990s at a rate of 5.7%, increasing to 21.2% by the late 2000s among speakers stratified by age, sex and class (Buchstaller 2011). In London, however, the frequencies are lower at only 0.5% reported in the Bergen Corpus of London Teenage Talk collected in 1993 (COLT, Stenström et al. 2002) and its use has not increased to the same extent as elsewhere (as shown in Table 1), although social class and ethnicity may be relevant factors yet to be explored within a British context.

 Sue Fox

In the following sections, I consider the social and linguistic constraints on the use of be like in London to see whether it is used in the same way in which it was first introduced in its earliest reported forms – with its “functional and social boots on” (Buchstaller and D’Arcy 2009: 292) – or if it has been reinterpreted during its incorporation into the local linguistic system. I will also focus on the newest quotative this is + speaker which, in Table 1, is shown to represent almost five percent (N = 61) of the total number of quotatives used among the adolescents and which is clearly a robust competitor in the quotative system of London adolescents. Again, the internal and external constraints operating on the use of this is + speaker will be considered to see whether any parallels can be drawn between this quotative and be like when its use was first reported. Some specific uses of this is + speaker will also be examined in order to consider the functions that new quotatives may fulfil in the discourse. The data are analyzed using Goldvarb X (Sankoff et al. 2005), a multivariate analysis application which reports the factors that significantly contribute to a given variant (here, a quotative form) being used. It also reveals the relative strength of each factor (reflected by the range), as well as the hierarchy of effects within each factor group. The data were coded for eight different factor groups: speaker, age, sex, ethnicity, grammatical person, content of the quote, tense/aspect and the lexical quotative used. These are factors that previous researchers have found to constrain quotative use (cf. Ferrara and Bell 1995; Tagliamonte and D’Arcy 2004; Tagliamonte and Hudson 1999). In the coding of grammatical person, first person contexts are taken to be all first person singular and plural pronouns. Third person contexts are taken to include all third person singular and plural forms of pronouns, and full noun phrases with the exception of tokens of existential it (N = 22) as in It’s like “I thought you had your own business” (Mark_2 5:58). Although it has been suggested that existential it forms may be playing a role in the grammaticalization of be like (Singler 2001: 261; see also Buchstaller 2004), it is generally agreed that the neuter pronoun, when it has existential status, is restricted to co-occurrence with be like and indeed this was the case in the data reported on here. On that basis, I follow other researchers (Buchstaller and D’Arcy 2009; Tagliamonte and D’Arcy 2004) and exclude the construction from the analysis. Also, all zero quotatives (N = 193) were excluded from this factor group and also from the tense and aspect group due to the fact that no overt grammatical person or verbal morphology occurs with the zero form. For the content of the quote, which was coded for speech, thought and non-lexicalized sounds and gestures, I excluded all tokens which were ambiguous (N = 9), as in I was like “what?” (Mark 1, 11:27), where it was unclear whether the constructed dialogue was direct speech or internal dialogue.



Performed narrative 

Table 2.  Contribution of external and internal factors on the use of be like among inner London Adolescents Input Total N

.30 1282 %

N

.42 .59 17

20 29

  655   627

.51 .50 .48

28 27 26

  521   485    42

.54 .26 .79

26 12 57

1045   204    21

32 29 29 44 40 51

  522   176   191     9    43    35

FW Sex

Male Female range Grammatical Person First Third Second Content Direct Speech Internal dialogue Non-lexicalized sounds/gestures range Tense/Aspect Simple past CHP Present Infinitive Habitual will Habitual would range

52 .52 .41 .47 .64 .57 .67 26

Table 2 displays the results of a multivariate analysis that tested for the effect of speaker sex, grammatical person, content of the quote and tense and aspect on the choice of be like among the inner London adolescents. The output of a multivariate analysis indicates the probability of occurrence of (in this case) be like versus all other quotatives as factor weights. Factor weights above .5 favour the occurrence of be like, while factor weights below .5 disfavour it and factor weights at or around .5 have little or no effect. Statistical significance of an effect is shown in bold type. As far as speaker sex is concerned, Ferrara and Bell (1995) predicted that females would use an innovative form such as be like more than males in the initial stages of use but that this difference would become neutralized over time, with males increasingly adopting the variable during the ongoing process of increasing grammaticalization towards its full use as a lexical verb. Baker, Cockeram, Danks,

 Sue Fox

Durham, Haddican and Tyler (2006) report that speaker sex was not selected as significant in their analysis of university students in York, thereby suggesting neutralization of sex difference and supporting Ferrara and Bell’s prediction. However, an analysis of Toronto Youth English (TYE) led Tagliamonte and D’Arcy (2004: 506) to suggest that “where be like is more frequent, it will have a sex differentiation”. In the Toronto study the sex effect was already statistically significant in the group of 10–14 year olds, with the girls ahead of the boys; however it was among the 17–19 year olds that the effect became the strongest factor, indicating a rise in female preference for be like. In inner London the results in Table 2 show that the sex effect is also statistically significant among the adolescents, with females leading in the use of be like in this area. However, it is a relatively weak factor that conditions the use of be like with a range value of only 17, in contrast to the strongest factor (content of the quote), which has a range value of 52. Furthermore, although not part of the research discussed here, the analysis of data from adolescents in outer London shows that it is the males (22%, FW .52) who have a higher frequency of use of be like than females (20%, FW .48), and while this is a statistically significant effect, the range value of only 4 suggests that it is also a relatively weak effect (Cheshire and Fox 2007). To summarize – and in line with Buchstaller and D’Arcy’s (2009: 309) conclusions that “speaker sex is not stable across space” – the results with respect to the sex effect are contradictory. Although this factor is shown to be statistically significant in both locations, it is females who are leading in inner London but males who are leading in outer London. I have already speculated that, with the lack of evidence to the contrary, be like appears to have emerged later in London than in other places and its use here does not seem to have followed the same path as the Toronto variety with respect to a strong female lead. The London results could indicate support for the prediction made by Ferrara and Bell (1995), who state that, over time, gender differences become neutralized. Alternatively, they could indicate that the use of be like simply does not carry the same social evaluation, as has been suggested by Tagliamonte and D’Arcy (2004) for Toronto. If the latter, the results provide support for the view that there are cross-societal differences in the changing social evaluation of be like (Baker et al. 2006; Buchstaller 2006b). Turning now to grammatical person, it can be noted that most studies have found that be like occurs most frequently in first person contexts. This is closely tied to its initial preferred use with internal dialogue – speakers quoting their own thoughts. Ferrara and Bell (1995: 270) claim that the expansion of be like into third person contexts would indicate an important step in the process of grammaticalization as this would suggest a transition from being predominantly an introducer of verbally non-committed thought or internal dialogue to introducing direct speech. However, there has been little evidence to suggest that this is the path that



Performed narrative 

be like is taking. For example, Tagliamonte and Hudson (1999: 161) found that be like was favoured in first person contexts among British and Canadian university students and, similarly, the study of Toronto youth revealed “no evidence of expansion into third person contexts” (Tagliamonte and D’Arcy 2004: 506). Carey (2004) and Baker, Cockeram, Danks, Durham, Haddican and Tyler (2006) also found that be like was still preferred in first person contexts. The results of this present study, then, are somewhat at odds with previous studies (although see Buchstaller 2011). In inner-city London, grammatical person is not statistically significant as a factor conditioning the use of be like. It is used in relatively equal proportions not only in first (FW .51) and third (FW .50) persons but also in second person contexts (FW .48). Although the number of tokens of use with second person is low, it is used with specific (8) and generic (9) reference. (8) no you was like “oh i wanna go back and see (name of person)”  (Ahmed, 9:54) (9) they say it so fast you’re like “whoa what did you just say?”  (Tina_1, 1:07:24) This, then, is the second piece of evidence which suggests that, when a quotative emerges in a new location, it does not necessarily do so with the same conditions attached to its use as in its earliest forms, thus supporting Buchstaller and D’Arcy’s (2009) cross-varietal evidence for this claim. Although be like has appeared as an innovation that has been adopted at the surface level, its users do not display knowledge of the original linguistic constraints or social evaluations attached to its use (Meyerhoff and Niedzielski 2003), a demonstration of an innovation having been “renegotiated” as it diffuses and situates itself into the local system (Britain 2002: 618). This is also in line with Buchstaller and D’Arcy’s view that there is “transformation under transfer” (Meyerhoff 2003) during the global diffusion of be like. Their comparative cross-variety study of American English, English English and New Zealand English demonstrated that while the surface form globalizes, the details of its social and linguistic constraints are recreated in locally specific ways (Buchstaller and D’Arcy 2009: 323). Do the results for the content of the quote provide us with further evidence that be like has diffused in this way and is more entrenched in the local linguistic system in London? If it is used in third person contexts, then we would expect its pragmatic functions to have expanded to include quoting direct speech as well as expressing internal thought associated more with first person contexts. Indeed, this is clearly the case. While be like is strongly favoured with non-lexicalized sounds and gestures (FW .79), we can see from Table 2 that it is also slightly favoured with direct speech (FW .54). Interestingly, it is strongly disfavoured with the use of internal dialogue (FW .26). The use of be like here is again in line with

 Sue Fox

Ferrara and Bell’s (1995: 279) suggestion of a developmental continuum, emerging as a quotative to frame non-lexicalized sounds, gestures and internal dialogue and then spreading to an introducer of direct speech as it grammaticalizes. This prediction is also supported by Carey (2004) who, in her analysis of York narratives, finds a significant change in the hierarchy of pragmatic functions and reports that, unlike earlier reports, be like is most favoured with direct speech followed by nonlexicalized sounds and then internal dialogue. The data presents additional evidence that be like is further entrenched in the linguistic system in this inner area of London. It may also suggest that, as innovations are adopted by new or younger speakers and as they spread to new locations, the constraints conditioning their use become weakened. Finally, let us look at the use of be like according to tense and aspect. Earlier research has tended to show that be like occurs more frequently with the present tense (Blyth et al. 1990; D’Arcy 2004; Romaine and Lange 1991; Singler 2001). However, not all of these studies have distinguished between the use of present tense to mark present temporal reference and the use of present tense to mark past temporal reference, that is, the conversational historical present (CHP). Importantly, Singler (2001: 271) notes that occurrences of present tense found in his corpora were almost always cases of historical present. Tagliamonte and D’Arcy (2007), in tracking the use of be like across three generations in Toronto, make the distinction and report that among the 30–39 year olds be like is favoured for all cases of present tense (i.e. present and CHP), whereas for younger speakers under the age of twenty it is clearly favoured in the historical present. The middle group of speakers, those aged 20–29 years old, are doing something in between the two; they favour be like with the historical present, while the present tense contexts fall on the median (Tagliamonte and D’Arcy 2007: 207). So can we predict from these findings that be like is most likely to be used with the CHP once it is separated from other present morphology contexts? The answer is not straightforward. Macaulay (2001: 10) and Carey (2004: 7) both report that in their British data, there is a greater tendency for be like to be used in the simple past. Similarly, Levey (2007) finds in his corpus of London preadolescents that be like occurs more frequently in the past tense. In the analysis presented here, past tense, present tense and CHP were separated and, in addition, be like was coded for uses with habitual will, habitual would and infinitive forms, as in Examples (10) to (12). (10) he would be like “aah shut up shut up...” (Serena, 1:07:12) (11) it’ll be like “why you always chatting English for?” (Fatima_2, 14:4) (12) do you get me.. be like “no my daughter it’s a good girl she just goes to college and does her work and comes home” (Fatima 1b, 1:12:55)



Performed narrative 

These three contexts all favour the use of be like, although the numbers of tokens are low and, as a result, I do not investigate this aspect of variation further. As for the past, present and CHP contexts, the frequencies are all around 30% and there is no strong favouring or disfavouring effect. If anything, the CHP is the least favoured context with a factor weighting of .41. Could this point to a difference between North American varieties and British varieties? If so, why should this be the case? What is clear is that reference to past events favours the use of be like (at least, in its early stages of use and in many of the varieties studied) regardless of whether or not present or past morphology is used to mark the past temporal reference. What seems important here is that both past tense and CHP often alternate in narratives which occur in everyday spontaneous conversations, and some studies (Tagliamonte and Hudson 1999; Winter 2002) have found that most examples of be like identified in the data have occurred in narratives. Indeed, Tagliamonte and D’Arcy (2007: 19) have suggested that narratives are crucial in explaining the rapid rise of be like. They demonstrate that the emergence of be like coincides with a rise in the use of inner thought processes in narratives among young speakers, as mentioned earlier. They suggest that the use of internal monologue in the telling of stories was expanding before be like entered the system and the newcomer was therefore able to fill a niche that already existed. It subsequently specialized as a device for foregrounding the narrator as a participant in his or her own story. The use of the CHP is also a well-known device for adding a performance element to a narrative and to add structure and drama to the story (Wolfson 1978: 217). Seen in this light, it may be the case that, in the early stages of innovation, new quotatives emerge with the pragmatic function of dramatising a ‘performed narrative’ and for this reason may have a close association with the CHP. This would certainly seem to account for the use of be like within American varieties, where it is assumed to have first arisen (Blyth et al. 1990: 224; Dailey-O’Cain 2000: 76; Tagliamonte and D’Arcy 2007: 212) as well as other quotative markers such as German und ich so (Golato 2000), Norwegian bare (Opsahl et al. this volume) and Dutch van (Foolen et al. 2006), which also seem to be associated with the pragmatic function of turning a quote into a performance. This is an important point that I will come back to when considering the emergence of another new quotative, this is + speaker, in London. To return to the results of be like in London, how can we account for the difference with respect to the tense and aspect effect reported in Northern American varieties? There seem to be two possible explanations. The first could be that different narrative strategies are employed by British speakers when compared to their Northern American counterparts. On the face of it, this looks to be a plausible explanation since, in their analysis of British and Canadian data, Tagliamonte and Hudson (1999) show that the British narratives have more introspection

 Sue Fox

overall (28%) than their Canadian counterparts (12%). For differences in tense and aspect, Buchstaller and D’Arcy (2009) show that be like is favoured with the CHP in both American and New Zealand English but disfavoured in English spoken in England; they point out that these differences could also be due to different narrative styles (although the corpora from which their data were extracted do not consist entirely of narratives). A comparative study which looks at the frequency of the CHP together with the use of be like specifically in spontaneous everyday narratives across comparable age cohorts would, therefore, be worthy of further investigation to highlight any such differences. Secondly, this chapter has demonstrated that the use of be like in London does not seem to have emerged with the same constraints as regards sex, grammatical person or content of the quote compared to its first emergence in other varieties; perhaps there is no reason to expect that tense and aspect should conform to the same constraints either. As suggested previously, these differences could come about as a result of the reorganization of constraints when new forms are adopted by new speakers in new locations, who may not have the same social evaluations towards them or, alternatively, who might restructure an innovation in order to imbue it with a distinct local, social and linguistic meaning (Buchstaller and D’Arcy 2009). 5. This is + speaker Let us now look at the new quotative which is found in the data from the innerLondon adolescents. The construction this is + speaker does not occur in the outer-London data and has not been reported in any other English variety. It is used with a range of subjects, as shown in Examples (13) to (17). (13) this is them “what area are you from . what part?” this is me “I’m from (inner London)” (Alex_1, 20:20) (14) this is her “that was my sister” (Tina_1, 26:20) (15) this is him “don’t lie . if I search you and if I find one I’ll kick your arse”  (Dom, 8:50) (16) this is my mum “what are you doing? I was in the queue before you”  (Alex_1, 36:40) (17) this is my mum’s boyfriend “put that in your pocket now”  (Alex_1, 1:18:05) What is the source of this is +speaker? As a quotative form, it is unremarkable, even though it has not been attested before for English. It presents the speaker in the end-focus position of the clause, where new information typically occurs (this assumes, perhaps simplistically, that the expression can be seen as a full clause, albeit



Performed narrative 

one that is perhaps now grammaticalizing into a semi-fixed expression). Quotatives that present the speaker in this way are common in the world’s languages and, in particular, are widely attested in African languages (Güldemann this volume). Furthermore, deictics are not unusual in quotative expressions, and other deictic forms do occur in varieties of English in quotative expressions. Irish English, for example, has here’s (speaker), as in Example (18), and the deictic form of I’m sitting there has been observed in a variety of southern American English, as in Example (19). (18) Here was I “Then I must be hard of hearing or something – you rapped the door and I didn’t hear you”... out the back and everywhere they were. Here’s me “Have youse took leave of your senses? He says – uh – “get everybody up, everybody up” (Milroy and Milroy 1977: 54) (19) I’m sittin’ there, “Wow, dude! Slap bracelets!” (Stein 1990: 303) Güldemann has pointed out that a similar presentational clause using evo (t)here to introduce reported discourse can be found in Serbian and Croatian. Interestingly, in the context of the discussion here, this construction is commonly used in spoken narratives; “It (...) occurs especially in contexts where the reporter wants to vividly perform a dynamic verbal interaction between narrative characters, in particular, involving quote content that is unexpected and remarkable before the background of the previously available information.” (Güldemann this volume) and, as we will see, in this way parallels the use of this is + speaker. The German quotative expression und ich so also contains a deictic element, so (Golato 2000: 43, this volume). These deictic forms can be seen as examples of empathetic deixis (Lyons 1977) or emotional deixis (Lakoff 1974), a use of deixis that is common in speech, especially in high involvement contexts such as narratives told within adolescent peer groups (Cheshire 1989), where the pace of speech is typically fast (Stenström et al. 2002). In fact, there is some evidence that this is + speaker may be an example of what Gordon and Trudgill (1999) term an “embryonic form”. There is one example of its use in Mark Sebba’s (p.c.) recordings of London Jamaicans, made in the 1980s and, while there is not a full analysis available as of yet, there is further evidence of this quotative’s existence in the COLT corpus, which was collected in 1993 and consists of the spoken language of thirteen to seventeen year old teenagers from different London boroughs. Examples (20) and (21) come from a fourteen year old female whose first language is British English and who is described as “ethnic minority” (no further details given). (20) he goes “this is for you” this is me “thanks”

 Sue Fox

(21) this is Jane to me the other day “throw your kitten off my floor and see if it lands on its feet.  (both examples extracted from COLT REF: B1327075) The variety of languages in contact in London prevents me from stating with any certainty whether the quotative is directly derived from the London Jamaican of the first generation of immigrants or whether it is due to the influence of the African languages (or, indeed, any of the other languages) spoken by some of the speakers but it would seem likely that the contact setting itself has played a role in the innovation of this quotative form (Cheshire et al. 2011; Mufwene 2001). Although this is + speaker is a minor part of the quotative system, the fact that it has a distinct linguistic and sociolinguistic profile (to be discussed below) suggests that its use is robust and that it is likely to maintain its position or that it may even gain ground on its competitors. Furthermore, its use occurs frequently enough to have been noticed by writers of TV shows, so that it is used by television personalities, usually in comedy sketches, to portray disaffected youth (e.g. the RAF pilot sketches in which the comedians Alexander Armstrong and Ben Miller6 use a mismatch of RP accents with colloquial and non-standard expressions). Dion and Poplack (2007) have suggested that the media has a limited role in the diffusion of such innovations, but the possibility of this kind of portrayal in the media playing a role in the spread of this innovation cannot be discounted. Table 3 displays the results of a multivariate analysis that tested for the effect of speaker sex, grammatical person, content of the quote and tense and aspect on the choice of this is + speaker among the inner London adolescents. As already acknowledged, the application value is low but – similar to Singler (2001) and Buchstaller, Rickford, Traugott, Wasow and Zwicky (2010) – I would like to argue that a multivariate analysis is nevertheless worthwhile, since a variationist analysis can provide us with important insights into the patterning of an innovation when it first emerges and thus allows a comparison between these results and earlier analyses of be like. It also presents an exciting opportunity to investigate the constraints on a change in progress occurring in a complex variable such as the quotative system (as a whole). Table 3 shows that the sex effect is statistically significant, with females leading in the use of this new quotative (FW .62). This is perhaps unsurprising and supports Ferrara and Bell’s (1995) prediction that an innovation in the quotative system of English would be introduced and used more by females than males in the initial stages. In this respect, this is + speaker seems to run parallel with be like 5.

Thanks to Jessica Mardell for supplying these examples

6. For an example of one of these sketches illustrating the use of this is + speaker, go to http:// www.youtube.com/watch?v=lwNQf08Kxsw



Performed narrative 

Table 3.  Contribution of external and internal factors on the use of this is + speaker among inner London Adolescents Input Total N

.04 1282 %

N

.38 .62 24

 2  7

655 627

.65 .35 30

 8  4

521 507

 1 34

191 176

FW Sex Male Female range Grammatical Person First Third range Content Direct Speech Tense/Aspect Simple present CHP range

100 .10 .92   82

when it was first introduced. The new form is also favoured in first person contexts (FW .65) in much the same way that be like has been reported (as discussed earlier). It is also strongly favoured in the CHP and once more runs parallel with be like in its earliest forms, which suggests that narratives play an important role in the introduction of new quotatives (Blyth et al. 1990; Buchstaller 2011; Tannen 1986). However, the noticeable difference between be like and this is + speaker is that the latter is used categorically with direct speech, whereas be like was said to be used initially for dramatic effect to introduce non-lexicalized sounds and for expressing the inner thoughts and emotions of the speakers (Butters 1982; Tannen 1986), as captured in expressions such as I’m like “Oh my god” (Ferrara and Bell 1995: 283). A more detailed examination of the uses of this is + speaker (below) reveals that new quotatives fulfil a specific function in the discourse. 6. Pragmatic functions of new quotatives As was already noted, narratives appear to play an important role in the introduction of new quotatives and, in fact, this is + speaker only occurs in narratives of personal experience in these data. Moreover, these narratives are often performed,

 Sue Fox

where “the function of performance is to structure the experience from the point of view of the speaker and to dramatise it” (Wolfson 1978: 216). At least fifty of the sixty-one occurrences of this is + speaker in these data occur in narratives of this kind. Wolfson argues that the degree to which a story can be said to be performed depends on the number of performance features it contains and on the amount that each performance feature is used. The list of performance features she gives are 1) direct speech, 2) asides, 3) repetition, 4) expressive sounds, 5) sound effects and 6) motions and gestures, all of which are features that are similar to those in an actual theatrical performance. She further hypothesizes that the CHP and, more specifically, the alternation between the CHP and the past tense is an additional performance feature which gives structure and drama to the story being performed (see also Schiffrin 1981). I would argue that, by extension and by their association with direct speech, expressive sounds, sound effects and motion and gestures, the use of quotatives, or to be more precise, the alternation between different quotatives, could be added to the list of performance features. This is because alternation between the different forms can be viewed as a performance feature which structures the narrative by drawing attention to the different characters in the story while also dramatizing the story being performed. Shifts between one quotative form and another can also emphasize dramatic high spots in the narrative. Perhaps, then, the function of new quotatives entering the system is to highlight a particularly dramatic peak in the performance. This is also suggested by Buchstaller’s (2003: 8) use of the term mimesis markers for quotatives, a term that highlights the enacted element of a narrative. She suggests that new quotatives are stylistically marked and speakers therefore choose them as “focus quotatives, to introduce quotes with emotionally-heightened material rendered by mimetic enactment” (Buchstaller 2003: 9). Subsequently, as each new quotative becomes more and more grammaticalized and starts to fulfil more general functions in the discourse, a gap for a new quotative starts to emerge. This explanation may also go some way to understanding why quotatives such as go sometimes lie dormant in the system and are then picked up again according to ambient trends (Buchstaller 2006a: 4); their resurgence may act in the same way as a new quotative entering the system. The fact that new quotatives appear to be introduced at heightened moments in the discourse also accords with claims that speakers innovate in order to be more expressive (Lehmann 1985) or “extravagant” (Haspelmath 1999: 1057) in order to achieve social success. According to Haspelmath (1999), some speakers do not only want to be clear but also want to be imaginative and vivid in order to be noticed; what better point to introduce such an innovation than when performing a story? If the innovation is later adopted by other speakers and if it spreads throughout the linguistic community (as in the case of be like), there comes a point when the notion of extravagance is no longer valid, once again leaving a gap for a brand-new innovation to emerge and



Performed narrative 

take over this role. Let us now look at some specific uses of the quotative this is + speaker to demonstrate both its role in the discourse and the fact that it seems to occur at a particularly dramatic moment in the narrative. The Bike Incident (Extract 1) is a narrative extracted from a two and a half hour recording of two boys, Alex and Zack, who are friends and both aged sixteen. The recording took place in a small room situated close to the student common room of the college where the two boys were attending a course in painting and decorating. They were talking to the fieldworker about their school experiences and Zack was responding to being asked why he had been kicked out of school. I follow Labov and Waletzky (1967) and Labov’s (1997) narrative analysis framework. This framework differentiates between orientation clauses that set the scene for the story (in the Bike incident lines 1–5, 18), complicating action clauses that give the blow-by-blow account of what happened (lines 6–27), a resolution to tell the listener what finally happened (line 28) and an evaluation (line 29), which makes the point of the story clear. Extract 1: The Bike Incident

(Alex_Zack 1, 1:26:15)

  1 Zack: no it was like . it was the end of school yeh so that school’s finished yeh   2 and everyone was going home   3 and I was getting my bike from the bike rack   4 and I was going out   5 and I was riding my bike   6 and he stopped my bike.   7 I was like “yeh”   8 and he goes “get off the bike”   9 I was like “why am I getting off the bike I’m going home like I’ve gotta go home” yeh 10 he was like “no get off the bike walk the bike outside of school” 11 I was like “what’s the point?” 12 yeh cos like it’s quite far like to get out the school from the entrance like in the school yeh 13 and he goes “ah no get off the bike” yeh 14 so like he kind of shoved me off the bike 15 so I dropped it but I didn’t fall over like but I kind of stumbled yeh 16 and he put his. he tried to take my bike up to his office like he was gonna keep my bike there. 17 I was like “no” like 18 and this time everyone was gathering round cos we were shouting at each other yeh 19 he was like “no I’m taking your bike upstairs”

 Sue Fox

20 I was like “what’s the point in that when I’m just gonna take it back downstairs” 21 so I must have pulled the bike off him yeh 22 and I put it . I put it I leant it up against the wall yeh 23 and I walked over to him 24 and this is me “what. what’s your . what’s your problem?” 25 and he goes “I don’t like you” 26 I was like “I don’t like you” yeh 27 so I just swung for him and then we like . 28 but we had a fight though. [S: did you] and I got kicked out of school . 29 like I weren’t allowed into any school that’s why I came here last year According to Wolfson’s (1978: 216) criteria (and if we accept her hypothesis that the alternation between CHP and past tense provides a seventh performance feature), we can see that this narrative can be analyzed as a performance rather than a simple report of past events; it contains instances of constructed dialogue, repetition and several switches of tense. In performance narratives such as this we can also draw on Wierzbicka’s (1974) view that direct speech is characterized by its theatrical nature and that quotations can be seen as imaginary speech performances where the speaker performs utterances as a way of allowing the hearer “to SEE for himself ” what the originator of the quotation said (Wierzbicka 1974: 272). In other words, the quotation becomes a demonstration “of what a person did in saying something” (Clark and Gerrig 1990: 769). In this narrative we can also see a mixture of quotative forms being used: be like, go and this is + speaker. Of course, variation of quotative forms is quite typical in spontaneous storytelling and it is proposed here that no narrative will be told using one quotative all the way through. In fact, I would argue that the use of the different quotatives can be seen in much the same way as the alternation between CHP and past tense operates in that no narrative is told entirely in CHP no matter how “performed” the story is (Wolfson 1978: 218). Hence, the use of the different quotatives is not of great importance in itself. Rather, just as others have claimed for the alternation between CHP and past tense (Schiffrin 1981; Wolfson 1978), it is the switch itself which is the significant feature. In Extract 1, Zack switches between past tense be like and CHP go in lines 7–20 of the “complicating action clauses” (Labov 1972; Labov and Waletzky 1967), those clauses which are temporally ordered and tell the listener what happened next. However, on line 21, the story reaches a dramatic climax when the situation moves from verbal confrontation to physical confrontation. At the dramatic moment when Zack walks over to the teacher and comes face to face with him, he switches to this is me (line 24). The use of deictic this brings the speaker in the story into focus and involves the audience more directly in the recounting of the events, producing a “mutual focus of speaker and co-participant” (Golato 2000: 43).



Performed narrative 

If we consider this is me to be a short clause, then me (or the speaker) is in end position and, we might argue, this construction has more rhetorical force than simply personal pronoun plus quotative verb as in I said/she said. As a deictic quotative, it places emphasis on who is talking rather than the focus being on the content of what is being said, while at the same time it has the effect of detaching the narrator from the character in the story. As such, the speaker is presented as a character in a theatrical performance, and in this way the listener is invited to see the event through the eyes of the speaker. In doing so, the narrator (here Zack) seems to invoke the support of his listeners (in this instance the fieldworker and his friend) and invites them to adopt his point of view and support his actions during this confrontation. Let us now look at another example. The second extract comes from a later part of the recording of the same two boys. Before this extract, Alex and Zack had been talking between themselves for the previous two minutes about going on holiday and ways of bringing banned goods back into the country. This time it is Alex speaking and although the fieldworker is present, Alex is principally telling his friend Zack (demonstrated by the use of the address terms bruv on lines 1, 12 and 27 and blud on line 5) about an incident at the airport when his friend had concealed goods in a deodorant can. Extract 2: At the Airport

(Alex_Zack 2, 0:56:35)

  1 Alex: /he had like/ half ounces in the deodorant can bruv.   2 cos where the punk’s so compressed yeh and it weighs heavy [Z: mm] they just wrap it up .   3 bang and it just goes right in the middle of the liquid   4 they put a l. the. the spray bit the hole they make a hole right through the middle of it yeh   5 blud they move it like that. so they make it like that make a little hole. they just stick it in .   6 bang. done everything. weld that little bit on. smack   7 all you have to do is go crack like that with a chiv gone   8 but I see him come back yeh and he opened .   9 one time he sprayed the. thing yeh cos [Z: punk] the export people innit 10 they sprayed the spray yeh [S: mm] 11 like just to check that it weren’t anything. 12 bruv when I say they were smelling weed 13 this is him. this is them “what’s that smell that’s coming out?” 14 this is him “oh i dunno like it must be d.d. thing” 15 this is them. but on the back yeh what he didn’t see yeh it had weed spray.. yeah?

 Sue Fox

16 but not like weed spray what you use to spray the weed like . 17 like it was called weed spray that’s the smell of it yeh [Z: mhm] 18 so the man’s looking at it now proper staring at it 19 and he’s thinking “weed spray”? 20 and this is him “yeh yeh like you know weed spray give you a nice smell like” 21 and the man just was so dumb 22 he went “yeh yeh my friend” 23 bang put it in a bag. 24 let them go. 25 some of them they can’t be bo. in amsterdam 26 it’s . you can get through their customs like that 27 cos they all. they smoke weed in the airport bruv . 28 I’ve seen them smoke weed /in the airport/ they are mad 29 Zack: /in the café/ Again, this narrative has all the elements of a performance: constructed dialogue, asides, repetitions, expressive sounds and sound effects. Perhaps it is even more of a performance because Alex is performing for a friend who he wants to impress with his story, a way of showing off about a topic which holds a lot of currency among his peer group. Alex sets the scene at the airport. At the dramatic point in the narrative (line 13) when the protagonist is challenged by the customs officer about the contents of the deodorant can, he uses this is them and this is him. In so doing, he puts the characters in opposition to one another and provides a theatrical staging for the event being recounted. We are led via the deictics to consider the actions of the two juxtaposed characters in the story. Indeed, lines 13 and 14 represent a narrative peak and I would like to argue that, by using these new quotative expressions, Zack is indicating that the reported speech and the situation that it was a part of is of great importance to the story; he is accentuating the strong impact of what was said or what took place by this dramatic enactment (Foolen 2008: 123). Note that as soon as the narrator reports that the customs officer appeared satisfied that the contents of the can were innocuous, Alex reverts to using past tense go (line 22) to conclude the story. To summarize, use of the new quotative this is + speaker and especially its alternation with other quotative forms in these narratives supports the argument that the alternation itself acts as an additional performance feature in stories that are performed. Extracts 1 and 2 are just two examples from the data that demonstrate the fact that new quotatives – when they first enter the quotative system – can be used by speakers to highlight a particularly dramatic moment in the performed narrative.



Performed narrative 

7. Discussion and conclusions The analysis presented here has demonstrated that the extensively reported new quotative be like is now well established in its use among London adolescents. The frequency of be like is still low compared to some US and Canadian varieties and, indeed, some varieties within the UK. Be like seems to have emerged in London after its use in other British locations such as Glasgow and York.7 Crucially, however, I have discussed an innovation in the quotative system that does seem to have emerged in London, namely this is + speaker. This innovation appears to be a very recent phenomenon and catching it at such an early stage of its use has provided an ideal opportunity to analyze how it is used in the discourse. An examination of the contexts in which it occurs shows that this is + speaker is used in narratives of personal experience which are “performed” (Wolfson 1978) and that quotative alternation in general may help to structure and add drama to a performed story. Furthermore, I have suggested that new quotatives, when they first emerge, may fulfil the function of highlighting a particularly dramatic peak in the performing of a story. Note that the above claim would also seem to provide an explanation for the emergence of be like. It has already been established in previous studies that the role of narratives is crucial in explaining its rapid rise and that, initially at least, be like was associated with narratives recounting inner thought processes, particularly among young speakers (Tagliamonte and D’Arcy 2007). Indeed, early reports stated that the “prototypical case of be + like is a theatrical, highly conventionalised utterance which makes the inner state transparent to the audience” (Ferrara and Bell 1995: 283), which again emphasizes the dramatic function of a new quotative. Similarly, the particle so in the German quotative construction und ich so is said to allow the speaker to give a visual representation of what was going on during the reported event, so that again the event becomes transformed from a mere telling of a story to a performance to be “watched and listened to” by the interlocutor (Golato 2000: 45). Norwegian bare is also said to have an intensifying function in that it, too, frequently introduces quotes with a dramatic character or in narrative peaks (Opsahl et al. this volume). Above all, the dramatic function of new quotatives might explain why the time is right for such an innovation to emerge in London. I have suggested that, as a quotative becomes more established and spreads to new users and particularly when it spreads to new locations, the constraints attached to its use can become weakened. This may be because the new 7. This is assuming that frequency rates are a true reflection of when be like emerged in the different locations. As mentioned previously, frequency of occurrence and diachronic entrenchment need not be necessarily related and, of course, there may be other social dimensions at play which have yet to be investigated.

 Sue Fox

speakers do not have the same social evaluations towards the quotative or they simply do not pick up on those constraints. The pragmatic functions of the quotative thus become more generalized, leaving a gap in the system for a new quotative to emerge that specifically fulfils the function of highlighting a particularly dramatic peak in the narrative. In some cases the process could involve the resurgence of a quotative that has been latent in the system, such as go (Buchstaller 2006a), or it might involve the emergence of a brand-new quotative, as in the case reported here of this is + speaker. I have demonstrated that the quotative be like is used in London within a wide range of grammatical contexts: it does not have a strong female lead, it is associated more with direct speech than with internal thought and it is not associated strongly with first person contexts nor with the CHP. This suggests that be like is not predominantly associated with highly dramatic peaks in narratives, thus leaving a gap to be filled by another quotative which does fulfil this function, here this is + speaker. It may be the case that the life span for this is + speaker as a quotative construction will be short-lived. Alternatively, it may cease to have the specific function of introducing constructed dialogue at the dramatic peak of a narrative and start to broaden its function in the discourse. The latter path would parallel the attested development of be all, which emerged in the 1980s, peaked in the 1990s and then declined in use as a quotative but continued to expand as an intensifier (Rickford et al. 2007). We do not yet know whether another construction (either innovative or recycled) will take the place of this is + speaker or whether its use will be taken up globally in the same way that be like has been spreading to other speakers and other locations. Whatever the future of this is + speaker, it has provided a rare opportunity of capturing and investigating a feature during the initial stages of its development and this analysis has highlighted the importance of taking into account the role of pragmatic function in the process of language change. References Bailey, Guy, Wikle, Tom, Tillery, Jan and Sand, Lori. 1991. “The apparent time construct.” Language Variation and Change 3: 241–264. Baker, Zipporah, Cockeram, David, Danks, Esther, Durham, Mercedes, Haddican, Bill and Tyler, Louise. 2006. “On the diffusion of be like: Real-time evidence from York, England.” Paper presented at NWAV 35, The Ohio State University, November. Blyth, Carl, Jr., Recktenwald, Sigrid and Wang, Jenny. 1990. “I’m like ‘Say what?!’: A new quotative in American oral narrative.” American Speech 65: 215–227. Britain, David. 2002. “Space and spatial diffusion.” In The Handbook of Language Variation and Change, J.K. Chambers, Peter Trudgill and Natalie Schilling-Estes (eds), 603–637. Oxford: Blackwell.



Performed narrative  Buchstaller, Isabelle. 2003. “The co-occurrence of quotatives with mimetic performances.” Edinburgh Working Papers in Applied Linguistics 12: 2–10. Buchstaller, Isabelle. 2004. The Sociolinguistic Constraints on the Quotative System: British English and US English Compared. Unpublished PhD thesis. University of Edinburgh. Buchstaller, Isabelle. 2006a. “Diagnostics of age-graded linguistic behaviour: The case of the quotative system.” Journal of Sociolinguistics 10(1): 3–30. Buchstaller, Isabelle. 2006b. “Social stereotypes, personality traits and regional perception displaced: Attitudes towards the ‘new’ quotatives in the U.K..” Journal of Sociolinguistics 2006 10(3): 362–381. Buchstaller, Isabelle. 2011. “Quotations across the generations: A multivariate analysis of speech and thought introducers across 4 generations of Tyneside speakers.” Corpus Linguistics and Linguistic Theory. Special issue: Corpus Linguistics and Sociolinguistic Inquiry. 7(1): 59–92. Buchstaller, Isabelle and D’Arcy, Alexandra. 2009. “Localized globalisation: A multi-local, multivariate investigation of quotative be like.” Journal of Sociolinguistics 13(3): 291–331. Buchstaller, Isabelle, Rickford, John, Traugott, Elizabeth, Wasow, Thomas and Zwicky, Arnold. 2010. “The sociolinguistics of a short-lived innovation: Tracing the development of quotative all across spoken and internet newsgroup data.” Language Variation and Change 22: 1–29. Butters, Ronald. 1982. “Editor’s note (on ‘be + like’).” American Speech 57: 149. Carey, Lucy. 2004. “I was like ‘Oh God’. The changing quotative system in British English.” Newcastle and Durham Working Papers in Linguistics 10. Cheshire, Jenny. 1989. “Addressee-oriented features in spoken discourse.” York Papers in Linguistics 13: 49–64. Cheshire, Jenny and Fox, Sue. 2007. “This is me, this is him: Quotative use among adolescents in London.” The Second International Conference on the Linguistics of Contemporary English (ICLCE 2), Toulouse, July. Cheshire, Jenny, Kerswill, Paul, Fox, Sue and Torgersen, Eivind. 2011. “Contact, the speech community and the feature pool: The emergence of Multicultural London English.” Journal of Sociolinguistics 15(2): 151–196. Clark, Herbert. H. and Gerrig, Richard. J. 1990. “Quotations as demonstrations.” Language 66 (4): 764–805. COLT: The Bergen Corpus of London Teenage Language. Aksis, Allegaten 27, N-5007 Bergen, Norway. Dailey-O’Cain, Jennifer. 2000. “The sociolinguistic distribution of and attitudes toward focuser like and quotative like.” Journal of Sociolinguistics 4(1): 60–80. D’Arcy, Alexandra. 2004. “Contextualising St John’s youth English within the Canadian quotative system.” Journal of English Linguistics 32(4): 323–345. Dion, Nathalie and Poplack, Shana. 2007. “Linguistic mythbusting: The role of the media in diffusing change.” Paper presented at NWAV 36, University of Pennsylvania, October. Du Bois, John. 2007. “The stance triangle.” In Stancetaking in Discourse, Robert Engelbretson (ed.), 139–182. Amsterdam: John Benjamins. Ferrara, Kathleen and Bell, Barbara. 1995. “Sociolinguistic variation and discourse function of constructed dialogue introducers: the case of be + like.” American Speech 70(3): 265–290. Foolen, Ad. 2008. “New quotative markers in spoken discourse.” In Empirische Forschung und Theoriebildung: Beiträge aus Soziolinguistik, Gesprochene-Sprache-und Zweitspracherwerbs­ forschung – Festschrift für Norbert Dittmar zum 65. Geburtstag (Taschenbuch), Bernt

 Sue Fox Ahrenholz, Ursula Bredel, Wolfgang Klein, Martina Rost-Rot, Romuald Skiba (eds), 117–128. Frankfurt: Peter Lang. Foolen, Ad, Van Alphen, Ingrid, Hoekstra, Eric, Lammers, Henk, Mazeland, Harrie and Pascual, Esther. 2006. “Het quotatieve van. Vorm, functie en sociolinguïstiche variatie.” Toegepaste Taalwetenschapin Artikelen, no 76. 5(2): 137–149. Goffman, Erving. 1986. Frame Analysis. An Essay on the Organization of Experience. Boston, MA: Northeastern University Press. Golato, Andrea. 2000. “An innovative German quotative for reporting on embodied actions: Und ich so/und er so ‘and I’m like/and he’s like’.” Journal of Pragmatics 32: 29–54. Gordon, Elizabeth and Trudgill, Peter. 1999. “Shades of things to come: Embryonic variants in New Zealand sound changes.” English World-Wide 20(1): 111–124. Haspelmath, Martin. 1999. “Why is grammaticalization irreversible?” Linguistics 37 (6): 1043–1068. Labov, William. 1963. “The social motivation of a sound change.” Word 19: 273–309. Labov, William. 1972. Language in the Inner City. Philadelphia: University of Pennsylvania Press. Labov, William. 1997. “Some further steps in narrative analysis.” Journal of Narrative and Life History 7: 395–415. Labov, William and Waletzky, Joshua. 1967. “Narrative analysis: oral versions of personal experience.” In Essays on the Verbal and Visual Arts. Proceedings of the 1966 Annual Spring Meeting of the American Ethnological Society, June Helm (ed.), 12–44. Seattle: University of Washington Press. Lakoff, Robin. 1974. “Remarks on this and that.” Proceedings of the Tenth Regional Meeting of the Chicago Linguistics Society, 345–356. Lehmann, Christian. 1985. “Grammaticalization: synchronic variation and diachronic change.” Lingua e Stile 20(3): 303–318. Levey, Stephen. 2007. The next generation. Aspects of grammatical variation in the speech of some London preadolescents. Unpublished PhD thesis. Queen Mary, University of London. Lyons, John. 1977. Semantics, 2 vols. Cambridge: Cambridge University Press. Macaulay, Ronald. 2001. “You’re like ‘why not?’ The quotative expressions of Glasgow adolescents.” Journal of Sociolinguistics 5(1): 3–21. Matthis, Terrie and Yule, George. 1994. “Zero quotatives.” Discourse Processes 18: 63–76. Meyerhoff, Miriam. 2003. “Formal and cultural constraints on optional objects in Bislama.” Language Variation and Change 14(3): 323–346. Meyerhoff, Miriam and Niedzielski, Nancy. 2003. “The globalisation of vernacular variation.” Journal of Sociolinguistics 7(4): 534–555. Milroy, James and Milroy, Lesley. 1977. “Speech and context in an urban setting.” Belfast Working Papers in Language and Linguistics. 1: 1–85. Mufwene, Salikoko. 2001. The Ecology of Language Evolution. Cambridge: Cambridge University Press. Rickford, John, Wasow, Thomas, Zwicky, Arnold and Buchstaller, Isabelle. 2007. “Intensive and quotative all: Something old, something new.” American Speech 82(1): 3–31. Romaine, Suzanne and Lange, Deborah. 1991. “The use of like as a marker of reported speech and thought: A case of grammaticalization in progress.” American Speech 66(3): 227–279. Sankoff, David, Tagliamonte, Sali and Smith, Eric. 2005. “Goldvarb X: a multivariate analysis application.” Department of Linguistics, University of Toronto. Available at http://individual.utoronto.ca/tagliamonte/goldvarb.htm Schiffrin, Deborah 1981. “Tense variation in narrative.” Language 57(1): 45–62.



Performed narrative  Singler, John. 2001. “Why you can’t do a VARBRUL study of quotatives and what such a study can show us.” University of Penn Working Papers in Linguistics.Vol.7(3): 257–278. Stein, Ellen. 1990. “I’m sittin’ there: Another new quotative?” American Speech 65: 303. Stenström, Anna-Brita, Andersen, Gisle and Hasund, Ingrid Kristine. 2002. Trends in Teenage Talk. Amsterdam: John Benjamins. Tagliamonte, Sali and D’Arcy, Alexandra. 2004. “He’s like, she’s like: The quotative system in Canadian youth.” Journal of Sociolinguistics 8(4): 493–514. Tagliamonte, Sali and D’Arcy, Alexandra. 2007. “Frequency and variation in the community grammar: Tracking a new change through the generations.” Language Variation and Change 19: 199–217. Tagliamonte, Sali and Hudson, Rachel. 1999. “Be like et al. beyond America: The quotative system in British and Canadian youth.” Journal of Sociolinguistics 3(2): 147–172. Tannen, Deborah. 1986. “Introducing constructed dialogue in Greek and American conversational and literary narrative.” In Direct and Indirect Speech, Florian Coulmas (ed.), 311–332. Berlin: Mouton. Wierzbicka, Anna. 1974. “The semantics of direct and indirect discourse.” Papers in Linguistics 7: 267–307. Winter, Joanne. 2002. “Discourse quotatives in Australian English: Adolescents performing voices.” Australian Journal of Linguistics 22(1): 5–21. Wolfson, Nessa. 1978. “A feature of performed narrative: The conversational historical present.” Language in Society 7: 215–237.

Appendix 1: Transcription Conventions [S: mm] Small feedback responses in square brackets, speaker is indicated by use of initial Paralinguistic features and transcriber comments in angled brackets /xxxxx/ Overlapping speech is marked with a single slash at each end of the overlapping parts of speech, for both speakers . Short audible pause of up to approximately 0.5 second .. Audible pause of between approximately 0.5 and 1.0 second (further points indicate further time segments of up to approximately 0.5 second) All names used in examples are pseudonyms

Dutch quotative van Past and present Peter-Arno Coppen and Ad Foolen Radboud University Nijmegen

The quotative use of the Dutch preposition van ‘of ’ can be considered as the Dutch counterpart of English like, German so and similar items in other languages, the use of which has increased significantly in many languages of the world in the last four decades. Stylistically, quotative van occurs most frequently in informal spoken language, less in more formal spoken language and infrequently in written language (the written language in the new media included). In this sense, Dutch quotative van fits in to a worldwide trend. Yet, quotative van shows some characteristics that seem to differentiate it from its counterparts in other languages. Firstly, van not only combines with direct speech, but also occurs regularly in combination with indirect speech. Secondly, van is not restricted to youth language; it is widely used in all age groups and regions (also in Belgian Dutch). Thirdly, based on written informal documents and dairies, we show that the quotative use of van has old roots in the history of Dutch, going back to at least the 17th century. Keywords: Quotative van, similative meaning, direct speech, indirect speech, synchrony, diachrony, sociolinguistic profile

1. Introduction Quotative markers in the languages of the world provide at least two puzzles. Firstly, utterances with quotative marking can easily be translated from one language into the other (e.g., English He was like “oh no!”, Dutch Hij zei van “o nee!”, German Und er so “o nein!”). However, the lexical and syntactic means for constructing such utterances vary quite a lot between languages, as the examples just given illustrate: English has a copula verb + like, Dutch a verb of saying + van (a preposition), and German uses a verbless construction with the deictic modal adverb so ‘thus’. The question, then, is what is universal and what is language-specific?

 Peter-Arno Coppen and Ad Foolen

Foolen (2008: 123) provides a typology of the kind of items that develop into quotative markers in the world’s languages: (i) verbs of speaking (see e.g. Bakker and Wessler 1999; Klamer 2000; Lord 1993), (ii) demonstrative and modal deictic markers: for example German so, Norwegian sånn (see Hasund et al., this volume), Sanskrit iti ‘thus’ (see Be_Slayed 2010), (iii) quantifying elements: for example, English all and just1, Norwegian bare (see Hasund et al., this volume), and (iv) words with similative meaning: for example English like, Sanskrit yatha ‘like’ (see Be_Slayed 2010). As we will show in this chapter, Dutch van belongs to this fourth category. Some of these quotative markers (e.g. English like) have recently increased in frequency of use (see Buchstaller 2011; Fox Tree and Tomlinson 2008; Tagliamonte and D’Arcy 2007) to such an extent that native speakers are aware of it. Foolen (2008) interviewed speakers from a range of languages in which innovative quotatives have been reported, such as Puerto Rican Spanish (estar ‘to be’ or asi ‘so’), Finnish (niinku ‘as like’ and ihan et ‘quite that’) and Turkish (gibi ‘as if, like, similar’). Time and again, the informants reported that these markers belong to the innovative language of younger speakers in their language. Older speakers in particular tended to hold negative attitudes towards the use of these markers (cf. Buchstaller 2006; D’Arcy 2007; Dailey-O’Cain 2000 for attitudes regarding English like). Yet, quotative markers seem to be ‘of all times’. Verbs of saying, deictic modal adverbs and words with ‘similative meaning’ have grammaticalized into quotative markers in different languages and in different times (see, e.g., Be_Slayed 2010 for Sanskrit; Buchstaller and Van Alphen, this volume; Güldemann 2008 for African languages). It will take time to solve these two issues (universal function versus language specific ‘solutions’, ‘new’ versus ‘of all times’), something we are not intending to do here. First and foremost, however, any attempt at solving these puzzles needs to be based on a careful and detailed description of the language-specific facts. The present chapter is meant to contribute to this aim by providing a synchronic and diachronic description of the Dutch quotative marker van ‘like’. Linguistic attention for quotative van started with a squib by Verkuyl (1976), followed by some other squibs and smaller papers by different authors (see Foolen et al. 2006 for an overview of the Dutch literature). The 5th Dutch sociolinguistic conference (2006) devoted a workshop to quotative van (see Foolen 2006; Mazeland 2006; Van Alphen 2006). Coppen (2010) focused on the synchrony and diachrony of the more specific construction zoiets hebben van (lit. ‘have such + something of ’, roughly meaning ‘be like’). Here, we will focus on three issues: 1. Just can be paraphrased as ‘nothing but’, which comes close to ‘all’. Gaudy-Campbell (2010) shows that just occurs in I just thought followed by direct speech. That is why it is included here. Norwegian bare also means ‘just’.

Dutch quotative van 



i. The structural profile of the marker, that is, the way the quotative marker fits in the structure of utterances will be discussed in Section 2. ii. In Section 3, we discuss the sociolinguistic and stylistic profile of van. Who uses this quotative marker and in which speech styles does it occur? Is the marker typical for narrative colloquial style as produced by young speakers (such as English like and German so; cf. Golato 2000), or is its use more widespread? iii. Section 4 is devoted to its diachronic development. Typically, quotative markers develop from elements that are already available in the language and fulfill other functions (cf. Güldemann 2008; Vandelanotte and Davidse 2009). Reconstructing this development for a specific quotation marker through time can help us understand how it has developed its present spectrum of uses. The data in this chapter are taken from two existing corpora: the CGN (Corpus of Spoken Dutch: see http://lands.let.kun.nl/cgn/ehome.htm) contains about 9 million words of spoken Dutch – one third from Belgium, two thirds from The Netherlands – compiled between 1998 and 2004. Different styles are represented, from informal to formal. The second is the DBNL (Digital Library of Dutch Literature: see www.dbnl.nl), which to date contains 7845 Dutch texts from the Middle Ages through to the 21th century. The DBNL – the collection of which started in 1999 – aims at digitizing Dutch texts which are relevant to the cultural history of Dutch, both in Belgium and the Netherlands. Furthermore, we use examples which have been individually collected by the authors over the past 10 years and which contain examples from spoken and written sources. We use these data-sets as sources for examples without intending to make quantitative claims on the use and distribution of quotative van. 2. The structural profile of quotative van Matrix clauses introducing direct speech in Dutch can take different forms. An unmarked way of quoting is when the matrix sentence contains a verb of saying or thinking (as in Jan zei “Ik blijf ”/John said “I will stay”). In Dutch, it is not unusual to insert van between verb and quote: cf. (1), a typical example of quotative use of van: (1) (CGN) Bij voorbaat werd ook al gezegd van “uh we spreken er de zorg over uit hoe het gaat lopen.” In advance was too already said van “eh we speak there the worry out about how it goes running” In advance, it was said already too like “we utter our worries about how things will develop.”

 Peter-Arno Coppen and Ad Foolen

It is important to note that, besides van, there are other ways to mark quotative material in Dutch. Sometimes the deictic modal adverb zo is inserted immediately before the quote, leading to the construction personal pronoun + quotative verb + zo. Speakers sometimes omit the matrix verb (see Hermann and Steinbach, this volume; Oshima and Sano, this volume for contextualization of quotation in German Sign Language and Japanese). These different forms are illustrated in the following fragment: ‘say’ + zo in line 3, verb-less zo in lines 4 and 5, and ‘say’ in line 6. (2) L1 L2 L3 L4 L5 L6

(CGN) was de vorige keer lachen met dat broertje van Tom. was the previous time laugh with that brother + dim, of Tom. wij zaten zo en we hadden net lootjes getrokken (...) we sat ZO and we had just lots drawn (...) dan zeg ik zo “ik weet lekker wie iedereen heeft en jullie weten niet wie ik heb”. then say I ZO “I know nice who everyone has and you know not who I have.” dus wij zo “nou Bram wie heeft wie dan?” so we ZO “now Bram who has who then?” Dus hij zo “ja jij hebt uh die” (...) So he ZO “yes you have uh that” (...) Ik zeg “oh dus dan heb jij die?” I say “Oh so then have you that?” ‘Last time, it was funny with Tom’s brother. We were sitting there and we just had drawn lots. (...) then I say ZO I know who has whom and you don’t know whom I have. Then we ZO well Bram, who has whom? Then he ZO yes you have eh him (...) I say oh then you have that one?’

Later in this chapter, we will see that zo and van can co-occur in quotative constructions. However, our focus will be on van. As its primary uses are non-quotative, we will start with an overview of these. 2.1

Non-quotative use of van

Van is the most common preposition in Dutch, comparable to English of. It can be used with a variety of meanings, ranging from prototypical possessive, partitive and causative meanings (3–5) to source meanings (where English would use from): cf. (6–8). (3) Dat boek is van mij (possessive) That book is van me ‘That book is mine’

Dutch quotative van 



(4) Een van mijn tantes woont in die straat (partitive) One van my aunts lives in that street ‘One of my aunts lives on that street’ (5) Ze zat te bibberen van de kou (cause) She sat to shiver van the cold ‘She was shivering from the cold’ (6) Hij liep van zijn bureau naar het raam (locational source) He walked van his desk to the window ‘He walked from his desk to the window’ (7) De tentamenperiode loopt van 19 tot 31 oktober (temporal source) The exam period walks van 19 to 31 October ‘The exam period ranges from October 19 to October 31’ (8) Zij is een kind van rijke ouders (relational source) She is a child van rich parents ‘She is a child of rich parents’ Note that prepositional van has also become, due to semantic bleaching, a semantically devoid function word, contributing almost no meaning. As such, it is quite commonly used for noun complements, especially the complements of nominalized verbs (as can be seen in Examples (9) and (10)). (9) Een schilderij van de ondergaande zon A painting van the setting sun ‘A painting of the sunset’ (10) Het schrijven van boeken The writing van books ‘The writing of books’ In complements like these, van can also mark the agent or the subject of the nominalized verb, as in (11). (11) De kritiek van de oppositie The criticism van the opposition ‘The opposition’s criticism’ One specific use deserves special attention in the context of the present paper: namely, where van has similative or approximative meaning (cf. 12–13): (12) Hij heeft iets van zijn vader He has something van his father ‘He looks like his father’

 Peter-Arno Coppen and Ad Foolen

(13) Ik zag iets van blauw in dat schilderij I saw something van blue in that painting ‘I saw something blue in that painting’ Example (12) is a construction with van indicating some kind of source (such as a genetic property). Importantly, it also indicates a similarity between the subject and his father. Compare this example to (13), which contains van, an existential quantifier (iets ‘something’) and an adjective. Here, the construction iets van blauw does not just signify something which is blue, but rather something ‘bluish’, some color resembling blue. The same meaning can also be expressed in a construction with a genitive (cf. 14): (14) Ik zag iets blauws in dat schilderij I saw something blue + gen in that painting ‘I saw something blue in that painting’ Many researchers have pointed out that quotative markers are often associated with a comparative, similative or approximative meaning (Buchstaller 2004; Güldemann 2008, this volume; Meyerhoff 2002), and the same seems to hold for Dutch van, which, in its pre-quotative function, has similative meaning. 2.2

Quotative van

In quotative constructions, two parts can be distinguished, optionally linked by a quotative marker. They have been labeled in different ways, depending on the theoretical viewpoint regarding the syntactic relationship between the two parts. Vandelanotte and Davidse (2009: 785) make a distinction between the classical matrix + complement view and a constructional interclausal perspective, “which takes the reporting clause (as a whole) and the reported clause as the primary component structures of quoting constructions” (see also Vandelanotte, this volume). The advantage of the latter view is that it easily accommodates a variety of matrix clauses (the first component of the construction), transitive but also intransitive clauses, and even rather loose connections between the two components, in which the quoted part looks more like an adverbial adjunct to the sentence as a whole. Nevertheless, in most cases, it is possible to identify an element in the matrix clause – usually a verb or a noun – that is the main trigger for the quotative part and of which the quotative part can be seen as a thematic argument. As we will show below, the contexts in which quotative van can be found vary substantially, both regarding the trigger component, as well as the quoted part. Crucially, it is not only verbs that function as the main trigger or anchor; in addition, nouns frequently are triggers (see Section 2.2.1). As regards the quoted material,

Dutch quotative van 



the most striking observation is that, contrary to like and other quotatives with similative semantics, van can combine with indirect speech (see Section 2.2.2). 2.2.1 Van introducing direct quotes Quotative van occurs in three types of structures.2 It can be found in a reporting clause that also contains a trigger, which can be a verb or a noun. The reporting clause can contain a light verb construction or, in the third type, the quotation is linked to a preceding clause by van, optionally preceded by the modal adverb zo ‘so’, but without a trigger verb or noun. We will now discuss these three types in turn. In the first type, the most frequent trigger verbs are zeggen ‘to say’ and denken ‘to think’. Besides semantic variants of these two, we also find verbs of perceiving and feeling (such as horen ‘to hear’ and zich schamen ‘to be ashamed’). Quotative van is optional. Typical examples are (15–16). (15) Papa zei (van) “ik ga de zolder opruimen en veel weggooien” Papa said van “I go the attic clean and much throw away” ‘Papa said like “I am going to clean the attic and throw away a lot.”’ (16) Dan denk je (van) “waar ben je nou mee bezig?” (Van Alphen 2006: 29) Then think you van “where am I PART with busy” ‘Then you think like “what am I doing?”’ The trigger nouns we find in this type are mostly deverbative nouns like vraag ‘question’, discussie ‘discussion’, but also semantically similar nouns like idee ‘idea’, gevoel ‘feeling’ (18) and sometimes even nouns only vaguely reminiscent of feelings, such as tendens ‘tendency, general feeling’ (20) and smile ‘smile’ (19). The syntactic function of the noun varies. It can be an object (17), a complement of a preposition (18–19) or a subject (20). If the trigger is a noun, quotative van is obligatory (indicated with an asterisk attached to the brackets, i.e. non-optionality). (17) Nu kreeg ik een mailtje *(van): “je was een succes” Now received I an email van: “you were a success” ‘Now I received an email like “you were a success.”’ (18) Ik deed het met een gevoel *(van) “had dit zo gemoeten?” I did it with a feeling van “had this so must?” ‘I did it with a feeling like “was this right?”’ (19) En dan zit je met ’n smile *(van) “ik heb je geholpen”. And then sit you with a smile van “I have you helped” ‘And then you are sitting there with a smile like “I helped you.”’ 2. We thank Emar Maier and Wessel Stoop for their collaboration, in particular regarding the typology of quotative constructions as presented here.

 Peter-Arno Coppen and Ad Foolen

(20) Helaas is de tendens nu *?(van) “het gaat niet meer.” Unfortunately is the tendency now van “it goes not anymore”. ‘Unfortunately, the tendency is like “it is not possible anymore.”’ The trigger noun is often the complement of a light verb like hebben ‘have’, creating a complex verb, like het idee hebben, lit. ‘have the idea’, meaning ‘think’, as in Example (21). (21) We hadden ook nog steeds het idee van “ja ze kennen ons” We had also still the idea van “yes they know us” ‘We still thought like “Yes they know us.”’ The second type in which quotative van occurs does not contain a semantically rich verb or noun. Instead, quotative van is triggered by a light verb (zijn ‘be’ or hebben ‘have’) and is always preceded by elements like demonstrative zo ‘such’ and/or iets ‘something’, often written as one word zoiets (cf. Coppen 2010). The content of such quotes is, typically, a ‘feeling’, cognitive state or attitude, as can be seen in (22), which expresses an attitude, (23), which reproduces a wish, and (24), which demonstrates inner thought (see Spronck, this volume for the encoding of attitudes in quotation). More examples can be found in the extracts of adolescent talk in Lamerichs and te Molder (2009). (22) Dat is zoiets van: “U kunt wel meer uitgeven als u het maar zelf betaalt” That is such something van “you can mod more spend if you it mod yourself pay” ‘That is something like “you can spend more if you only pay it yourself.”’ (23) Toen had ik zoiets van “Ja daar wil ik ook aan meedoen” Then had I such something van “yes there want I also to participate” ‘Then I felt like “I also want to participate in that.”’ (24) Ik heb zoiets van “Als het echt niet kan, dan horen we ’t wel.” I have such something van “if it really not can, then hear we it for sure” ‘I have a feeling like “If it really is impossible, we will hear it.’” It is, in particular, the combination of hebben ‘to have’, zoiets ‘such a thing’ and van ‘like’, which is targeted by critics of modern language use, the criticism being that this usage is considered ‘vague’. In recent years, this construction has repeatedly ended up in the top five (and in some cases as the ‘winner’) in popular polls for the most irritating word or construction of the year, both in the Netherlands and in Belgium.3 It is one of the favorite pet peeves in web forums and in popular language

3.

See for 2008 http://irritantstewoord.nl/?p=20.



Dutch quotative van 

magazines such as Onze Taal.4 Despite this negative attitude, the construction seems to have gained in considerable frequency since it was mentioned for the first time in 1986 in Onze Taal.5 Whereas the two constructional types just discussed both contain some kind of trigger (a full verb or noun, or light verb construction with demonstrative), in the third type, no such trigger is present. Rather, the quotation is added to an utterance in an asyndetic way, as a kind of elaboration or illustration of the foregoing sentence (cf. Mazeland 2006). In this type of construction, the quotative marker van is also often accompanied by the demonstrative zo ‘so’ (cf. 27). (25) Moeten jullie ook wel eens mensen teleurstellen van “ja ’t is op?” Must you also mod sometimes people disappoint van “yes it is up?” ‘Do you also have to disappoint people sometimes like “nothing is left?’” (26) Toen gaf ze mij een doosje met brieven van “bewaar jij dat maar.” Then gave she me a box-dim with letters van “take care you that mod” ‘Then she gave me a little box with letters like “you better take care of that.”’ (27) Veel studiegenoten reageerden laconiek, zo van “ja, ja, zal wel” Many co-students reacted laconically, so van “yes, yes, will be” ‘Many co-students reacted laconically, like “yes, yes, if you say so.”’ In these examples, the quotes cannot be complements of the matrix verbs, as these already have a complement (mensen ‘people’ in 25, doosje met brieven ‘little box with letters’ in 26) or they are intransitive (reageerden ‘reacted’ in 27). Rather, the function of the quote seems to be to paraphrase or illustrate the content of the sentence in a more informal, involved way (see Golato this volume). 2.2.2 Van introducing indirect quotes Like and a range of other new quotative markers in other languages have long been claimed to be restricted to direct quotes; indeed the focus of attention in the literature tends to be on direct quotatives. However, for English be like and go, Vande­ lanotte (this volume) points out that “[s]ome infrequent inroads into the indirect 4. Onze Taal is a monthly journal, widely read by people who are interested in the Dutch language and in language use in general: see www.onzetaal.nl. 5. There must have been earlier uses of this specific construction, cf. the following passage from Louter Leugens (1951) by Simon Carmiggelt, an author who had a ‘good ear’ for informal spoken language: ‘t Is ver van me bed’, sprak de andere man. Hij had zo iets van: Piet eet tòch wel, en goed ook. ‘I don’t have much to do with it’, said the other man. He felt like: Piet will eat anyhow, and good too. It thus took 35 years, until 1986, before the construction was observed explicitly in Onze Taal.

 Peter-Arno Coppen and Ad Foolen

domain have (...) been reported”, although “[it] remains to be seen whether [this] type of usage (...) takes root and expands beyond relatively rare occurrences”. A remarkable syntactic property that differentiates Dutch quotative van from its English counterpart is the fact that it can be used with indirect clauses. This can be seen in Examples (28) and (29), in which the indirect clause starts with the complementizer dat ‘that’ and the wh-word wat ‘what’ respectively. Note also the fact that these constructions are verb-final, with clause final finite verbs kwam ‘come’ and doen ‘do’, a clear diagnostic of subordination. (28) Net zoals dat ze nu al roepen van dat ’t griepvirus ook uit de ruimte kwam Just like that they now already shout van that the influenza virus also from the space came ‘Just like they now claim like that the influenza virus came from outer space too.’ (29) Ik vroeg van wat ze ’s middags wou doen weet je wel I asked van what she afternoon wanted do know you well ‘I asked like what she planned to do in the afternoon, you know.’ As has been stated in the literature (cf. Banfield 1973; Coulmas 1986; Jiang 2009 inter alia), the distinction between direct and indirect reported discourse is not absolute. As van is used in both contexts, it would narrow the perspective if we were to exclude cases of indirect reports from our overview. Verkuyl (1976) observed the use of van + indirect discourse, commenting that he would never use it himself. However, a search in the Corpus of Spoken Dutch (CGN) provided about 300 examples of van preceding an indirect clause.6 Examples include cases with a verb as the trigger (as in (28) roepen ‘shout’ and (29) vragen ‘ask’), but also – parallel to the construction with direct quotation – with nouns (cf. (30) and (31)). (30) ...deze theorie van dat de populaire cultuur eigenlijk in dit opzicht ook een soort voedingsbodem is geweest voor de vrouwenbeweging die in de jaren zeventig echt heel sterk is opgekomen. ...this theory van that the popular culture actually in this respect also a sort nurture-soil is been for the women’s-movement that in the years seventy really very strongly is arisen. ‘...this theory like that popular culture has actually been in this respect a kind of fertile soil for the women’s liberation movement which strongly developed in the seventies.’

6. This figure indicates that the construction is not uncommon, although its relative importance compared to direct reported speech has not yet been determined.

Dutch quotative van 



(31) ...en dat die bovendien ook aanvaardde ’t argument van dat uh dat het eigenlijk niet bewezen was dat zij het gedaan had. ...and that he moreover also accepted the argument van that eh that it actually not proven was that she it done had ‘...and that he, moreover, accepted the argument like that eh that it had not been proven actually that she had done it.’ These are not examples of direct quotation, yet the parallel with the use of van in relation to direct quotes is remarkable. Quotative van with indirect quotation also occurs in the second type of construction, with light verbs such as hebben ‘have’ or zijn ‘be’ accompanied by zo ‘so’ and/or iets ‘something’, again parallel to the direct quotation. (32) Bij Arthur was ’t altijd zo van als we uitgingen dan was ’t echt zo van dat hij ja een beetje ons zat te pushen. With Arthur was it always so van if we went-out then was it really so van that he yes a little us sat to push ‘With Arthur, it was always the case that when we went out, then it always was like that he was pushing us.’ (33) Ik had zelf van dat hij heel veel vragen stelde. I had self van that he very many questions put ‘I was like that he asked very many questions.’ The third type of construction we discussed above for van + direct speech can also be found with indirect quotation. Here, quotative van occurs without triggers, loosely connecting the previous sentence to an indirect quotation, which again has an elaborative, illustrative meaning. This is demonstrated in (34), taken from the CGN. (34) en ze heeft ’t ook wel gestimuleerd van dat zij zei van “ik had in ’t begin ook heel erg dat ik m’n stem dat dat fout ging.” And she has it also mod evoke van that she said van “I had in the beginning also very much that I my voice that that wrong went” ‘And she also evoked it, like that she said like “I also had it very badly in the beginning that I, my voice, that that went wrong.”’7 We can thus conclude that van followed by an indirect clause occurs in exactly the same syntactic types of construction as quotative van followed by direct clauses, that is, with a full noun or verb as the trigger in a reporting clause, with a light verb construction in a reporting clause, and in loosely connecting illustrative 7. In this example, two quotative vans occur. The second one is with direct quotation as indicated by the quotation marks.

 Peter-Arno Coppen and Ad Foolen

elaborations. This parallel is a strong argument for considering direct and indirect variants as belonging to the same family of (quotative) constructions. We now turn to the second profile of Dutch quotative van, taking a sociolinguistic and stylistic perspective. 3. Sociolinguistic and stylistic aspects of quotative van In most languages, the use of a new quotative marker shows sociolinguistic and stylistic variation. English like and other quotative markers (be all, go, this is + NP) occur predominantly in informal speech of younger speakers, and demonstrate regional and ethnic variation (cf. Buchstaller and D’Arcy 2009; Fox this volume; Kohn and Franz 2009). For Dutch van, only a few sociolinguistic studies are available. Van Alphen (2006) analyzed spoken corpus data from Amsterdam speakers recorded in 1975 and 1993.8 The percentage of direct quotes marked by van increased from 39% to 58%, and at the same time, the diversity of verbs and nouns that acted as triggers in the matrix clause also increased. In addition, the factors social class and sex played a role. In 1975, the use of van was more frequent amongst higher educated male speakers, whereas by 1993 the higher educated women and the lower classes had caught up with the trend. Overall, van-quotes were spread relatively evenly across all social classes, ages, genders and ethnicities and Van Alphen (2006) concluded that sociological factors ceased to be relevant for the variability in their usage. In the Corpus of Spoken Dutch (CGN 1998–2004), we do not find any difference between Dutch as used in the Netherlands (Netherlandic Dutch) and in Belgium (Flemish), two varieties that differ in many other syntactic, lexical and phonetic aspects. The occurrences of van + indirect clause in the CGN, which has a ratio of 2:1 Netherland Dutch to Flemish, corresponds exactly to the relative corpus size of the two sub-corpora. We expect the same balance to hold for van + direct quotes. Gender, social class, age and ethnicity do not seem to have a significant bearing on the use of the van quotative in the CGN corpus examples. There are, however, individual differences. Indeed, the frequency of using van seems to be an aspect of personal style. Some speakers hardly ever use it, whereas others – even sometimes with the same social profile – use it in nearly every context, in which it can be used. It might even be the case that some speakers consciously suppress the use of van on the basis of their awareness of the negative attitudes against quotative 8. The corpus consists of Amsterdam informal conversations between friends. All interviews were conducted in dyads and triads and the collection contains speakers between 20 and 40 years who come from all social classes and both genders. Token numbers of all instances of direct reported discourse were in 1975 N = 211; in 1993 N = 262 (Van Alphen 2006: 32).



Dutch quotative van 

van in the hebben-zoiets-van context (see Section 2.2.1) and that they generalize the avoidance of quotative van to all contexts. As to stylistic variation, we can say the following. Firstly, the use of van belongs to the register of spoken language, primarily occurring in colloquial speech. Yet, it is not difficult to observe it in more formal discussions or in interviews on the radio and on TV, or even in discussions that take place in the Dutch Parliament; as such, there is no absolute restriction to informal styles. Jones and Schieffelin (2009) have shown that between 2003 and 2006, the use of be + like in ‘Instant Messaging’ increased. In order to compare their findings with Dutch computer mediated language use, we investigated the use of quotative van in data from ‘new media’ such as ‘chats’ and internet fora.9 We were surprised to see that quotative van was used infrequently in this type of discourse. It seems that the medium – namely, spoken versus written (in this case computer-mediated) communication – is the main variable that conditions the occurrence of van quotes. It also seems that the real time, spontaneous and thus unprepared production of speech in spoken discourse is the main factor favouring the use of van. As such, the distribution of van in spontaneous speech depends partly on the cognitive needs of speakers reporting demonstrating (see Clark and Gerrig 1990), or acting out, (see Wierzbicka 1974) a quote. In computer-mediated text, such online ‘hold on’ signals are less necessary (see Jaeger 2010 on the management of syntactic density) or can be replaced by dots. Indeed, as regards the prosodic marking of van-framed quotation, Verkuyl (1976) claimed that quotative van was often followed by a (filled) pause. While pauses, especially filled ones, have the obvious advantage of buying the speaker valuable time before producing the relevant content, van itself also provides such extra time and of course, the same holds for interjections (such as ja ‘yes’ and nou ‘now’), which often mark the beginning of the direct quotation. However, in the more recent CGN corpus, (filled) pauses after van are infrequent. Van Alphen (2009) similarly found a strong decrease in the frequency of van followed by the hesitation marker eh between 1975 and 2009: In the corpus collected in 1975, 27% of van-quotations were followed by eh, compared to 11% in 1993.10 We take these findings to mean that over time, the need to produce such elements before the quotative content has reduced. By now, the quick switch to direct discourse after van seems to have become routine in spoken language and speakers need much less time to cognitively process the quotative construction. 9. These data were collected in 2010 by Steven Westelaken, student assistant for the ‘Newspeak’ project, a corpus project that aims at putting together a corpus of language use in the new media. 10. The ratio of van-quotes per number of direct reported discourse is as follows: 1975: 82/211 (39%), 1993: 152/262 (58%), 2009 = 34/155 (22%).

 Peter-Arno Coppen and Ad Foolen

The sociolinguistic and stylistic profile of quotative van seems to suggest that, despite the negative attitudes attached to its use, the form has spread across the whole linguistic community, in all regions in the Netherlands and Flemish-speaking Belgium. Although it seems clear that the use of quotative van has spread rapidly during the last quarter of the 20th century, there is ample evidence that it has been present in Dutch for a much longer time, as we will show in the next section. 4. The diachronic development of van Recent diachronic studies on quotative markers have shown that their quotative use is typically the result of a grammaticalization process, involving an item that exists already in the language. This can be seen, for example, in English like (Vandelanotte and Davidse 2009: 795), as well as Saramaccan táa (Lefebvre and Loranger 2008) and Bislama olsem (Meyerhoff 2002). A similar process also seems to hold for Dutch van. The exact transition from non-quotative uses to quotative uses is an issue that needs further investigation. In this section, we intend to show that the use of van as a quotative marker is much older than the perception of its recent increase in use suggests. Van den Toorn (1997: 530) similarly remarks: “Although the construction can be observed with certainty in the 19th century already, it is not an exaggeration to assert that the use of this expletive (superfluous) or performative (i.e., relating to a speech act) van is increasing”. Indeed, Coppen (2010) presents examples of quotative van in Dutch literature going back to 1894. A construction with quotative van, notably van ja ‘VAN yes’ and van nee ‘VAN no’ can be found in 17th century diaries. We encountered it several times in the diaries of Willem Frederik (1613–1664), stadholder of Friesland, Groningen and Drente (from 1648, see Example 35), and other diary-like or letter-like documents from that time. (35) Willem Frederik, Gloria parendi, diaries of Willem Frederik, stadhouder of Friesland, Groningen and Drente, 1648. L1 Gemmenich bij mij geweest, sprack mij van de cornetzplaetz, of hij nae Wesel soude gaen; Gemmenich with me been, spoke me from the Cornetzplaetz, whether he to Wesel should go; L2 ick seide van “jae, hij most sich laeten voorstellen”; I said van “yes, he should himself let introduce” ‘Gemmenich, who had come to me, spoke to me of the Cornetzplaetz, whether he should go to Wesel. I said like “yes, he should let himself be introduced”.’

Dutch quotative van 



Although van ja and van neen may be considered rather limited constructions, it should be noted that in (35), the following clause hij most sich laten voorstellen ‘he should let himself be introduced’ is in fact the object of the verb seyde ‘said’, and shows features of both direct and indirect speech. Like direct speech, it lacks a complementizer dat (which is obligatory in finite subordinate clauses in Dutch) and has a word order with verb second (which is indicative for main clauses). Like indirect speech, however, it shows tense agreement with the main clause (most ‘had to’ is simple past, whereas in direct speech, present tense mot would be expected) and the pronouns are also shifted to a deictic center different from the speaker at the original time of utterance (hij and sich 3rd sg. rather than gy and U 2nd sg.: Gy mot U laten voorstellen). As such, the construction resembles the German Konjuktiv II (Er sagte, er sollte sich vorstellen lassen ‘he said he should let himself be introduced’). In any case, the answering particle jae and the following clause must be analyzed as the object of the verb. Together, they constitute an elaborate quotation that straddles the boundary of direct and indirect speech. Another use of van still a century earlier (1568), is observed by Van der Horst (2008). This concerns the construction with van and an infinitival clause. This infinitival clause can be a verb complement or a noun complement, as seen in Examples (36–37). (36)

(Anonymous, possibly Coornhert 1568, Van der Horst 2008: 870) dat se vresen van berispt te worden that they fear van reprimanded to become ‘that they fear to be reprimanded’

(37) (Coornhert, Zedekunst dat is wellevenskunste, 1585, Van der Horst 2008: 941) uyt vreezen van in ghebreck te komen from fear van in shortage to come ‘from fear of becoming in need’ Syntactically, this construction differs from the construction types we observed in modern-day Dutch involving quotative van. It resembles a verb complement construction which occurs relatively frequently in Flemish (also modern-day Flemish) but which is almost absent in modern Netherlandic Dutch. This particular construction, van + verb or noun complement, has been analyzed extensively for modern Flemish by Van Craenenbroeck (2002), who convincingly argues that examples such as (36) and (37) cannot be analyzed as a normal quotative construction since, for instance, elements from this type of van-construction can be extracted, an impossibility in other constructions with quotative van. Although the constructions in (36–37) may appear quite unlike other quotative constructions, they do provide evidence that in earlier stages of Dutch, the

 Peter-Arno Coppen and Ad Foolen

preposition van could be used to link infinitival complements to verbs and nouns like vresen ‘to fear’ in (36) and uyt vreezen ‘out of fear’ in (37). Moreover, this has only been observed with deverbative nouns (apart from fear, Van der Horst 2008 mentions danger, reason, need, want, trust and thankfulness) and verbs that encode thoughts and feelings. What this effectively means is that the infinitival van-complements we observe in earlier stages of Dutch are related to the same semantic field in which modern-day quotatives occur. The construction with van followed by a finite clause that acts as the complement of a verb, noun or adjective is already attested in the 16th century (see 38). From then on, it occurs throughout the centuries. (38) (Willem Boonen, Geschiedenis van Leuven, 1593–1594) in danckbaerheijt van dat zij hunne pelgrimagie soo geluckichlijcken volbracht hadden in thankfulness van that they their pilgrimage so luckily accomplished have ‘in thankfulness of having accomplished their pilgrimage so luckily’ (39)

(Baruch de Spinoza, Nagelate schriften. n.p., 1677) Gy zijt zonder twijffel verwondert van dat ik u zo lang heb doen wachten Thou art without doubt wondered van that I you so long have done wait ‘Thou hast wondered without doubt, that I have kept you waiting for so long’

In Example (39), the finite clause is the complement to the past participle verwondert ‘wondered’, which clearly denotes a feeling. Yet, early examples with verba dicendi also occur, for example, seggen ‘say’ (line 2 in 40) and schryft ‘write’ (in 41): (40) (Constantijn Huygens, Diaries 1695) L1 Daernae quamen daer de vrouw v. Gendt met haer schoonsuster joff. van Oosterweel, Thereafter came there the woman v. Gendt with her sister-in-law miss. van Oosterweel, L2 en begosten wat te seggen van dat ick haer visite geexcuseert had, and began something to say van that I her visit excused had, L3 en geseght dat ick sanderen daeghs thuys soude wesen, maer gingh soo al lachende door. and said that I the-other day home would be, but went so mod laughing on. ‘After that, Mrs. v. Gendt with her sister in law Miss van Oosterweel arrived, and began to say that I had permitted her visit, and that I had said that I would be at home the other day, but they (or I) went on laughing.’

Dutch quotative van 



(41) (Wolff en Deken, Sara Burgerhart, 1782) Och! hy schryft nog al een hope moois van zyne liefde; en van dat ik lief ben, geloof ik; O! he writes mod mod a lot beautiful-gen of his love; and van that I sweet am, believe I; ‘O! He writes quite a lot of beautiful things about his love, and that I am sweet, I think.’ In Example (41), van dat ik lief ben ‘VAN that I am sweet’ is coordinated with the direct object of the trigger verb schryft ‘writes’: een hope moois van zyne liefde. Note that this object, een hope moois van zyne liefde, is in itself a noun phrase with the head noun moois, which is derived from the genitive of the adjective mooi. Indeed, there is reason to believe that historically, the quotative van construction is related to a genitive construction. Crucially in modern Dutch, most genitive constructions (especially the partitive genitives, such as in 42) have been replaced by constructions with the preposition van (as in 43). (42) Ik zag iets blauws in dat schilderij I saw something blue + gen in that painting ‘I saw something blue in that painting’ (43) Ik zag iets van blauw in dat schilderij I saw something van blue in that painting ‘I saw something blue in that painting’ The general replacement of the genitive by van in this particular (similative) construction might be an important contributing factor to the quotative use of the preposition van. Indeed, from the beginning of the 19th century, the construction of existential quantifier iets ‘something’ and alternatively a genitive or van complement is often used in the context of expressions of feelings and inner states. It is also often accompanied by the adverb zoo ‘so’. At first, the construction seems limited to nouns and adjectives, as in (44 – 45): (44)

(N. Beets, Camera Obscura, 1839) Nooit in zijn leven had hy zoo iets moois gezien of gehoord. Never in his life had he so something beautiful-gen seen or heard ‘Never in his life had he heard or seen something that beautiful.’

(45) (De Gids, 1898) Men krijgt dan zooiets van weeheid over al dit zinneloos gemor One gets then so-something van sloppiness about all this senseless moaning ‘One becomes a bit tired then with all this senseless moaning’

 Peter-Arno Coppen and Ad Foolen

Yet, from the end of the 19th century, sentential quotations with van also begin to occur, and Example (46), dated from 1894, is structurally indistinguishable from the present day Dutch construction zo iets van (line 2): (Tine van Berken, Een klaverblad van vier, 1894) Veel menschen, de mooie geslepen glazen, wijn! Many people, the beautifully cut glasses, wine! Pa rood en glimmend van pret, mij telkens een knipoogje gevend, zoo iets van: “Hoe voel je je nu, meid?” Dad red and shining from fun, me repeatedly a wink giving, so something van “how feel you you now, girl?” ‘Many people, beautifully cut glasses of wine! Dad red and shining with fun, repeatedly winking to me, something like “How do you feel now, girl?”’ (46) L1 L2

Note that Example (46) is one of the first clear examples of zoiets van with a full sentence as quotation. In the course of the 20th century, the construction with zo, iets and van becomes more frequent, and from the second half of the century, it also occurs with the light verb hebben ‘to have’, thus giving rise to another frequent form zoiets hebben van ‘to have something like’, which is nowadays subject to much language criticism. In the early 20th century examples of van with elaborate quotations as the object of verba dicendi, an approximative meaning tends to be apparent. Many occurrences contain verbs like brommen ‘growl’, mompelen ‘mumble’, prevelen ‘murmur’, which literally indicate unclear speech. However, with time this literal approximation of reported speech disappears and van generalizes across all sorts of verba dicendi and even in verbless constructions. What remains is a kind of typification of the quotation. Most of these attestations occur in genres such as diaries, daily reports, personal letters (or literary suggestions of these), which obviously contain a wealth of features typical of the spoken register; other attestations appear in children’s literature with a lot of direct speech (from the second half of the 19th century onwards). This corresponds to the observation made in Section 3 that quotative marking is a typical property of spoken discourse or narratives (see Golato this volume; Steinbach and Herrmann this volume). Indeed, sources like these can be argued to approximate spoken discourse more closely than written language. They are obviously less likely to be constrained by literary convention and resemble (or aim to resemble) colloquial, everyday speech. Since quotative van can be found in such sources over many centuries, it seems to have been present in Dutch for a long time, albeit restricted to informal domains. For this reason, historical data are sparse; yet, the following four lines of development can be reconstructed for quotative van in Dutch:

Dutch quotative van 



i. Van as a marker for complements expressing thoughts and feelings: From the 16th century onwards, verbs, nouns and adjectives expressing thoughts and feeling have finite and infinite complement clauses with van. ii. Van as marker of literal quotation: From the 17th century onwards, verba dicendi have complements with van and literal quotation ja/nee ‘yes/no’, elaborated with quotation-like clauses. From the beginning of the 20th century onwards, verba dicendi expressing unclear speech occur with van and reported speech with approximative meaning. During the course of the 20th century, the literal approximative meaning gradually disappears, making way for typifying meaning. iii. Van with similative meaning in the context of iets ‘something’: From the beginning of the 19th century onwards, existential iets ‘something’ occurs with van and genitive adjectives and nouns; the construction usually has similative meaning and refers to feelings. From the end of the 19th century onwards, existential iets ‘something’ with similative meaning also occurs with van and sentential quotations. iv. Finally, from the second half of the 20th century onwards, zo, iets and van are combined with the light verb hebben to a quotative construction approximately expressing inner feelings and other nonverbal expressions. It seems that in the course of history, these lines of development merge. Somewhere in the 19th century, the older construction with van as the complement of thoughts and feelings merges with its similative function. In the course of the 20th century, the verba dicendi construction seems to merge with this line of development, giving rise to the modern construction with quotative van.11 5. Conclusion Quotative van has a strong place in the Dutch way of speaking, both in Belgium and in the Netherlands. It occurs in three types of syntactic construction: (i) within a reporting clause, triggered by a full verb or noun, (ii) within a reporting clause, triggered by a light verb construction involving similative and/or demonstrative elements, and (iii) in an asyndetic construction, in loose connection between an utterance and an elaborative or illustrative quotation. Furthermore, quotative van occurs in all three types with a direct quotation as well as with indirect speech.

11. Van den Toorn et al. (1997) also assume a relation with constructions like en van je hela hola, roughly translated as ‘and then we go hela hola’, and with onomatopoeic quotations like dat ging van auw ‘that went ow’. This construction belongs to the family of van as a marker of literal quotation.

 Peter-Arno Coppen and Ad Foolen

From a sociolinguistic perspective, the use of quotative van seems to have increased since Verkuyl first observed it in 1976. Nowadays, it is frequently used in spoken language in all Dutch-speaking regions and in all sociolinguistic layers of the language community in all spoken settings. It can be observed to a lesser degree in the new media and in informal written language. The increase sharply contrasts with a negative attitude which pertains especially to the constructions of the second type (namely, ik heb zoiets van). In contrast to the perception (of native speakers and linguists) of van as an innovative feature, quotative van has old roots. Some types of construction (van with infinitive indirect quotation, van with ja ‘yes’ and nee ‘no’) can be found as early as the 16th century, mainly in personal documents. Further examples suggest an origin in noun and verb complementation and in similative constructions with a quantifying expression (iets van blauw ‘some of blue’). It is thus hard to determine when quotative van did not occur in spoken Dutch. However, there clearly has been a recent increase in frequency of use. Indeed, we would like to suggest that the informalization of the written language might have contributed to its visibility and attestation by linguists. It is time to give it a proper place in descriptive grammars of Dutch. References Bakker, Peter, and Wessler, Heinz Werner. 1999. “The etymology of the Romani quotative xåcē, xåtē ‘I/you/he said’.” Grazer Linguistische Studien 51: 1–9. Banfield, Ann. 1973. “Narrative style and the grammar of direct and indirect speech.” Foundations of Language 10: 1–39. Be_Slayed. 2010. “English ‘like’ can, like, function like Sanskrit “इित”. http://staefcraeft.blogspot. com/2010/09/english-like-can-like-function-like.html (accessed 10 March 2011). Buchstaller, Isabelle. 2004. The Sociolinguistic Constraints on the Quotative System: British English and US English Compared. Unpublished PhD thesis. University of Edinburgh. Buchstaller, Isabelle. 2006. “Social stereotypes, personality traits and regional perception displaced: Attitudes towards the ‘new’ quotatives in the U.K.” Journal of Sociolinguistics 10: 362–381. Buchstaller, Isabelle. 2011. “Quotations across the generations: A multivariate analysis of speech and thought introducers across 4 generations of Tyneside speakers.” Corpus Linguistics and Linguistic Theory. Special issue: Corpus Linguistics and Sociolinguistic Inquiry. 7(1): 59–92. Buchstaller, Isabelle and D’Arcy, Alexandra. 2009. “Localized globalization: A multi-local, multivariate investigation of quotative be like.” Journal of Sociolinguistics 13(3): 291–331. Clark, Herbert and Gerrig, Richard. 1990. “Quotations as demonstrations.” Language 66: 764–805. Coppen, Peter-Arno. 2010. “Bericht van de innerlijke stem. Synchronie en diachronie van de heb-zoiets-van-constructie.” [Message from the inner voice. Synchrony and diachrony of the heb-zoiets-van-construction] Nederlandse Taalkunde 15(1): 33–53.



Dutch quotative van  Coulmas, Florian. 1986. “Reported speech: Some general issues.” In Direct and Indirect Speech, F. Coulmas (ed.), 1–27. Berlin: Mouton de Gruyter. D’Arcy, Alexandra. 2007. “Like and language ideology: Disentangling fact from fiction.” American Speech 82(4): 386–419. Dailey-O’Cain, Jennifer. 2000. “The sociolinguistic distribution of and attitudes toward focuser ‘Like’ and quotative ‘Like’.” Journal of Sociolinguistics 4(1): 60–80. Foolen, Ad. 2006. “Het performatieve van opnieuw beschouwd.” [The performative van revisited] In Artikelen van de vijfde sociolinguïstische conferentie, T. Koole, J. Nortier and B. Tahitu (eds), 163–174. Delft: Eburon. Foolen, Ad. 2008. “New quotative markers in spoken discourse.” In Empirische Forschung und Theoriebildung. Festschrift für Norbert Dittmar zum 65. Geburtstag, Bernt Ahrenholz, Ursula Bredel, Wolfgang Klein, Marina Rost-Roth and Romuald Skiba (eds), 117–128. Frankfurt am Main: Peter Lang. Foolen, Ad, Van Alphen, Ingrid, Hoekstra, Eric, Lammers, Henk, Mazeland, Harrie and Pascual, Esther. 2006. “Het quotatieve van. Vorm, functie en sociolinguïstische variatie.” [The quotative van. Form, function and sociolinguistic variation] Toegepaste Taalwetenschap in Artikelen 76: 137–149. Fox Tree, Jean E. and Tomlinson, John M. 2008. “The rise of like in spontaneous quotations.” Discourse Processes 45: 85–102. Gaudy-Campbell, Isabelle. 2010. “Just dans la collocation just thought: Marque de codage d’oralité à l’oral.” Neuphilologische Mitteilungen 111(4): 465–477. Golato, Andrea. 2000. “An innovative German quotative for reporting embodied action: Und ich so/und er so ‘and I’m like/and he’s like’.” Journal of Pragmatics 32: 29–54. Güldemann, Tom. 2008. Quotative Indexes in African Languages: A Synchronic and Diachronic Survey. Berlin: Mouton de Gruyter. Jaeger, T. Florian. 2010. “Redundancy and reduction: Speakers manage syntactic information density.” Cognitive Psychology 61: 23–62. Jiang, Haowen. 2009. “Reported speech and thought in Kavalan.” Rice Working Papers in Linguistics 1: 142–150. Jones, Graham M. and Schieffelin, Bambi B. 2009. “Enquoting voices, accomplishing talk: Uses of be + like in Instant Messaging.” Language and Communication 29: 77–113. Klamer, Marian. 2000. “How report verbs become quote markers and complementisers.” Lingua 110: 69–98. Kohn, Mary E. and Franz, Hannah A. 2009. “Localized patterns for global variants: The case of quotative systems of African American and Latino speakers.” American Speech 84(3): 259–297. Lamerichs, Joyce and te Molder, Hedwig F.M. 2009. “And then I’m really like...: ‘preliminary’ self-quotations in adolescent talk.” Discourse Studies 11(4): 401–419. Lefebvre, Claire and Loranger, Virginie. 2008. “A diachronic and synchronic account of the multifunctionality of Saramaccan táa.” Linguistics 46(6): 1167–1228. Lord, Carol. 1993. Historical Change in Serial Verb Constructions. Amsterdam: Benjamins. Mazeland, Harrie. 2006. “VAN as a quotative in Dutch: Marking reported speech as a typification.” In Artikelen van de vijfde sociolinguïstische conferentie, T. Koole, J. Nortier and B. Tahitu (eds), 354–365. Delft: Eburon. Meyerhoff, Miriam. 2002. “All the same? The emergence of complementizers in Bislama.” In Reported Discourse. A Meeting Ground for Different Linguistic Domains. T. Güldemann and M. von Roncador (eds), 314–359. Amsterdam: Benjamins.

 Peter-Arno Coppen and Ad Foolen Tagliamonte, Sali and D’Arcy, Alexandra. 2007. “Frequency and variation in the community grammar: Tracking a new change through the generations.” Language Variation and Change 19: 199–217. Van Alphen, Ingrid. 2006. “Ik had zoiets van ‘doei’. Interactioneel sociolinguïstische aspecten van van-citaties.” [I was like ‘bye’. Interactional sociolinguistic aspects of van-quotations] In Artikelen van de vijfde sociolinguïstische conferentie, T. Koole, J. Nortier and B. Tahitu (eds), 29–42. Delft: Eburon. Van Alphen, Ingrid. 2009. “Quotatives in global perspective.” Paper given at the OAP–Symposium, 18th December 2009, Amsterdam Center of Language and Communication, University of Amsterdam. Van Craenenbroeck, J. 2002. “Van as a marker of dissociation: Microvariation in Dutch.” In Studies in Comparative Germanic Syntax, C.J.W. Zwart and W. Abrahams (eds), 133–163. Amsterdam: Benjamins. Vandelanotte, Lieven and Davidse, Kristin. 2009. “The emergence and structure of be like and related quotatives: A constructional account.” Cognitive Linguistics 20(4): 777–807. Van den Toorn, M.C. 1997. “Nieuwnederlands (1920 tot nu).” [New Dutch (1920 till now)] In Geschiedenis van de Nederlandse taal, M.C. Van den Toorn et al. (eds), 479–562. Amsterdam: Amsterdam University Press. Van der Horst, Joop. 2008. De Geschiedenis van de Nederlandse Syntaxis. [The history of Dutch syntax] Leuven: Universitaire Pers Leuven. Verkuyl, H.J. 1976. “Het performatieve van.” [The performative van] Spektator 6 (7/8): 481–483. Wierzbicka, Anna. 1974. The semantics of direct and indirect discourse. Papers in Linguistics 7: 267–307.

appendix

Glossary of specialist terms for research in quotation Addressee: The discourse entity who is represented as the discourse participant being addressed in the current discourse situation, the addressee of the reported message. (see also hearer) Affective nonmanuals: Nonmanual features like facial expressions that are used for affective purposes, such as fear, disgust, disbelief or surprise. Agreement verbs: A class of verbs in sign languages that agree with one or two of their arguments by modifying the beginning and endpoint of the path movement and the orientation of the hands according to referential loci in signing space, which are linked to the respective arguments. Approximative meaning: A specific similative relation between A and B, where resemblance is the target (rather than the source). Assessment: An evaluation, either positive or negative. Attitude (of the speaker): A point of view or evaluation of the speaker about the information encoded in the utterance. Booster: The subset of degree modifiers (see intensifier) that are associated with high degree. They amplify the property denoted by their head, moving the property up an imaginary scale. Boosters are also used to affirm the speakers’ confidence in a claim they are putting forward. Bridging context: Context allowing two interpretations, one old and one new, for a linguistic item undergoing change. Causative meaning: A relation between A and B, where A causes B. Claim-backing: A conversational move in which speakers support, warrant, or illustrate a position they have taken. Co-constructing talk: Two (or more) speakers jointly produce a stretch of talk, for instance, one speaker starting an utterance and the other speaker finishing it.

 Quotatives

Complementation: A relation between component structures in which a conceptually more autonomous component elaborates a salient substructure of a conceptually more dependent component. Conduit metaphor: A dominant metaphor for conversational interaction whereby speech or ideas (feelings, meanings, thoughts, concepts, etc.) are considered objects and linguistic expressions are containers. Communication thus consists of inserting ideas and words into containers and sending them to an interlocutor. Considered this way, language is viewed as a ‘conduit’ conveying verbal and mental content between people. Constructed dialogue: A term used to refer to reported speech or direct quotation in storytelling or conversation that has (probably) never been (and often could not have been) actually produced, instead being ‘constructed’ by the storyteller. (see hypothetical discourse) Construction: A conventional form-meaning pairing. Construction Grammar: Family of linguistic theories using the notion of construction as their cornerstone. Constructionalization (or schematization): The emergence of a new construction. Continuum hypothesis: The hypothesis that direct and indirect reported discourse are not categorically distinct but rather form a continuum. Coparticipant: Conversational partner of the current speaker. Current speaker: The discourse entity making an utterance at the present discourse moment (also referred to as speaker). Decategorialization (and recategorialization): Change from one linguistic category to another; typical of grammaticalization processes. Deontic: see Modality (Deontic) Deixis: The phenomenon whereby the understanding of the meaning of certain words and phrases (referring to space, time and (inter)personal components, such as here, now, I) relies on the context in which they are uttered. Deixis (Emotional): The use of deictic forms to express emotional solidarity between speaker and addressee, which presupposes shared familiarity with the referent (Lakoff 1974). Deixis (Empathetic): The use of deictic forms to encode emotional proximity (this), or distance (that) between a speaker and referent (Lyons 1977).



Appendix:  Glossary of specialist terms for research in quotation 

Demonstrations: Verbal or nonverbal actions which show selected aspects of a referent to the conversational partner as if they were experiencing them first-hand. Deverbative noun: A noun derived from a verb. Direct speech/thought: Form of speech and thought representation in which the deictic center shifts to that of the represented speaker and the current speaker purports to re-enact the represented speaker’s presumed original utterance or thought. Discourse entity: A participant in a discourse situation (hypothetical or real), irrespective of their conversational role (such as the speaker, addressee, person being talked about, person present in the discourse situation). Discourse of another speaker: Information for which the discourse of a discourse entity who is not the speaker is represented. Prototypically, this information is presented as having been uttered verbally, but in some cases the speaker may also have inferred the information through gesture or on the basis of other non-verbal signs. (see speaker) Double-voiced utterance: A concept first discussed in Vološinov (1929), in which a speaker’s utterance does not only reflect her/his own views, perspective or evaluation, but anticipates or co-expresses the views, perspective or evaluations of some other speaker/discourse entity. (see multiple-perspective construction) Downgrader: See Downtoner Downtoner: The subset of degree modifiers (see intensifiers) that are associated with low degree. They reduce the property denoted by their head, moving the property down an imaginary scale. Downtoners are also used to de-emphasize the illocutionary force of a proposition or the strength of a claim put forward. Doxic: See Modality (Doxic) Dubitative: See Modality (Dubitative) Elaboration: A relation of filling in a salient substructure of a component structure in a complex construction. Elliptic quotation: A quotative construction where the quotative predicate, such as say, is elided. (see Zero quotation) Embryonic form: The existence of a variant which occurred with low frequency at an earlier stage of its use. Enactment: Acting out a situation; acting out talk; performance.

 Quotatives

Epistemic: See Modality (Epistemic) Evaluative meaning (in quotation): A view or evaluation about the reported message, either by the speaker (current speaker) or some other discourse entity as reflected in a quotative construction. Evidentiality: A grammatical category encoding how the speaker or another discourse entity came to know the information s/he is talking about. The category is often expressed through morphemes co-encoding other grammatical categories (most notably epistemic modality) or multi-word constructions. Evidential value: The semantics of a particular expression of evidentiality. (see Evidentiality) Existential quantifier: A word (or constituent) selecting one or more elements from the set of all possible elements (e.g. some, any or something). Extravagance: Speaker’s use of unusually explicit forms in order to attract attention (Haspelmath 1999). Fact clauses: Clauses following so-called ‘factive’ verbs whose meaning is not affected by negation. Focus marker: Item that (i) marks the information that is at the center of the speaker’s communicative interest and/or (ii) strengthens the speaker’s commitment to the proposition, expressing certainty and exactness. Frustrative: See Modality (Frustrative) Grammaticalization: “The change whereby lexical items and constructions come in certain linguistic contexts to serve grammatical functions, and once grammaticalized, continue to develop new grammatical functions” (Hopper and Traugott 2003: 18). Grammatical nonmanuals: Nonmanual features in sign languages, such as raised eyebrows and head nods, that are systematically used to encode grammatical information such as topicalization, sentence types, and negation among others in sign languages. Headhood: The status of a semantically primary component in a composite construction, identified on the basis of profile equivalence (the profile of the head is the same as that of the whole construction, such that the whole construction is a more specific instance of the head). Hearer: Discourse entity in the role of listener/perceiver of a speaker’s utterance. (see also addressee)



Appendix:  Glossary of specialist terms for research in quotation 

Hearer-oriented functions (of quotatives): Linguistic features whose function is to acknowledge the presence of the interlocutor, often in the form of appeals for agreement/backchannel or involvement by the interlocutor. Hearsay: Information that the reporting person does not have direct knowledge of, but that is based on second-hand evidence. Hedge: Item that reduces the speaker’s commitment to the proposition, expressing approximation and inexactness. Hypothetical discourse: Talk that has not been (and often could not have been) uttered in prior talk but that is still reported on; invented talk; (often used synonymously with constructed discourse). Indirect speech/thought: Reported speech and thought in which deictic elements in the reported message are construed largely from the current speaker’s deictic center and perspective. This type of quotative construction is generally associated with a more ‘subjective’ representation of the reported message, i.e. with a more clearly expressed current speaker evaluative meaning. Intensifier: Adverb that modifies the concept denoted by the head it modifies. Intensifiers are degree modifiers that can be semantically differentiated with respect to the value which they assign to their heads. They can diminish the property denoted by their head (see downtoner), they can moderate it, or they can amplify it (see booster). They can also be used to strengthen or diminish the affective commitment of the speaker towards the proposition. Irrealis: A modal category that may describe a range of non-factual events. Konjunktiv II: A verb form marking the subjunctive mood in German, used to form the conditional tense and for marking hearsay/indirect speech. Loci in signing space: Specific points in signing space that are used to refer to topographic locations (spatial use of signing space) and discourse referents (deictic and anaphoric use of signing space). Logophoricity: A pronominal category used for reference tracking in subordinate clauses, where a pronoun in a subordinate clause indicates co-referentiality with the referent in the main clause. In some languages, these have been found to grammaticalize to reportative markers, similar to other types of words occurring at clause boundaries in quotative constructions, such as complementizers. Mimesis: Mimesis is the direct representation and the total imitation of an event, including “delivery aspects” (Clark and Gerrig 1990), such as voice effects, gestures,

 Quotatives

inarticulate sounds etc. (as opposed to summarized representation or a mere synthesis of the original event). Mirative: See Modality (Mirative) Modality: A grammatical category predominantly encoded on verb phrases which roughly expresses meanings about the knowledge state of some evaluating entity or the necessity/desirability of a particular situation/event according to some evaluating entity. Modality in quotation: The grammatical encoding of an evaluation of the reported message by a discourse entity involved in quotation. Modality (Deontic): Modality type pertaining to the necessity/desirability of a particular situation/event according to some evaluating entity. Modality (Dubitative): A subcategory of epistemic modality pertaining to doubt. Modality (Doxic): Modality type concerning all truth meanings (see Spronck this volume) Modality (Epistemic): Modality type pertaining to the knowledge state of some evaluating entity. Modality (Frustrative): A modal category that describes intended actions that failed to materialize, most accurately translated into English as ‘to do X in vain’. It describes two moments in time: a moment at which someone wants something and a moment at which that something has not been attained. Modality (Mirative): A modal category that indicates unexpected and new information. It shares with frustrative modality that it describes two moments in time; a moment at which someone has a certain expectation or no expectation at all and a moment at which an unexpected event/an event contrary to the previous expectation has happened. Modality (Timitive): Mood encoding that the speaker fears something expressed in what is said. Modality (Oral-auditory): The marking of modality in spoken languages which is produced orally with the vocal tract and perceived auditorily with the ears. Modality (Visual-manual): The marking of modality in sign language which is perceived visually with the eyes and produced manually and nonmanually with the hands and with the upper part of the body, the head, and the face.



Appendix:  Glossary of specialist terms for research in quotation 

Modality (Volitive): A subcategory at the intersection of epistemic and deontic modality pertaining to wanting something. Mood: Constructions encoding speakers’ attitude as regards the content of an utterance. Multiple-perspective construction: A term coined in Evans (2006) to characterize constructions encoding the views, perspectives or evaluations of two or more discourse entities. (see double-voiced utterance) Narrative: An account of a sequence of events that culminates in a climax (see story). Narrator: The speaker who is telling a narrative. Nondiscourse uses (of markers): Purely grammatical or syntactic uses of markers, in contrast to their discourse uses. Nonmanual articulators: Sign language articulators such as the upper part of the body, the head, and specific parts of the face (i.e. eye brows, mouth, eye gaze) that can be produced simultaneously to manual signs to convey linguistic information. Paradigmatization: The process in which a linguistic item undergoing change becomes part of a pre-existing paradigm; typical of processes of grammaticalization. Parataxis: The joining of two clauses of equal syntactic status. Partitive meaning: A relation between A and B, where A is a part of B. Possessive meaning: A relation between A and B, where A is the owner of B, as in: A’s B. Performance features: Structural features, such as direct speech, asides and repetitions, which function to give structure and drama to a story being performed. Performative: Relating to a speech act. Perspective: The point of view, or vantage point, from which a quote is being reported. The perspective is one of the major differentiating factors between direct quotes (usually reported from the perspective of the original speaker) and indirect quotes (usually reported from the perspective of the narrator). Plain verbs: Class of verbs in sign languages that are lexically specified for the beginning and endpoints of the path movement. Unlike agreement verbs, plain verbs cannot agree with one or two of their arguments.

 Quotatives

Quotation: A discourse situation in which a speaker presents the information s/he is conveying as originating in the discourse of another speaker (or the same speaker at another point in time), thereby not claiming authority over the content of the utterance. Quotative: A word/construction signaling reported discourse (also referred to as ‘quotative marker’ or ‘quotative index’). Quotative construction: A construction that encodes quotation (also referred to as ‘quotative index’). Quotative phrase: A syntactic constituent (for example a complement clause) that represents a quote or the content of reported speech/thought. Quotative predicate: A predicate of communication or attitude that selects for a quotative phrase (e.g. say, think). Also known as verbum dicendi. Quotative predicate ellipsis (QPE): The phenomenon whereby a quotative predicate such as say is elided. Quote: A quotation, that is speech, thought, gesture or actions attributed to someone other than the current reporting speaker. Recategorialization: see Decategorialization Renegotiation: The process of a feature being adopted by new speakers who then adapt it to fit the local linguistic structure and sociolinguistic repertoire. Reportative (evidential): A grammaticalized evidential marker encoding quotation. Reported discourse: See Quote. Reported speaker: The discourse entity who is presented as having authority over the reported message (also ‘source’ or ‘narrative source’). Reporting speaker: The discourse entity who is presenting a reported message (also ‘speaker’). Reported speech: Quotation of verbal (linguistic and sometimes non-linguistic) material. (see quote) Reported thought: Quotation of non-uttered, mental material, prototypically selfreporting from the speaker. Signing space: The three-dimensional space in front of the upper part of the signer’s body that is used for the production of signs.



Appendix:  Glossary of specialist terms for research in quotation 

Similative meaning: Relation between A and B where A resembles B, as in: A looks like B. Spatial verbs: Class of verbs in sign languages that express topographic relations in the signing space and agree with spatial locations by modifying the path movement of the verb. Speaker-oriented functions (of quotatives): Displays of the speaker’s affective or epistemic stance towards the quote. Stance (Affective): Speakers’ involvement in the topic, as displayed by evaluative devices, displays of emotion, dramatization etc. Stance (Epistemic): Speakers’ commitment to the propositional content or speakers’ display of their knowledge status. Story(-telling): See narrative Suspensive QPE: A subtype of quotative predicate ellipsis, where a quotative predicate heading a gerund clause is elided. Timitive: See Modality (Timitive) Transitivity: The property of taking direct objects. Unquote: Features used to mark the end of reported discourse. Depending on whether a language is spoken or signed, they can be lexical devices, particles, silences, prosodic features, body shift or shift in gaze. Verbum dicendi: See Quotative verb. Voice: Quotation contains multiple speakers (i.e., the narrator and the person quoted), who each have their own perspective and whose voices overlap in quotation (note the concepts of “multivoicedness” and “polyphony” raised by Bakhtin 1986 and Vološinov 1973). Zero quotative: The quoted material is uttered without introduction, without a verbum dicendi and/or quotative marker/construction (see Elliptic quotation).

Author index

A Antaki, Charles,  xvi, 6, 21, 31

Davies, Mark,  174, 185, 194 Du Bois, John W.,  8, 237

B Bell, Barbara,  xx, 173, 185, 189, 231, 238–40, 242, 246–47, 253 Bolden, Galina,  xix, 5, 32 Bucholtz, Mary,  xvi, 56 Buchstaller, Isabelle,  vii, xi–xxx, 40, 47, 49, 52–54, 62–63, 154, 162, 168, 173–75, 182–83, 191–93, 195–96, 231, 235–38, 240–41, 244, 246–48, 254, 260, 264, 270 Butters, Ronald R.,  xi, xiv, 174–45, 182, 196, 247

E Ebert, Karen,  14, 20, 119 Evans, Nick,  71–72, 78–79, 81, 84–85, 89, 96, 100, 107, 185, 287

C Cameron, Richard,  xviii, xx, 121, 154, 168 Chafe, Wallace L.,  xx, 8, 82, 85 Cheshire, Jenny  xi, 131–32, 232, 240, 245–46 Clark, Herbert H.,  xv, xvii, xx, 7, 11, 20, 51, 56, 58, 82, 148, 178, 183–84, 213, 250, 271, 285 Clift, Rebecca,  xvi, xix, 6–7, 11, 24, 82, 86 Coppen, Peter-Arno,  vii, xii– xiii, 5, 14, 32, 47, 56, 175, 231, 259–280 Coulmas, Florian,  xx–xxi, 6, 19, 78, 81, 85, 95, 108, 146, 148, 208, 268 Couper-Kuhlen, Elizabeth,  xx, 6, 11, 20, 24, 30–31, 40, 179 Croft, William,  73, 110, 181, 184 D D‘Arcy, Alexandra,  xii, xx, 38, 40, 52, 173, 190–91, 231, 236–38, 240–44, 253, 260, 270 Davidse, Kristin,  x, xx, 174, 177, 179, 181, 186, 189, 261, 264, 272

F Ferrara, Kathleen,  xx, 173, 185, 189, 231, 238–40, 242, 246–47, 253 Fleischman, Suzanne,  xiv, xix– xx, 5, 32, 84 Foolen, Ad,  vii, xii, xx, xxiii, 5, 14, 32, 47, 56, 71, 175, 231–32, 243, 252, 259–80 Fox, Sue,  vii, xi, xv, xxiii, 14, 46, 131–2, 187, 193, 222, 231–258, 270 Frajzyngier, Zygmunt,  71, 73, 76, 86–87, 89, 92, 96, 106–07, 109–10, 154, 168 G Gerrig, Richard J.,  xv, xvii, xx, 7, 11, 20, 51, 56, 58, 82, 148, 178, 183–4, 213, 250, 271, 285 Goffman, Erving,  xix, 14, 19, 23–24, 79, 236 Golato, Andrea,  viii, xiv, xvi, xix, xxiii, 3–36, 45, 50, 56, 134, 183, 204, 222, 232, 243, 245, 250, 253, 261, 267, 276 Goldberg, Adele,  73, 110, 184–85 Goodwin, Charles,  10, 14, 30 Goodwin, Marjorie Harness,  5, 10–11, 30 Güldemann, Tom,  viii, xvi– xxi, xxiii, 47, 81–82, 101, 117–42, 154, 161–3, 168, 175, 187, 245, 260–1, 264

H Haberland, Hartmut,  3, 5–6, 32, 204 Halliday, M.A.K.,  179–80, 185, 187 Haspelmath, Martin,  xv, 71, 89, 92, 98, 106, 248, 284 Hasund, Ingrid K.,  viii, xv– xvi, xxiii, 37–67, 175, 191, 260 Herrmann, Annika,  viii, xv, xxi, xxiii, 45, 49, 203–28, 276 Heritage, John,  8, 11, 20, 31 Holt, Elizabeth,  xix, 5–7, 11, 20, 25, 30 Hopper, Paul J.,  46–47, 76, 178, 187–88, 284 Hudson, Rachel,  xii, xx, 173, 180, 185, 189, 190–91, 231, 237–8, 241, 243 K Klewitz, Gabriele,  xx, 20, 30, 179 L Labov, William,  xi, 234, 249–50 Lakoff, George,  xvi, 63, 174, 183, 245, 282 Langacker, Ronald W.,  180–81, 185–86, 189 Lange, Deborah,  xv, 39, 42, 62, 173–74, 176–77, 181, 188–89, 195–96, 231, 242 Lee, Benny P.H.,  20, 42, 98, 208–9 Lehmann, Christian,  188–9, 191–92, 248 M Macaulay, Ronald,  xv, xviii, 173, 189–93, 231, 237, 242 Mathis, Terrie,  xvii, xix, 7, 14, 24, 30, 45, 180

 Quotatives McGregor, William B.,  73, 77, 82–3, 107, 121, 179–80, 183–84, 187 Meehan, Teresa,  174, 176, 182 Meyerhoff, Miriam,  xiv, xvii, xxii, 47, 54, 175, 187, 241, 264, 272 Myers, Greg, xix,  5, 10, 24, 29 N Noël, Dirk,  175, 190, 196 O Opsahl, Toril,  ix, xv, xxiii, 37–67, 191, 232, 243, 253 Oshima, David Y.,  ix, xii, xxi, xxiii, 96, 145–172, 211, 216, 262 P Partee, Barbara H.,  xx, 148, 184 Pascual, Esther,  xv, xx, 20 R Rickford, John R.,  173, 191–93, 246, 254 Romaine, Suzanne,  xv, 38, 42, 62, 173–74, 176–77, 188–89, 195–96, 231, 242

S Sano, Shin-ichiro,  ix, xxi, xxiii, 96, 145–172, 211, 216, 262 Schegloff, Emanuel,  7, 14, 23 Schourup, Lawrence C.,  xv, xvii, 20, 175 Singler, John,  xx, 194–95, 237–38, 242, 246 Spronck, Stef,  ix, xvi, xxi, xxiii, 38–39, 49, 56, 71–116, 162, 184, 224, 266, 286 Steinbach, Markus,  ix, xv, xxi, xxiii, 45, 49, 203–228, 262, 276 Streeck, Jürgen,  xviii, xx, 3, 14, 222 Svennevig, Jan,  x, xv, xxiii, 37–67, 92, 191 T Tagliamonte, Sali,  xii, xx, 40, 52, 173, 185, 189–91, 231, 236–38, 240–243, 253, 260 Tannen, Deborah,  xv, xviii, xx, 5–6, 11, 39, 222, 235, 247 Thompson, Sandra A.,  xix, 8, 178, 187, 208

Traugott, Elizabeth Closs,  46– 47, 76, 174–76, 186, 188, 190–91, 246, 284 V Van Alphen, Ingrid,  x, xi–xxx, 47, 175, 260, 265, 270–71 Vandelanotte, Lieven,  x–xi, xx, xxiii, 47, 53, 57, 73, 79, 110, 173–202, 261, 264, 267, 272 Vlatten, Andrea,  5, 14, 25, 30 Vološinov, Valentin,  xv, 71–72, 77–78, 283, 289 von Roncador, Manfred,  xx– xxi, 78, 134, 138 W Wierzbicka, Anna,  xv, xx, 7, 20, 71, 78, 183, 250, 271 Wooffitt, Robin,  8, 11, 31 Y Yaguello, Marina,  xix, 5, 32 Yule, George,  xvii, xix, 7, 14, 24, 30, 45, 180, 235

Index of terms

A Adolescent,  xii, 37, 40, 131, 193, 231–32, 234–40, 242, 244–46, 253, 266 Adposition,  163, 165 Adult,  40, 60 Adverb,  41–43, 46–47, 57–58, 62, 88, 90, 92–93, 97, 132, 150, 183 189, 192, 259–60, 262, 265, 275, 285 focus adverb,  42–43, 46 restrictive adverb,  42–43, 46–47 Adverbial, vii,  63, 76, 86, 89, 149–152, 157, 160, 165–66, 183, 205, 216, 264, Age,  8, 40–41, 44, 46–47, 58–60, 173, 193, 234, 237–38, 242, 244, 259, 270 Akkadian (Old),  xxi, 126–30, 134, 136–37 Analogy,  126, 128, 182, 184–85, 190–91 analogous,  150, 152, 167–68 analogical,  173, 189–92 analogization,  183, 191 Appealing marker, appealing function  53 Approximation,  xiv–xvii, 14, 37, 47–48, 54, 56, 63, 174, 182, 276, 285 approximative,  183, 195, 263–64, 276–77, 281; see Glossary Argument,  6, 19–21, 31, 93, 112, 152, 207, 221, 264, 281, 287 argumentation,  xvi, 31 Assessment,  xviii, 3–6, 11–12, 14–15, 20, 23–25, 29–31, 43, 46, 74, 173, 281; see Glossary

Attitude,  xii–xv, 30, 38–39, 49, 56, 71–76, 78–80, 83–84, 86–87, 91–92, 95–97, 103, 105, 107–08, 111–12, 145–47, 150, 152–154, 167, 206, 216, 224, 236, 260, 266–67, 270, 272, 278, 281, 287–88; see Glossary attitude predicate,  147, 150, 152–53, 167 negative attitude,  xii, 260, 267, 270, 272, 278 B be all,  173, 187–88, 193, 195, 254, 270 be like,  xii, xvii–xviii, 173–97, 231–32, 235–250, 253–54, 260, 267 Belgium,  xii, 203, 261, 266, 270, 272, 277 biclausal,  120, 131 Bininj Gun-Wok,  xxi, 89–90 Bislama,  xvii, 187, 272 Bulgarian,  xxi, 89–90, 104 C Cataphoric,  xvii, 187 Claim-backing,  6–8, 10, 14, 21, 23, 31, 281; see Glossary Combination (of markers, quotatives),  46, 56, 58, 61–62, 73, 76–77, 95, 98, 101, 104, 107, 158, 183, 191, 216–18, 266 Comical hypothetical,  25 Commitment,  xvi, 39, 43, 49, 56, 58, 71, 84, 86, 91, 94, 100–01, 111, 284–85, 289 affective,  43, 285 epistemic,  39, 56, 58

Complement,  147, 150, 152, 158, 173, 176–82, 187, 196, 210, 212, 263–67, 273–75, 277–78, 288 complementation,  76, 105, 177–81, 196, 278, 282; see Glossary complementizer,  xviii, 86, 88, 89, 92, 94, 107, 128–29, 146, 173–4, 176–77, 187, 196, 210, 268, 273, 285 Constructed dialogue,  xv, 209, 235, 238, 250, 252, 254, 282; see Glossary Construction,  xiii, xv–xix, xxii–xxiii, 5, 7, 9, 12, 14–15, 21, 23–25, 29–30, 38, 45–63, 71–112, 117–19, 122–25, 128–140, 145, 147, 152–68, 173–97, 205, 238, 244–45, 251–54, 259–78, 282–85, 287–89; see Glossary concessive,  154, 165 grammatical,  79, 117, 175 idiomatic,  161, 166–67 multiple perspective,  71–72, 79, 283, 287 quotative,  xvi–xxiii, 5–9, 12, 14–15, 21, 23–25, 30, 38, 51–63, 71–112, 145, 147, 153, 157–58, 167–68, 173–74, 179, 188, 195–96, 231, 253–54, 262, 264–65, 270–73, 277, 283–85, 288 reported message construction,  79, 93–94, 109–10 source construction,  xiii, xv–xix, 74–76, 82, 88, 91–93, 96, 99–101, 109–11 constructionalization,  173, 188, 190, 196, 282; see Glossary

 Quotatives Continuum hypothesis,  148–50, 282; see Glossary Conversation(al),  xv–xvi, 5, 7–9, 14–15, 21, 31–32, 39–41, 117, 131, 183, 206, 211, 216, 242–43, 270, 281–83 conversation analysis (/analytical),  xxii, 3, 6–7 conversational historic present,  242–43, 248 Corpus,  7–10, 25, 30, 32, 40–41, 44, 46–47, 58–61, 112, 122, 174, 194, 203–04, 210, 214, 217, 219, 232, 237, 242, 245, 261, 268, 270–71 Co-teller,  29–30 Croatian,  xiv, xxi, 132, 136–7, 245 Current speaker evaluative meaning,  76, 78–80, 83, 86, 92, 95, 101–02, 105–08, 112, 285 D Danish,  xiv, xxi, 5, 32, 38, 41, 47, 53, 60, 209 de dicto,  96, 108–09 de re,  108–09 Decategorialization,  188–89, 196, 282; see Glossary Deictic,  xiii, xiv–xv, xviii, 14, 30, 53–54, 57–58, 97, 118, 132, 148–50, 176, 180, 187, 207, 245, 250–52, 259–62, 273, 282–83, 285; see Glossary expression,  150 shifts,  148–49 Deixis,  viii, 56, 245, 282 Demonstration,  xv–xvii, 9, 11, 56–58, 63, 183–84, 241, 250, 283; see Glossary Demonstrative,  xiv–xv, 37, 53, 55, 58, 61, 118, 131, 187, 260, 266–67, 277 Diachrony/diachronic,  xviii, 40, 97, 117, 124–25, 167, 182, 187, 234, 253, 259–61, 272 Direct object,  145–46, 151–54, 158, 167, 178–79, 275, 289 Direct speech,  xxi, 19–20, 46, 50, 60, 85, 95, 110, 122, 175–77, 180, 183–90, 193–97, 212–13, 238–42, 247–50, 254, 259–61, 269, 273, 276–77, 283, 287; see Glossary

Discourse blended,  148 constructed,  6, 285 direct reported,  122, 126, 130–40, 145, 147–50, 212, 222, 235, 268, 270–71, 282 hypothetical,  3–15, 20–32, 204, 282, 285; see Glossary indirect reported,  19, 127–30, 137–39, 145–50, 184, 203, 268, 282 medium/semi-direct reported,  148 modeling,  7–8, 31 Discourse marker,  xii, 14, 38–48, 52–54, 57–62, 147, 176, 192 Distribution,  xiii, 39, 58, 61, 124, 163, 210, 232, 235, 237, 261, 271 Dramatic peak,  43, 46, 193, 231, 248, 253–54 Duna,  xxi, 89–90, 102, 109 Dutch,  vii, xi–xii, xiv, xviii– xix, xxi, xxiii, 5, 32, 56, 231, 243, 259–78 E Egyptian (Old),  xxi, 124–30, 133–37 Ellipsis,  121, 145, 153–56, 159–61, 164–68, 184, 216, 288–89 Elliptic quotation,  145, 152–54, 168, 283; see Glossary English,  xi, xiv, xvi–xix, xxi, 3, 5, 7–8, 10–11, 14, 20–21, 23, 25, 29–32, 38–39, 42, 44, 49, 52–54, 56, 62–63, 73–75, 79–81, 83, 86–87, 97, 100, 106, 110, 117, 119–21, 123–24, 131–32, 134, 136–37, 147–50, 152, 154, 159, 166, 168, 173–77, 179, 183, 187–88, 190, 192, 194–96, 231–32, 234–35, 237, 240–42, 244–46, 259–62, 267–68, 270, 272 Evaluation (evaluative),  xvi, 13–14, 39, 43–46, 71, 73–80, 83–92, 95, 101–113, 121, 165, 217, 236, 240–41, 244, 249, 254, 281–89 evaluative component,  75 Evenki,  xxi, 89–90 Evidential value,  xvi, 101, 109–12, 284; see Glossary

Evidentiality,  xvii, 71, 74, 79, 82–92, 101, 107–112, 284; see Glossary Ewe,  xxi, 89–90, 96–97, 103, 153 Exclamation,  46, 56 Explaining,  21, 31 Explanatory talk,  31 F Finnish,  xiv, 260 Focus marker,  37, 56, 62–63, 127, 284; see Glossary Footing,  14, 19, 23–25, 30, 138 Foreground,  xvi, xviii–xix, 78, 125, 134–35, 139–40, 243 French,  xiv, 5, 32, 112 Frequency (frequent),  xv, xvii, 7, 10–11, 14, 21, 37–38, 46, 48, 52–53, 58–61, 67, 87, 92, 96, 120–21, 125, 129, 131, 183, 185, 192–6, 208, 211, 217, 219–221, 236–37, 240, 242–46, 253, 259–60, 264–65, 267, 270–73, 276, 278, 283 G Generation,  xii, xvi, 37–40, 58–60, 234, 237, 242, 246 Genitive,  132, 264, 275, 277 Georgian,  xviii, xxi German,  xi, xiv, xviii, xxi, xxiii, 3–14, 19, 21, 25, 30–32, 56, 83, 86, 89–90, 97, 134, 203–04, 209, 213, 222, 232, 243, 245, 253, 259–62, 273, 285 Gesture,  xv, xviii, xx, xxii, 20, 138–39, 158, 179, 204, 206, 214, 220, 222–23, 235, 238–42, 248, 283, 285, 288 representational,  138 go,  xii, xiv, xvi–xviii, 150, 173–96, 235–37, 241, 246–54, 267, 270, 272, 277 Grammaticalization/grammaticalized,  xviii–xxii, 39, 47, 49, 62, 72, 76, 79–84, 87–92, 95–96, 99, 102, 105–12, 118–19, 123–27, 130, 136–40, 173–76, 188–91, 196, 207, 220–23, 238–42, 245, 248, 260, 272, 282, 284–85, 287–88; see Glossary Greek,  xi, xiv, xvii, 108



Index of terms  H Hausa,  xxi, 119–20, 129–30, 134, 136 Hearer expectancy,  112 Hearer-oriented,  37, 39, 53, 63, 285; see Glossary hebben,  260, 266, 269, 271, 276–77 Hebrew,  xi–xiv, xviii, 56, 121, 127 Hedge,  37, 39, 42–43, 47, 49, 51, 54, 62–63, 192, 285; see Glossary Hesitation marker,  14, 271 Honorific,  91–93, 97 Humorous stories,  3, 6–7, 25, 30–31 Hypothetical situation,  15, 21, 31, 51, 56 I Icelandic,  xi, xiv, 47 ideophone,  135, 138–40 Idiomatic phrase,  154, 161 Illocution,  103, 106–07, 161, 283 Illustration,  14, 19–21, 24, 29, 31, 49–52, 61, 63, 88, 101, 267 Imitation/imitative  138–39, 173–75, 182–85, 189–90, 196, 209, 285 Imperative,  43, 106, 148–49, 211 Indexicals,  150, 210, 212–13, 220–21, 223 Inner monologue,  51, 193 Innovative intransitive quotative (IIQ),  174, 186–88, 192, 196 Insubordinated speech,  95 Intensifier,  37, 39, 42–44, 46–47, 62–63, 189, 193, 254, 285; see Glossary Interactional linguistics,  37, 40 Interview,  xi, 20, 40–43, 210, 214, 218, 232, 234, 236, 260, 270–71 Involvement,  39, 44–47, 58, 63, 139, 189, 191, 245, 285, 289 Italian,  xi, xiv, 209, 221 J Japanese,  xi–xiv, xviii, xxi, xxiii, 71, 80, 85–86, 89–97, 105, 145–68, 211, 216, 262 Jarawara,  xxi, 89–90, 104

K Kanuri,  xxi, 122–23 Kera,  119–20, 153 Koasati,  xxi, 74 Kolyma,  xxi, 89–90, 105, 107 Korean,  xxi, 89–90, 93, 97, 104, 148, 153, 168 Koromfé,  xxi, 89–90, 95 Kunama,  xxi, 121, 126–27, 137 Kwaza,  xxi, 76, 80, 86, 89–92, 104 L Lamang,  xxi, 121 Lao,  xxi, 89–90, 96–97 Lele,  xxi, 71, 73, 76, 78–79, 84, 86, 89–90, 93–94, 103, 105, 110 Lezgian,  xxi, 89–90, 95–97, 106 Logophoric marker,  93, 96 London,  xiv, xxi, xxiii, 131, 187, 231–47, 253–54 M Manambu,  xxi, 89–90 Mimesis,  135, 138–40, 248, 285; see Glossary Modal value,  94, 109–11 Modality,  71, 74, 79, 83–95, 103, 106–12, 124, 203–04, 208, 210–13, 222–23, 286–87; see Glossary Deontic,  43, 85, 88, 90, 105 Doxic,  94–95 Dubitative,  76, 80, 84, 94 Volitive,  94–95 Frustrative,  88, 90, 92, 94, 103–04, 284, 286 Mode direct,  148, 157 indirect,  147–50, 157 Monoclausal,  119–20, 127, 130–31, 134, 136 N Narrative,  xv, xvii, 11, 21, 25, 43–46, 58, 77, 82, 112, 131–32, 193, 231, 236–54, 261, 276, 287; see Glossary Ndyuka,  xxi, 89–90, 97 Netherlands,  xii, xxi, 261, 266, 270, 272, 277 Ngarinyin,  xxi, 89–90, 95–96, 105 Non-clausal,  119–24, 127–31, 134, 137, 154

Non-topical agent,  123, 136–37 Norwegian,  xi, xiv, xvi, xxi, 37–43, 47–48, 52–54, 57–63, 92, 232, 243, 253, 260 O Old English,  176, 187 Onomatopoeia/onomatopoe(t) ic,  xiv, 46, 174, 182–83, 277 P Papuan,  96, 162 Performance,  xv–xviii, 20, 76, 139, 193, 231, 236, 243, 248, 250–53, 283, 287; see Glossary Person effect,  92, 99 Phonetic,  xx, 47, 189, 270 Politeness,  43, 84, 100 Portuguese,  xi, xiv, xxi, 132, 136 Presentational clause,  117, 125, 132, 245 Prosody/prosodic,  xv, 50, 55–57, 73, 100, 107, 111, 179, 204–05, 210–11, 222, 271, 289 Q Quantitative,  37, 42, 60, 231, 261 Quotation direct,  6, 14, 25, 32, 122, 183, 195, 212–13, 223, 268–69, 271, 277, 282 indirect,  6, 195, 269, 278 Quotative,  xi–xxiii, 5, 7, 9, 12–15, 19, 21, 23–25, 30, 37–41, 44–63, 71–112, 117–140, 145–68, 173–96, 216, 231–54, 259–78, 283–85, 288–89; see Glossary quotative construction, see Construction, quotative quotative index,  xvii, xix, 117–40, 154, 288 quotative verb,  19, 45, 49–52, 58, 60, 63, 119, 124–30, 237, 251, 262, 289; see Glossary zero quotative,  24–25, 30, 180, 235, 238, 289; see Glossary Quotative predicate/predicate of communication,  145–47, 151, 153–59, 167, 216, 283, 288; see Glossary Quotative predicate ellipsis (QPE),  145, 153–68, 216, 288; see Glossary sentence-final,  154, 159–62, 168

 Quotatives suspensive,  154–60, 167–68, 289; see Glossary Quote,  xv–xvi, xviii–xix, 5–7, 11–14, 20, 24, 30, 32, 37, 45–63, 72–74, 96, 105, 117–40, 145–50, 154, 160–63, 175–80, 183–84, 195, 204–06, 211–20, 223, 236–48, 253, 261–71, 287–89; see Glossary direct,  118, 122, 134, 137, 140, 146, 148, 160, 175–78, 184, 195, 265, 269–70, 287 indirect,  146, 148–49, 267, 287 quote orientation,  118–21 quote-initial,  60 quote-final,  52, 60, 63 R Reduction,  47, 164, 66, 189, 91 Reenactment,  55–57 Reportative,  xii, 77, 82, 85–86, 91, 94–95, 98, 101–02, 173–74, 180, 285, 288; see Glossary Russian,  xi, xiv, 5, 32, 71, 92, 183 S Semantic bleaching,  46, 263 Serbian,  xxi, 132, 136–37, 245 Shona,  xviii, xxi, 135, 140, 187 Sign language,  xv, xx–xxiii, 203–24, 262, 281, 284, 286–87, 289 German (DGS),  xxi, 203–24, 262 Similarity,  xiii–xiv, xvii, xix, 14, 37, 48–49, 62–63, 174–76, 182, 213, 264,

Similative meaning,  260, 264, 277, 289; see Glossary Spanish,  xi, xiv, xviii, xxi, 121, 154, 168, 234, 260 Speaker-oriented,  37, 39, 44, 58, 63, 289; see Glossary Speech verb,  xvii, 100, 111, 122, 126–27, 134 Staging,  7, 50, 55, 252 Stance,  xv–xviii, xxii–xxiii, 6–10, 20–21, 24, 31, 37–49, 56–63, 289; see Glossary affective,  37, 39, 63 epistemic,  xv–xvi, 37, 39, 63, 289 Storytellings,  7, 25, 29–30, 236, 250, 282 Subjectivity,  39, 72, 108 Swedish,  xi, xiv, xxi, 38, 47, 52–53, 58, 63, 134, 209 Synchrony/synchronic,  xxiii, 39, 80, 124, 128–29, 136, 174, 182, 191, 210, 259–60 Syntax/syntactic,  xx, 42, 47–49, 61, 73, 76, 86, 113, 117–20, 124–26, 129–37, 140, 150, 153, 158–59, 175–76, 179, 186, 189, 204–08, 211–12, 220–21, 259, 264–65, 268–73, 277, 287–88 T Tariana,  xxi, 77–78, 82, 85, 89–90, 102, 106 Thetic,  117, 136, 140 Tikar,  xix, xxi, 123, Tonga-Imhambane,  xxi Topic markers,  164–65

Toqabaqita,  xxi, 71, 89–90, 96, 104, 106 Transcription,  4, 8, 36, 41, 44, 67, 232, 257 Transitivity/transitive,  155, 173–74, 178–81, 186–88, 192, 196, 264, 267; see Glossary Tubu,  xxi, 122–23, 126–27, 138 Turkish,  xxi, 89–90, 104, 121, 260 Tuvaluan,  xxi, 89–90 Typification,  xv, xvii, 25, 276 V van,  xii, xiv, 260–61, 267, 272, 277 in combination with hebben/ zijn and zo(iets),  260, 262, 266, 269, 271, 275–78 Verbification,  126, 130 Viewpoint-sensitive expression,  149 W Wari,  xxi, 71, 89–90, 96 Warrwa,  xxi, 121 West-Greenlandic,  xxi, 89–90, 96, 103 Y Yukaghir,  xxi, 89–90, 105, 107 Z Zapotec Mayan,  xxi, 75 zijn,  266, 269 zo(iets),  xiv, 260, 262, 265–67, 269, 271, 275–78