Pragmatics and Corpus Linguistics: A Mutualistic Entente 9783110199024, 9783110195804

Citation preview

Pragmatics and Corpus Linguistics

≥

Mouton Series in Pragmatics 2 Editor Istvan Kecskes

Editorial Board Reinhard Blutner Universiteit van Amsterdam The Netherlands N. J. Enfield Max-Planck-Institute for Psycholinguistics Nijmegen The Netherlands

Ferenc Kiefer Hungarian Academy of Sciences Budapest Hungary Lluı´s Payrato´ University of Barcelona Spain

Raymond W. Gibbs University of California Santa Cruz USA

Franc¸ois Recanati Institut Jean-Nicod Paris France

Laurence R. Horn Yale University USA

John Searle University of California Berkeley USA

Boaz Keysar University of Chicago USA

Mouton de Gruyter Berlin · New York

Deirdre Wilson University College London Great Britain

Pragmatics and Corpus Linguistics A mutualistic entente edited by

Jesu´s Romero-Trillo

Mouton de Gruyter Berlin · New York

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter GmbH & Co. KG, Berlin.

앝 Printed on acid-free paper which falls within the guidelines 앪 of the ANSI to ensure permanence and durability.

Library of Congress Cataloging-in-Publication Data Pragmatics and corpus linguistics : a mutualistic entente / edited by Jesu´s Romero-Trillo. p. cm. ⫺ (Mouton series in pragmatics ; 2) Includes bibliographical references and index. ISBN-13: 978-3-11-019580-4 (cloth : alk. paper) 1. Pragmatics. 2. Corpora (Linguistics) 3. Discourse analysis. I. Romero-Trillo, Jesu´s. P99.4.P72P737 2008 4101.1⫺dc22 2008032962

Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de.

ISBN 978-3-11-019580-4 쑔 Copyright 2008 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin. All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Cover design: Martin Zech, Bremen. Printed in Germany.

Contents

Introduction: Pragmatics and corpus linguistics – a mutualistic entente Jesús Romero-Trillo At the interface between grammar and discourse – a corpus-based study of some pragmatic markers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Karin Aijmer The subjectivity of basically in British English – a corpus-based study Christopher S. Butler

1

11

37

Variation in advanced oral interlanguage: The effect of proficiency on style choice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 Jean-Marc Dewaele The use(fulness) of corpus research in cross-cultural pragmatics: Complaining in intercultural service encounters. . . . . . . . . . . . . . . . . . . . . . . 93 Ronald Geluykens and Bettina Kraft Hesitation markers among EFL learners: Pragmatic deficiency or difference? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 Gaëtanelle Gilquin Evidentiality in discourse: A pragmatic and empirical account . . . . . . . . . 151 Leo Francis Hoye Multi-modal corpus pragmatics: The case of active listenership. . . . . . . . 175 Dawn Knight and Svenja Adolphs Discourse markers and the pragmatics of native and non-native teachers in a CLIL corpus. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 Ana Llinares-García and Jesús Romero-Trillo A cross-linguistic study on the pragmatics of intonation in directives . . 205 M. Dolores Ramírez-Verdugo

vi

Contents

The discourse-grammar interface of regulatory teacher talk in the EFL classroom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235 Silvia Riesco Bernier

Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263 Index of subjects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269 Index of persons. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271

Introduction: Pragmatics and corpus linguistics – a mutualistic entente Jesús Romero-Trillo

The present volume is neither a book on corpus linguistics nor on pragmatics stricto senso. In fact, the subtitle “A mutualistic entente” intends to go beyond the limits of both disciplines to shed light on their intricate relationship. The unusual collocation, “a mutualistic entente”, is a composite based on the concept of mutualism insofar as it tries to depict the relationship between two independent linguistic disciplines of different species that benefits both, as it is commonly understood in ecology. The other term, “entente”, though its meaning also refers to an agreement between several parties in a general sense, often alludes to political pacts, i.e. the Balkan Entente (the mutual-defence agreement between Greece, Turkey, Romania, and Yugoslavia, signed in February 9, 1934), the Triple Entente (the association between Great Britain, France, and Russia, the nucleus of the Allied Powers in World War I) or the Baltic Entente (the mutual-defence pact signed by Lithuania, Latvia, and Estonia on September 12, 1934). These “ententes” indicate bellicose situations in which some countries have allied to make a stronger position against external forces. In fact, this is the rationale for the present volume: pragmatics and corpus linguistics have not only helped each other in a relationship of mutualism, but, they have also made common cause against the voices that have derided and underestimated the utility of working with real data to elucidate the patterns of language use. The intention of this volume is to present a collection of articles written by leading scholars that have merged pragmatics and corpus linguistics, with the conviction that some specific language issues can be better addressed through the combination of methodologies pertaining to both disciplines. Another non-otiose reason for the publication of this volume is the silver jubilee commemoration in 2008 of the publication of three of the books that have possibly influenced most the attraction to linguistics for many language scholars and language aficionados: Pragmatics by Stephen C. Levinson, Principles of Pragmatics by Geoffrey N. Leech and Discourse Analysis by Gillian Brown and George Yule, all published in 1983. The coincidence of the publication of the three books in the same year, far from an oracular

2 Jesús Romero-Trillo interpretation, represents the effervescent search for a new paradigm that could reconcile theoretical descriptions of language, some of classical ancestry, with the new possibilities of the electronic archiving of texts. With regard to corpus compiling, it is mandatory to mention, albeit briefly, two corpora that I consider the most representative in the English language for their precursor status: the Brown Corpus and the London-Lund Corpus. The publication in 1964 of the Standard Sample of Present-Day American English (the Brown Corpus), edited by W. N. Francis and H. Kucera, with a total number of 1,014,312 running words, preceded the advent of the London-Lund Corpus of Spoken English on the other side of the Atlantic. The London-Lund corpus combined the work of the Survey of English Usage (SEU) at University College London, launched in 1959 by Randolph Quirk, and the Survey of Spoken English (SSE), started by Jan Svartvik at Lund University in 1975. The complete collection of its texts – 100 – was published in 1980 by Jan Svartvik and Randolph Quirk and it contained such detailed prosodic annotation that it has guaranteed its emblematic value throughout the years. Both corpora are at the inception of corpus linguistics and studies based on their data have made history in terms of methodology and possibilities of corpus linguistics1. In the following decades many types of corpora appeared and many directions and fashions have affected corpus linguistics. To a great extent, for many years studies on corpora took a quantitative, down to earth, hands-on perspective, in contrast with the pragmatic approach that propounded a ‘language as it is’ vs. a ‘language as it means’ perspective. Obviously, these two approaches collided with the ‘language as a system’ and ‘void-ofcontext pollution’ approach of more traditional linguistics. It would be a Herculean task to reflect the immense production on different aspects of corpus linguistics applied to different fields of language use in recent years and its description is certainly beyond the scope of this volume, whose aim is to discover the net that pragmatics and corpus linguistics have woven together to sustain a new form of researching language and language use. For many years Corpus linguistics and Pragmatics have represented two paths of scientific thought, parallel but often mutually exclusive and excluding. As regards corpus linguistics, during its early days it was merely seen as a “bundle of methods and procedures that simply dealt with empiri1

See for example the seminal volume: Karin Aijmer and Bengt Altenberg (eds.) (1981). English Corpus Linguistics. London: Longman.

Introduction

3

cal data in linguistics” 2, primarily used for grammatical 3, lexicographic and typological purposes4. To fulfil this aim, corpus linguistics was soon influenced by mathematics and statistics, and was considered a very accurate discipline with a meticulous methodology. In fact, it was soon assimilated to a branch of what was termed “quantitative linguistics”5 Conversely, the main feature of Pragmatics was its indefiniteness, it was a discipline that sailed the sea between sentence meaning and intended meaning with an unclear rhumb. Nevertheless, the discipline soon had a tremendous impact on several areas of knowledge such as philosophy, ethnography6 and psychology7. In fact, recent years have seen an increase in empirical studies of pragmatics in areas such as power and politeness, speech acts, and intercultural pragmatics and, to substantiate the results, studies have resorted to corpus linguistics as a methodology. Likewise, many studies in the area of corpus linguistics now use pragmatics as a model for the interpretation of data. It looks as if the pass of time had sieved out the theoretical excesses and arcane fundamentalisms from both sides. The distance grown between both disciplines for some years often arose from the lack of understanding and knowledge of the mutual possibilities. Leech, for instance, asserted with aplomb: I wish to argue that computer corpus linguistics (…) defines not just a newly emerging methodology for studying language, but a new research enterprise, and in fact a new philosophical approach to the subject. The computer, as a uniquely powerful technological tool, has made this new kind of linguistics possible. So technology here (as for centuries in natural science) has taken a more important role than that of supporting and facilitating research: I see it as the 2

3

4

5

6

7

Nelya Koteyko (2006). Corpus Linguistics and the Study of Meaning in Discourse. The Linguistics Journal 2 (2): 132–157. See for instance: Edward L. Keenan (2001). A Quantitative Study of Voice in Malagasy. Oceanic Linguistics 40: 67–84. See for instance: John Myhill (1992). Typological Discourse Analysis: Quantitative Approaches to the Study of Linguistic Function. Oxford: Blackwell. For a detailed description of the immense productivity in quantitative linguistics see: Reinhard Köhler (1995). Bibliography of Quantitative Linguistics. Amsterdam: John Benjamins. The volume numbers over 6,500 titles of quantitative language research up to that year. See for instance: Ashley E. Maynard (2004). Cultures of teaching in childhood: Formal schooling and Maya sibling teaching at home. Cognitive Development 19: 517–535. See for instance: Arthur C. Graesser, Shane S. Swamer and Xiangen Hu (1997). Quantitative discourse psychology. Discourse Processes 23: 229–264.

4 Jesús Romero-Trillo essential means to a new kind of knowledge, and as an “open sesame” to a new way of thinking about language8. In sum, Leech bets for the inclusiveness of corpus linguistics and other theories or approaches to language, i.e., corpus linguistics as a tool for a better inspection of the language phenomenon. This conception was not shared by many linguists who dreaded what was felt the presumptuous omnipresence of statistics and corpus linguistics in all areas of language. For instance, Fillmore warned linguists of this danger by stating that corpus linguistics is not statistical linguistics 9. In other words, statistics should not be used to validate or refute language theories because the capacity of corpora to represent language was always limited, no matter the size of the corpus. He insisted on this idea by arguing against the fallacy of always asking for a wider corpus when some feature of language, discovered and accepted by means of the intuition of the linguist or language user, is not found. Here we encounter the debate that has kept vividly active in recent times of authentic-real data vs. introspectionism. Wallace Chafe, on the same wavelength 10, compares the value of introspection and elicitation vs. corpus data, understood as authentic records of language within a context. In his opinion, the essential point is that the statistical tabulation is not an aim in itself but the starting point of the analysis that follows and should be based on sound theoretical principles. In fact, he holds the suspicion that often corpora are assembled for their own sake, and then the compiler counts the distributions of an item without caring for what the distributions signify. Against this procedure, he mentions that science has two components: the systematic, which accounts for non casual, probing observations of small portions of reality, that we call data; and the imaginative, that is based on the creation of descriptions of parts of the universe that extend beyond the observed phenomena, and include causal explanations for those phenomena, what we call theories. Chafe ends his analysis of corpus linguistics by asserting that the object of corpus linguistics is not the explanation of what is in the corpus but of the nature of language. Therefore, linguists must not limit the data to a corpus sample, but 8

9

10

Geoffrey Leech (1992). Corpora theories of linguistic performance. In Directions in Corpus Linguistics. Proceedings of Nobel Symposium 82. Stockholm, 4–8 August 1991, Jan Svartvik (ed.), 105–122. Berlin /New York: Mouton de Gruyter. Charles J. Fillmore (1992). Corpus linguistics or Computer-aided Armchair linguistics. In Jan Svartvik (ed.) op. cit., 35–60. Wallace Chafe (1992). The importance of corpus linguistics to understanding the nature of language. In Jan Svartvik (ed.) op. cit., 79–97.

Introduction

5

they have to provide a framework to know what questions should be asked about language. These opinions by two linguists of repute give an idea of how the apparent scientific supremacy of corpus linguistics placed traditional linguistics in some difficulty, although at the same time theoretical linguists saw the importance of this discipline for the betterment of linguistic analysis. Following the debate, Michael Halliday 11 presents the concept of “probabilistic grammar”, one of the theoretical hallmarks of Systemic Functional Grammar, and suggests that grammatical features in a language can be better modelled through a probabilistic analysis. This hypothesis is at the origin of grammatical descriptions of language based on corpora that are now prevalent and had the publication of “A Comprehensive Grammar of the English Language”, by Quirk et al.12, as their most distinguished antecedent, succeeded by, amongst others, Biber et al’s “Longman Grammar of Spoken and Written English”13. In fact, Halliday states that it is only through the analysis of spontaneous conversation that one can obtain the full semantic and grammatical potential of the language system. This statement about the value of conversation could be also applicable to its capacity to capture the full-fledged pragmatic force of the system. As a corollary, the notion of probabilistic grammar has been developed, inter alii, by Romero-Trillo, with the incorporation of the notion of social variation (as described by Labov14) in the studies on the statistical modelling of discourse markers in conversation15. Furthermore, it can be argued that, in the same way as for Halliday grammar and vocabulary are two sides of the same process (the lexicogrammar), corpus linguistics and pragmatics are two versions of the same phenomenon: the mechanics – the subjectmatter – (corpus studies), and its interpretation and explanation (pragmatics), 11

12

13

14

15

Michael A. K. Halliday (1982). Language as system, language as instance: the corpus as a theoretical construct. In Jan Svartvik (ed.) op. cit. Randolph Quirk, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik (1985). A Comprehensive Grammar of the English Language. London: Longman. Douglas Biber, Stig Johansson, Geoffrey Leech, Susan Conrad and Edward Finegan (1999). Longman Grammar of Spoken and Written English. London: Longman. William Labov (1972). The study of language in its social contexts. In Sociolinguistic Patterns, 183–259. Philadelphia: University of Pennsylvania Press. Jesús Romero-Trillo (2001). A mathematical model for the analysis of variation in discourse. Journal of Linguistics 37: 527–550.

6 Jesús Romero-Trillo as for example Karin Aijmer has shown in her pragmatic analysis of discourse particles16. A traditional drawback for the pragmatic analysis of corpora, as Tony McEnery and Andrew Wilson point out, resides in the fact that there is often a deficit of information with regard to social and textual contexts, and thus, it is not always easy to infer the veracious context of the corpus 17. Nowadays this drawback is being addressed with the plethora of corpora that include refined descriptions of the social, geographical and age descriptions of the participants. This concretion in the pursuit of scientific credibility and control of the data for quantitative pragmatic analysis coexists with the possibilities of searching for data in the web, as Renouf and Kehoe (2006) indicate 18. In fact, for many researchers, the size of the corpus has preempted the control of its internal variables as there is no limit to the need of quantity of items to study linguistic phenomena in depth. The beginning of the 21th Century has propagated the increasing use of internet as a source for corpus building. The web has incorporated new text-types, such as e-mails, blogs, chat-room discussions, etc. which represent our real daily use of language. Hund et al.19 in their introduction to the volume ‘Corpus Linguistics and the Web’ discuss some of the advantages of using the web as the corpus, primarily because of the considerable time and financial resources used to compile data that is soon out of date in traditional corpus building. For compilation purposes, they distinguish two approaches: Web as Corpus and Web for Corpus Building. The former, in their opinion, has some methodological caveats such as the following: there is no indication of size, there is no possibility of replicability due to the ephemeral presence of texts, there is no indication of text types, webcrawlers prioritize sites which are closer to the computer so different results according to place of search, and web-crawlers create profiles for users. The Web for Corpus Building approach, on the other hand, enables the lin-

16

17

18

19

Karin Aijmer (2002). English discourse particles. Evidence from a corpus. Amsterdam /Philadelphia: John Benjamins. http://bowland-files.lancs.ac.uk/monkey/ihe/linguistics/contents.htm (access 5th May 2007). Antoinette Renouf and Andrew Kehoe (2006). The changing face of Corpus Linguistics. Amsterdam: Rodopi. Carolin Biewer, Nadja Nesselhauf and Hundt Marianne (eds.) (2006). Corpus linguistics and the Web. Amsterdam: Rodopi.

Introduction

7

guist to exert some control on content and text-types when culling data from the web, and allows the use of annotated off-line monitor corpora. The present volume is organised around the two topics that I deem essential for the coalescence of corpus linguistics and pragmatics: firstly, Corpus linguistics at the interface with the pragmatics of conversation and pragmantax, with four articles by Karin Aijmer, Chris Butler, Dawn Knight & Svenja Adolphs, Leo Hoye and Ron Geluykens & Bettina Kraft. Secondly, Corpus linguistics and Intercultural Pragmatics from a theoretical and language learning stance, with articles by Dolores Ramírez-Verdugo, JeanMarc Dewaele, Ana Llinares-García & Jesús Romero-Trillo, Silvia RiescoBernier and Gaëtanelle Guilquin. Aijmer’s contribution, At the interface between grammar and discourse – a corpus-based study of some pragmatic markers, introduces the notion of emotion and the meta-representation of involvement in the dialogic enterprise between speakers and hearers, specifically, through the analysis of certain pragmatic markers. The choice of the elements surprisingly, sadly, fortunately is explained by the author because, unlike other adverbs, they contribute to the pragmatic interpretation of the message beyond their semantic load. Aijmer characterises these evaluative elements under the aegis of evidentiality theory to make a clear-cut differentiation between pragmatic markers and commentary markers, an incisive proposal substantiated with different corpora that evidences the different pragmatic function of both types of elements. Knight and Adolphs’ chapter, Multi-modal corpus pragmatics: The case of active listenership, presents the possibilities that the multi-modal collection and analysis of corpora with technical facilities offer for pragmatics research. Their contribution focuses on the role of active listenership and use of backchannel signals to indicate that the listener is following the thread of speech. The novelty of their analysis lies in the multi-modal approach to discourse that includes gestures, eye-movement as a support – not ancillary but essential– for the correct interpretation of –especially – face to face conversation. In fact, the multi-modal approach underlines the potency of kinesics to describe language use. Also, the access to visual information eliminates the need to annotate with cumbersome glosses the gestures performed by speakers in a conversation. This article undoubtedly links corpus linguistics and pragmatics with the flourishing epoch of ethnomethodology and sheds new light on the understanding of conversational phenomena. Geluykens and Kraft discuss the utility of data triangulation in Cross Cultural Pragmatics in comparison with the traditional controlled elicitation

8 Jesús Romero-Trillo data, based on production questionnaires. The authors make an insightful revision of the most significant methods of data collection and analysis in pragmatics with especial attention to recent studies based on triangulation. To illustrate this methodology, Geluykens and Kraft present their study on complaint sequences in intercultural service encounters, with very interesting results that reinforce the adoption of a triangular methodology in Cross Cultural Pragmatics. Chris Butler’s contribution, The subjectivity of ‘basically’ in British English, a corpus- based study refers to the phenomenon of subjectivity in language as evidenced in the use of the adverb basically compared with fundamentally and essentially. After a thorough theoretical discussion on the notion of subjectivity from a synchronic and diachronic perspective, with an acute description of its cognitive implications, the author approaches the semantic and grammatical description of the different meanings of the three selected adverbs as they appear in the BNC. Butler completes his study with the analysis of these elements from a quantitative perspective and a comparison with other discourse markers that are pragmaticalised in speech. Leo Hoye makes an insightful contribution entitled ‘Evidentiality in Discourse: A Pragmatic and Empirical Account’ that surveys the theoretical evolution and present state- of- the-art of evidentiality studies. Hoye’s approach, eclectic as it is, succeeds in establishing a link between the Weltanschauung that animates any interaction and the language used to perform it, a view that permeates his hypothesis of the pragmatic principle of evidential substantiation. This principle, as the author avers, is not a rule but a probabilistic tool that mainly resorts to modality to establish and develop the rapport between speaker/writer and audience. With the aid of corpus examples of several genres, Leo Hoye constructs a model that certainly clarifies the role of evidentiality in communication. Dewaele’s chapter addresses the paradoxical issue of the difficult transition from formal to informal speech in non-native speakers of French. The hypothesis is that non-native speakers feel ‘safer’ in a formal register and only very good learners reach the competence to navigate the currents of informal language. Based on previous findings, the author investigates the effect of proficiency on style choice by considering the effects of amount and intensity of formal instruction on the target language, and the amount of authentic communication in the target language outside the classroom. To validate his hypothesis, Dewaele looks into the morpholexical accuracy of the participants with regard to their level of explicitness. Then, the sociobiographical variables selected for the analysis are used to give an accurate

Introduction

9

description of the relationship between linguistic competence in the L2 and the use of an informal register. Ana Llinares-García and Jesús Romero-Trillo examine the realisation of discourse markers of native and non-native teachers in a CLIL corpus and the pragmatic implications of the different uses in learners of English. Departing from a discourse-cognitive approach, the authors embark on a description of the most frequent functions of discourse markers in teacher talk in different Content and Language Integrated Learning (CLIL) contexts, to then descend into the mechanics of their appearance in classroom registers. Gaëtanelle Guilquin investigates the different appearance of hesitations in the speech of French non-native speakers of English compared to native speakers. The author hypothesises that the non-natives overuse non-lexical elements to the detriment of lexical ones. Although this disparity may not be considered important, with the proviso that hesitations are mutually interchangeable, the author argues that this fact may result in a pragmatic failure in communication with a negative opinion of foreign speakers of English. Silvia Riesco-Bernier analyses the discourse-grammar interface of regulatory teacher talk in the EFL classroom through a qualitative and quantitative examination of the lexicogrammatical realisations of fifteen regulatory functions in the UAM-Learner English Spoken Corpus, considering both native and non-native teachers of English. She concludes that there are three major findings to be highlighted: (i) a lack of “bi-uniqueness” between the regulatory function and their lexicogrammatical realisation, which questions the one-to-one function-form association; (ii) the existence of “dependency” between the two variables under study and (iii) the evidence of differences in the native and non-native teachers’ regulatory talk, a fact that calls for pedagogical implications in the EFL pre-school classroom. The last chapter by Dolores Ramírez-Verdugo reports a cross-linguistic study on the pragmatics of intonation in directives. The study explores the prosodic realization of directives and its pragmatic effects in classroom discourse. It analyses the communicative meaning transmitted by the intonation patterns used in scripted dialogues in comparison with the patterns used in natural classroom discourse by native and non-native (Spanish) teachers of English in the UAM-Learner English Spoken Corpus. The author confirms the existence of prosodic differences in the choice of intonation patterns produced by the two language groups both in experimental and naturalistic speech, and hence, differences on the pragmatic meaning transmitted in similar communicative contexts. As the reader will have noted, the articles in this volume represent the compatibility from various stances as the authors pertain to different lin-

10 Jesús Romero-Trillo guistic traditions, but all have a common interest in establishing connections between pragmatics and corpus linguistics. Also, the authors belong to different age generations and are therefore influenced by different traditions. It is important to emphasize that some of these scholars, in the past, provoked academic meanders, dissociated from the main linguistic riverbeds, that have now become plentiful flows of inspiration for the new generations of linguists. To conclude, I would like to quote the words expressed by Jacob Mey in a special volume of the Journal of Pragmatics devoted to this topic: Current studies in Corpus Linguistics are much more than, and very different from, the statistical rehashing of data that was standard in the ‘quantitative linguistics’ (as it was called then) of thirty-forty years ago; and further, that the study of the pragmatic aspects of language use need not preclude a rigo20 rous quantification of linguistic data.

20

Jacob Mey (2004). Introduction to special issue on Corpus Linguistics, Part III. Journal of Pragmatics 36: 15–19.

At the interface between grammar and discourse – a corpus-based study of some pragmatic markers 1 Karin Aijmer

1. Introduction Emotion has to do with our reactions to events. For example, an unexpected event may cause excitement or a high degree of surprise and cause a verbal or non-verbal reaction: Much ongoing activity can be explained in terms of our non-verbal and verbal reactions to unexpected events. We react by ‘excitement, sometimes anger, and sometimes aggression’ in the face of what is new. (Chafe 1994: 122)

We are not necessarily conscious about our emotions unless they are very strong: One can assume that there is a constant emotional involvement of a human being with everything in the world (s)he has knowledge of. As long as this emotional involvement remains within some limits of ‘strength’, it may very well go unnoticed. An individual can become aware of his/her emotions, however, because they grow too strong and ‘demand’ attention by explicitly referring to his or her emotions. (Nuyts 1990: 234)

The speaker’s or writer’s evaluation can be regarded as ‘metarepresentative’, it results from the speaker’s/writer’s attending to his/her emotions (cf. Nuyts 2001: 40). This raises the question: why do speakers and writers find it necessary to intrude into the discourse in order to express their emotional involvement and what further functions can be signalled by these expressions? When does it happen and in what types of text does it occur? What are the functions of the expressions and how are they related to the achievement of the speaker’s/writer’s goals in the communication? In this article I will look at elements which are generally regarded as adverbs (in particular disjuncts). The focus will be on evaluative disjunct adverbs such as surprisingly, sadly, fortunately but I will also consider adverbs 1

I am grateful to Göran Kjellmer for reading an earlier version of the text.

12 Karin Aijmer expressing other attitudes and discourse markers. Following Fraser (1996) I will refer to adverbs such as surprisingly or fortunately as a type of pragmatic markers. They are unlike other adverbs and contribute to the pragmatic interpretation of the utterance rather than its semantics (Section 2). In Section 3 a distinction will be made between commentary markers and discourse markers. The data which is taken from the British National Corpus is presented in Section 4. The following sections are devoted to a description of the factors or properties characterising the evaluative markers. A criterion for evaluative commentary markers is that they are ‘performative’ and refer to the speaker’s/writer’s current evaluation (Section 5). Section 6 discusses the parenthetical status of the evaluative markers. A note on evaluative markers and emergent grammaticalization is given in Section 7. Section 8 further characterises the evaluative markers in terms of their lack of informational salience in the utterance. In Section 9 it is argued that the evaluative markers can be described with regard to evidentiality as subjective (personal responsibility) or intersubjective (shared responsibility). Evaluative markers have also developed discourse strategic functions, e.g. to challenge or agree with assumptions, beliefs and expectations which are part of the common context. Disagreement with expectations is dealt with in Section 10 and agreement with hopes and expectations in Section 11. Combinations of discourse markers and commentary markers used discourse-strategically for extra rhetorical strength are discussed in Section 12. Section 13 contains the conclusion. 2. Pragmatic markers In the last few decades we have witnessed an expansion of research in the area of pragmatics characterised by new directions, methods, and perspectives (Fischer 2006). It is in line with this general trend that elements which have earlier been analysed in grammar have had a revival in pragmatics. A good example of this phenomenon is the adverbial categories. Both conjuncts and disjuncts are of interest in this context. Conjuncts and disjuncts are elements which are only loosely integrated with the rest of the sentence. Elements such as again, however, now (Greenbaum 1969: 25) indicate some connection with what comes before and are therefore termed conjuncts. On the other hand, the adverbs briefly, probably and frankly are designated disjuncts ‘a term suggesting their lack of integration within the clause to which they are subordinated’ (Greenbaum 1969: 25). Conjuncts and disjuncts have in common a number of restrictions on

At the interface between grammar and discourse

13

the type of clauses where they occur. For example, neither disjuncts nor conjuncts can be the focus of a cleft construction (*it is surprisingly that John left). Although they look like adverbs they are used with functions which are not typical of adverbs in general. When a word or a phrase like and then, first of all occurs it is classified as part of the nearest main clause, but that is essentially a convenience because these phrases have virtually no links with the clause to which they are assigned, and acquire names like “conjuncts” (conjunctive adverbs) and “disjuncts” which sounds almost disparaging. (Sinclair and Mauranen 2006: 75)

Disjuncts and conjuncts do not modify the propositional meaning conveyed. Instead, a sentence like Frankly Bill is fat seems to assert both that Bill is fat and that the speaker is frank in saying that (cf. e.g. Bellert 1977). This analysis does not fit in with the analysis of adverbs in general. Furthermore it is unclear what it means to say that a sentence simultaneously contains two assertions. Functionally, disjuncts and conjuncts can be regarded as a kind of pragmatic markers, signalling the meaning of a message. According to Fraser (1996), ‘these pragmatic markers, taken to be separate and distinct from the propositional content of the sentence, are the linguistically encoded clues which signal the speaker’s potential communicative intentions’ (Fraser 1996: 168). For example in Stupidly, Sara didn’t fax the correct form in on time (Fraser 1996: 168) stupidly is not a part of the propositional content but functions as a type of pragmatic marker. The basic message can be taken as a report ‘while the commentary message, signalled by stupidly, is that the speaker believes Sara’s failure to act to have been stupid’ (Fraser 1996: 168).

3. Commentary evaluative markers and discourse markers Messages or propositions can be of many different types depending on the marker introducing the message. I will be concerned here only with what Fraser refers to as commentary markers and the messages they convey. ‘Comment’ plays an important role in many accounts of how hearers and readers interpret utterances. It has been associated with procedural information contributing to the pragmatic interpretation of the utterance by pointing to properties of the context needed for the interpretation (Rouchota

14 Karin Aijmer 1998: 122). Fraser (1996) also refers to the procedural meaning of markers. According to Fraser, there are two types of linguistically encoded meaning depending on whether the meaning is explicitly encoded (representational or propositional meaning) or only signals how the message is to be understood (procedural meaning). Most words (cat, table) encode concepts, while other words (such as pragmatic markers) act as instructions to the hearer or to the reader about how to interpret the speaker’s or writer’s goals (Blakemore 1987). The commentary markers are ‘lexical expressions which have both a representational meaning specifying an entire message [the proposition], and a procedural meaning signalling that this message is to function as a comment on some aspect of the basic message’ (Fraser 1996). In Fraser’s analysis disjuncts are different from conjuncts which would be analyzed as discourse markers. In contrast to commentary markers, discourse markers do not contribute to representational meaning but only to procedural meaning. ‘They provide instructions on how the utterance to which the discourse marker is attached is to be interpreted’ (Fraser 1996: 186). However commentary markers and discourse markers can be difficult to distinguish from each other. While we have a fairly clear idea of what discourse markers mean (and the type of coherence relations between utterances they can express) commentary markers are more complex since the speaker’s and writer’s attitudes and feelings are also aspects of the context and part of their pragmatic interpretation. For example, the markers can also have the function to signal coherence with the preceding discourse in certain positions and they can be used strategically for argumentation with functions going beyond the commentary one. Thompson and Zhou (2000: 131) note for example that disjuncts (commentary markers) can signal a conjunct (discourse marker) relation and ‘that their removal would often result in a sense of incoherence’. One of their examples is: All of these sports and many others are dominated by the human urge to aim at something. Surprisingly, this aspect of sport is often overlooked when underlying motivations are being discussed. (Thompson and Zhou 2000: 131)

Surprisingly relates to expectation and is necessary to show how the sentence it introduces is associated with the preceding discourse. To begin with we need to consider commentary markers as a class. Commentary markers are a large and important group of pragmatic markers which includes markers with many different meanings. They are called assessment (evaluative) markers when they ‘signal the speaker’s

At the interface between grammar and discourse

15

evaluation of the state of the world represented in the proposition’ (Fraser 1996: 180). Following Fraser (1996) we can distinguish several classes of commentary marker. Certainly illustrates an evidential marker: Certainly, Harry will go Certainly signals a degree of confidence held by the speaker with regard to the truth of the basic message. There are also markers commenting on the manner of speaking (frankly, briefly), hearsay (reportedly), consequenteffect (to clarify), emphasis (mark my words) (Fraser 1996). The focus in this study is on evaluative markers. The evaluative markers represent a fairly open class of expressions. Fraser illustrates the class with the following markers: amazingly, amusingly, annoyingly, appropriately, artfully, astonishingly, cleverly, conveniently, cunningly, curiously, delightfully, disappointingly, disturbing, foolishly, hopefully, ideally, importantly, incredibly, inevitably, ironically, (in)correctly, justifiably, rightly, sadly, sensibly, shrewdly, significantly, stupidly, suspiciously. thankfully, tragically, (un)luckily, (un)expectedly, (un)fortunately, (un)happily, (un)reasonably, (un)remarkably, understandably, wisely, wrongly.

Commentary evaluative markers can refer to expectations (fulfilled or thwarted) (Thompson and Zhou 2000; Chafe 1986) as well as to hopes or wishes that something will take place or not take place. In this group belong surprisingly, amazingly, astonishingly but also fortunately, sadly, unfortunately, regrettably to give a few examples. The basic message (what is expressed in the proposition) can be unexpected, unusual or strange: the speaker’s or writer’s expectations can be fulfilled and information can be in conflict with expectations. Swan (1988) suggests the following classification for evaluative adverbs (EAs): a fact or event is regarded as (un)fortunate in the speaker’s opinion (positive or negative affect) (fortunately, unfortunately, luckily). EAs can also ‘express the speaker’s opinion that the fact, event, etc. is understandable or natural on the one hand, or strange or even miraculous on the other’ (Swan 1988: 476). These are referred to as [±strange] adverbs. There is also a class of miscellaneous adverbs which includes many new creations: ‘Innovative writers obviously create new EAs [Evaluative adverbs]; even if they have not been encountered earlier, they are immediately understood and interpreted as sentence-modifiers’ (Swan 1988: 476). In addition, new markers can be formed by modification of already existing markers. Surprisingly can for instance be modified by rather (37 exx), perhaps (perhaps surprisingly 48 exx, perhaps not surpris-

16 Karin Aijmer ingly 34 exx, (not) surprisingly perhaps 19 exx), somewhat (33 exx), more (32 exx).2 They can be pragmatic idioms (Fraser 1996) or routines, e.g. when we have amazingly enough. Markers which seem to be synonymous can be distinguished on the basis of the strength of emotion. Astonishingly and amazingly for example express stronger surprise than surprisingly. However they also have different properties as pragmatic markers. For example, unlike surprisingly, neither astonishingly nor amazingly can be negated and they occur less frequently as pragmatic markers.

4. Material The data for this study come from the British National Corpus and represent both spoken and written English (90 million words of written English and 10 million words of spoken English). The written texts in the corpus come from a variety of sources including ‘imaginative’, ‘arts’, ‘belief and thought’, ‘commerce and finance’, ‘natural and pure science’, etc. The spoken texts include both informal conversation and institutionalized texts. Evaluation markers were generally more frequent in writing than in speech as seen from Table 1 which gives the frequencies for some evaluation markers: Table 1. Frequencies in speech and writing of some common evidential evaluation markers (the markers include adverbial uses as intensifier or manner adverbial). The figure within parentheses is normalized to 1 million words. (The normalized figures are those given in the BNC.)

surprisingly (and ‘not surprisingly’) astonishingly amazingly sadly fortunately unfortunately

2

Statistics from the BNC.

SPEECH 10 million words

WRITING 90 million words

46 (4.44)

2536 (28.26)

8 21 74 107 549

(0.77) (2.03) (7.14) (10.32) (52.96)

203 360 1852 1540 4094

(2.26) (4.01) (20.64) (17.16) (45.62)

At the interface between grammar and discourse

17

Markers were generally infrequent in speech, especially in informal conversation, the only exception being unfortunately. The spoken examples come for example from public institutional talk, business meetings, discussions of the radio. These are situations which are characterised by little involvement in comparison with for example informal conversation. However it is possible that involvement is simply expressed differently in spoken and written language and that markers are used for different purposes depending on the medium. 5. Evaluative markers and performativity Evaluative markers involve the speaker’s/writer’s current evaluation of a state of affairs. Nuyts (1990, 2001) refers to the property that the adverbs express evaluative commitment as performative. However in the following example the adverb is not used performatively. In (1) amazingly is used to describe a mental act at some time in the past which does not involve the speaker/writer’s current commitment (cf. Nuyts 2001: 39): (1)

Some time ago I achieved a longstanding ambition – a lead of Slape Direct on Llanberis’s Clogwyn y Grochan. First climbed in 1954, this classic test-piece had inspired me for decades. Amazingly it was chalk-free. Yet all around, E4’s and E5’s were white-caked: obviously ascended far more often than my lowly objective. (CG2 545 written periodical leisure)

The author has carried out the climbing and describes his feelings on finding it chalk-free (I found to my amazement that it was chalk-free). In (2) amazingly could be interpreted either as including the speaker’s evaluative commitment or descriptively (to our amazement, we arrived unscathed). (Cf. Nuyts 2001: 39 and Thompson and Zhou 2000: 138). (2)

Alas, the streets are full of Czech drivers intent on displaying maximum aggression. The general standard of driving was only slightly above that of the average twelve-year-old at the dodgems. Amazingly, we arrived unscathed in Wenceslas Square and I set off to find the Čedok offices to arrange accommodation. (AE 8742 written book imaginative)

The marker is used to describe an emotion held by the third person subject which is not necessarily shared by the speaker/writer:

18 Karin Aijmer (3)

She leant lazily against the sofa and they talked a little and ate and drank and the evening progressed and Robyn found that, astonishingly, she was enjoying herself, relaxing back against the sofa, stretching out her long legs in front of her, listening to the deep, magnetic timbre of his voice. (HGT 1954 written book imaginative)

Robyn found that ‘to her surprise’ she was enjoying herself. Similarly in (4), the speaker’s comment is not included in astonishingly which describes instead the feelings of ‘Herr Nordern’. (4)

Herr Nordern sank on to the sofa and, astonishingly, found himself holding his wife’s hand, and feeling hers firmly gripping his. (A7A 2647 written book imaginative)

6. Evaluative markers and parenthetical status Evaluative markers are potentially flexible and can occur as a parenthesis in the clause in addition to having initial position. As parentheticals which fulfil a commentary function they can be described as ‘partly anchored’ (Wichmann 2001: 179) implying that there is some natural relation between the marker and the sentence. For example, in order to use amazingly one would expect a state of affairs to have certain surprise-causing properties. Initial position was the most frequent position with the markers I looked at: Table 2. Frequencies of different positions in a random sample of 100 examples of some evaluative markers Initial

Initial medial (between subject and finite verb)

Medial

Final

surprisingly (incl. not surprisingly)

76

9

12

3

astonishingly* amazingly sadly fortunately unfortunately

34 88 89 97 98

2 5 9 2 2

9 7 1 1 –

– 1 1 – –

*astonishingly as a pragmatic marker was only found in 45 examples

At the interface between grammar and discourse

19

In order to represent that the marker is only loosely integrated it is placed in a special slot which is maximally distinct from the rest of the utterance. Auer (1996: 310) refers to this slot as the pre-front field with the function ‘to frame the subsequent utterance, and its function to provide some information which is important for understanding’. It is a position appropriate for discourse-strategic functions such as commenting as well as for textual functions and it is also the ‘preferred locus’ for the grammaticalization process ‘by which adverbials turn into discourse markers’ (Auer 1996: 297). Figure 1 shows the position of surprisingly in front of the rest of the utterance analyzed topologically as the front, middle and end field. pre-front field

front field

amazingly

it is snowing

middle field

end field

Figure 1. The position of commentary markers in the pre-front field

Final position is less important from a textual point of view. In (5) amazingly is inserted as an afterthought and has a weaker link to the preceding context. (5)

He voted Labour even, amazingly; like no one they knew. (AOU 287written book imaginative)

7. A note on grammaticalization When an adverb functions as a pragmatic marker it is no longer an ‘ordinary’ adverb. However in a diachronic perspective the pragmatic marker has developed from an adverb. In (6) ‘she was amazingly beautiful’ amazingly functions as an intensifier before a degree adjective and in example (7) it is a manner adverbial. In (8), amazingly is placed in the pre-front field. The contact is lost between the adverb and the adjective and it is used in a wider function as marker signalling how the utterance should be taken. However amazingly still has some conceptual meaning, i.e. the same meaning it has as an intensifier: (6)

She was amazingly beautiful

20 Karin Aijmer (7)

These things alter amazingly you know, I mean it’s incredible that the, the sort of and yet that they both work. (spoken public institutional)

(8)

Amazingly, she was beautiful

In a grammaticalization perspective we have a development intensifier→ pragmatic marker which is parallel to the change of manner adverbs to pragmatic markers (discourse marker) which has been discussed in the literature (Traugott 1995). The development to a commentary marker with the meaning in (8) may also have been affected by the existence of impersonal speaker comments such as ‘it is amazing that’. Grammaticalization explains for instance that the markers have undergone subjectification and are subjective in the sense of Traugott and Dasher (2002). Subjectification is defined as: the semasiological process whereby SP/Ws [speakers/writers] come over time to develop meanings for Ls [lexemes] that encode or externalize their perspectives and attitudes as constrained by the communicative world of the speech event, rather than by the so-called “real-world” characteristics of the event or situation referred to. (Traugott & Dasher 2002: 30)

Subjectification and grammaticalization are closely associated with the procedural or signalling meaning of the marker. Grammaticalization into the pre-front field may also explain that the evaluative markers can be used for rhetorical or discourse-strategic functions as a discourse marker. These are less conventional meanings of the markers and closely linked to the initial position. 8. Evaluative qualification and the information structure In Fraser’s analysis, commentary markers have variations which are not adverbial, for example it was surprising that, what is surprising, that S was surprising (Fraser 1996: 180). As comments they convey the speaker’s evaluation of a fact or an event. However there are also differences. An important difference is that only the adverb can occur parenthetically. The differences can also be described in terms of the information structure (Nuyts 2001). We showed above that one of the characteristics of disjuncts and conjuncts is that they cannot occur as a cleft sentence (*it is surprisingly that).

At the interface between grammar and discourse

21

On the other hand, the adjective seems to be quite similar to a cleft variant of the adjective. cf. Nuyts 2001: 79, ‘[i]f one compares (10): probably they have run out of fuel and (11): it is probable that they have run out of fuel, the latter appears quite similar to a cleft variant of the former’. In (9) and (10) I compare surprising and surprisingly: (9)

Surprisingly it has started snowing

(10) It is surprising that it has started snowing Both the adjective and the adverb express an evaluation towards a fact or event. However as a kind of cleft the adjective is informationally salient in the sentence. As a result the adjective can be used in negative polarity epistemic expressions (it is not surprising that it has been snowing) and it can be questioned (is it surprising that it has started snowing). It is not surprising can for instance be used when two different qualifications are contrasted (it is not surprising vs. it is surprising) (Nuyts 2001). Surprisingly cannot be negated or questioned which shows that it is not in a salient position. However it is possible to challenge the qualification but not the truth of the assertion: (11) There is actual evidence that he did do it in his childhood, and th so they’re not erm, they’re kind of building everything on a single sentence like Leo does, you know, amazingly enough, Leo ‘s book starts with entry of Who’s Who in a single phrase, where she calls herself daughter of, her father. (HE 2249 spoken lecture) In (11) the hearer could object that this (the entry in Leo’s book) is not amazing on the basis of the facts which are known in the case (cf. Rouchota 1998: 115). Thus examples such as the following are fully acceptable according to Rouchota: (12) A: Unfortunately, the party is over B: That’s not true; you’re not at all sad about it In other examples the adjective and adverb can be shown to be used differently. For example, when the marker has a discourse strategic use to reject or deny expectations it is never similar to the adjective which only has a commentary function. In the following example surprisingly has a function where it is similar to but or however.

22 Karin Aijmer (13) The loco went down the embankment taking some of the tanks with it, and the wreckage together with trees standing in the field were enveloped in flames which took several hours to extinguish. The aftermath is seen, but it was to be several days before the damaged locomotive could be removed. Surprisingly it was given an overhaul and returned to service six months later. (AMR 269 written world affairs). 9. Evaluative markers and evidentiality According to Chafe (1986) evidentiality covers the matching and nonmatching of knowledge with previous expectations. Chafe (1986) has for example analysed markers such as oddly enough as an evidential indicating a conflict while of course suggests that something is in line with expectations. We also need to consider how easily the information or knowledge can be integrated in the speaker’s mind. Nuyts (2001) makes a distinction between two type of evidence. For example if the information is shared or known it can be more easily integrated with existing assumptions and expectations. Nuyts (2001: 66) refers to shared evidence as ‘intersubjective’ – it refers to intersubjectively shared rather than objectively verifiable information and is therefore used in particular discourse contexts for example in texts where it is important to go beyond subjective evaluation in order to strengthen the argumentation. On the other hand, when the speaker/writer presents the evidence as his/her personal responsibility the evidentiality is subjective. Factors which influence how the speaker/writer chooses to present something are the speaker’s/writer’s rhetorical purpose and aspects of the communication situation (e.g. if something can be assumed to be general knowledge). By choosing a particular marker the speaker/writer can furthermore influence the hearer/reader’s opinion of whether something is good or bad, sad or strange. The markers can also be chosen in order to manipulate the hearer’s or the reader’s inferential processes. By not revealing his/her real feelings and attitudes the speaker/writer can for instance ‘seduce the addressee into believing the content of the proposition’ (cf. Corum 1975). The distinction which is drawn between subjectivity and intersubjectivity has a close affinity with the distinction between “prepared mind” and the speaker’s “unprepared mind” i.e. unexpected new information, and concomitant surprise in some languages (Aikhenvald 2004: 195). For example, in Turkish terms such as prepared vs. unprepared mind are used to describe

At the interface between grammar and discourse

23

an evidential category ‘which signals experience which can vs. cannot easily be assimilated to existing assumptions and expectations in the language user’s mind’ (Nuyts 2001: 36 with reference e.g. to work by Aksu-Koç and Slobin 1986). Moreover, in the Albanian verbal system there is a special ‘admirative mood’ expressing surprise somehow referring to ‘the speaker’s present or past nonconfirmation of the truth of the statement’ (Friedman 1986: 180). Another example, referred to by Nuyts, is the ‘habitual experiential’ vs. ‘gnomic’ evidential category in Sherpa (Woodbury 1986): in this case the former indicates that the speaker has direct experiential evidence to be compared with the case where the evidence refers to what is generally known. The observations about these languages are sometimes subsumed under the category mirativity. Mirativity covers speaker’s “unprepared mind”, unexpected new information, and concomitant surprise’ (Aikhenvald 2004: 195; DeLancey 2001). In English the distinction between prepared and unprepared minds is present in different functions of evaluative markers. For example an expression such as surprisingly can involve the speaker’s assumptions about what is shared knowledge or it can mark surprise resulting from the situation that the information is new and unexpected. In English the type of evidence has consequences for whether the marker has primarily a textual or interpersonal function. Example (14) is ambiguous. (14) Astonishingly, the fish is considered a delicacy in Japan where it is known as “fugu” but because of its toxicity it can only be prepared by licensed chefs. (CJ3 1376 written book natural and pure sciences) It can be paraphrased: ‘Although the fish is considered a delicacy it can only be prepared by licensed chefs because of its toxicity’. In this interpretation astonishingly refers to old and shared information. However the information can also be considered to be new (surprising) information (‘the fish is considered to be a delicacy but because…’). It is also possible that markers based on hopes and desires could be regarded as evidential although these meanings never seem to be grammaticalized as obligatory evidentials in languages with evidential systems. For example, we may feel and express regret if something we had wished for does not take place (regrettably, sadly) and happiness if what happens corresponds to our wishes (happily). However it is difficult to compare Nuyts’ category of intersubjectivity with evidential categories in other languages

24 Karin Aijmer since the evidential meanings are described in terms of the speaker’s perception of the evidential source rather than what the speaker thinks the hearer knows.

10. Evaluative markers and countered expectations In an interactional perspective, evaluation as well as modality or evidentiality have to do with taking up a stance to a proposition, assumptions or expectations rather than with the expression of emotions. Compare Sandhöfer-Sixel (1990: 276): Emotionalität im linguistisch relevanten Sinne meint die emotionale Bewertung eines Gegenstandes durch den Sprecher und wird als eine Kategorie der M [Modalität] verstanden. M als Ganzes ist die subjektive Stellungnahme eines Sprechers zu einem in P [Proposition] dargestellten Gegenstand, aber auch zu modalen Stellungnahmen selbst.

Because of their evaluative meaning the commentary markers are different from discourse markers such as after all, moreover, thus, well, now whose only function is to guide the hearer or reader to the relationship of the utterance to the preceding discourse. What makes the evaluative markers difficult to analyze is that they have both a textual and interpersonal function. With evaluative markers the interpretation must take into account both the attitude expressed by the writer/speaker and their textual function. As a result of their interpersonal function the evaluative markers can be used strategically in the discourse. Modality, evaluation and evidentiality are part of the argumentative or rhetorical resources used by the speaker or writer to establish various goals in the discourse. Speakers use them [epistemic expressions] to make different types of strategic ‘adjustments’ to an utterance, in function of their interpersonal relationship with the listener/reader (e.g. for reasons of politeness, in order to boast or appear modest, in order to downplay matters, etc.) and/or in view of the properties of the actual discourse situation. Such uses will be covered here under the term ‘discourse strategy’. (Nuyts 2000: 44)

The strategic development of new meanings goes hand in hand with the negotiation in the dialogue. In this perspective the evaluative markers provide a means for the speaker to adopt a stance towards real or imagined points-of-view in order to agree or disagree with them (White 2003). Thompson and Zhou (2000: 39 ) speak instead of ‘evaluative coherence’

At the interface between grammar and discourse

25

which involves ‘amongst other things the recognition of the conjunctive function of disjunct’. Thompson and Zhou continue: ‘A single word such as plainly or unfortunately may at the same time tell us what the writer thinks and construct a dialogue between writer and reader and move the argument from one stage to another’ (Thompson and Zhou 2000: 122). Certainly is a typical disjunct but can also be used with the meaning concession. In the dialogic perspective the rhetorical function of surprisingly can be described as a kind of rejection (Martin and White 2005: 120). Just like a denial it responds to an expectation or assumption which may be explicit or implicit in order to replace, supplant or counter it (‘But you would not expect this’). (15) A play really demands to be read aloud -it needs the sound of the human voice to bring it alive. It takes imagination and a lot of practice to read a play to yourself in the same time as it would take to see it on the stage. Surprisingly, this is true of both the classical and modern play, though you might expect that the obscurities and difficulties of words and expression would prove more of a stumbling block in, say, Shakespeare or Ben Jonson. (A06 33 book written arts) The expectation which is denied is that it might be easier to read a classical play than to read it aloud because of the many obscurities. The writer could also have written ‘Even though this is true of both the classical and modern play, you might expect that the obscurities and difficulties in a classical play would prove more difficult when read aloud’ (but they aren’t). Surprisingly conveys that something is unexpected (to people with some knowledge of classical and modern plays) and therefore warns the reader to expect something new that he/she may not be prepared for. In a similar vein Thompson and Zhou (2000: 139) argue that ‘coherence in text can only be adequately understood if the concept of coherence is complemented by evaluative coherence, and that, amongst other things, this involves recognition of the conjunctive function of disjuncts’. The disjuncts contribute to the textual organisation because of their frequent placement first in the sentence but they also function as a window on what the speaker (or writer) thinks and feels. In the discussion below I will distinguish between markers disagreeing with expectations and markers agreeing with expectations (Section 11). In (16), astonishingly can be replaced by ‘however’ or ‘but’ and is a commentary marker signalling a conjunctive relationship to the preceding discourse as well as stance (cf. Thompson and Zhou 2000: 131):

26 Karin Aijmer (16) Stonehenge was for sale: The government could have it for £125000; failing that, Sir Edmund was open to offers, from a showman perhaps, or an advertising contractor (“ The Pear’s Soap Stone”?), or an American who would ship it away to the United States. Astonishingly, no law existed to prevent any of these things happening to what many people considered to be the “Frontispiece to British History”. That was not entirely for lack of trying (B7A 410 written periodical applied science) Astonishingly is used to counter a proposition attributed not necessarily to the speaker or hearer but to ‘normative viewpoints characterizable as “People say (think) that x”. Both discourse markers and commentary markers are contextualized, i.e. that is they are indexically linked to features of the discourse context such as the speaker’s or writer’s stance as well as to features of the preceding discourse (what is shared knowledge, what has been said) (Ochs 1996). However it is important to distinguish between the functions of commentary markers and discourse markers. Thompson and Zhou (2000: 130–131) point to ‘one crucial difference between disjuncts and conjuncts, at least in the case of concession [a relationship where the speaker/writer asserts something in order to make a counter-assertion]: whereas conjuncts signal the relation being established, disjuncts provide a potential context without in themselves signalling it’. For example, in (17) the writer is more likely to refer to existing assumptions or expectations than to comment on new information. The speaker takes up a position of disalignment towards a preceding viewpoint (an assumption that many building assumptions would offer a share dealing service): (17) Surprisingly, N & P seems to be the only building society to offer a permanent share dealing service, though others have offered temporary privatisation arrangements. (AHJ 755The Guardian, written) Additional evidence for this interpretation is that surprisingly is followed by a though-clauses and can be replaced by a conjunct (but, however). When the marker is premodified it is more closely related to the commentary marker than to the discourse marker: (18) The Habitat-Mothercare group reduced the risk of offending against local custom when it entered the Japanese market by operating a franchise through an established Japanese retail group called Siebu.

At the interface between grammar and discourse

27

The main problem was that all the furniture had to be scaled down to suit the smaller stature of the Japanese and to fit their smaller homes. Perhaps surprisingly, Conran’s management style and operating methods have changed little in the light of his vast experience. (A6L 470 written commerce and finance) However in (19) the context shows that astonishingly is used with a textual function rather than with a commentary function: (19) MPs yesterday accused NatWest chief Lord Alexander of behaving like a loan shark by charging penal rates. Astonishingly he defended his policy of imposing sky high 36.3 per cent rates, saying it had cut unauthorised loans by almost a third in a year. But the Commons Treasury Select Committee, investigating banks, said it had also closed tens of thousands of businesses. (CBF 5674 written periodical world affairs) 11. Markers agreeing with expectations The pragmatic marker cannot normally be negated since it does not occupy a position in the sentence where it would be salient (cf. Section 7). However, we find not surprisingly (849 matches in the BNC) with the meaning to express that something is self-evident or according to expectation i.e. as a positive evaluation (cf. obviously, naturally, of course). In other words it can be regarded as grammaticalized in this form just like we have rather surprisingly, surprisingly enough as pragmatic markers. Similarly not unnaturally (50 examples in the BNC) and not unexpectedly (25 examples in the BNC) were only commentary markers. Not unnaturally, for example, suggests ‘as might be expected according to standards of normality or naturalness’. Not surprisingly is illustrated in (20): (20) And there with a written in biro by Mrs Harrison was that small amount, 1 pence.And just to add insult to injury, Doreen’s husband, Mike, had died two years ago.Not surprisingly, she’s very annoyed. (KRM 2684 tv news spoken) There is no presupposition that something unexpected is denied in (21).

28 Karin Aijmer (21) Unfortunately, although not surprisingly, it has proved difficult to devise a suitable control discrimination. Some use has been made of a control suggested by Kurtz (1955), in which the subjects received pre-training on a same – different task. (APH 1097 book written applied science) The expectations can be explicitly mentioned (‘it is not surprising since’): (22) Kark Weschke, at the Redfern Gallery, is Expressionist in a more orthodox fashion – not surprisingly, since he was born in Germany in 1925 and did not come to England until 1948. (AIJ 111 The Independent, written) Kark Weschke could be expected to be an orthodox expressionist since he was born in Germany and did not come to England until late. Not surprisingly signals that the evaluation is in line with expectations and has the function to make it easier for the reader to accept an argument. (23) Inland Revenue figures show that housing now accounts for 40% of the total value of inherited estates. It looms large in middle-sized estates; not surprisingly, as the big increase in home ownership and house prices in the past 30 years has mainly benefited the comfortably-off. (ABD 1800 written commerce and finance) Not surprisingly has the function to make the transition to however more smooth. (24) Colleagues, great credibility should go to John Prescott, Shadow Transport Secretary, who has demanded that all foreign registered tankers should provide details of routes when they set off from port so that all operators can be dealt with. Not surprisingly however, John McGregor the Transport Secretary, although he did not oppose John Prescott’s submissions, made it quite clear that this would be extremely expensive. (HDU 257 Trade Union Annual Congress spoken) In the following example not surprisingly comes close to of course because the reference can be understood as ‘what everyone would expect’. The meaning is emphatic after and:

At the interface between grammar and discourse

29

(25) erm until some of the people who are putting a lot of pressures on schools to do different things are prepared to resolve some of these contradictions, I think the teachers feel that they’re being continually criticized by different groups for not doing different things, many of which are mutually contradictory, erm and not surprisingly they feel a little defensive in this situation. (KRH 3678 radio broadcast spoken) Not surprisingly occurs in the pattern not surprisingly therefore (19 examples). It is argued that since human beings have a large number of wants marketing efforts should not concentrate on those rather than people’s needs: (26) The point is that human beings have relatively few needs, but can generate an enormous number of wants. Not surprisingly, therefore, most marketing efforts concentrate predominantly on satisfying people’s wants. In addition to this, some marketing is directed specifically at creating or changing people’s wants. (GUY 1434 book written commerce and finance) Not surprisingly could be omitted but not therefore. However in (27) not surprisingly has the function of therefore: (27) A somewhat different system simulated the effect of momentarily jamming an iron bar across the terminals in the substation. Not surprisingly, it caused lights and TV sets to flicker. A more refined technique called Cyclocontrol replaced this system in the early 1970s. (B7H 1708 written New Science applied science) 12. Combinations We also need to consider combinations of an evaluative marker with a conjunct or some other marker (surprisingly though, but surprisingly, surprisingly however, for surprisingly, indeed surprisingly). Combinations provide evidence that the markers refer to information which has already been integrated with preceding knowledge and therefore lost some of its surprise effect. In the examples below the combinations are used strategically for greater force.

30 Karin Aijmer The textual function of the marker can be seen as a blend or combination of the textual and interpersonal function. The same idea is expressed by the term ‘conjuncts with attitude’ referred to in Thompson and Zhou (2000: 124). Compare also Swan (1988), ‘In a sense all those EA [evaluative adverbs; my commentary markers] may be seen as blends of EA and conjuncts in that they evaluate, but also refer back to previous discourse’ (Swan 1988: 488). In (28), the marker is premodified by more which suggests that something has already been described as surprising in the preceding discourse (cf. Swan 1988: 492 for similar examples): (28) Support for the Alliance was weaker amongst readers of the tabloids than readers of the quality press: all perhaps as expected. Rather more surprisingly perhaps, Sun and Star readers (taken together) were at that time split fairly evenly between Labour and Conservative with the Alliance a fairly close third. (A 627 33 written world affairs) Astonishingly as a pragmatic marker can be premodified by most referring to what is most astonishing in comparison with other astonishing events: (29) Most astonishingly, it has recently been discovered that many trees send out messages of danger when under attack from insects. On receiving these alarm chemicals, the sedentary neighbours increase the amount of tannin in their leaves, making them less palatable. (FEU 1315 written book natural and pure sciences) The examples illustrate that pragmatic markers can combine an attitude with a cohesive function. ‘The importance of this insight is that, since expectations inherently draw on the interaction between writer and reader, it places interpersonal, evaluative judgements on a par with more accepted “external” logical signals as complementary ways of establishing cohesion in text’ (Thompson and Zhou 2000: 132). The writer signals his/her attitudes explicitly as a way of reinforcing an objection (but, yet, however), an elaboration (and) or explanation (for). The combinations are above all a way of strengthening the speaker’s/writer’s involvement. If the marker is left out the reader/hearer is given no guidance whether the message should be taken as expected or unexpected, good or bad, important or unimportant, serious or non-serious, obvious to everyone or whether it agrees or does not agree with what the speaker/writer believes or expects.

At the interface between grammar and discourse

31

The meaning explanation +attitude is found after for: (30) This stipulation, supported by an Act of Parliament, is still in force, and, as far as treatment is concerned, puts the sexually transmitted diseases in a unique position. For, surprisingly, it is not illegal for other conditions to be treated by medically unqualified people as long as such people do not claim to be doctors. (ARH 173 written book natural and pure sciences) Sexually transmitted diseases have against all expectation a unique position since it is illegal for unqualified people to treat them. The evaluative marker cannot be omitted because of its ‘extra’ attitudinal information. In (31) amazingly adds the information that something is amazing (unexpected). Something is unexpected against the background that Kylie is a well-known writer. Both but and amazingly signal unexpectedness and reinforce each other: (31) Ashley and Mushroom Records wanted Kylie’s new song released under license to a major British record company. But amazingly every company approached rejected Kylie. (ADR 422 written book leisure) In (32) yet amazingly similarly has the meaning counter-expectation. The function of the marker is to emphasise that the writer is involved while yet shows that a state of affairs is against expectations: (32) A Chinese master takes the full force of a spear against the throat. To the left of the picture, two students are pushing the spear: yet amazingly, the sharp point does not penetrate his throat, because his chi gung training will not allow it do so. (GVF 182 written book leisure) In (33) the writer’s presence adds emphasis and involvement: (33) Yet the large majority of our teenagers said they had not been taught the practical realities of avoiding AIDS until they were at least 15. They wanted to know much more, much earlier. And surprisingly only one in five teenagers questioned had been taught about homosexuality and lesbianism in school. (CH 16974 written the Daily Mirror)

32 Karin Aijmer In (34) amazingly is used emphatically after and. The author refers explicitly to a particular group of people as the evidential source: (34) From Tom Lofthouse I had heard he was not popular in the Doctors’ House, yet, conversely and amazingly to anyone who had been his junior, our Dr Jones rated as a favourite pin-up in the Staff Nurse’s Home. (CKO 2745 written book imaginative) In its combinations (though, however, but), sadly, signals a relation to not fulfilled expectations as well as regret: (35) Sadly, though, the gaols do not stand half-empty. Convicted prisoners are outnumbered by the euphemistically named “undertrials” – remanded prisoners for most of whom trial is months (if not years) away and many of whom, detained pending police investigations, have not yet been charged with an offence (Diaz 1977). (CRT 419 written book social science) (36) It had undergone detailed and lengthy examination by a Select Committee of this House. Sadly, however, a few Opposition Members sought to block the Bill as it neared the end of its passage through Parliament. (HHW 1387 written miscellaneous published) (37) Maybe we’ll see you in the summer, but, sadly enough, the firm cancelled the trip to Moscow that Elaine and I were going to take in June, to deliver papers on ELT dictionaries. (written leisure miscellaneous unpublished) (GXL 211 written miscellaneous unpublished leisure) In all these examples the writer’s attitudes are involved and add some information to the conjunct (discourse marker). In another group of examples a new stage in the narrative or argumentation can be additionally signalled by an evaluative marker (then astonishingly, now amazingly): (38) He handed her the file, watching her flick through it quickly, again as if she were looking for something specific. Then, astonishingly, she looked up at him, a great beam of a smile on her face. (GUG 1890 written book imaginative)

At the interface between grammar and discourse

33

(39) Now, amazingly, the government prosecution team in Miami admits it has “no hard evidence linking Noriega to drugs”. (ARW 1326 written periodical social science) (40) Then, sadly, subsequent drummer Eric Carr succumbed to cancer last year. (C9K 1072 written periodical arts) 13. Conclusion The evaluative markers have certain characteristic functional and distributional properties. For example, they have a special position which has been described in topological terms as the pre-front field because it is separated from the rest of the utterance. This position is suitable for discoursestrategic (e.g. rhetorical) purposes as well as for discourse-organizing or framing functions. It is also a grammaticalization (or pragmaticalization) position providing an end-point for the development of intensifying adverbs into pragmatic markers. As a result of their placement in the pre-front field they can develop further discourse-organizational and strategic properties. The evaluative markers have both interpersonal and textual functions and can be used strategically in the communication. In their discourse strategic uses they can be used both to reject expectations and to agree with the hearer/reader. The evaluative markers can also combine with discourse markers for greater rhetorical effect. In order to describe evaluative markers. we need to take into account a number of different factors such as performativity or lack of salience. We get evidence for these factors ‘intralinguistically’ for example by comparing evaluative adjectives and adverbs but also from comparisons with other languages. Surprise is for example one of the most basic emotions and evaluative markers expressing surprise can be related to mirativity and the distinction between ‘prepared’ and ‘unprepared’ minds in the verbal systems of some Non-European languages. Evaluative markers do not only refer to subjective evidence but they can express intersubjectivity or shared evidence. Intersubjective evidence is hearer-oriented rather than based only on what the speaker knows or is responsible for. When evaluative markers are intersubjective they refer to expectations or assumptions which have already been assimilated as common ground by the discourse partners.

34 Karin Aijmer Evaluative markers are used more in writing than in speech suggesting that emotional involvement is realised differently especially in informal conversation. They often combine with discourse markers for more emphasis but they can also be sneaked in as a comment to show how the speaker or writer wants to the text to be interpreted. References Aikhenvald, Alexandra Y. 2004 Evidentiality. Oxford: Oxford University Press. Akzu-Koç, Ayhan and Dan Slobin 1986 A psychological account of the development and use of evidentials in Turkish. In Evidentiality: The linguistic coding of epistemology, Wallace Chafe and Johanna Nichols (eds.), 159–167. Norwood: Ablex. Auer, Peter 1996 The pre-front field in spoken German and its relevance as a grammaticalization position. Pragmatics 6 (3): 295–322. Bellert, Irene 1977 On semantic and distributional properties of sentential adverbs. Linguistic Inquiry 8: 337–351. Blakemore, Diane 1987 Semantic constraints on relevance. Amsterdam/Philadelphia: John Benjamins. Chafe, Wallace 1986 Evidentiality in English conversation and academic writing. In Evidentiality: The linguistic coding of epistemology, Wallace Chafe and Johanna Nichols (eds.), 261–272. Norwood: Ablex. 1994 Discourse, consciousness, and time: the flow and displacement of conscious experience in speaking and writing. Chicago: University of Chicago Press. Corum, Claudia 1975 A pragmatic analysis of parenthetic adjuncts. In Papers from the 11th Regional Meeting of the Chicago Linguistic Society, R. E. Grossman et al. (eds.), 131–141. Chicago: Department of Linguistics, University of Chicago. DeLancey, Scott 2001 The mirative and evidentiality. In Evidentiality, Patrick Dendale and Liliane Tasmowski (eds.). Special issue of Journal of Pragmatics 33 (1): 369–382. Fischer, Kerstin 2006 Towards an understanding of the spectrum of approaches to discourse particles: introduction to the volume. In Approaches to dis-

At the interface between grammar and discourse

35

course particles, Kerstin Fischer (ed.), 1–20. Oxford /Amsterdam: Elsevier. Fraser, Bruce 1996 Pragmatic markers. Pragmatics 6 (2): 167–190. Friedman, Victor A. 1986 Evidentiality in the Balkans: Bulgarian, Macedonian, and Albanian. In Evidentiality: The linguistic coding of epistemology, Wallace Chafe and Johanna Nichols (eds.), 168–187. Norwood: Ablex. Greenbaum, Sidney 1969 Studies in English adverbial usage. London: Longman. Martin, James R. and Peter R. White 2005 The language of evaluation: Appraisal in English. Basingstoke & New York: Palgrave Macmillan. Nuyts, Jan 1990 Emotions and the functionality of language. Grazer Linguistische Studien 33/34: 227–240. 2001 Epistemic modality, language, and conceptualization. A cognitivepragmatic perspective. Amsterdam / Philadelphia: John Benjamins. Ochs, Elinor 1996 Linguistic resources for socializing humanity. In Rethinking linguistic relativity. John J. Gumperz and Stephen C. Levinson (eds.), 407– 437. Cambridge: Cambridge University Press. Rouchota, Willy 1984 Procedural meaning and parenthetical discourse markers. In Discourse markers: Descriptions and theory, Andreas H. Jucker and Yael Ziv (eds.), 97–126. Amsterdam /Philadelphia: John Benjamins. Sandhöfer-Sixel, Judith 1990 Emotionale Bewertung als modale Kategorie. Grazer Linguistische Studien 33/34: 267–278. Sinclair, John McH. and Anna Mauranen 2006 Linear Unit Grammar: Integrating speech and writing. Amsterdam/ Philadelphia: John Benjamins. Swan, Toril 1988 Sentence adverbials in English: a synchronic and diachronic investigation. Oslo: Novus. Thompson, Geoff and Jianglin Zhou 2000 Evaluation and organization in text: the structuring role of evaluation. In Evaluation in text: Authorial stance and the construction of discourse, Susan Hunston and Geoff Thompson (eds), 121–141. Oxford: Oxford University Press. Traugott, Elizabeth Closs 1995 The role of the development of discourse markers in a theory of grammaticalization. Paper presented at the 12th International Conference on

36 Karin Aijmer Historical Linguistics, Manchester, August 1995. (www.stanford.edu/ ~traugott/papers/discourse.pdf). Traugott, Elizabeth and Richard B. Dasher 2002 Regularity in semantic change. Cambridge: Cambridge University Press. White, Peter 2003 Beyond modality and hedging: a dialogic view of the language of intersubjective stance. Text 23 (2): 259–284. Wichmann, Anne 2001 Spoken parentheticals. In A wealth of English. Studies in honour of Göran Kjellmer, Gothenburg Studies in English 81, Karin Aijmer (ed.), 177–193. Göteborg, Sweden. Woodbury, Anthony C. 1986 Interactions of tense and evidentiality: A study of Sherpa and English. In Evidentiality: The linguistic coding of epistemology, Wallace Chafe and Johanna Nichols (eds.), 188–202. Norwood: Ablex.

The subjectivity of basically in British English – a corpus-based study Christopher S. Butler

1. Introduction The adverbs basically, essentially and fundamentally tend to be treated as (near) synonyms in dictionaries of English, so much so that one is often defined in terms of the other(s), as illustrated by the following entry from the Compact Oxford English Dictionary 1: basically • adverb 1 fundamentally. 2 in fact; essentially: I basically did the same thing every day.

In earlier work using the British National Corpus (Butler in press) I showed that although the three adverbs do share a number of properties, they are clearly differentiable in terms of their preferred positions in the clause and their collocational profiles. The aim of the present paper is to obtain evidence for the claim that basically often projects, through its core semantics, further meanings which are strongly subjective or, putting a slightly different slant on things, pragmaticalised, in the sense that its meaning is concerned not just (or even mainly) with commenting on the propositional content to which it is attached, but also with the speaker or writer’s own attitudes and aims in relation to the interaction at that point2. I will also show that there is evidence that fundamentally does not pattern in this way, and that essentially is somewhere between the two, though closer to fundamentally than to basically.

1 2

Online at http://www.askoxford.com, accessed 16.6.2006. I am indebted to Anne-Marie Simon-Vandenbergen for sowing the seeds of the idea on which this work is based, as well as for comments on an earlier draft. I also wish to thank Lachlan Mackenzie and Francisco Gonzálvez García for their comments and suggestions.

38 Christopher S. Butler 2. Different interpretations of subjectivity In an early discussion of subjectivity in language, Benveniste (1971: 225) observes: “Language is marked so deeply by the expression of subjectivity that one might ask if it could still function and be called language if it were constructed otherwise”. Here, Benveniste is concerned with the ability of speakers to regard themselves as subjects, particularly with regard to person deixis. Lyons (1982: 102), also writing about subjectivity in relation to deixis, describes subjectivity as “the way in which natural languages, in their structure and their normal manner of operation, provide for the locutionary agent’s expression of himself and of his attitudes and beliefs”. This theme is also taken up in Traugott’s work on the diachronic dimension of language: Traugott (1989: 35) defines subjectification as a process whereby “meanings become increasingly based in the speaker’s subjective belief state/attitude toward the proposition”, and studies this process in relation to grammaticalisation (see Traugott 1995, 2003), while pointing out that it is also relevant to matters of lexical change (Traugott 2003: 634). Subjectification thus involves a process of ‘pragmatic strengthening’, by which original propositional meanings come to have an expressive function (Traugott 1982: 256, 2003: 633), and indeed this process is sometimes referred to as pragmaticalisation (see e.g. Erman and Kotsinas 1993; Aijmer 1997: 2). While Traugott’s approach is concerned primarily with changes through time, and is linked to the concept of grammaticalisation, it is also possible to view subjectivity synchronically, in terms of different degrees of subjectivity displayed by alternative linguistic construals of a situation, or different uses of particular expressions. This is the approach taken by Langacker (1985, 1989, 1990). Working within the framework of Cognitive Grammar, Langacker analyses subjectivity in terms of viewing relationships between a perceiver and what is perceived. In the ‘optimum viewing arrangement’, the inherent asymmetry of the perceptual relationship is maximised, in that perceiver and object of perception are completely distinct, and the viewer’s attention is focused solely on what s/he is perceiving. In this situation, the viewer is seen as maximally subjective, the object of perception as maximally objective. This arrangement is contrasted with the ‘egocentric viewing arrangement’, in which the locus of viewing attention is expanded to include the viewer. Langacker then generalises these situations to the non-perceptual case of the linguistic utterance: perception has its analogue in the construal relationship between a conceptualiser and what is conceptualised, the conceptualiser being equated with the participants in the speech act, and the conceptualisation with the meaning of the expression (i.e. the predication).

The subjectivity of basically in British English

39

What we are contrasting here, then, is the situation in which the speaker and hearer remain totally implicit, or act only as a point of reference, and that in which a speech act participant is made explicit, as a salient, ‘onstage’ entity. Compare the following sentences 3: (1)

(= Langacker’s 6a, 1990: 326) Vanessa jumped across the table.

(2)

(= Langacker’s 6b, 1990: 326) Vanessa is sitting across the table from Veronica.

(3)

(= Langacker’s 9a, 1990: 328) Vanessa is sitting across the table from me.

(4)

(= Langacker’s 9b, 1990: 328) Vanessa is sitting across the table.

In 1, the movement across the table is objective, though the role of speech act participants is maximally subjective, since it is confined to that of conceptualiser in the construal of a situation in which there is physical movement across the table. In (2), there is no objective movement, and the conceptualiser must mentally trace a path across the table, this movement being purely subjective. This, for Langacker, represents one kind of (synchronic) subjectification. (3) and (4) represent a second type of subjectification. (3) is analogous to (2), but with the difference that the speaker rather than some third party is the reference point for the mental trajectory. The speaker is here regarding him/herself as participating in the event in the same way as Veronica in (2), and so is taking a detached view of him/herself. In (4), however, the role of the speaker is implicit, and purely subjective. The two types of subjectification thus combine in the interpretation of path prepositions in English. The interpretations of Traugott and Langacker are clearly linked through the concept of the degree to which the speaker is involved in the meaning of what is said: indeed, Langacker (1990: 325) concedes that Traugott’s claim for a shift, during grammaticalisation, from ‘propositional’ to ‘textual’ and ‘expressive’ functions “amounts to subjectification, under a broad interpretation of that term”. Some scholars have found it fruitful to combine the broader interpretation espoused by Traugott with the narrower one 3

Page references to Langacker (1990) are to the version of the paper as given in Concept, Image, and Symbol: see reference list.

40 Christopher S. Butler of Langacker: see e.g. Kemmer (1995) on -self constructions in English, Carey (1995) on the development of the English perfect, Taverniers (2005) on a semiotic-functional interpretation of subjectivity inspired by Halliday’s systemic functional linguistics. In the current paper, subjectivity will be interpreted in the wider Traugottian sense. I shall, however, introduce a distinction between a weaker type of subjectivity, covering the use of linguistic expressions to convey the speaker or writer’s ‘take’ on the propositional material expressed in the same utterance, and a stronger type, concerned with indicating the aims, attitudes and points of view, with respect to the discourse at that point, which the speaker or writer wishes to convey to the addressee. These two kinds of subjectivity will be clarified and exemplified in §3. It is now abundantly clear that subjectivity in the wide Traugottian sense is crucial to language and its use. Two studies stand out, apart from those already mentioned, both focusing on the synchronic dimension. The work of Scheibman strongly supports the hypothesis that “linguistic items (constructions of all sizes) that most frequently appear in conversation are those that participate in subjective expression” (Scheibman 2002: 2), and convincingly demonstrates that “the varied relations between linguistic structure and speaker point of view are neither exceptional nor unique, but rather the norm in conversational English” (Scheibman 2002: 7). Verhagen (1995, 2005), starting from Langacker’s conception of the construal relationship, shows that in the prior treatment of a number of linguistic phenomena, too much has been located at the level of the object of conceptualisation rather than at that of the conceptualiser. His work is based on a slightly different concept of subjectivity from those summarised above, namely the recognition by human communicators that they may have “thoughts and beliefs that may differ from those of other people” (Tomasello 1999: 14–15). This gives rise to the concept of ‘intersubjectivity’, i.e. “the mutual coordination of cognitive systems” (Verhagen 2005: 28), in terms of which Verhagen analyses in detail three areas of the grammar: negation, finite complementation and discourse connectivity4. Verhagen’s main thesis is that in such constructions the speaker presents content to the hearer in such a way that the latter is led to particular interpretations. 4

Note that Verhagen’s concept of intersubjectivity differs from that of Nuyts (2001), who discusses subjectivity in relation to types of evidentiality. In Nuyts’ account, the term ‘intersubjective’ is used for evidential markers which show that the appropriate evidence is shared by a group, as opposed to those indicating that the evidence is available only to the speaker, which are termed ‘subjective’.

The subjectivity of basically in British English

3.

41

The meanings of basically

3.1. The core semantics: ‘weak’ subjectivity We must first differentiate between two types of function which adverbs can have (see Quirk et al 1985: 439–441; Biber et al 1999: 538). Firstly, they can modify adjectives or other adverbials, as in (5): (5)

Our European colleagues’ vision of Europe is fundamentally different from ours and a shotgun marriage would be in nobody’s interests… (BNC HHV 07076)5

In this function, they are elements of a phrase. Secondly, they can be elements of the clause, as in (6)–(8): (6)

Thereafter the situation changed fundamentally, and, despite the interference of the Corn Laws, grain came to figure high on the list of imports in years of bad harvest. (BNC HR0 1158)

(7)

More fundamentally, no one knows why sunspots should occur at all… (BNC CET 1780)

(8)

Consequently, they’re always coming in a bit behind the boys because the boys have been playing for a bit longer, … (BNC FL5 412)

According to the classification presented in Quirk et al (1985: 440–441, 501–647), fundamentally in (6) is an adjunct, closely integrated into the structure of the clause and able to be focused in clefting, negation and alternative interrogation. On the other hand, the adverbials in (7) and (8) are less integrated with the clause, and cannot be focused: more fundamentally in (7) is a content disjunct, while consequently in (8) is a conjunct. Quirk et al (1985: 621) classify basically, essentially and fundamentally within the group of content disjuncts which “state the sense in which the speaker judges what he says to be true or false”, and more narrowly, that they “claim that what is being said is true in principle”. These characterisations are very similar to that of Greenbaum in earlier work, where he writes that basically, essentially and fundamentally “assert that what is being said 5

All examples are taken from the original version of the British National Corpus (BNC).

42 Christopher S. Butler is true in principle, despite minor qualifications that might be made” (Greenbaum 1969: 206). It is clear, then, that not only basically, but also essentially and fundamentally, contain, even in their straightforward meanings, a component which involves the speaker or writer, in that the propositional content is presented in the light of his or her assessment that although it is true in principle, there may be aspects of it which would not be true under certain quite minor circumstances which are irrelevant to what is being claimed. Thus even in its most straightforward uses, basically, like the other two adverbs, has an element of subjectivity about it. When acting as a clause adverbial, it is what is commonly referred to in the literature as a ‘stance adverbial’, defined by Biber et al (1999: 853) as having “the primary function of commenting on the content or style of a clause or a particular part of a clause”. Curiously, Biber et al do not mention basically, essentially and fundamentally, and Powell’s (1992) work on stance adverbs also fails to deal with them. Although the discussion in Quirk et al is focused on the clause adverbial use of the adverbs, a similar semantic content can be proposed for them in their modifying function. Consider example (9): (9)

I thought it was a basically good shape …(BNC HSK 010)

Here, the writer is saying that in his/her opinion the proposition ‘it was a good shape’ was true in all the respects which really mattered with respect to the purposes under consideration (functional, aesthetic, or whatever). Such examples represent the smallest scope of the adverb, limited to just the following adjective. A further example of this kind is shown in (10): (10) … human beings are neither basically bad, evil, and anti-social, no more are they basically good, altruistic and cooperative. (BNC HUN 243) Here, the speaker is claiming that human beings are neither bad nor good as a central and essential property.

3.2. Pragmatic extensions: ‘strong’ subjectivity As we move from modifying to clause adverbial function, we increase the scope of the adverb (or, more properly, adverbial phrase, since the adverb can be premodified, as in (7)) to include the whole clause or even the whole

The subjectivity of basically in British English

43

sentence. As we have seen, the adverb becomes more loosely attached to the propositional content, and this offers possibilities for reorientation of the meaning6. In many such instances, the straightforward core meaning of basically is still the predominant one. Consider example (11): (11) Since, as we have seen, Adorno overestimates the homogeneity of culture under advanced capitalism, he is led to a similar interpretation of popular musical form. His key concepts here are standardization and – in order to explain the apparent elements of variety – pseudoindividualization. Basically his argument is that all aspects of musical form – Adorno instances overall structure (the thirty-two-bar chorus), melodic range, song-types and harmonic progressions – depend on pre-existing formulae and norms, which have the status virtually of rules, are familiar to listeners and hence are entirely predictable. (BNC FB3 0669-0670) Here, although the scope of basically is the whole of the complex sentence which follows, the writer is indeed using the adverb with its straightforward meaning, since s/he is attempting to summarise the most important aspects of Adorno’s complex argument. Similarly, in (12), the function of basically is to indicate that the main job of the female person being referred to is to answer the telephone, though she may on occasion perform other (subsidiary) tasks too. (12) She sits there and answers the telephone basically. (BNC KD5 6416) But now consider the following, which forms part of a discussion on changes in Europe: (13) A. So basically there was no no point in it what so ever John B. Well I’m I’m not some I wouldn’t I wouldn’t say that I mean I’m saying that erm 6

Although we are dealing here with relationships between synchronically coexisting meanings of an expression, it is of interest to note that Traugott (2003) makes a similar point with respect to subjectification as a diachronic process, pointing out that changes from, for example, nominal complexes to clause connectives (e.g. instead of) or discourse markers (e.g. indeed) are accompanied by an increase in structural scope. This is, of course, just one example of the intimate relationship between the synchronic and diachronic dimensions of language.

44 Christopher S. Butler A. Well you shouldn’t say that John, you’d be out of a a job, wouldn’t you? B. Well I I don’t have any er mandates, I mean it’s not my er A. Mm B. I I’m basically just erm giving information about it I don’t have any mandates (BNC HUV 1012–1017) Here, A is engaging in face-threatening behaviour by pointing out that ‘there was no point in it’ and claiming that B ‘shouldn’t say that’. A’s use of basically can be seen as equivalent to ‘when it comes down to the most central points of the argument’, but B’s is part of a defensive strategy, distancing himself from A’s claims by saying that he has no mandates but is simply providing information. We can still relate basically (and also just: see further the discussion in §5) to the proposition ‘I’m giving information’ in the sense ‘All I’m really trying to do is …’, but the overriding discourse function is concerned with the speaker’s attitude towards the interaction between himself and his interlocutor at this point. A similar function can be seen in the following example: (14) I’m sending a tape recorder to my dad. I I I was delayed a little in that. Basically I I’ve got a l a l a whole list of things that I didn’t get done this week either. (BNC F8U 0221–0223) Again, we can relate basically to its core meaning through a paraphrase such as ‘The main thing is that there were lots of other things I didn’t get done this week either’, but in order to understand why the speaker would say this, we have to assume that s/he was presenting an excuse for the delay in sending the tape recorder: the speaker presumably was so busy that all manner of jobs got delayed, rather than just the sending of the machine. Again, then, basically is being used within a context where the dominant discourse function is strongly related to the speaker’s interactional aims at this point, rather than just to the content of the proposition in which basically appears. In (15), taken from a transcript of a committee meeting, the function of the utterance is not one of defensive excuse as in (14), but rather one of persuasive recommendation (note the occurrence of surely 7 and should), in which the use of basically serves primarily to tone down the recommenda7

For an analysis of surely as a marker of evidentiality and stance, see Downing (2001, 2006).

The subjectivity of basically in British English

45

tion, though still via the implication that there may be limits to ‘changing it to whatever they want’: (15) This only applies to teaching so surely we should change it to whatever they want basically. (BNC KM4 1247) Example (16) shows yet another type of context in which basically has a primary function concerned with the speaker’s attitude rather than simply his or her assessment of the propositional content: (16) Most of the, what has happened I think, over the last two and a half years, that I’m aware of, erm, is basically happening a long time before that, I, I can’t say yes or no. (BNC JSN 330) The speaker is talking about what has happened in his/her company during the two and a half years since s/he joined it. Note the hedges I think, that I’m aware of and I can’t say yes or no here, which show that the speaker is distancing him/herself from the claim that what has happened in the firm recently was happening long before that. Basically adds to this pattern of distancing by implying that what is being claimed is true in principle if not in every detail. By an extension of this idea of distancing, we can perhaps explain why some speakers appear to sprinkle their language so liberally with basically. Consider (17), taken from a student seminar: (17) he, he, he lists er a wide range of things that basically you know should be banned or, or basically erm sh should be sort of cut down upon. Erm I, I won’t go into them because you know basically erm I think the three main ones reflect the fact that he’s saying the peasants are rising up and they’re not only changing their political views erm but are obviously changing their cultural values as well according to, to, according to what he suggests. (BNC KGN 0899) Again, the hedges I think and sort of indicate that the speaker wishes to distance him/herself from full commitment to what s/he is saying, and the frequent hesitations and repetitions suggest some degree of insecurity. Indeed, it is likely that over time, repeated usage of basically by some speakers can become a habit, a kind of linguistic tic in which the originally strong intended subjective meaning has become revalued into an involuntary indicator of insecurity, an expressive rather than an intentionally communicative element, so leading to the interpretation, by some commentators, that the word

46 Christopher S. Butler is meaningless 8. This ‘semantic bleaching’ effect is an example of what Haiman (1994: 7) has described as “decline in the tendency to respond to stimuli that have become familiar due to repeated or persistent exposure”. To summarise the main point of this section, basically can be used not only to express the language user’s assessment of the propositional content in terms of its truth value, but also, through this central meaning, to project even more strongly subjective meanings concerned with, for example, the rhetorical function of an utterance or group of utterances (e.g. excusing oneself), the toning down of a face-threatening act, or the lessening of commitment to what is being said. It is, however, possible that habitual use may result in the degradation of the originally strong subjective force of basically into an involuntary indication of insecurity. In the remainder of this paper, I shall present further evidence of a more indirect nature, using the whole BNC corpus rather than just discussing selected examples. 4.

Evidence from differences between spoken and written English

4.1. Frequency Table 1, derived from data presented in Butler (in press), shows the frequencies of basically, essentially and fundamentally in the whole BNC and in the spoken and written components taken separately. From the normalised frequencies, it can be seen that basically is over 7 times more frequent in the spoken part of the corpus (138.9 per million words) than in the written part (19.2 per million words). On the other hand, fundamentally is 2.8 times more frequent in the written component (9.3 per million words) than in the spoken (3.3 per million), whereas essentially lies between the two, being just 1.3 times more frequent in the written part (37.2 per million) than in the spoken (29.2 per million)9. 8

9

Wallace Chafe (personal communication) has suggested to me that the frequency of basically may differ considerably from one speaker to another, and observes that in 60 excerpts from the Santa Barbara Corpus of Spoken American English there are 42 occurrences of basically, 22 of which come from just 3 speakers, while in 15 of the excerpts there was someone who said it only once. Chafe (1985) shows that a group of adverb(ial)s,categorised as indicating ‘concern for statistical reliability’, and consisting of basically, essentially, generally, in some sense, invariably, normally, primarily, and virtually, were more common in a written sample of American English consisting of personal letters and

The subjectivity of basically in British English

47

The great discrepancy between the frequency of basically in the two modes in itself suggests that there might be good reason to investigate whether this adverb might have certain functions in spoken English which are not present, or present only with low frequency, in the written language. (Scheibman 2002: 171) observes that “[i]n interactive discourse, Englishspeaking participants are more likely to explicitly mark their contributions as based in their own viewpoints than in more formal genres”, and Biber et al (1999: 859) describe the general communicative features of conversation in terms of “the focus on interpersonal interactions and the conveying of subjective information”. These characteristics of spoken language give us some (fairly weak) grounds for the hypothesis that at least part of the elevated frequency of basically in spoken English may be related to subjectivity. It is no coincidence that examples (13)–(17), showing clear evidence of strong subjectivity, are all taken from the spoken subcomponent of the corpus 10. Table 1. Frequencies of basically, essentially and fundamentally in the spoken and written components of the BNC

basically essentially fundamentally

Spoken component

Written component

Total

1389 [138.9] 292 [29.2] 33 [3.3]

1727 [19.2] 3347 [37.2] 835 [9.3]

3116 [31.2] 3639 [36.4] 868 [8.7]

Figures in brackets are frequencies per 1 million words of text.

10

academic writing than they were in a spoken sample taken from dinner table conversation and lectures. However, given that the frequency of basically was lumped together with that of the other adverbs, also in view of the differences in the composition and geographical origin of the samples as compared with the BNC, Chafe’s data are not comparable with those reported here. Chafe (1985) argues that speakers, under pressure from the constraints of online planning, generally make categorial distinctions about the truth or falsity of their propositions, while writers have time to make more nuanced judgments about relative degrees of truth and falsity. It may perhaps still be possible to reconcile this claim to some extent with my frequency data, if basically in spoken language often has functions which are not primarily concerned with truth or falsity as such, but with the expression of strong subjectivity.

48 Christopher S. Butler 4.2. Position We have seen that we may predict a relationship between the degree of subjectivity of basically and its two possible syntactic functions, as modifier, where it occupies a fixed position relative to an adjective or other adverb, and as a clause adverbial, which shows considerable mobility, being able to occur in initial, medial and final positions in the clause (see Quirk et al 1985: 490–501). We argued that the looser relationship between clause adverbial and propositional content can be exploited to allow pragmatic extensions of meaning in the discourse. Since subjectivity is known to be evidenced most strongly in spoken rather than written language (see the quotation from Scheibman 2002: 171 above), we may predict that those positions which accommodate clause adverbials will be more common in spoken language than in written, whereas those in which the adverb acts as a modifier will be more frequent in written language. In Butler (in press), the frequency of basically, essentially and fundamentally was studied in relation to the various positions which these adverbs could take in the clause. However, the whole BNC was sampled, no attention being given to differences between spoken and written components. In the present study, samples of basically from the two components were examined separately. The methodology employed was as follows. A concordance 11 of all 1389 occurrences of basically in the spoken part of the BNC was sorted, first on the word 5 positions to the left of the headword, then on the word 5 positions to the right. Since it has been shown (see Sinclair, Jones and Daley 1969) that statistically significant collocation tends not to occur outside these limits, this procedure is tantamount to randomising the entries. Subsequently, the first and then every sixth entry were taken for analysis, giving 232 occurrences in all. A similar procedure was adopted with the 1727 occurrences of basically in the written component: the entries were resorted and then the first and every eighth subsequent entry taken for analysis, giving 216 in all. It was found that 21 categories were required for a full positional analysis of the samples: of these, 16 were based on those required for the analysis of the original mixed sample in Butler (in press), the others being required for 11

Concordances were prepared using WordSmith Tools, written by Mike Scott and marketed by Oxford University Press. For details of this program see http://www.lexically.net/wordsmith/.

The subjectivity of basically in British English

49

rare positional types in the new spoken sample. The 21 categories are listed below and exemplified from the samples used in the study. A.

Initial

(18) Basically they go to bed about half seven … (KBW 02097) The adverb is put in this category even if it is preceded by a coordinating or subordinating conjunction or by a discourse item such as you know, I mean, etc.: (19) … but yeah, basically nobody’s punished at all. (BNC KPV 7657) The category also includes cases where the first clause in a sentence is subordinate, since here the adverb can be seen as initial to the whole sentence: (20) … in other words basically, as my parents spend a lot of time up in their top garden in the summer time, in the sunshine and all, you get the continual hum from that bloody … thing (KC1 1589) B.

Between left-dislocated element and clause proper

(21) Right this basically it’s too simplistic, … (BNC J95 026) C.

Between fronted object and subject

(22) But anything that we covered basically I’ve got in them notes there. (BNC KLG 0011) D.

Between fronted adjunct and subject

(23) … and in our response to that, basically, we have said the nineteen seventy-four arrangements for the staff commission seemed to have worked very well … (BNC J9D 619) E.

Between subject and first verbal element

(24) We basically are not special people, … (BNC KRL 1925) F.

Between auxiliary and main verb, or between two auxiliaries

This includes the case where an adverbial intervenes between basically and the main verb. (25) … in Texas the audience have basically come to see a gig. (BNC CD6 0619)

50 Christopher S. Butler (26) The ship owner will basically only be denied limitation if the casualty was due to “An act or omission done with intent to cause loss or reckless to such loss occurring” – Merchant Shipping Act 1979. (BNC HB4 0394) G.

After a relational verb and before AdjP or NP

By relational verb is meant some form of BE, SEEM, APPEAR or similar verbs, which take a subjective complement. Categories G, H and I include the case where an adverbial, or occasionally another class of word, separates the relational verb from basically. Examples in this category are ambiguous as between a reading where basically modifies the AdjP or NP, and one in which they are sentence adverbs (see later for discussion). (27) Many French dishes are basically simple but may take a long time to prepare in the traditional way. (BNC ABB 1636) (28) The RoboCop is basically a “man in a suit”; … (BNC FB8 696) (29) But the objects that physicists study are still basically simple objects … (BNC J52 0037) H.

After a relational verb and before a clause

(30) … but I think what it is, is basically they’re cramming them now. (BNC GYX 1560) I.

After a relational verb and before a PrepP or adverbial phrase

(31) I mean they’re all basically for different schools anyway. (BNC JNW 093) J.

After a non-relational verb and before another constituent

(32) … it is because of that view that we think whether there is a need for a new settlement depends basically on the numbers game into which we can not go. (BNC HVK 246) K.

After a postverbal constituent and before another constituent

(33) I was telling them just basically about Claire in general… (BNC KPV 3628)

The subjectivity of basically in British English

L.

51

Final

(34) So any anybody else that’s going on board is bored out their brains basically! (BNC KBD 1744) M.

Clearly within an AdjP, or after a determiner and before adj in NP

(35) … seen by many as basically hostile to the Czechs and to the Hungarian minority in Slovakia; … (BNC HKV 1195) (36) I thought it was a basically good shape … (BNC HSK 010) N.

Within a phrase but not before an adjective

(37) The colonies as a group quickly emerged as a key element in British strategy because they were net dollar earners (basically because of Malayan rubber and West African products such as cocoa) … (BNC A6G 0863) O.

Before a subordinating conjunction in a non-initial subordinate clause

(38) The metropolitan counties, such as Tyne and Wear, in which I live, have been disbanded, basically because the Government did not like their political complexion and the fact that people in those areas consistently elected Labour authorities. (BNC HHX 05815) P.

Before the verb of a non-finite clause

(39) Er basically stuffing yourself to excess for two weeks. (BNC F88 288) Q.

Between to and an infinitive, or before to + infinitive

This category excludes any infinitive clauses which come after a relational verb, which are put in category H. (40) And right now the term competitiveness is used to basically get by safety regulations and anything else concerning safety. (BNC HMG 168) R.

Before or after a phrase (NP, AdjP, PrepP) standing alone

(41) After Easter basically. (BNC J8D 1158)

52 Christopher S. Butler S.

Standing alone, or just with Yes/No

(42) Yeah basically. (BNC GYX 0734) T.

Between nominal head and appositional phrase or clause

(43) … and that’s the reason basically that I want to be your Euro candidate. (BNC G5G 121) U.

Ambiguous or unfinished

(44) Erm and there are basically a short term pension is available. (BNC FUF 284) We may generalise over these specific positions in terms of the modifying and clause adverbial functions of the adverb. The modifying function is most clearly seen in category M in which the adverb modifies an adjective (see examples (35) and (36) above). Still clearly modifying is category N, in which basically appears in a phrase, but not before an adjective: in example (37), basically can be seen as modifying the preposition because of or (perhaps better) the whole PrepP because of Malayan rubber and West African products such as cocoa. At the other end of the scale, we have initial and final basically (categories A and L respectively), in which the adverb has a whole clause in its scope, or even possibly a whole sentence, where there is more than one clause in a paratactic or hypotactic relationship. Such examples, as we have seen, are classified by Quirk et al (1985: 621) as content disjuncts. These elements are characterised by their mobility: they can occur not only initially and finally, but also in the intermediate positions of categories B–F, H, J and K. Interestingly, there are two categories, G and I, in which it is not clear whether basically is acting as a modifier within a phrase, or as a content disjunct in the clause. Consider again example (27), repeated for convenience as (45) below: (45) Many French dishes are basically simple but may take a long time to prepare in the traditional way. Here, one possible interpretation (and intuitively the most probable) is that basically modifies directly the adjective simple. Another possibility, however, is that basically is a disjunct at clause level, in positional variation with the following:

The subjectivity of basically in British English

53

(46) Basically many French dishes are simple … (category A) (47) Many French dishes basically are simple … (category E) (48) Many French dishes are simple basically … (category L) Although grammars of English do not generally admit the modification of a noun phrase by an adverb, it is clear that at least in terms of semantic modification, (49) and (50) are parallel: (49) He is basically foolish. (50) He is basically a fool. In these terms, at least, we can see basically in example (28) as modifying the NP a “man in a suit”. Again, however, there is an alternative reading, in which basically is a content disjunct, with positional variants as shown in (51), (52) and (53): (51) Basically the RoboCop is a “man in a suit”. (52) The RoboCop basically is a “man in a suit”. (53) The RoboCop is a “man in a suit” basically. And in (29), basically could be regarded as modifying simple, or again as a disjunct in commutation with the following variants: (54) But basically the objects that physicists study are still simple objects… (55) But the objects that physicists study basically are still simple objects… (56) But the objects that physicists study are still simple objects basically… Similar arguments apply to example (31), in category I. Now let us relate these positional categories to the predicted degree of subjectivity of basically. The modifying function of the adverb is seen in cases where it has scope over just the following element. This corresponds to categories M and N of the above categorisation of positions, the former dealing with cases where the adverb unambiguously modifies an adjective, the latter

54 Christopher S. Butler with other instances of clearly modifying function. On the other hand, initial (category A) and final (category L) basically are the clearest cases of the adverb having scope over the whole clause, or even a whole multi-clausal sentence, though categories B–F, H, J and K also represent positions in which the adverb clearly acts as a disjunct within the clause. We have also seen that instances in categories G and I are ambiguous as between modifying and disjunct uses. In accordance with the arguments presented so far, we may predict that categories M and N will be significantly more frequent in written language than in spoken language, whereas categories A and L, and probably also B–F, H, J and K, will be more prevalent in spoken language than in written. Table 2 shows the frequencies of the 21 positional categories in our samples of spoken and written English from the BNC. It can be seen that categories M and N are, as predicted, more numerous in the written sample than in the spoken: indeed, there are no examples whatever of category M (modification of an adjective within an AdjP or in an NP after a determiner) in the spoken sample. Category A (initial) is somewhat more frequent in the spoken than in the written sample, while category L (final) is very much more numerous in the spoken language. Regarding the other categories for which we made predictions, B, C, D and K have frequencies which are too low for any reliable conclusion to be drawn, while E, H and J are slightly more frequent in the spoken sample and F slightly greater in the written sample. Apart from F, all the predictions are borne out. Note that category G, where the adverb follows a relational verb and precedes an AdjP or NP, is much more common in the written language than in the spoken: that is, the frequency pattern aligns it with premodifying interpretations rather than with the alternative interpretation as a mobile disjunct. In Table 3, the categories are grouped, where necessary, according to our predictions. A chi-square test on this table shows that there is a highly significant association between category group and mode (spoken/written). Comparison of observed frequencies with those expected on the null hypothesis of no association between category type and mode supports our predictions. The association of basically in final position with the spoken rather than the written language is in agreement with the findings reported in Butler (in press), in which it was shown that 12 of the 16 occurrences of basically in final position, in a sample taken from the whole of the BNC, came from spoken texts. It is also of interest to compare the findings for basically with those for fundamentally and essentially in Butler (in press). In that study, it was shown that fundamentally, which as we can see from Table 1 is more common

The subjectivity of basically in British English

55

Table 2. Frequencies of basically at various positions within the clause in samples from spoken and written components of BNC Position

Spoken Written

A. Initial B. Between left-dislocated element and clause proper C. Between fronted object and subject D. Between fronted adjunct and subject E. Between subject and first verbal element F. Between auxiliary and main verb, or between two auxiliaries G. After a relational verb and before NP or AdjP H. After a relational verb and before a clause I. After a relational verb and before PrepP or AdvP J. After a non-relational verb and before another constituent K. After a postverbal constituent and before another constituent L. Final M. Clearly within AdjP, or after a determiner and before adj in NP N. Within a phrase but not before an adjective O. Before subordinating conjunction in non-initial subord clause P. Before verb of a non-finite clause Q. Between to and infinitive, or before to + infinitive R. Before or after a phrase (NP, AdjP, PrepP) standing alone S. Standing alone, or just with Yes/No T. Between nominal head and appositional phrase or clause U. Ambiguous or unfinished Total

83 1 1 1 23 13 20 7 5 8 2 28 – 6 – 3 2 5 11 1 12

62 – – – 18 18 69 2 3 6 – 2 16 10 3 – 2 2 – 1 2

232

216

Table 3. Chi-square analysis on category groups vs. spoken or written mode Category M, N A L B–F, H, J, K G, I O–U Total

Spoken

Written

6 [16.6] 83 [75.1] 28 [15.5] 56 [51.8] 25 [50.2] 34 [22.8] 232

26 [15.4] 62 [69.9] 2 [14.5] 44 [48.2] 72 [46.8] 10 [21.2] 216

Figures in square brackets are expected frequencies on the null hypothesis of no association between category and mode. χ2 = 74.90, df = 5, p < 0.0001

56 Christopher S. Butler in written than in spoken English, has a preference, in comparison with the other two adverbs, for pre-adjectival position, either within an Adj or NP, or after a relational verb. Furthermore, it was found that all of the instances of fundamentally in final position acted as adjuncts rather than as disjuncts. These findings strongly suggest that fundamentally retains its straightforward meaning, ‘in a fundamental manner’, rather than being subjectified as with many occurrences of basically. For essentially, it was found that when compared with the other two adverbs, it tended to prefer the position after relational verbs and before an AdjP or NP, and also within phrases not containing adjectives. Since the latter is a clearly modifying position, and as we have seen that for basically the former position fits into the frequency pattern shown by modifying rather than by disjunct uses, it seems that essentially may behave more like fundamentally than like basically, in that modifying interpretations are dominant, and subjectivity minimal. However, further research on fundamentally and essentially is needed in order to firm up the tentative proposals made here. 4. Evidence from co-occurrence with other strongly subjective elements Östman (1981: 42), in his study of the routinised collocation you know, observes not only that one occurrence of this ‘pragmatic particle’ often triggers other occurrences of the same expression in the same speech turn (see example 17, discussed earlier), but also that “[y]ou know triggers off, and is triggered off by, other pragmatic particles, too”. In other words, such elements tend to co-occur, within the limits of their compatibility. We might expect, then, that if basically is often used in informal spoken English with strongly subjective meaning, it will at times co-occur with other expressions which express the speaker’s own attitudes and aims with respect to the interaction at a particular point. Of these expressions, you know, I mean and I think have received especial attention in the literature. Fox Tree and Schrock (2002) present a useful review of previous claims about the meaning and discourse function of you know and I mean, and then go on to postulate that each has a basic meaning, from which the various uses attested in the literature can be derived. These individual uses fall into five types: interpersonal, turn management, repairing, monitoring and organising. For you know, Fox Tree and Schrock (2002: 737) adopt the proposal of Jucker and Smith (1998: 196), to the effect that this expression “invites the addressee to complete the argument by drawing the appropriate inferences”.

The subjectivity of basically in British English

57

We may note that although this definition is hearer-oriented in the sense that it is the addressee who is invited to fill in further interpretation, it is still strongly subjective in the sense intended here, in that it expresses the speaker’s own aims in producing the utterance in a particular way. The interpersonal functions are particularly relevant here. Fox Tree and Schrock show that the basic underlying meaning proposed is consistent with a range of observations made in the literature, some of them even conflicting, about the speaker’s confidence or the lack of it, and the expression of politeness. Such areas are clearly strongly subjective in their orientation. For I mean, Fox Tree and Schrock (2002: 741) concur with Schiffrin’s (1987: 304) claim that this expression indicates upcoming adjustments, which may be just at the level of changing a particular word, or may even be concerned with the (re)negotiation of the meaning to be conveyed. This basic meaning is clearly strongly subjective in its orientation, being concerned with what the speaker really wishes to say/convey. In discussing interpersonal uses, Fox Tree and Schrock (2002: 741) suggest that I mean “may be linked with positive politeness because using it reminds conversational participants of more casual talk”, and that “it may be linked to negative politeness by decreasing face threat: saying I mean may be like saying ‘I’m not committed to what I just said and will adjust if you are offended’”. I think, like I mean, is clearly subjective. A brief summary of previous work on this expression is provided by Kärkkäinen (2003: 110 –115), who goes on to give a detailed analysis of its uses in the sequential and activity contexts in which it occurs in her conversational American English data. Kärkkäinen demonstrates that “I think is not just a marker of tentativeness or deliberativeness, or of negative and positive politeness, as has been claimed in much earlier research” (2003: 172), but rather that it has a multiplicity of functions in conversation. She shows that there is a clear difference in function between I think when it occurs at the beginning of an intonation unit and when it is encoded as a separate intonation unit. The former case is much more frequent, and shows three main types of function. Two of these are described as a starting-point function of I think, to routinely bring in the speaker’s personalized perspective in the discourse at a given point, either to mark boundaries and act as a frame in discourse (at points of transition), or to display that the upcoming turn will contain a new or different perspective to what was said in the prior turn (in second pair parts). (Kärkkäinen 2003: 171)

The third function of initial I think is concerned with the management of material which the speaker regards as a face threat to one of the participants.

58 Christopher S. Butler When I think is encoded as a separate intonation unit, its function is concerned with turn taking rather than with interpersonal matters. Let us now return to the prediction that if basically has strongly subjective functions in spoken English, it is likely to co-occur with other markers of subjectivity. In fact, some of the examples given in §3 illustrate such cooccurrences: in example (13), one instance of putatively subjective basically co-occurs with I mean, in (16) we have I think, and in (17) both I think and you know. A rather more systematic investigation using WordSmith Tools reveals that in the spoken sub-corpus, basically co-occurs with I think 40 times, I mean 49 times and you know 34 times 12. In other words, 8.9% of the occurrences of basically co-occur with one or more of these expressions. If we take into account the fact that there are many other strongly subjective expressions besides these three (see, for instance, example (16), discussed earlier), and also the fact that many of these expressions would be expected to occur only in informal subgenres of spoken English such as conversation, which makes up only half of the spoken component of the BNC, we can see that these co-occurrence patterns are by no means a minor phenomenon. This conclusion is reinforced by comparison of the spoken and written components of the corpus. In the written part of the BNC (which, of course, includes quotations from speech), there are just 8 co-occurrences of basically with I think (0.5% of the total occurrences of basically), 1 with you know and none with I mean. Some further examples from the spoken material, showing co-occurrence of basically with two other elements, are given below: (57) … you know I think basically final salary pension schemes are good for the employee … (BNC K77 193) (58) … but you know I mean basically you’ve gotta go back to basics … (BNC KGN 0918) (59) It’s basically, I mean, you know, who’s really gonna want one for less than ten years. (BNC JT3 0686) 12

The criteria used were that the first word of each expression should occur within 5 words of basically, and that the two should be clearly in the same utterance. Cases where the expressions formed the subject and verb of a main clause, such as I think you’re crazy, I mean John not Jim, you know what it’s like, were discarded.

The subjectivity of basically in British English

59

A further interesting observation is that in the BNC as a whole, the most frequent lexical collocate of basically is just (see Butler in press). In the spoken subcorpus, the two words collocate (within a distance of 5 words) 68 times, representing 4.9% of all the occurrences of basically. Just also collocates with 2.4% of the occurrences of basically in the written corpus. In a study in which just is examined in some detail in the London-Lund corpus of spoken English, Aijmer (2002: 158) states that “[f]rom a discoursepragmatic point of view just is an intensifying or modal discourse particle modifying a proposition” and that “[i]t has an indexical relation to the speaker’s attitudes or emotions towards a discourse event”. A more recent study by Aijmer (2005), using parallel corpora of English and Swedish, reviews some of the previous work on just, and goes on to show that “[j]ust has propositional, textual, attitudinal and social meanings, which may overlap” (Aijmer 2005: 45), and that this particle “is an interactive resource which is used to take up a position to the text, to ideas, assumptions, wishes and expectations which can also be implicit in the discourse” (2005: 37). Östman (1981: 35), who makes reference to Eisenberg (1980), also argues that just is involved in face-saving and politeness. Clearly, then, just can have a strongly subjective orientation. Some examples of co-occurrence with basically and other subjective expressions in the spoken subcorpus are given below: (60) I mean that’s just basically a chemist that innit? (BNC KD1 0928) (61) Aha, just basically I mean how do you find working and, you know rearing children, bringing up a family and doing your housework, … (BNC H03 111) Finally, it is instructive to examine the behaviour of fundamentally and essentially with regard to co-occurrences with I think, I mean, you know and just. In the case of fundamentally in the spoken subcorpus there is just one example with I mean, none with the others; for essentially, the figures are 4 for I think, 2 for I mean, 1 for you know and 6 for just. We see, then, that fundamentally does not pattern at all like basically with respect to cooccurrence with other strongly subjective elements, while essentially shows only a very weak tendency for such co-occurrence. The differences between fundamentally and essentially, on the one hand, and basically, on the other, show that although the percentage frequencies of co-occurrence for basically are quite low in an absolute sense, they indicate an important element in the behaviour of that adverb.

60 Christopher S. Butler 5. Conclusions The purpose of this paper has been to show that although the core semantic meaning of basically itself includes an element of subjectivity, this adverb, as it occurs in informal spoken English, can project, through its core semantic meaning, further meanings which are of a more strongly subjective nature in that they serve to indicate the aims and attitudes, with regard to the interaction and to his/her co-interactants, which the speaker wishes to convey. Such meanings were exemplified from the spoken component of the British National Corpus (original edition). Corroborative evidence was then obtained from two sources. Firstly, it was shown that basically behaves differently in spoken English, where one might expect strong indications of subjectivity, and in written English. Basically is much more frequent in spoken than in written English, and occurs more often in those positions where the adverb does not modify a particular element in the clause, but is more loosely attached, as a content disjunct, allowing the development of a range of more subjective meanings. Secondly, it was demonstrated that basically co-occurs with expressions with clear subjective function, such as I think, I mean, you know and just. Finally, it was shown that fundamentally shows none of the properties which define the behaviour of basically, and that essentially shows only a weak tendency towards such properties.

References Aijmer, Karin 1997 I think – an English modal particle. In Modality in Germanic Languages, Toril Swan and Olaf Jansen Westvik (eds.), 1– 47. Berlin / New York: Mouton de Gruyter. 2002 English Discourse Particles: Evidence from a Corpus. Amsterdam / Philadelphia: John Benjamins. 2005 Just and multifunctionality. In Contexts – Historical, Social, Linguistic. Studies in Celebration of Toril Swan, Kevin McCafferty, Tove Bull and Kristin Killie (eds.), 31–47. Bern et al.: Peter Lang. Benveniste, Emile 1971 Problems in General Linguistics. Translated by M. E. Meek. Coral Gables, FL: University of Miami Press. Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad and Edward Finegan 1999 Longman Grammar of Spoken and Written English. Harlow: Pearson Education Limited.

The subjectivity of basically in British English

61

Butler, Christopher S. in press ‘Basically speaking’: a corpus-based analysis of three English adverbs and their formal equivalents in Spanish. In Current Trends in Contrastive Linguistics: Cognitive and Functional Perspectives, María de los Ángeles Gómez-González, J. Lachlan Mackenzie and Elsa González-Álvarez (eds.). Amsterdam /Philadelphia: John Benjamins. Carey, Kathleen 1995 Subjectification and the development of the English perfect. In Subjectivity and Subjectivisation: Linguistic Perspectives, Dieter Stein and Susan Wright (eds.), 83–102. Cambridge: Cambridge University Press. Chafe, Wallace L. 1985 Linguistic differences produced by differences between speech and writing. In Literacy, Language and Learning: The Nature and Consequences of Reading and Writing, David R. Olson, Nancy Torrance and Angela Hildyard (eds.), 105–123. Cambridge: Cambridge University Press. Downing, A. 2001 ‘Surely you knew!’ Surely as a marker of evidentiality and stance. Functions of Language 8 (2): 251–282. 2006 The English pragmatic marker surely and its functional counterparts in Spanish. In Pragmatic Markers in Contrast, Karin Aijmer and AnneMarie Simon-Vandenbergen (eds.), 39–58. Amsterdam: Elsevier. Eisenberg, Ann R. 1980 Indicating sentential mode through linguistic particles: The case of just. Ms. Dept. of Psychology, UC Berkeley. Erman, B. and U.-B. Kotsinas 1993 Pragmaticalization: the case of ba’ and you know. Studier i Modern Sprakvetenskap, Acta Universitatis Stockholmiensis, New Series 10: 76–93. Fox Tree, Jean E. and Josef C. Schrock 2002 Basic meanings of you know and I mean. Journal of Pragmatics 34: 727–747. Greenbaum, Sidney 1969 Studies in English Adverbial Usage. London: Longman. Haiman, John 1994 Ritualization and the development of language. In Perspectives on Grammaticalization, William Pagliuca (ed.), 3–28. Amsterdam /Philadelphia: John Benjamins. Jucker, Andreas H. and Sara W. Smith 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 /Philadelphia: John Benjamins.

62 Christopher S. Butler Kärkkäinen, Elise 2003 Epistemic Stance in English Conversation. Amsterdam /Philadelphia: John Benjamins. Kemmer, Suzanne 1995 Emphatic and reflexive –self: expressions, viewpoint, and subjectivity. In Subjectivity and Subjectivisation: Linguistic Perspectives, Dieter Stein and Susan Wright (eds.), 55–82. Cambridge: Cambridge University Press. Langacker, Ronald W. 1985 Observations and speculations on subjectivity. In Iconicity in Syntax. Proceedings of a Symposium on Iconicity in Syntax, Stanford, June 24–26, 1983, John Haiman (ed.), 109–150. Amsterdam /Philadelphia: John Benjamins. 1989 Subjectification. Series A, Paper 262. Duisburg: Linguistic Agency University of Duisburg (L.A.U.D.). 1990 Subjectification. Cognitive Linguistics 1: 5–38. Revised version as Chapter 12 (pp. 315–342) of Concept, Image, and Symbol: The Cognitive Basis of Grammar. Berlin /New York: Mouton de Gruyter. Lyons, John 1982 Deixis and subjectivity: Loquor, ergo sum? In Speech, Place, and Action: Studies in Deixis and Related Topics., Robert J. Jarvella and Wolfgang Klein (eds.), 101–124. Chichester /New York: John Wiley. Nuyts, Jan 2001 Epistemic Modality, Language, and Conceptualization: A CognitivePragmatic Perspective. Amsterdam /Philadelphia: John Benjamins. Östman, Jan-Ola 1981 You know: a Discourse-Functional Approach. Amsterdam /Philadelphia: John Benjamins. Powell, Mava Jo 1992 The systematic development of correlated interpersonal and metalinguistic uses in stance adverbs. Cognitive Linguistics 3(1): 75–110. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik 1985 A Comprehensive Grammar of the English Language. London: Longman. Scheibman, Joanne 2002 Point of View and Grammar: Structural Patterns of Subjectivity in American English Conversation. Amsterdam /Philadelphia: John Benjamins. Schiffrin, Deborah 1987 Discourse Markers. Cambridge: Cambridge University Press. Sinclair, John McH., Susan Jones and Robert Daley 1969 Report to OSTI on Project C/LP/08.

The subjectivity of basically in British English

63

Taverniers, Miriam 2005 A semiotic-functional perspective on subjectivity. Paper given at FITIGRA: From ideational to interpersonal: Perspectives from grammaticalization, University of Leuven, 10–12 February 2005. Tomasello, Michael 1999 The Cultural Origins of Human Cognition. Cambridge, MA / London: Harvard University Press. Traugott, Elizabeth Closs 1982 From propositional to textual and expressive meanings: some semantic-pragmatic aspects of grammaticalization. In Perspectives on Historical Linguistics, Winfred P. Lehmann and Yakov Mikael (eds.), 245–271. Amsterdam /Philadelphia: John Benjamins. 1989 On the rise of epistemic meanings in English: an example of subjectification in semantic change. Language 65: 31–55. 1995 Subjectification in grammaticalization. In Subjectivity and Subjectivisation: Linguistic Perspectives, Dieter Stein and Susan Wright (eds.), 31–54. Cambridge: Cambridge University Press. 2003 Constructions in grammaticalization. In A Handbook of Historical Linguistics, Brian D. Joseph and Richard D. Janda (eds.), 624–647. Oxford: Blackwell. Verhagen, Arie 1995 Subjectification, syntax, and communication. In Subjectivity and Subjectivisation: Linguistic Perspectives, Dieter Stein and Susan Wright (eds.), 103–128. Cambridge: Cambridge University Press. 2005 Constructions of Intersubjectivity. Oxford: Oxford University Press.

Variation in advanced oral interlanguage: The effect of proficiency on style choice Jean-Marc Dewaele

1. Introduction Quantitative sociolinguists and pragmaticists working in the interlanguage (IL) paradigm measure synchronic variation in linguistic systems in terms of the distance to the norm or target-likeness. Occurrences of phonological, morphological or syntactical variants are counted allowing the computation of frequencies and, in IL studies, accuracy rates. The objective of the present analysis is the uncovering of the source(s) of the variation in speech styles. These factors can be linguistic, situational, or psycho-social in nature. The concept of speech style, often defined in terms of attention to form (Labov 1972), is used frequently but rarely constitutes the object of the investigations. The probable reason for this is the absence of quantified measure of style (Azuike 1992). Using a measure of style called the degree of explicitness-implicitness developed in Dewaele (1995a), I devoted a number of studies to the effects of age, gender, social class, and degree of extraversion on style shifting in native Dutch and in French IL (Dewaele 1998a, 2001b, 2004; Dewaele and Furnham 2000; Dewaele and Heylighen 2002; for an overview see Dewaele 2007). I propose in the present study to analyse the effect of proficiency on style choice in a corpus of advanced French IL produced by 32 Dutch L1 speakers. 
2. Measuring speech style variation IL studies dealing with the phenomenon of style-shifting often used accuracy rates as a dependent variable and task as an independent variable (Tarone 1983, 1988). In her seminal study Variation in interlanguage (1988), Tarone lists four possible causes for IL variation in accuracy rates: “(1) linguistic context, (2) a psychological processing factor whose precise nature remains to be defined (…) (3) a set of social factors so far seemingly limited to interlocutor, topic and social norms and (4) the differing functions which a given form performs in communicative discourse” (1988:

66 Jean-Marc Dewaele 135). Tarone has argued consistently that any second language acquisition theory must describe and explain systematic variation in IL performance from one social context to another (Tarone 1997, 2000). Her early research interest lay especially in the second cause1, namely task-related variation: Different tasks may tap different styles. A grammaticality judgment task provides information about only the more formal, or careful style; there is no sense in which this style ‘underlies’ other styles. Other tasks, like reading or word lists or connected texts, the combining of sentences, the description of entities, or narration of story must be assumed to tap different styles of the IL continuum. (Tarone 1988: 40)

Tarone chose the term ‘chameleon’ for her model because a speaker’s IL adapts itself to the environment. She defines interlanguage as follows: The abstract system which guides the regular language behavior of the second-language learner. This capability is hypothesized to be an abstract linguistic system consisting of a continuum of styles; the system exists apart from any particular insistence of its use. That portion that underlies a particular learner behavior may be determined by the degree of attention which the learner pays to language form in that instance. (Tarone 1983: 156)

Systematic observation of a speaker would create a formal context and the speaker would pay more attention to form (1988: 40), resulting in higher accuracy rates (Tarone 1985). Faced with complex patterns of style-shifting, she later had to admit that this hypothesis was wrong: The hypothesis that the IL style with least grammatical accuracy would be the vernacular style (that produced on the narrative task) and the style with the most grammatical accuracy would be the careful style (that produced on the grammar tests) was not upheld. (1988: 100).

She concluded that attention to form “cannot be an end cause of variation; at best, it is only intermediary in its function” (1988: 43). The problem with this approach is that no independent evidence exists of the “formality” of the speech style. The task is supposed to induce a certain level of formality affecting psychological processing resulting in variation in accuracy rates. Defining speech-style by referring to the situation in 1

Tarone’s later research concentrated more on the third cause of IL variation, namely the influence of social context on psycholinguistic processes (Tarone 2000).

Variation in advanced oral interlanguage 67

which it was produced is empirically unsound (Dewaele 1995a, 1996a, 2003) as the result is a circular definition stating that a formal style is a style produced in a formal situation. Labov (1972) was conscious about this problem as he stated: “the most immediate problem to be solved in the attack on sociolinguistic structure is the quantification of the dimension of style” (1972: 245). Labov acknowledges the fact that the setting of a particular context is not enough to observe a certain speech style: “We also look for some evidence in the type of linguistic production that the speaker is using a speech style that contrasts with Style B” (1972: 94–95). A second problem with defining speech style through the situation in which it is produced is that measuring the position of individuals on the speech style continuum is impossible, hence excluding any analysis of intra- and interindividual variation on this dimension. Accuracy rates cannot be used to define the position on the speech-style continuum when they are the dependent variable in an analysis. Moreover, I argued that the dimensions of accuracy and speech style are orthogonal (Dewaele 1995a). Labov has equally rejected the use of phonological variables in the definition of his styles: “because the values of these variables in Styles A and B are exactly what we are trying to determine by isolation of styles (1972: 95). Labov has suggested that one possible solution to this problem is to analyse the “channel cues” to determine a style: a vernacular style would be characterized by increases in speech rate, in loudness, in laughter and by a decrease in pausing (1972: 95). As Giles and Powesland (1975) regret however that: “Labov does not provide any information as to what proportion of the speech in the five contexts failed to meet the strict criteria for a concomitant change in one of the five channel cues in order to constitute casual speech” (1975: 131). The concept of attention to form as a cause for style-shifting has been rejected by many linguists (Graddol and Swann 1989; Bell 1997, 2001) but its influence lingers on as the following definition of “formal style” in the Dictionary of Language Teaching and Applied Linguistics shows: “the type of speech used in situations when the speaker is very careful about pronunciation and choice of words and sentence structure. This type of speech may be used, for example, at official functions, and in debates and ceremonies” (Richards, Platt and Platt 1997: 144). This definition gives us an idea of what a formal situation is, but does not define formal speech as such; it just offers a hypothesis of what a speaker pays attention to in certain situations. The main criterion for formality in speech is thus non-linguistic. A better definition has been proposed by Argyle et al. (1981) who observed that “formal speech appears to be directed mainly towards efficient communication of questions, information, etc., so that a lot of ma-

68 Jean-Marc Dewaele terial is conveyed, with precision” (p. 297). Prototypical informal speech would be produced in a relaxed conversation among close friends or family members. The authors note that informal speech is more than pure transfer of information: (it) “is concerned with keeping people happy and interested, or on other interpersonal goals, in addition to pure task goals” (id.). Argyle et al.’s description of formal speech is better than that by Richards, Platt and Platt (1997), but it still falls short of an empirical definition. The lack of quantification of the dimension of style has hampered sociolinguistic and pragmatic research as Labov (1972) had foreseen. Rickford and McNairKnox (1994) point out that the decline of interspeaker or stylistic variation as a focus of research in quantitative sociolinguistics was precisely due to “the fact that investigators found it difficult to separate ‘careful’ from ‘casual’ speech in reliable and objective ways” (p. 265). Rickford (2001) underlines the limitations of unidimensional models like the “attention to speech” and advocates multidimensionality, and a variety of approaches (p. 228). One such multidimensional approach is defended by Preston (2001) for whom variation is characterized by “a funnel of influences” which are ordered hierarchically, with linguistic factors outweighing social and, finally, stylistic factors. Coupland (2001) also argues that the scope of approaches to style need to be extended: My contention is that sociolinguistic approaches to style can and should engage with current social theorizing about language, discourse, social relationships and selfhood, rather than being contained in one corner … of one disciplinary treatment of language. (Coupland 2001: 186).

Labov (2001) acknowledges the criticism leveled against the concept of attention to form as sole cause for style-shifting: The organization of contextual styles along the axis of attention paid to speech was not intended as a general description of how style-shifting is produced and organized in every-day speech, but rather as a way of organizing and using the intra-speaker variation that occurs in the [sociolinguistic] interview. (Labov (2001: 87)

The terms “implicit-explicit”, and “involvement-detachment” crop up regularly in research dealing with speech styles in oral and written language (cf. Olson 1977; Chafe 1982; Redeker 1984). Cummins and Swain (1986) for example, argue that communication can be defined in terms of context dependence. It can be either context-reduced communication where “shared reality cannot be assumed, and thus linguistic messages must be elaborated precise-

Variation in advanced oral interlanguage 69

ly and explicitly so that the risk of misinterpretation is minimized” (1986: 153); or context-embedded communication, which “derives from interpersonal involvement in a shared reality which obviates the need for explicit linguistic elaboration of the message” (1986: 153). Cummins and Swain (1986) do not offer any specific measure of context-embeddeness but underline that speech fragments will be situated on the continuum between both poles. Hasan (1984) develops the pragmatic aspects of explicit and implicit styles, defining them “in terms of what a normal person needs in order to interpret an utterance, as is intended by the speaker” (p. 109). Although she states that the degree of implicitness is determined purely quantitatively, “by comparing the proportion of explicit units to the implicit ones” (p. 110), she finally rejects quantitative grading and thus fails to present a workable empirical definition: “A more valid criterion for the grading of implicitness would be by reference to the requirements for interpretation. The greater the ease in interpreting the speaker’s intended meanings, the less implicit the device” (p. 125). Leckie-Tarry (1995) develops Hasan’s theory and states that in an “explicit” text the context of situation is not immediately apparent and that “a greater load falls on the linguistic structures to convey meaning” (p. 123). According to Leckie-Tarry, words can be classified on a continuum of explicitness, and range from non-explicit core words which are not very field- or genre specific to highly explicit words, like technical words that have strong associations with particular fields and hence particular registers. Leckie-Tarry’s theoretical explanation of the causes of style choice is most enlightening but no empirical measure is proposed to compare the degree of explicitness of speech extracts. I have suggested (Dewaele 1995a) that such an empirical measure could be developed from Levelt’s observation that “Interlocutors anchor their contributions steadily in the spatio-temporal context of their conversation” (Levelt 1989: 58). He thus echoes Lyons (1977) who pointed out that deixis is the most fundamental and universal type of linguistic reference corresponding to what has been named the “indexical” function of language (Bar-Hillel 1954; Barnes and Law 1976). By anchoring their utterances in a shared spatio-temporal context, the so-called ego-nunc-hic origo, speakers can transmit much more information than what is literally said. There is less need for explicitness since the context shared by sender and receiver will provide the additional information lacking in the linguistic expression itself. If on the other hand senders and receivers do not share the same spatial or temporal context, like a journalist talking to an invisible audience or writing for imaginary readers, they will have to add more details to make tacit understandings explicit, their style will thus become more explicit and less

70 Jean-Marc Dewaele context-dependent. Levelt (1989: 45) distinguishes four types of deixis: 1) person, 2) place, 3) time, and 4) discourse. This last type could also be interpreted as anaphoric reference, i.e. reference to something that has been said earlier in the exchange. The words that will be used to produce this “deictical” speech belong mainly to the classes of pronouns, adverbs and verbs. The relative proportion of these word-classes is therefore likely to be higher in more informal speech. Using De Jong’s (1979) frequency dictionary of spoken Dutch as a guide (120,000 words from 80 speakers), I calculated the proportion of deictical word tokens in the grammatical classes that contain both deictical and non deictical words. Most pronouns (94.3%) are deictical (person and discourse deixis), as well as about two thirds of adverbs (63%) (place and time deixis). More than two thirds of verb tokens (71%) are finite verbs (time deixis). Specifiers are generally non-deictical: 79% of tokens of adjectives and determiners in De Jong’s corpus. Deictical specifiers include demonstratives, reflexives and possessives. Nouns and prepositions form a homogeneous group of non-deictical words. I assume that the proportion of deictical words within the different grammatical classes are comparable in other Romance and Germanic languages. This assumption is based on my discovery of identical patterns of variation in word-class proportion between different genres of written speech in native Italian and native Dutch and between different styles (ranging from informal to formal) in native French, and in French IL (Dewaele 1996a, 2001c). The variation found was not one of content words versus function words as predicted by Halliday (1989) but, rather, cut across both categories as was demonstrated earlier by Fielding and Fraser (1978) and by Moscovici and Humbert (1960). It appears that the proportions of nouns, specifiers (articles and adjectives) and prepositions are higher in the corpora of written and formal oral speech, whereas pronouns, verbs and adverbs are found to be more frequent in the corpora of oral and informal speech. The variation appears to be significant for every category except for the conjunctions. I conclude that written speech or a more formal context leads to an increase in nouns, specifiers and prepositions. The values for the different word classes do not constitute an easy measure of formality or explicitness, but they do offer a quantitative indication of the degree of context-dependence or contextuality for expressions that are ambiguous when considered on their own, but where the ambiguity can be resolved by taking into account additional information from the context (cf. Heylighen 1999; Heylighen and Dewaele 2002). The term “contextuality” encompasses both the case of deixis, where a connection is to be made with a distinct part of the spatio-temporal setting, and the more abstract case of implicature, where the information to be added

Variation in advanced oral interlanguage 71

must be inferred from unstated background assumptions. It also includes reference to information expressed earlier (discourse deixis). More generally, the context of an expression can be defined as everything available for awareness which is not part of the expression itself, but which is needed to correctly interpret the expression. One possible single measure of contextdependence (called the F-measure) weights the proportion of deictical versus non-deictical word classes (Heylighen and Dewaele 2002). The more formal the language excerpt, the higher the value of F has been found to be (Nowson 2006; Nowson, Oberlander and Gill 2005; Oberlander and Gill 2006). A number of researchers developed more sophisticated quantitative approaches to measure speech styles. Fielding and Fraser (1978), for example, used factor analysis to extract dimensions from linguistic material collected through Parents and Krauss tasks. The variables they entered in the analysis were proportions of parts of speech, word class ratios, frequency of various types of personal pronouns, and ego- and alter-referencing sequences (“I think”, “you know”). The most important dimension was labeled nominalverbal and accounted for approximately 15% of the total variance in each task (p. 223). The variables that loaded on the nominal pole of the dimension were noun-verb ratios, noun-pronoun ratios and proportions of nouns, adjectives and articles, while personal pronouns referring to the speaker and the listener, ego- and alter-referencing sequences, auxiliary verbs, and adverbs loaded on the verbal pole of the dimension. The speaker’s choice of speech style on this dimension has important syntactical consequences: for instance, in nominal constructions extensive use is made of non-finite rather than finite verb forms, modification is rather adjectival than adverbial, the construction is likely to be longer than the equivalent verbal structure, and the variety of structures allowed is rather restricted. (Fielding and Fraser 1978: 226)

Fielding and Fraser (1978) suggest that “This noun versus verb distinction, together with its related word classes, represents a fundamental and perhaps universal grammatical distinction” (p. 223). After considering the objective differences between verbal and nominal styles, the authors concentrate on the subjective differences: the nominal style is likely to be more monotonous, less personal, and more formal. It appears to be a careful considered and closely monitored production. The verbal style, on the other hand is characteristic of spontaneous, unreflective speech. It is immediate, informal and varied. (Fielding and Fraser 1978: 226)

72 Jean-Marc Dewaele Fielding and Fraser (1978) then describe an experimental setting where the speaker exploits the listener’s response to the two styles in order to manipulate the speaker-listener relationship. They assume that when one likes the listener “the vivid, personal, and varied qualities of the verbal style would be appropriate. However, when addressing a disliked listener, the impersonal, unemotional, formal characteristics of the nominal style would be preferable” (p. 226). The factor scores on the nominal-verbal dimension allowed the authors to analyse interindividual variation through ANOVAs on the factor scores. They found that, as predicted, “differences in speakers’ liking of their listeners had a significant effect in both experimental tasks (…) speech to liked listeners was significantly less nominal (F = 8.30, p < .05)” (p. 225). The most comprehensive analysis of variation in style or register has been carried out by Biber (1988, 1995; Biber and Finegan 1994; Biber, Conrad and Reppen 1994, 1998; Finegan and Biber 2001). Using large corpora of different types of written and oral language, Biber developed a multidimensional analysis of register variation using sophisticated statistical techniques in order to isolate the effects of many factors other than the traditional attention paid to speech and conscious manipulation. Biber (1994) claims, quite rightly, that “in previous work, dimension interpretations have been based on qualitative assessments of the underlying communicative functions (or situational parameters) and linguistic dimensions” (p. 49). Biber submits 67 heterogeneous linguistic variables to a factor analysis, from which six dimensions emerge2. Biber’s Dimension 1 (1988) is interpreted as representing “Involved versus Informational Production”. The naming of this dimension is a bit confusing, because it implies that involved production contains less information, being more general and more ambiguous which is not necessarily true (Dewaele 2001c). There does seem to be a parallel between Biber’s Dimension 1 and the notion of a continuum ranging from implicit to explicit production. Dimension 1 comprises such linguistic features as first and second person pronouns, contractions, that-deletions, hedges, and emphatics versus nouns, prepositional phrases, and type/token ratio. In his correlational analysis between linguistic dimensions and situational parameters, Biber (1994) finds that production circumstances (non/careful) have the strongest 2

Biber argues that the heterogeneity of his features is a strength as the factor analysis does the sorting (personal communication). I am not entirely convinced by this argument but entering into this debate would involve a highly technical discussion outside the scope of the present study.

Variation in advanced oral interlanguage 73

relation to Dimension 1 (r2 = 74%), followed by interactiveness (r2 = 59%), physical mode (r2 = 58%), and informative purpose (r2 = 36%) (p. 49). Biber (1994) concludes that his analytical techniques can “usefully be applied to both register and dialect studies” (p. 50) and these could lead to a more “comprehensive sociolinguistic theory” (p. 50). Indeed, his approach allows a global analysis, beyond that of individual (socio)linguistic variables, which “have typically been further restricted to include only phonological and morphological features with semantically equivalent alternants” (p. 51). A more complicated, but potentially better single measure can be obtained through a principal components factor analysis on the proportion of nouns, specifiers, prepositions, verbs, pronouns, adverbs, and conjunctions. By ensuring that the variables submitted to the factor analysis are of the same nature (contrary to Fielding and Fraser (1978) and Biber (1988, 1995) for example), the homogeneity and the validity of the results can be increased. This could, in turn, boost the total variance explained in the data. A quartimax rotation yielded two components explaining 75% of the variance in two styles of French IL of 27 Flemish university students (Dewaele 1995a). The first component3, which explains 59% of the variance, was called “implicitness-explicitness” (see table 1). The nouns, specifiers and prepositions obtained strong negative loadings on this factor, as opposed to the pronouns, adverbs, and verbs which obtained high positive loadings. The nouns, specifiers and prepositions are thus situated near the explicit end of this dimension, in contrast to the pronouns, adverbs, and verbs on the implicit end of the continuum. Table 1. Overview of loadings of the components (PC) of the grammatical variables in the French IL corpus after quartimax rotation (the highest loadings in bold) (Dewaele 1995a: 234) VARIABLES: Nouns Specifiers Prepositions Pronouns Verbs Adverbs Conjunctions 3

COMPONENT 1 – 0.92 – 0.87 – 0.81 0.88 0.56 0.86 – 0.16

The factor analysis also yielded a second factor. As this is not of interest to the present discussion it will be ignored here.

74 Jean-Marc Dewaele Using individual factor scores, the degree of explicitness/formality of individual speech extracts can be measured and located on the continuum of styles. These factor scores are reliable and empirical second order measures to analyse style-shifting. I have argued in previous studies (Dewaele 1995a, 2001b; Dewaele and Heylighen 2002) that the decision to opt for a certain speech style in a certain situation is guided by the speaker’s interpretation of Grice’s (1975) maxim of quantity. A speaker will only use a more formal, explicit style – which is cognitively more demanding – if s/he deems it necessary to be unambiguously understood. In that case s/he will provide the highest possible amount of information without violating the maxim of quantity. In informal situations, on the other hand, where speakers share the same spatio-temporal context and presuppositions and where there is a possibility of correction of communicational intentions, a more informal, implicit style can be used which is more economical. In this situation the speaker will provide the lowest possible amount of information without violating the maxim of quantity. For example, friends who see each other regularly have extensive shared knowledge and clear mental models of each other (cf. G. Green 2000). They can therefore interact using highly implicit speech when talking about familiar topics. The exact interpretation of Grice’s maxim of quantity in a particular situation will vary between individuals. This means that some will produce a more explicit speech style than others in the same situation. One can assume that this choice is determined by a number of psychological and sociobiographical factors. Indeed, Labov (1990, 1994) pointed out that no single independent variable can account for all the variation in style-shifting. Explicitness scores were used to analyze patterns of style-shifting in two styles of advanced French IL in Dewaele (2001b). The results show complex interaction between objective and subjective variables. The formality of the situation in one-to-one conversations clearly affected the degree of explicitness but a lot of intra- and interindividual variation was observed. Gender, social class and degree of extraversion significantly affected the speaker’s perception of the formality of the situation in the French IL corpus (Dewaele 1996a, 1998a, 2001a, b). Male speakers, speakers from lower social classes and introverts were found to use more explicit speech in the informal conversations. The degree of extraversion appeared to be the main predictor of style-choice in the formal situation, with the introverts using the most explicit speech styles (Dewaele 1996a; Dewaele and Furnham 2000). These results were interpreted in terms of social distance and fear of ambiguity, which pushed the speakers to opt for a maximal interpretation of the maxim of quantity.

Variation in advanced oral interlanguage 75

The present study concentrates on another, more learner-specific factor which may determine style choice, namely the level of proficiency of the learner. It is often said that less proficient IL speakers feel communicatively handicapped and do not yet possess the full range of styles on the continuum. They might therefore be unable to translate their communicative intentions accurately in the IL, hence their use of compensatory strategies (Poulisse 1997). The fear that their words could be misinterpreted might push them towards more explicit speech styles. The balance between being clear and saving effort will thus be different in IL production compared to L1 production. Learners have been found to favour formal speech styles in the IL. This has been attributed to the type of input and the lack of authentic interactions in the target language (TL). Tarone and Swain (1995) analysed oral speech produced by adolescent native speakers of English enrolled in French immersion programs in Canada. The authors found that the participants were in fact mono-stylistic in their French IL. They used relatively formal French in an academic context and switched to vernacular English styles when addressing peers. Their lack of contact with native French speakers of their own age limited the development of their sociolinguistic competence in French, and more specifically the use of more informal speech styles. Similarly, Blanco-Iglesias, Broner and Tarone (1995), Foster (1998) found that learners have very few authentic interactions with their peers within the classroom. Moreover, teachers themselves do not engage in enough negotiated interactions with the learners in the TL (Lyster 1998). Not surprisingly, teachers’ talk was found to be more formal. Mougeon et al. (2002) analysed 30 hours of recordings of seven teachers of French in their classrooms in Ontario schools and found that the teachers systematically avoided colloquial variants and heavily favoured formal variants in their speech. My own analyses of the sociolinguistic competence in French of nonnative speakers confirmed Tarone and Swain’s finding. IL speakers were found to have a clear preference for standard variants. Stigmatized or simply non-standard structures or forms, like the omission of ‘ne’ in the negation, non-standard interrogative structures, the use of informal personal pronouns and colloquial words were clearly avoided (for an overview, see Dewaele 2007). Finally, learners rely heavily on written language, which tends to be more formal. Mougeon et al. (2002) analysed a sample of French coursebooks used in the Toronto area and looked specifically at texts which more or less reflected oral French (dialogues, interviews) and texts which reflected written French (extracts from novels, newspaper articles). As expected

76 Jean-Marc Dewaele the texts reflecting written French in the course books did not contain a single informal variant. More surprising, informal variants were also almost absent in the texts supposed to reflect oral French. 3. Rationale for the present study The multiple regression analyses that were carried out in Dewaele (2001b) allowed me to pinpoint the predictive power of several independent variables in the choice of speech style but they do not permit analysis of intergroup variation. Level of proficiency turned out to be a strong predictor, but only ANOVAs can throw a light on the exact differences between low, medium and high proficiency groups 4. The present study will thus focus on the effect of proficiency on style choice, and check whether there is any significant linear relationship between proficiency and implicitness. I will also look at the effects of amount and intensity of formal instruction on the TL and amount of authentic communication in the TL outside the classroom. 4. Research Hypotheses 1.) More proficient IL speakers will possess a wider range of speech styles. These speakers will be able to shift down the speech style continuum if the situation requires it. Less proficient IL speakers on the other hand will always use more explicit (“formal”) styles. 2.) Interindividual variation will be linked to the participants’ linguistic history (amount and intensity of instruction in the TL and frequency of authentic communication outside the classroom). 5. Methodology 5.1. Participants Thirty-two university students from the Vrije Universiteit Brussel, 13 female and 19 male, aged between 18 and 21, participated in the experiment. They had taken French at a high school level (3 to 5 hours a week) for 6 to 8 years and had been following intensive French courses (150 hours) for 5 4

A MANOVA design was impossible because of the small sample size.

Variation in advanced oral interlanguage 77

months with the researcher as their teacher. Both the participants and the researcher were trilingual (Dutch-French-English) although the participants’ French was weaker. It could be described as a “pre-advanced to advanced interlanguage” (Bartning 1997). Teacher and students communicated in French but the students knew that the teacher had native competence in Dutch. The participants were administered a sociobiographical questionnaire which included questions about the type and frequency of contact with the TL.

5.2. Recordings The researcher and the participants were recorded sitting face to face in a classroom in an informal situation. It involved one-to-one conversations between the researcher and participants in a relaxed atmosphere. They were told that the purpose of the conversation was merely to have a relaxed informal chat about their studies, hobbies, politics etc. Efforts were made to make the participants feel at ease, and to this end it was stressed that content more than the form of their speech was important. Errors were not corrected and a coherent and spontaneous discussion was thus maintained. There was no time-restriction. In all 13 hours of speech (55,223 words) were recorded. The recordings were transcribed by the researcher into orthographical French. These transcriptions were then coded at the word level according to their grammatical nature and possible lexical or morphological errors.

5.3. The variables General morpholexical accuracy rate is the independent variable in the present study, explicitness scores the dependent variable. In Dewaele (1994, 1998c) morpholexical accuracy rates for all word classes were calculated using the following formula: (Total of words tokens – (incorrect + omitted + oversupplied words)) X 100 / Total number of word tokens.

Among the morphological errors 6 classes were distinguished: violation of gender and number, and for verbs, violation of tense and aspect, of mode and of person. Taken into account at the lexical level were: lexical inventions, words that were superficially right but that did not fit the context (semantic errors), the absence of a word in an obligatory context, and the

78 Jean-Marc Dewaele suppliance of a word where it was not required. The corpus contains 3,400 morpholexical errors. The decision to choose morpholexical accuracy rates as main indicator of level of proficiency is based on the fact that these reflect the state of the IL grammar (Selinker 1992). Moreover, accuracy rates were found to correlate positively and significantly with speech rate (Dewaele 1998b). A significant negative correlation was found between accuracy rates and the proportion of filled pauses (Dewaele 1996b). As high accuracy is linked with high fluency and as these two aspects are generally considered to be the main components of proficiency in IL (Ely 1986), the first variable was used to divide the speakers in three proficiency levels. The first group contained 8 participants with mean accuracy rates that were more than one standard deviation (2.8) below the group mean (93.4%). The second and largest group contained 16 participants whose mean accuracy scores lie within one standard deviation above and below the group mean. The third group consisted of 8 participants with mean accuracy rates that were more than one standard deviation above the group mean.

6. Analysis A one-way analysis of variance (ANOVA) with proficiency as main effect revealed that the difference in degree of implicitness between low proficiency, medium proficiency and high proficiency groups is highly significant (F (2, 29) = 15.44, p < 0.001, eta2 = .139) (see figure 1).

Figure 1. The effect of proficiency on the degree of explicitness

Variation in advanced oral interlanguage 79

A closer analysis of the mean implicitness scores for the three proficiency groups shows that the low proficiency group used more explicit speech styles (M = .81, SD = .27) than the medium proficiency group (M = .07, SD = .70), which, in turn, used more explicit speech styles than the high proficiency group (M = –1.21, SD = .25). A Fisher’s PLSD post-hoc analysis revealed that the difference between the high and low proficiency groups is significant or highly significant (p < .0001). A similar picture emerges for the difference between high and medium proficiency (p < .002) and for the difference between low and medium proficiency (p < .019). Two one-way ANOVAs revealed that the independent sociobiographical variables “amount and intensity of formal instruction in the TL” and “frequency of use of the TL outside the classroom” have an effect on style choice5 (see figure 2).

Figure 2. The effect of sociobiographical variables on the degree of explicitness

The participants who had had French as an L2 and had consequently enjoyed more formal instruction (n = 26) produced less explicit speech styles than those who had had French as an L3, with less formal instruction (n = 6) (M = –.20, SD = .94 versus M = .90, SD = .84). This difference is significant (F (1,30) = 7.00, p < .013, eta2 = .071). A Fisher’s PLSD post-hoc analysis revealed that the difference between the two groups is significant (p < .013).

5

I used two one-way ANOVAs rather than a two-way ANOVA because of a problem in the design (only one case of L3 + Regular communication in the TL). The two-way ANOVA shows however that there is no interaction between both independent variables (df = 1, 28; F = .11, p = .73).

80 Jean-Marc Dewaele Those who reported speaking the TL regularly (n = 7) used less explicit speech styles than those (n = 25) who reported infrequent use of the TL (M = –.74, SD = .80 versus M = .21, SD = .98). This difference is significant (F (1,30) = 5.45, p < 0.027, eta2 = .062). A Fisher’s PLSD post-hoc analysis revealed that the difference between the two groups is significant (p < .027). 7. Discussion These results suggest that level of proficiency is indeed linked to style choice in advanced French IL. A number of possible causes can be imagined to account for the specific effect of level of proficiency on style choice. The preference for more explicit, heavier speech styles by less advanced speakers can be related to the lower degree of grammaticalisation of their IL (Housen 1998). They might not yet possess the linguistic means to produce cohesive, highly grammaticalised IL. The use of multiple deictical pronouns and adverbs, for example, present notorious difficulties in terms of word order. The less advanced speakers’ use of more explicit speech styles is not the result of a conscious choice but rather the consequence of their lower level of proficiency in the TL. It is not surprising, therefore, to find that those learners having benefited from less formal instruction in the TL and those who use the TL infrequently outside the classroom should use more explicit speech styles. An illustration of this phenomenon can also be found in Chini (2005) who analysed reference to person in corpora of postbasic, intermediate and advanced Italian IL from 8 learners and compared them with the speech of 13 native speakers of Italian. Using a film-retelling task, she found that learners at lower levels of proficiency tended to be over-explicit in their use of reference maintaining devices, relying heavily on full NPs and full pronouns, before using more native-like grammatical co-referential means with more zero pronouns and clitics. She also reports a progress from lexical anaphoras (bare nouns, full noun phrases (NPs) to grammatical ones which she interprets as an illustration of the evolution from a pragmatic mode to a more syntactic based coreferential style (Givón 1979). Chini concludes that her Italian IL extracts were generally less cohesive and more redundant than the natives’ extracts. A second possible cause for the specific effect of level of proficiency on style choice might be of a synchronic nature. Speakers might be capable of producing a more implicit speech style but might still prefer not to use it because of a lack of confidence and a fear of the risk of misinterpretation of

Variation in advanced oral interlanguage 81

their communicative intentions. Hence the avoidance of short-cuts that multiple deictical pronouns, adverbs and verbs offer because of the fear of getting lost along the way, or making errors, and a preference for a more nominal and explicit speech-style, which may be more heavy and less dynamic but less prone to misinterpretation by their interlocutor. The amount of effort IL speakers are prepared to put into their speech production depends on the type of task and of situational variables. Poulisse (1997) found that her Dutch L1 subjects were more brief and economical in naturalistic tasks in English interlanguage compared to more controlled tasks where they used detailed analytic strategies. They clearly felt in the latter case that the communicative goal was important enough to warrant an extra processing effort. Situation and task thus affect the balance between the principles of clarity and economy. Interlanguage speakers capable of producing implicit speech might also be unable to sustain this speech style during the whole of the interaction because of the high cost in cognitive resources; hence the (temporary) lapse back into more explicit speech styles, and lower overall scores on the implicitness dimension. The higher cost would be explained by the fact that their production of implicit styles is not yet fully automatized, hence the constant need for conscious intervention. The higher frequency of verbs and pronouns could also present extra syntactic and morphological difficulties. The more explicit speech style might be heavier but it is probably more automatized (cf. Towell, Hawkins and Bazergui 1996; Towell and Dewaele 2005) and thus considered to be more “reliable” by the learner. A final – and probably related – explanation of the more explicit speech styles of the less proficient speakers is the type of input they have been exposed to and the type of output which they are expected to produce. Tarone and Swain (1995) labelled their IL speakers in French immersion classes as monostylistic: they used relatively formal speech styles and shun more vernacular styles. They avoided non-standard sociolinguistic variants, which reveals an incomplete pragmatic and sociolinguistic competence6 (Dewaele and Mougeon 2002; Mougeon and Dewaele 2004). Even if the learners know more vernacular forms and structures, they might still feel unsure about its appropriateness in certain situations. There might therefore be both a real, and a self-perceived, lack of sociopragmatic competence in the second language. While the real lack of competence is a matter of developing skills in the IL, the perceived lack depends much more on the personality of 6

Defined by Lyster (1994: 263) as the capacity to recognise and produce discourse that is socially appropriate in a given situation.

82 Jean-Marc Dewaele the learner. Learners in a classroom environment tend to transfer written styles to oral interaction creating an impression of lack of naturalness (Debrock, Flament-Boistrancourt and Gevaert 1999). A popular idea in research on bilingualism is that speakers need scripts in order to talk about various delicate and taboo subjects (Ranney 1992; D. Green 1998; Dewaele and Pavlenko 2002). It could be argued that the use of vernacular, implicit speech is also directed by specific scripts which are shared by native speakers (Kitayama and Markus 1994) but which are culture-specific. Hall (1976) has distinguished different types of cultures according to their dependence on context for communication. On the one hand low-context cultures like American and Northern European cultures where communication is more explicit and overt, stating the facts exactly and in detail; and, on the other hand, high-context cultures (Mediterranean and Eastern cultures), where communication is more implicit, and information is conveyed more by the context than by the verbal expression). The scripts probably include an indication of context-dependence. This indication is going to vary within a certain range in a particular culture as scripts are flexible and are adapted to the sociocultural and semiotic environment of the individual. Scripts which are used within one certain community of practice (Wenger 1998) are not necessarily shared by other groups but there is enough overlap in interpretive frames to coordinate practices and activities. Language learners probably quickly realize that their native vernacular scripts cannot be easily “translated” in the TL. Learners might be more aware of lexical differences in different speech styles of their native tongue and of the target-language than of conceptual differences. It is precisely at the level of conceptual representations that highly fluent non-native speakers may differ from equally fluent non-native speakers with higher levels of cultural competence. Pavlenko (1999, 2000) found that foreign language (FL) and L2 Russian learners of English were able to understand and define the language- and culture-specific American-English notions of “privacy” and “personal space”. This proved that both groups had semantic representations of the two words. However, a closer analysis revealed that only the L2 users, whose classroom learning was supplemented by authentic interactions, used these words in a manner similar to that of native speakers of American English. Pavlenko (1999, 2000) argues that only the L2 learners (and not the FL users) had linked the conceptual representations (imagery and scripts) to the semantic representations. A study by Dewaele and Regan (2001) on the use of colloquial words in the advanced French IL of 29 native speakers of Dutch and 6 Anglo-Irish speakers revealed very similar patterns: the IL contained significantly fe-

Variation in advanced oral interlanguage 83

fewer colloquial words than that of 6 native speakers French doing the same task. A multiple regression analysis revealed that the amount of classroom instruction did not have any predictive value on the use of colloquial vocabulary in advanced French IL. However, authentic communication in the TL was found to predict the use of colloquial vocabulary. It was hypothesised that less proficient speakers may simply not know the colloquial words, or lack the necessary morpho-phonological information at the lexical level. Incomplete semantic representation of the words could prevent the production of colloquial words in more advanced speakers. Even highly fluent speakers might lack information at the conceptual level; the semantic representation might not be linked to the TL concept, or the scripts where these words could appear might be incomplete or lacking. It seems that only prolonged authentic contact with the TL community might allow learners to develop implicit, proceduralised sociopragmatic knowledge. Dewaele (2004) pursued the investigation into the use of colloquial vocabulary using a corpus of interviews between 62 non-native and native speakers of French. Statistical analyses revealed a positive relation between the use of colloquial words and extraversion level, frequency of contact with French and proficiency level in French. Proficiency appears to be a prerequisite, but not the only factor, for actual use of colloquial vocabulary. Indeed, native speakers were found to use only marginally more colloquial words than non- native speakers I would like to argue that all scripts carry specific indications with regard to style /register and degree of explicitness. Native speakers will select a script to translate their communicative intentions and will automatically adapt the value for explicitness according to their perception of the situation. However, non-native speakers may experience difficulties at different levels. Firstly, the script in the TL could be absent or incomplete, and, secondly, the required explicitness value might be absent. The result could be conceptual transfer from the L1 or other ILs in order to make up for the missing scripts in the TL, and maybe also transfer of the explicitness value (register transfer). However, as I have no reason to believe that French and Dutch have different explicitness values for comparable scripts, I would argue that in the present study, the higher explicitness values for the less proficient speakers are not the result of register transfer, but rather a universal tendency among non native speakers to err on the side of caution. In other words, less proficient speakers, unsure about the exact explicitness value for a particular script, might automatically choose a value that is high enough to guarantee an unambiguous interpretation of their communicative intentions.

84 Jean-Marc Dewaele 8. Conclusion The quantitative analysis of individual scores of 32 French IL learners on the explicitness-implicitness dimension revealed a significant relationship between level of proficiency and speech style. Lower levels of proficiency are linked to more explicit speech styles. It was argued that in order to adopt a more implicit speech style, learners need to have reached a certain level of grammaticalisation in their IL because the use of implicit and deictical styles requires knowledge of a set of complex grammatical and syntactical rules. I found that learners using more explicit styles had had less formal instruction in the TL and used the TL less frequently outside the classroom. The learners’ lack of contact with the full range of speech styles in the TL – they had learnt French through formal instruction only – may be responsible for more limited conceptual representations. Crucially, explicitness values attached to certain scripts may be lacking. The default option in that case is to choose an explicitness value that is high enough to ensure an unambiguous interpretation of the communicative intentions. Those who use these explicit speech styles also underuse emotional and colloquial vocabulary, the most salient characteristics of vernacular speech styles.

References Argyle, Michael, Adrian Furnham and Jean Ann Graham 1981 Social Situations. Cambridge: Cambridge University Press. Azuike, Macpherson Nkem 1992 Style: theories and practical application. Language Sciences 14: 109– 127. Bar-Hillel, Yehoshua 1954 Indexical expressions. Mind 63: 359–379. Barnes Barry and John Law 1976 Whatever should be done with indexical expressions. Theory and Society 3: 223–237. Bartning, Inge 1997 L’apprenant dit avancé et son acquisition d’une langue étrangère. Tour d’horizon et esquisse d’une caractérisation de la variété avancée. Acquisition et interaction en langue étrangère (AILE) 9: 9–50. Bell, Allan 1997 Language style as audience design. In Sociolinguistics. A reader and coursebook, Nikolas Coupland and Adam Jaworski (eds.), 239–250. London: Macmillan Press.

Variation in advanced oral interlanguage 85 2001

Back in style: Reworking audience design. In Style and Sociolinguistic Variation, Penelope Eckert and John R. Rickford (eds.), 139–160. Cambridge, UK: Cambridge University Press. Biber, Douglas 1988 Variation across speech and writing. Cambridge: Cambridge University Press. 1994 An analytical framework for register studies. In Sociolinguistic Perspectives on Register Variation, Douglas Biber and Edward Finegan (eds.), 31–56. New York /Oxford: Oxford University Press. 1995 Dimensions of Register Variation. A Cross-Linguistic Comparison. Cambridge: Cambridge University Press. Biber, Douglas and Edward Finegan 1994 Register and social dialect variation: An integrated approach. In Sociolinguistic Perspectives on Register Variation, Douglas Biber and Edward Finegan (eds.), 315–347. New York /Oxford: Oxford University Press. Biber, Douglas, Susan Conrad and Randi Reppen 1994 Corpus-based approaches in applied linguistics. Applied Linguistics 15 (2): 169–185. 1998 Corpus linguistics. Investigating language structure and use. Cambridge: Cambridge University Press. Blanco-Iglesias, Susana, Joaquina Broner and Elaine Tarone 1995 Observations of language use in Spanish immersion classroom interactions. In The Current State of Interlanguage, Lynn Eubank, Larry Selinker and Michael Sharwood Smith (eds.), 241–254. Amsterdam / Philadelphia: John Benjamins. Chafe, Wallace L. 1982 Integration and involvement in speaking, writing, and oral literature. In Spoken and Written Language: Exploring orality and literacy, Deborah Tannen (ed.). Norwood, NJ: Ablex. Chini, Marina 2005 Reference to person in learner discourse. In The Structure of learner varieties, Henriette Hendriks (ed.), 65–110. Berlin / New York: Mouton de Gruyter. Coupland, Nikolas 2001 Language, situation and the relational self: Theorizing dialect-style in sociolinguistics. In Style and Sociolinguistic Variation, Penelope Eckert and John R. Rickford (eds.), 185–210. Cambridge, UK: Cambridge University Press. Cummins, Jim and Merrill Swain 1986 Bilingualism in Education: Aspects of theory, research and practice. London /New York: Longman.

86 Jean-Marc Dewaele Debrock, Marc, Danièle Flament-Boistrancourt and Raymond Gevaert 1999 Le manque de ‘naturel’ des interactions verbales du non-francophone en français. Analyse de quelques aspects à partir du corpus LANCOM. Faits de langue. Revue de linguistique: Oral – écrit: formes et théories 13: 46–56. De Jong, Evelyn 1979 Spreektaal. Woordfrequenties in gesproken Nederlands. Utrecht: Bohn, Scheltema and Holkema. Dewaele, Jean-Marc 1994 Variation synchronique des taux d’exactitude. Analyse de fréquence des erreurs morpholexicales dans trois styles d’interlangue française. International Review of Applied Linguistics 32: 275–300. 1995a Style-shifting in oral interlanguage: Quantification and definition. In The current state of interlanguage, Lynn Eubank, Larry Selinker and Michael Sharwood Smith (eds.), 231–238. Amsterdam /Philadelphia: John Benjamins. 1995b Variation synchronique dans l’interlangue: analyse critique du Modèle du Caméléon de E. Tarone. ITL Review of Applied Linguistics 109–110: 1–18. 1996a Variation dans la composition lexicale de styles oraux. International Review of Applied Linguistics 34: 261–282. 1996b Les phénomènes d’hésitation dans l’interlangue française: analyse de la variation interstylistique et interindividuelle. Rassegna Italiana da Linguistica Applicata 28: 87–103. 1998a The effect of gender of the choice of speech style. ITL Review of Applied Linguistics 119–120: 9–25. 1998b Speech rate variation in 2 oral styles of advanced French interlanguage. In Contemporary Approaches to Second Language Acquisition in social context: crosslinguistic perspectives, Vera Regan (ed.), 113– 123. Dublin: University College Academic Press. 1998c Lexical inventions: French L2 versus L3. Applied Linguistics 19: 471–490. 2001a Stéréotype à l’épreuve: le discours des femmes est-il plus émotionnel et déictique que celui des hommes? In La langue française au féminin. Le sexe et le genre affectent-ils la variation linguistique?, Kate Beeching, Cécile Bauvois and Nigel Armstrong (eds.), 123–147. Paris: L’Harmattan. 2001b Interpreting the maxim of quantity: interindividual and situational variation in discourse styles of NNS. In Cognition in Language Use: Selected Papers from the 7th International Pragmatics Conference, Vol. 1, Eniko Németh (ed.), 85–99. Antwerp: International Pragmatics Association. 2001c Une distinction mesurable: corpus oraux et écrits sur le continuum de la deixis. Journal of French Language Studies 11: 179–199.

Variation in advanced oral interlanguage 87 2003

Review of Penelope Eckert and John R. Rickford (eds.) (2001) Style and Sociolinguistic Variation. Journal of Language and Social Psychology 22: 234–238. 2004 Individual differences in the use of colloquial vocabulary: The effects of sociobiographical and psychological factors. In Learning Vocabulary in a Second Language: Selection, Acquisition and Testing, Paul Bogaards and Batia Laufer (eds.), 127–153. Amsterdam: Benjamins. 2007 Diachronic and/or synchronic variation? The acquisition of sociolinguistic competence in L2 French. In Handbook of French Applied Linguistics, Dalila Ayoun (ed.), 208–236. Amsterdam: Benjamins. Dewaele, Jean-Marc and Adrian Furnham 2000 Personality and speech production: A pilot study of second language learners. Personality and Individual Differences 28: 355–365. Dewaele, Jean-Marc and Raymond Mougeon (eds.) 2002 Appropriation de la variation en français langue étrangère. Special issue of Acquisition et Interaction en Langue Etrangère (AILE) 17. Dewaele, Jean-Marc and Aneta Pavlenko 2002 Emotion vocabulary in interlanguage. Language Learning 52: 265–324. Dewaele, Jean-Marc and Vera Regan 2001 The use of colloquial words in advanced French interlanguage. EUROSLA Yearbook 1: 51–68. Ely, Christopher M. 1986 An analysis of discomfort, risktaking, sociability, and motivation in the L2 classroom. Language Learning 36: 1–25. Fielding, Guy and Colin Fraser 1978 Language and Interpersonal Relations. In The Social Context of Language, Ivana Markova (ed.), 217–232. Chichester: J. Wiley. Finegan, Edward and Douglas Biber 2001 Register variation and social dialect variation: The register axiom. In Style and Sociolinguistic Variation, Penelope Eckert and John R. Rickford (eds.), 235–67. Cambridge, UK: Cambridge University Press. Foster, Pauline 1998 A classroom perspective on the negotiation of meaning. Applied Linguistics 19: 1–23. Giles, Howard and Peter F. Powesland 1975 Speech style and social evaluation. London: Academic Press. Givón, Talmy 1979 On understanding grammar. New York /London: Academic Press. Graddol, David and Joan Swann 1989 Gender Voices. Cambridge: Basil Blackwell. Green, David W. 1998 Mental control of the bilingual lexico-semantic system. Bilingualism: Language and Cognition 1: 67–81.

88 Jean-Marc Dewaele Green, Georgia M. 2000 The nature of pragmatic information. In Grammatical Interfaces in HPSG, Studies in Constraint-Based Lexicalism 8, Ronnie Cann, Claire Grover and Philip Miller (eds.), 113–138. Stanford: CSLI Publications. Grice, H. Paul 1975 Logic and conversation. In Syntax and Semantics, vol 3: Speech Acts, Peter Cole and Jerry L. Morgan (eds.), 41–58. New York: Academic Press. Hall, Edward T. 1976 Beyond Culture. New York: Anchor Press. Halliday, Michael Alexander Kirkwood 1989 Spoken and written language. Oxford: Oxford University Press. Harley, Birgit, Jim Cummins, Merrill Swain and Paul Allen 1990 The nature of language proficiency. In The Development of Second Language Proficiency, Birgit Harley, Paul Allen, Jim Cummins and Merrill Swain (eds.), 57–81. Cambridge: Cambridge University Press. Hasan, Rugaiva 1984 Ways of saying: ways of meaning. In The semiotics of Culture and Language, Vol. 1, Language as Social Semiotic, Robin Fawcett, Michael Alexander Kirkwood Halliday, Sydney Lamb and Adam Makkai (eds.), 105–162. London /Dover: Pinter. Heylighen, Francis 1999 Advantages and limitations of formal expression. Foundations of Science 4: 25–56. Heylighen, Francis and Jean-Marc Dewaele 2002 Variation in the contextuality of language: An empirical measure. Foundations of Science 7: 293–340. Housen, Alex 1998 Le processus de grammaticalisation dans le développement de l’interlangue. Travaux de linguistique 36: 209–222. Kitayama, Shinobu and Hazel Rose Markus 1994 Introduction. In Emotion and Culture. Empirical studies of mutual influence, Shinobu Kitayama and Hazel Rose Markus (eds.), 1–19. Washington: American Psychology Association. Labov, William 1972 Sociolinguistic Patterns. Pennsylvania: University of Pennsylvania Press. 1990 The intersection of sex and social class in the course of linguistic change. Language Variation and Change 2: 205–254. 1994 Principles of linguistic change. Oxford: Blackwell. 2001 The anatomy of style-shifting. In Style and Sociolinguistic Variation, Penelope Eckert and John R. Rickford (eds.), 85–108. Cambridge, UK: Cambridge University Press.

Variation in advanced oral interlanguage 89 Leckie-Tarry, Helen 1995 Language and context. A functional linguistic theory of register. London /New York: Pinter. Levelt, Willem J. M. 1989 Speaking. From Intention to Articulation. Cambridge, MA / London: ACL / MIT Press. Lyons, John 1977 Semantics. Cambridge: Cambridge University Press. Lyster, Roy 1994 The effect of functional-analytic teaching on aspects of French immersion students’ sociolinguistic competence. Applied Linguistics 15 (3): 263–287. 1998 Recasts, repetition, and ambiguity in L2 classroom discourse. Studies in Second Language Acquisition 20: 318–323. Moscovici, Serge and Claudine Humbert 1960 Etudes sur le comportement verbal. Langage oral et langage écrit. Psychologie française 3: 175–183. Mougeon, Raymond and Jean-Marc Dewaele (eds.) 2004 Patterns of variation in the interlanguage of advanced Second Language learners. Special issue of the International Review of Applied Linguistics 42 (4). Mougeon, Raymond, Terry Nadasdi and Katherine Rehner 2002 État de la recherche sur l’appropriation de la variation par les apprenants avancés du FL2 ou FLE. Acquisition et Interaction en Langue Etrangère (AILE) 17: 7–50. Nowson, Scott 2006 The Language of Weblogs: A study of genre and individual differences. Ph.D. thesis, University of Edinburgh. Nowson, Scott, Jon Oberlander and Alister J. Gill 2005 Weblogs, genres and individual differences. In Proceedings of the 27th Annual Conference of the Cognitive Science Society, 1666– 1671. Oberlander, Jon and Alister J. Gill 2006 Language with character: A stratified corpus comparison of individual differences in e-mail communication. Discourse Processes 42: 239– 270. Olson, David R. 1977 From utterance to text: The bias of language in speech and writing. Harvard Educational Review 47: 257–281. Pavlenko, Aneta 1999 New approaches to concepts in bilingual memory. Bilingualism: Language and Cognition 2: 209–230. 2000 What’s in a concept? Bilingualism: Language and Cognition 3: 31–36.

90 Jean-Marc Dewaele Poulisse, Nanda 1997 Language production in bilinguals. In Tutorials in Bilingualism: Psycholinguistic Perspectives, Annette M. B. de Groot and Judith F. Kroll (eds.), 201–224. Mahwah, NJ: Lawrence Erlbaum. Preston, Dennis 2001 Style and the psycholinguistics of sociolinguistics: The logical problem of language variation. In Style and Sociolinguistic Variation, Penelope Eckert and John R. Rickford (eds.), 279–304. Cambridge, UK: Cambridge University Press. Ranney, Susan E. 1992 Learning a new script: An explanation of sociolinguistic competence. Applied Linguistics 13: 25–50. Redeker, Gisela 1984 On differences between spoken and written language. Discourse Processes 7: 43–55. Richards, Jack, John Platt and Heidi Platt 1997 Dictionary of Language Teaching and Applied Linguistics. London: Longman. Rickford, John R. 2001 Style and stylizing from the perspective of a non-autonomous sociolinguistics. In Style and Sociolinguistic Variation, Penelope Eckert and John R. Rickford (eds.). 220–231. Cambridge, UK: Cambridge University Press. Rickford, John R. and Faye McNair-Knox 1994 Addressee – and topic– influenced style shift. A quantitative sociolinguistic study. In Sociolinguistic Perspectives on Register Variation, Douglas Biber and Edward Finegan (eds.), 235–276. New York /Oxford: Oxford University Press. Selinker, Larry 1992 Rediscovering interlanguage. London: Longman. Tarone, Elaine 1983 On the variability of interlanguage systems. Applied Linguistics 4: 143–163. 1988 Variation in interlanguage. London: Edward Arnold. 1997 Analyzing IL in natural settings: A sociolinguistic perspective on second-language acquisition. Communication and Cognition 30: 137– 149. 2000 Still wrestling with ‘context’ in interlanguage theory. Annual Review of Applied Linguistics 20: 182–198. Tarone, Elaine and Merrill Swain 1995 A sociolinguistic perspective on second language use in immersions classrooms. Modern Language Journal 79: 166–178.

Variation in advanced oral interlanguage 91 Towell, Richard, Roger Hawkins, and Nives Bazergui 1996 The development of fluency in advanced learners of French. Applied Linguistics 17: 84–119. Towell, Richard and Jean-Marc Dewaele 2005 The role of psycholinguistic factors in the development of fluency amongst advanced learners of French. In Focus on French as a Foreign Language: Multidisciplinary Approaches, Jean-Marc Dewaele (ed.), 210–239. Clevedon: Multilingual Matters. Wenger, Etienne 1998 Communities of practice: learning, meaning, and identity. Cambridge: Cambridge University Press.

The use(fulness) of corpus research in cross-cultural pragmatics: Complaining in intercultural service encounters Ronald Geluykens and Bettina Kraft

1. Introduction Within cross-cultural pragmatics, there is a growing awareness of the necessity to get away from collecting data through controlled elicitation procedures, and in particular through production questionnaires such as Discourse Completion Tasks (DCTs). Due to the highly influential pioneering work of the Cross-Cultural Speech Act Realization Project (CCSARP; see BlumKulka et al. 1989), DCTs have become the much favoured data collection procedure within cross-cultural and interlanguage pragmatics. In recent years, however, questions have been raised, both with regard to the scope of CCP research (which was largely focused on the realization of speech acts) and with regard to the type of empirical data employed for pragmatic analysis (see Geluykens 2007 for an extensive review). These days, it is at least implicitly acknowledged that a combination of data from different sources may yield more reliable results. However, while such a combination of data collected through different means, also referred to as data triangulation, may be generally considered the way forward within CCP, there is still a shortage of empirical studies putting this principle into practice. In section 2, we will review some recent studies dealing with this issue. One thing that also becomes clear, even in studies that use more than one procedure, is that data triangulation does not necessarily imply the use of spontaneous, uncontrolled data. Nevertheless, when one wants to investigate the linguistic implications of some pragmatic factor, it should be clear that this cannot be done without some reliance on spontaneously occurring discourse data. Assuming that the objective of CCP is to uncover what interactants do in actual situations of use, controlled elicitation data cannot, by themselves, yield reliable results, since one can never be sure that whatever correlation is found is not the by-product of the data collection procedure. In other words, it would be very dangerous indeed to extrapolate findings based on artificial data to other contexts. Data triangulation does

94 Ronald Geluykens and Bettina Kraft not necessarily provide an answer here, since converging results from two ‘artificial’ data procedures (say, DCTs and roleplays) does not necessarily confirm greater reliability. It would appear, then, that using more than one data collection procedure does not, per definition, provide any added value, unless at least one data set consists of more authentic material. It is only through comparing the artificial data with authentic material that the researcher can assess whether any regularities found in the former are significantly different from those encountered in the latter. This raises the question as to the potential sources for such authentic material; it seems rather obvious that one potential source for authentic material lies in using corpora of spontaneous speech material. The remainder of the paper is organized as follows: In the following section, we will review some recent empirical CCP research which attempts to combine and/or compare different data collection methods. In doing so, we will show that these attempts have been fragmentary, and have very rarely included a corpus of spontaneous interaction. In section 3, we will go on to discuss the potential usefulness of existing English corpora for cross-cultural research, and argue that they are generally not very suitable for the detailed (comparative) investigation of interactional phenomena, especially not when dealing with low-frequency, confrontational phenomena such as complaint sequences. We will then present, in section 4, a qualitative analysis of complaints in service encounters based on our own database of TV documentary material, and illustrate through one extensive example why there is no substitute for such authentic material. Finally, in section 5, we will discuss the wider implications of our findings for CCP methodology.

2. Data triangulation in cross-cultural pragmatics: A brief review Even in relatively early DCT-based research, a growing awareness can be observed of the limitations of data collected through DCTs. Some authors have specifically attempted to evaluate the usefulness of DCTs by comparing them to other elicitation methods, ranging from slight variants to completely different formats. Some methodologically oriented studies are geared specifically towards testing the use of different types of DCTs, for instance oral versus written DCTs. Others attempt to evaluate whether responses elicited through DCTs systematically resemble controlled data elicited through other methods, in particular, role-plays.

The use(fulness) of corpus research in cross-cultural pragmatics

95

Of course, this does not answer the question as to how ‘authentic’ DCTelicited speech is, given that role-plays are also non-authentic. A third category of comparative studies that therefore deserves being mentioned consists of those comparing DCTs with naturally occurring speech. And, finally, there are a few studies that attempt to validate more than two elicitation methods empirically. This brief discussion does not attempt to give an exhaustive overview of all possible data procedures; we will merely focus on some attempts to verify their usefulness empirically. For more complete reviews, mention must be made of the overviews of research methods provided in Kasper & Dahl (1991) and, more recently, Kasper (2000). In a comparative study which belongs to the CCSARP-based studies reported in Blum-Kulka et al. (1989), Rintell & Mitchell report on a contrastive analysis based on DCTs and role-plays. Their intention was to assess to what extent the two procedures would produce comparable data sets. They conclude that the elicited language is in fact “very similar whether collected in written or oral form” (Rintell & Mitchell 1989: 270). However, as Rose (1996) points out, they downplay the differences which they do encounter, in that significant differences in the use of some strategies were found between the two data sets. More puzzlingly, a significant difference could be observed between native speaker and non-native speech production in the DCT and role-play data, in that the native data were found to be much more similar than the non-native data. In other words, the two data elicitation methods appear to have affected the two populations in different ways. This is worrying, in that it calls the comparability of L1 and L2 data into question. As Rose remarks, “most of the ILP research (…) has been based on the implicit assumption that the instruments used for data collection produce comparable data for both populations” (Rose 1996: 115). Should this basic assumption turn out to be incorrect, the entire rationale for using such controlled data would turn out to be invalid. Similar problems are recorded in Eisenstein & Bodman’s study (1986) of expressions of gratitude, in which they note that, while both native and non-native speakers sometimes felt uncomfortable with some DCT scenarios employed, native speakers “were always able to respond appropriately to the situations presented, whereas non-natives were not always able to do so” (Eisenstein & Bodman 1986: 173). Rose (1992a, 1992b) reports on an investigation in which two forms of DCTs were administered to a population of native American English users, the ultimate goal being the testing of the reliability of using triangulation rather than the classification of actual speech production. One type of DCT included hearer responses (HR), the other type did not (NoHR). Rose con-

96 Ronald Geluykens and Bettina Kraft cludes that “although responses on the NoHR DCT tended to be slightly longer and use slightly more supportive moves and downgraders, inclusion of hearer response did not have a significant effect on requests elicited” (Rose 1992a: 49). While this may seem to indicate that it does not matter much whether HR or NoHR DCTs are used; therefore, it does not answer the fundamental question of whether either of these elicitation methods yields speech production that resembles authentic speech patterns in any way. In another study, Rose (1994) compares an ‘open’ DCT study and one employing ‘closed’ multiple choice questionnaires (MCQ), as part of a contrastive project on American English and Japanese, and concludes that “DCTs may be inappropriate for collecting data on Japanese” (Rose 1994: 1). A follow-up study by Rose & Ono (1995) found similar, significant differences in request realizations in MCQ and DCT data, the MCQ data exhibiting more opting out strategies and more indirect realizations. A similar comparison between DCT and MCQ data is offered by Hinkel (1997) in a cross-cultural study comparing advice giving produced by native (American) English speakers to that of (highly proficient) Chinese English interlanguage speakers. Hinkel finds that, while the MCQ results are more or less congruent with the existing body of research (Chinese respondents producing more options in their advising, both of the direct and hedged varieties), the situation in the DCTs is, in fact, reversed. Needless to say, this casts serious doubts on the reliability and degree of authenticity of the production data obtained in either data set. Sasaki (1998) used production questionnaires and role-plays for investigating two types of (L2 English) FTAs, requests and refusals, produced by Japanese students at three different levels of proficiency. Not surprisingly, both data sets do not completely converge: “Role plays induced longer responses, and a larger number and greater variety of strategies/formulas, than production questionnaires” (Sasaki 1998: 457). Sasaki explains this difference in production in terms of the interactional properties of the roleplays, a plausible enough explanation, given that the same respondents switched strategies when confronted with a different elicitation procedure. Additionally, native speaker evaluations of the responses in terms of their appropriateness correlated only minimally; Sasaki suggests that these differences may be attributable, at least in part, to the additional audio-visual information present in the (recorded) roleplays, but it does raise the question as to whether either of the elicitation methods employed yielded appropriate results in the first place, from a perception point of view. Sasaki’s conclusion that “production questionnaire scores cannot be simply substi-

The use(fulness) of corpus research in cross-cultural pragmatics

97

tuted for role play scores” (Sasaki 1998: 457), although true, misses the point, and does not answer the question as to which method yields the most authentic speech production. A combination of DCT data and roleplays can also be found in Trosborg’s (1995) detailed study of requests, apologies and complaints. While the studies mentioned above primarily discuss the merits and drawbacks of different types of controlled eliciation data, there are other studies which have compared data obtained through DCTs with more or less naturally occurring data. In a study on native and non-native speaker rejections, Hartford & Bardovi-Harlig (1992) found significant differences between their DCT data and naturally occurring rejections as found in university advising sessions, mostly with regard to the type, range and relative frequencies of the semantic formulas used. Similarly, in their study on refusals, Beebe & Cummings (1995) got comparable differences for refusals collected via DCTs and via telephone recordings. Nevertheless, they argue that DCTs can be useful for creating an initial classification of the semantic formulas used in the realization of Face Threatening Acts (FTAs). To what extent such a DCT-based classification would then be applicable is open to debate, however, given the diverging results in uncontrolled data. The study by Yuan (2001) investigates compliments and compliment responses in Mandarin Chinese. She compares written DCTs, oral DCTs, field notes, and transcripts of oral interviews, analyzing and contrasting a large sample of speech production. Unsurprisingly, while the oral DCTs share some of the drawbacks of the written DCTs (cf. supra), they nevertheless generate a significantly larger number of ‘natural’ speech features than the written DCT does. While observational fieldwork notes enjoy the unarguable advantage of being realistic, the time lapse between the speech event and its transcription means that the actual wording may not be one hundred percent accurate. Perhaps the most important conclusion Yuan draws is that “the choice of a data gathering method for a particular study should be made based on the research questions and objectives of the researcher” (Yuan 2001: 271). Golato (2003) provides one of the most recent and most detailed studies comparing naturally occurring data with DCT data, while also providing a critical evaluation of other methods of data elicitation, such as roleplays, field observation, and recall protocols. The empirical study focuses on compliment responses in German that were elicited partially through DCTs and partially through recordings of naturally occurring talk-in-interaction. The spoken corpus consists of 6 hours of telephone and 25 hours of face-to-

98 Ronald Geluykens and Bettina Kraft face conversations. These data were compared to written questionnaires completed by 30 respondents. Golato’s investigation confirms the observations made by others about the different nature of data elicited through DCTs, and comes to the, hardly surprising or controversial, conclusion that “a DCT is not an on-line task in which a person uses language spontaneously and without consciously focusing on linguistic output, but is instead an off-line task in which a person has time for introspection” (Golato 2003: 110). As a result, she continues, naturally occurring data is much better suited for finding out about turn-taking mechanisms and the organization of language, but questionnaires can be helpful tools for learning more about people’s perceptions of their language behaviour. However, “DCTs do not cleanly and reliably inform us as to how talk-in-interaction is organized and realized in natural settings” (Golato 2003: 111). However, Golato’s paper does not inform us as to how these two data collection methods could be usefully combined through triangulation. While the studies reviewed above are helpful in that they reveal the shortcomings of relying on one single data collection method, they do not really offer a way to successfully integrate different data types. In particular, one can find precious little information on whether a combination of authentic and controlled data is desirable and, if so, how such added value would be generated. One thing appears uncontroversial, however, viz. the analysis of authentic talk in interactional settings being a prerequisite for any claims about the pragmatic choices interactants make in real life. 3. The usefulness of (existing) corpora An insistence on employing naturally occurring interactional data does not necessarily entail that the researcher personally carries out a fieldwork investigation. In theory at least, use could be made of existing discourse corpora. These could be either relatively small-scale corpora developed for a specific research purpose, or larger-scale corpora developed for more general use. The main problems associated with small corpora collected for specific purposes seem clear. First of all, as they are limited in size, they might not contain a sufficient number of tokens of the linguistic feature under investigation to enable the CCP researcher to carry out a systematic quantitative investigation (though they could then obviously still be used for qualitative analysis). Secondly, the fact that they were developed with other research objectives in mind may put serious limits on their usability.

The use(fulness) of corpus research in cross-cultural pragmatics

99

For instance, limitations in the transcription system may prevent us from identifying our intended variables. A final problem, which they share with larger databases, is that such corpora were rarely developed for the explicit purpose of doing cross-cultural or even contrastive research. In most cases, they contain one single variety of one single language (English being collected most often, either in its native form or in some interlanguage variant). Even if one were to find another corpus of similar size in another language, variety, or interlanguage, it is very unlikely that this corpus would have been collected under similar circumstances or transcribed using similar techniques. This problem is even more outspokenly present in larger-scale corpora, for the simple reason that (i) fewer of them exist, and (ii) the fact that they were collected for more general use means that their potential usefulness for (cross-cultural) pragmatic purposes might be even more limited. The internal organization of such a large-scale database will be influenced to a large extent by the fact that the compilers are attempting to collect a representative sample of various discourse genres, both written and spoken. This is, for instance, the case for the British National Corpus (BNC), the largest database currently available for (L1) English, where some care was taken not only to compile a sociologically representative sample (in terms of regional dialects, gender, age, and social class of the respondents), but also to collect a varied sample of different discourse modes (both spoken and written, both interactional and not). Potential drawbacks for the use of interactional spoken data may further include a less than ideal transcription system for pragmatic analysis. In the London-Lund corpus, for instance (see Svartvik & Quirk 1980), turn taking and overlapping are transcribed rather awkwardly. Furthermore, interactions are not always recorded surreptitiously (the BNC data are consistently non-surreptitious, for instance), which makes it hard to judge the potential impact of the researcher on the interaction (where the researcher is one of the interactants, which is often the case in the London-Lund data), This also means that the speakers’ awareness of being recorded might change their speech production, and one is thus confronted with Labov’s observer’s paradox: “the aim of linguistic research in the community must be to find out how people talk when they are not being systematically observed; yet we can only obtain these data by systematic observation”. (Labov 1972: 209) There is some evidence, however, that people’s awareness of being recorded wears off a little after getting used to having recording devices around (Kasper 2000: 319ff). An additional drawback of using existing corpora for sociolinguistic purposes is the fact that background information

100 Ronald Geluykens and Bettina Kraft on the speakers may be very limited, which makes it very hard to measure the effect of even very ‘traditional’ social variables such as gender, power and distance, let alone more specific discourse-sensitive ones. What makes matters worse is that the original recordings are generally not made publicly available, usually for reasons of confidentiality (there are exceptions, such as the Santa Barbara Corpus of American English, which is also available online; see DuBois et al. 2000). And once again, even disregarding these issues, it is extremely unlikely that similar or comparable corpora can be found for two different languages, for native and non-native versions of the same language, or for different cultural varieties of the same language (though, at least for English, some progress is being made with regard to the last-mentioned aspect; witness the International Corpus of English compiled at University College London, cf. Greenbaum 1996). It would appear, then, that the CCP researcher has little choice but to compile his/her own database, or at the very least employ one which was set up specifically with cross-cultural purposes in mind. In this regard, mention should be made of corpora of interlanguage which have emerged in recent years, such as the International Corpus for Learner English (ICLE) for written language and the Leuven International Database of Spoken English Interlanguage (LINDSEI) for spoken discourse, both started at by the Université Catholique de Louvain (see Granger 1993, Granger et al. 2002, among others). In most cases, however, such corpora are, once again, not ideal for doing contrastive research (i.e. comparing L1 realizations across cultures) or for studying intercultural interactions between native speakers and interlanguage users. Compiling a corpus of spontaneous intercultural interactions is a labour-intensive enterprise, and presupposes some ethnographic fieldwork, at the very least. This, of course, explains to a large extent why early studies have generally relied on less time-consuming controlled elicitation data in the first place, even in the face of a growing awareness of the limitations and inherent problems associated with DCTs. One way of reducing the effort involved in collecting data could be to employ data that have been recorded for other purposes. Candidates would be, essentially, anything recorded for radio, television, or other audio-visual media. Once again though, one must tread very carefully here: while there is a huge amount of data material, even interactional material, potentially available, not all of it is equally suitable for ethnographic or ethnomethodological analysis. First of all, a significant proportion of this material is institutional discourse, such as political debates, legal discourse, and the like. While this is not a problem per se (institutional talk being a legitimate object of study in its own right, cf. Drew & Heritage 1992, and many others), one should

The use(fulness) of corpus research in cross-cultural pragmatics

101

of course be very careful when trying to extrapolate regularities found in this type of talk to spontaneous, informal conversation. Secondly, what looks like spontaneous interaction might actually be scripted, either entirely or partially (and this might not be easy to determine, due to a lack of background information). It should be noted here that scripted dialogue might still have its uses, not just as baseline data, but also as an aid in constructing DCT scenarios and developing video-prompted elicitation procedures and metapragmatic judgment tasks. However, one must keep in mind that this type of data does not constitute authentic interaction by any stretch of the imagination. A promising candidate as a data source for more spontaneous interaction is documentary material, in particular the interactions provided in fairly new television formats such as docusoaps (see also Kraft 2007 for a discussion and application to complaint sequences). Documentary soaps (Kilborn et al. 2001) or fly-on-the-wall documentaries (Bruzzi 2001) are recent genres that have enjoyed huge commercial success. Fly-onthe wall documentaries purport to simply record everyday events as they unfold, without any script or manipulation, the impression being created that “everything you see really happened just as if the camera was not there at all” (Bruzzi 2001: 128). This is perhaps a somewhat naïve assumption, given the outspoken presence of the camera, and the influence that the documentary maker has in terms of choosing which events to record and how to edit them. The events themselves, however, are not initiated or influenced by the filmmaker. The term documentary soap, or docusoap for short, is more recent, and suggests a hybrid of documentaries and soap operas, and hence perhaps a less spontaneous, more edited format which aims at maximizing entertainment value. The two terms are often used interchangeably, however. Drawbacks include: (i) the fact that often several narrative strands are interwoven, so that it might be hard to follow an interaction from beginning to end; and (ii) the presence of the camera, which could potentially affect people’s behaviour significantly (though it is probable that the effect of the camera’s presence erodes over longer stretches of speech). On the other hand, presence of the camera may prompt people to offer metacomments on their own (verbal) behaviour. The biggest advantage seems to be that we are indeed dealing with unrehearsed, unscripted interaction, often involving situations for which ethnographic data might be hard to obtain (e.g. social conflicts; cf. Kraft & Geluykens 2008). At any rate, this is a potential data source worth pursuing further. In the following section, we will demonstrate how such data may be usefully combined with controlled data for enhancing the reliability of an analysis.

102 Ronald Geluykens and Bettina Kraft 4.

Complaints in Service Encounters

4.1. Definitions Complaints are complex speech acts which usually occur as substrategies of conflict talk. They are infinitely more complex than, for instance, apologies or requests, as they do not have clearcut components which are discernible as the complaint proper. A complaint can be uttered in various manners, and can consist of various turns containing other speech acts. Therefore, it is necessary to take this discursive nature of complaints into account when setting up research into complaint behaviour. It is important to note that complaints in the private domain are different in nature and goal-orientation from complaints in service encounters. In the past two decades, pragmatic research has taken an increasing interest in complaints (Cupach & Carson 2002; Frescura 1995; Geluykens & Kraft 2003, 2007; House & Kasper 1981; Kraft & Geluykens 2002, 2004; Kuha 2003; Laforest 2002; Murphy & Neu 1996; Newell & Stutman 1989; Olshtain & Weinbach 1993; Tatsuki 2000; Trosborg 1995). The focus, however, has been on private complaints; furthermore, most of the data in these studies are taken from DCTs and role-plays. This section focuses on complaints as part of conflict-solving processes in service encounters, using data from a corpus of authentic interactions. Service encounters take place in institutional settings which directly and unmistakably shape the interaction. The institution directly determines the participants’ roles, as pointed out by Thornborrow: “(…) institutional discourse can perhaps be best described as a form of interaction in which the relationship between a participant’s current institutional role (that is, interviewer, caller to a phone-in programme or school teacher) and their current discursive role (for example, questioner, answerer or opinion giver) emerges as a local phenomenon which shapes the organisation and trajectory of the talk” (Thornborrow 2002: 5). This presupposes that the institution puts constraints on participants’ behaviour from the onset (they have a specific role within or vis à vis the institution) and in their ongoing discursive role within any given conversation. In our corpus, this translates into the participants being the company representative and a customer, with clearly defined discursive roles, where the customer usually has a problem and/or reason to complain, and the company representative is expected to solve the problem and defuse the customer’s complaint. The institutional frame not only positions and aligns the interactants (Thornborrow 2002: 4), but also serves to put constraints on their possible contributions within the interaction (Drew & Heritage 1992: 22). The institution does not only determine the roles and

The use(fulness) of corpus research in cross-cultural pragmatics

103

role relationships of the interlocutors, however: it also entails that communication will be mainly oriented to a specific task: “(...) in these various contexts for talk there is an orientation towards a specific task – the business of the talk as it unfolds is to ask questions, to provide answers (or to resist providing them), to have a discussion, to make a complaint, amongst others” (Thornborrow 2002: 3). Institutional discourse therefore is mostly goaloriented or transactional, and the interpersonal relationship between the interactants is of secondary importance. In our corpus, the interactions happen for the purpose of solving a problem, with the interactants usually meeting for the first and probably last time, and therefore their interpersonal relationship figuring low on their agenda. What is true for institutional discourse in general is even more relevant for service encounters, which are interactions that are strictly transactional and goal-oriented in nature. “Service encounters are clearly drawn instances where participants’ goals, be they more or less complementary, have to be achieved by means of verbal interaction” (Iacobucci 1990: 85). The verbal interaction in service encounters as observed in our corpus has the sole purpose of attaining the interactional goal, which in the first instance means to solve the problem at hand. As our focus lies on conflict talk, however, the participants’ goals are usually not complementary: most of the negotiation in conflict talk, and most of the complaints uttered therein, revolve around the problem that the participants’ goals are at odds. The challenge thus consists in finding a suitable compromise that makes both parties attain their goal at least approximately. A slightly more comprehensive definition of service encounters is given by Merritt (1976: 321), according to whom service encounters are an instance of face-to-face interaction between a server who is “officially posted” in some service area and a customer who is present in that service area, that interaction being oriented to the satisfaction of the customer’s presumed desire for some service and the server’s obligation to provide that service. Although at first glance Merritt’s definition of service encounters seems to encompass all their facets, we find that his postulation that they are face-to-face interactions does not always hold true. Especially in recent years, more and more service encounters happen via media such as the telephone or even the internet. This of course also affects the second aspect of Merritt’s definition, according to which service encounters have to be conducted in a clearly defined space, namely in some service area, with both the server and the customer physically present. In service encounters which are conducted over the telephone, usually only the service provider is present on the company premises (although even this is not a prerequisite), whereas the customer probably calls

104 Ronald Geluykens and Bettina Kraft from his home or place of work. We therefore suggest broadening the definition of service encounters to include telephone conversations and putting less emphasis on spatial orientations (thus making the term ‘encounter’ somewhat inappropriate). Our main focus lies on conflict resolution in service encounters, and more specifically, on customer complaints, which are one subcategory of conflict talk. Although within pragmatics there is an increasing interest in complaints, most of these studies deal with private complaints, many with a focus on interlanguage (e.g. Arent 1996; DeCapua 1989, 1998; Geluykens & Kraft 2003, 2007; Kraft & Geluykens 2002, 2004; Nakabachi 1996; Tamanaha 2003; Trenchs 1995), and only very few investigating a professional context (Brünner 2000; Clyne et al. 1991; MoralesLópez et al. 2005; Trosborg 1998). Complaints in service encounters in particular have received very little attention in linguistics (Bendel 2001; Morales-López et al. 2005; Zabava Ford 2001) but are of great importance in marketing (e.g. Forsyth 1999; Kelley & Davis 1994; Mattson et al. 2004; Maxham III & Netemeyer 2002; Shaw 2001). Complaints in service encounters therefore remain underinvestigated from a linguistic perspective. 4.2. Findings from Controlled Elicitation Data Complaints are part of conflict discourse, inherently highly face-threatening, and as such they are difficult to study systematically, due to their relative infrequency. For this reason, many studies dealing with complaints rely almost exclusively on constructed data. However, complaints are interactive in nature, and it seems near impossible to look at them from a speech-act perspective as utterances in isolation. The various stages of the complaint event depend to a large extent on the reactions of the interlocutors, their choice of conversational strategies, and how they position themselves in an exchange. By investigating complaints within the greater complex of conflict exchanges, we want to shed light on the question how complaint exchanges evolve in service encounters and which typical strategies in the realization of complaints can be observed. Compared to other speech acts, such as for example requests or apologies, complaints are more complex and difficult to define. Moreover, it is often difficult to delineate a complaint from related concepts such as criticisms or reproaches. A complaint can be identified as an utterance or set of utterances, which identifies a problem or trouble source and seeks remediation, either from the person responsible for the trouble source, or a third party who has the power to affect the situation (Schaefer 1982: 8). In a service encounter this usually entails the customer

The use(fulness) of corpus research in cross-cultural pragmatics

105

as the one who is annoyed by the trouble or problem, and who is the one who seeks remediation. Complaints do not consist of one particular utterance or ‘head act’ (Blum-Kulka et al. 1989); as a result, it is often difficult to determine which utterances actually qualify as complaints. In the analysis of our corpus, one of our aims is to find out which strategies can be identified as prototypical complaint strategies, as with complaints we can observe almost no fixed lexical expressions. Unlike other speech acts, complaints almost never co-occur with a performative verb (‘I hereby complain’), and IFIDs (Illocutionary Force Indicating Devices) are very rarely used as an initiation (‘I would like to complain’). We approach our data from a cognitive-pragmatic point of view. Cognition plays an important role determining the expectations participants bring to a particular encounter. The main assumption in this context is that the ’complaint-in-a-service-encounter frame’ (Goffman 1974) triggers a mental representation of events and actions in the participants’ minds and that they will behave accordingly. The participants in this speech event, therefore, will match their expectations and actions to previous similar experiences and adapt their actions accordingly. However, in service encounters, members of the institution in question often seem to have expectations and views which are competitive with those of the customers. This may lead to (further) conflict and misunderstandings. In service encounters, the participants‘ roles are predefined (customer vs. company representative), there is inequality in terms of power, and the interactants usually have clearly defined task-oriented goals which they bring to the interaction. In the course of the interaction, these parameters may change and realign in subtle ways, depending on the choice of discourse strategies the interactants make. It seems to be imperative, when analyzing institutional discourse, to take the situational context, the role relationship and the interactants’ goals and expectations into account. A close sequential analysis of utterances as well as of facial expressions, gestures, and spatial alignment, reveals the inner mechanisms at work in conflict exchanges. Discourse completion tasks cannot, realistically, provide any insight into the discursive nature of complaints, as they consist of one turn only. They are very useful, however, in revealing prototypical complaint strategies and lexical choices. Comparing the DCT data with naturally occurring complaints, we find very similar conversational strategies which are typical for complaints. As the DCT was developed before the other parts of the corpus were gathered, the situational context is slightly different, but still comparable. The subjects were asked to respond to scenarios such as the following:

106 Ronald Geluykens and Bettina Kraft You are at an expensive restaurant. After you have waited three quarters of an hour for your meal, it turns out to be cold and much too salty. You stop the waiter and say to him: Amongst the reactions which could be evaluated as complaints, we found the following general strategies (in isolation or in combination with each other, allowing for several different permutations): – Use of apologetic expression (as politeness marker and attention-getter): “excuse me” – Statement of the problem: “it‘s cold and quite frankly too salty” – Request for remedial action: “can you do something about this meal?” – Request to speak to the manager: “can I speak to the management” – Expression of anger/dissatisfaction: “I‘m not happy with this at all” – Threat: “I‘m not paying for it.” These strategies, uttered in a context where a complaint is warranted, serve as a complaint, whether they appear in isolation or in combination with any number of the other strategies. The DCT is a useful tool if one wants to find out what people regard as typical strategies and expressions in certain contexts (see Kraft & Geluykens 2002, 2004; Geluykens & Kraft 2003, 2007 for full details). This mainly reveals what they regard to be the norm, i.e. what counts as appropriate behaviour. Bearing in mind that DCTs cannot reveal actual (interactive) speech behaviour, the advantages of this data collection method are well known: DCTs allow relatively easy access to a large number of test subjects, the researcher can rigidly control the conditions, and, most importantly, DCTs allow for large-scale cross-cultural comparisons, which would otherwise be nearly impossible. However, because of the restricted, non-interactional context in which these complaints were elicited, any findings cannot necessarily be extrapolated to actual language use. In the following, therefore, we will analyze an extract from our corpus of naturally occurring complaints. We will see how two factors aggravate the conflict: on the one hand this is due to the institutional context and the conflicting goals of the participants; on the other hand, friction arises because of different cultural backgrounds which apparently involve different expectations concerning politeness and appropriate behaviour.

The use(fulness) of corpus research in cross-cultural pragmatics

107

4.3. Cross-Cultural Interaction in a TV Data Corpus: An Illustration The intercultural interactions in our corpus involve conversations between native speakers of English and native speakers of other languages (predominantly French, Greek, Arabic, and Italian), the latter using interlanguage English. Their foreign language background seems to influence the interaction and quite often even tends to aggravate the existing conflict. This is mostly due to the fact that these speakers, despite being fully conversant in English, tend to employ strategies derived from their native language. Such transfer may lead to misunderstandings and misinterpretations of speaker intentions. Here especially, a multi-modal analysis is vital in understanding the mechanisms at work in cross-cultural conflict interactions, as the above-mentioned problems seem to be rooted in suprasegmental features, intonation, gestures, etc. far more than in the actual verbal utterances. With speakers at an advanced level of proficiency, any cross-cultural misunderstandings are barely discernible, and usually open to misinterpretation by the interlocutors themselves. What is judged as perfectly appropriate behaviour by the interlanguage user might be perceived as rude by the native speaker, without either of them noticing the reasons for their evaluative judgment. In the following, we will discuss an example from our data, in which the customer (Nicolas), a native speaker of French, has to deal with two British EasyJet employees (Becky and Leo). Apparently Nicolas’ flight had been delayed, but by less than the half hour which was originally announced. Nicolas thought it would be acceptable just to wait for half an hour and then come back for check-in. Meanwhile, the delay has turned out to be shorter, and he has missed his flight. Now Nicolas wants a refund, because he thinks that he missed the flight through no fault of his own. The airline, on the other hand, thinks it is Nicolas’ own fault that he missed the flight, and as a general rule does not give any refunds. The following extract represents the first 22 turns of the exchange: 01 Nicolas: ((slapping on the desk with his hand)) I need a refund I need a refund. 02 Becky:

we don’t do refunds [we don’t do refunds in our company.

03 Nicolas: ((with raised eyebrows, looking indignant)) you don’t do refunds] ok (.) what about our luggage 04 Becky:

it’s up to passengers to watch the screen for departures=

108 Ronald Geluykens and Bettina Kraft 05 Nicolas: ((agitated, raising his voice, leaning in, pointing with his left hand index finger)) = NO it’s not up it’s not fair 06 Leo:

((sitting at next desk, extending his arm towards the customer; very loud voice)) EXCUSE ME

07 Nicolas: !no I’m very sorry sir you’re not involved ((slapping the desk with his newspaper))you’re dealing with this people!= 08 Leo:

!I’m the manager! ((with outstretched right hand))

09 Nicolas: [I don’t care I] 10 Leo:

[speak to Becky like this] ((pointing to the door with his right hand index finger)) and you’ll be out of the [airport straight away]

11 Nicolas: ((not shouting anymore)) [I’m not talking not nicely] to the [lady I’m explaining the situation] 12 Leo:

[I’m I’m the assistant manager]

13 Nicolas: it does [not say anywhere-] 14 Leo:

[excuse me]

15 Nicolas: we should keep checking the screen 16 Leo:

[Excuse me]

17 Nicolas: [when it said] half an hour late it’s now half an hour late 18 Leo:

ok would you like to go tomorrow morning?

19 Nicolas: I can't I have to be there tonight 20 Leo:

I can't get you there tonight

21 Nicolas: I have to be there tonight 22 Leo:

I can't help you

When assessing the actual conversation, we find that, throughout, Nicolas appears to come across as highly aggressive and threatening. This, however, is not due to the inherent content of his utterances, but rather to his non-verbal cues, in particular his body language and his tone of voice. In general, it seems that in France for customers it is normal and advisable to approach a conflict situation like this one more aggressively than a native speaker of British English would (based on informal French native speaker judgments). The main conflict in the first part of the interaction is that Nicolas and the EasyJet personnel have different evaluations of what is going on and what obligations they have vis à vis the other. Nicolas thinks that it is EasyJet’s fault that he missed the flight, and believes it to be their

The use(fulness) of corpus research in cross-cultural pragmatics

109

obligation to either give him a refund (turn 1) or put him on the flight tonight (as he wants to reach his destination on that same day, not the following day, turns 19 and 21). His request for a refund is turned down by Becky, one of the EasyJet employees, who explains to him that EasyJet do not give any refunds to anybody (turn 2). Nicolas’ entire body language expresses indignation and anger at this, and he then proceeds to ask what is going to happen to their luggage (turn 3), which they had checked in already. Obviously he is worried now that his luggage might travel without him. Becky chooses not to enlighten him about the luggage, but indirectly blames Nicolas for what is going on by pointing out to him that it is ‘up to passengers to watch the screen for departures’ (turn 4). This clearly upsets Nicolas even more, mainly because he disagrees with Becky’s assessment of the situation. According to him this is clearly not up to the passengers. His indignation prompts him to raise his voice and to cut her off (turn 5), and to point out that he feels treated unfairly. His indignation and anger are very clearly expressed in his body language, as he leans onto the counter and moves his arm towards Becky. Leo, Becky’s colleague, who is sitting on the left of her, perceives this as very menacing. He feels it his responsibility to come to her rescue by telling the customer off. It is vital here to point out that Leo’s interference was not triggered so much by what the customer was saying, but rather by the manner in which he delivered it (raised voice, agitated body language), all of which was perceived by Leo as menacing (although it is doubtful whether he himself was aware of that). Leo first intervenes with a loud and forceful ‘excuse me’ (turn 6), which is interpreted by Nicolas as an intrusion upon his conversation with Becky. He points out to Leo that he is ‘not involved’ because he is dealing with other people at the moment (turn 7). Leo in his next turn justifies his interference and shows his authority by pointing out that he is the manager (turn 8), although it later transpires that he is only an assistant manager (turn 12). Nicolas is clearly unimpressed by Leo’s status (cf. ‘I don’t care’ in turn 9), a reaction which prompts Leo to utter a very direct threat: ‘speak to Becky like this and you’ll be out of the airport straight away’ (turn 10). This seems to sober Nicolas up slightly, and he reacts defensively by saying that he is merely ‘explaining the situation’ (turn 11). This explanation is crucial to understanding what exactly is happening in this interaction. Nicolas clearly feels that his behaviour is perfectly reasonable and non-threatening. He is not aware of the threatening effect his raised voice and agitated body language have on Becky and her colleague Leo. Nicolas himself judges his behaviour as perfectly polite, as he is ‘not talking not nicely’ (turn 11), by which he means that he is not using any

110 Ronald Geluykens and Bettina Kraft swear words or any kind of personal attack on Becky. This shows that impoliteness can be conveyed in various ways, with tone of voice and body language contributing significantly. This part of the interaction concludes with Nicolas further explaining his point of view. He thinks that he is not at fault, as he trusted a half-hour delay to be 30 minutes, not less (turns 13–17). Several important points should be noted here. First of all, when one closely examines the verbal strategies employed in this entire exchange, the types of strategies turn out to be surprisingly similar to the ones found in the DCT data. Even a cursory glance at the data reveals striking similarities, such as the following: – – – – –

Use of apologetic expression: “I’m very sorry sir“ (turn 7) Statement of the problem: “I need a refund” (turn 1) Request for remedial action: “you’re gonna have to do something” (turn 31) Expression of anger/dissatisfaction: “it’s not fair” (turn 5) Threat: (inherent in body language and intonation)

However, these similarities with regard to the lexical and syntactic choices made by the speaker actually mask profound differences on other levels. Firstly, as was already pointed out, the authentic data reveal a level of emotional involvement which is, of necessity, missing from the controlled data. Such involvement is mostly marked through prosodic, paralinguistic and non-verbal means such as intonation, loudness, and body movements. Secondly, it goes without saying that the sequential ordering within this highly confrontational exchange is relevant: each step in the negotiation process is, to some extent, shaped by what precedes, and in turn influences what follows. This also means that the speaker’s actual complaint chain is partially dependent on the addressee’s contributions. Thirdly, the fact that this is an institutional context, in which the participants involved have asymmetrical roles, cannot be ignored; in particular, the differences with regard to power, with the customer being completely dependent on the service provider for a successful outcome, has an impact on how the conflict develops. And finally, the fact that we are dealing here with an intercultural encounter means that the interactants have different expectations as to what can be regarded as the socially acceptable norm for such service encounters. While space does not permit us to go into all this in more detail, we hope it is clear that, at the very least with regard to these four dimensions – emotional involvement, sequential ordering, interactional asymmetry, and intercultural confrontation – we need a corpus of authentic interactional material to put the findings from our DCT material into a larger context. What

The use(fulness) of corpus research in cross-cultural pragmatics

111

is more, it shows that it would be dangerous to restrict oneself to an exclusively quantitative analysis of complaint features, since the pragmatic implications of such features are dependent on the context in which they occur; this is true both in terms of the macro-context (e.g. service encounters) and the micro-context (i.e. the local sequential organization of the complaint exchange). 5. Conclusions (Im)politeness is not inherent in certain utterances, but negotiated in the course of any given interaction, and ratified by the interactants. Clearly, not only what people say, but the way they say something, and the way in which an utterance is interpreted by the interlocutor(s), are all factors which contribute to the impact of an utterance and any subsequent utterances. In order to make truly informed judgments about how people communicate, a multimodal analysis of actual speech behaviour seems to be imperative. Such an analysis presupposes (i) a reliable corpus of authentic material, and (ii) an accurate and detailed transcription of both verbal and non-verbal features in that corpus. This does not mean that other data collection methods cannot provide valuable insights into norms and perceptions of language use. It is important, however, for researchers to be aware of the intrinsic limitations of these other procedures. It is our belief to date that the way forward in cross-cultural pragmatics is to employ a mixed method approach to data collection and data analysis (Geluykens 2007). We maintain that constructed data, such as for example DCTs, provide useful insights into prototypical linguistic representations, in this instance of complaints, as well as an invaluable opportunity to make cross-cultural comparisons. Naturally occurring data, on the other hand, reveal the influence emotion and involvement have on actual speech behaviour. A combination of data elicitation methods therefore helps understand people‘s perceptions, expectations and actions. This does not mean, incidentally, that triangulation should be restricted to combining data collecting procedures: ideally, a true mixed method approach should also involve a combination of both quantitative and qualitative methodologies (with corpus research allowing, at least in principle, for both methodologies). Once again, we refer the reader to the extensive discussion in Geluykens (2007) for more details. We suggest using DCTs only on the basis of previously observed, realistic scenarios, and to avoid constructed or invented situations. It is important

112 Ronald Geluykens and Bettina Kraft to bear in mind, when using DCTs as a data collection tool, that they can only provide non-authentic, non-interactive data. However, DCTs remain useful when the object of research is to find out about appropriateness and norm judgments of speakers, as well as for insights into prototypical realizations of certain strategies. For complaints, DCTs help reveal people’s evaluations of what a typical complaint realization in a given context might look like. Furthermore they can be useful in assessing L2 users’ proficiency and pragmatic awareness for specific contexts, as well as in systematic crosscultural comparisons of speaker norms and evaluations. In pragmatics research, however, analyzing natural speech is still the only way to find out about people’s actual speech behaviour. Only authentic behaviour can show the mechanisms at work when people communicate. We found that, especially for research into complaints, it is most useful to work with a corpus of multi-modal data that provides additional information such as facial expressions, spatial alignment, and meta-comments. Therefore we found the relatively new TV format of docusoaps a useful source for actual complaint behaviour, allowing us to investigate complaints in intercultural encounters, produced by both native speakers and English interlanguage users, and the emotional and motivational background that led to it.

Transcript Notation [] = CAPS ! utterances! * Italics ? . ! (.), (..), (…) :, ::, ::: But((coughs)) (only) (xl)

overlap latching (no interval between turns) louder speech stretches of louder speech quieter speech emphasis intonation rising intonation falling animated tone pause (short, medium, long) elongation of utterance (short, medium, long) abrupt cut-off point false start transcriber’s comments or descriptions uncertain/unintelligible utterances untranscribable material

The use(fulness) of corpus research in cross-cultural pragmatics

113

References Arent, R. 1996

Sociopragmatic decisions regarding complaints by Chinese learners and NSs of American English. Hong Kong Journal of Applied Linguistics 1: 125–147. Bardovi-Harlig, K. 1993 Refining the DCT: Comparing open questionnaires and dialogue completion tasks. Pragmatics and Language Learning 4: 143–165. Beebe, L. M. and M. C. Cummings 1996 Natural speech act data versus written questionnaire data: How data collection method affects speech act performance. In Speech Acts Across Cultures. Challenges to Communication in a Second Language, S. M. and J. N. Gass (ed.), 65–86. Berlin /New York: Mouton de Gruyter. Bendel, S. 2001 Die interaktive Bearbeitung von Servicefehlern: Problemgespräche und Gesprächsprobleme zwischen Gästen und Angestellten an der Hotelreception. Gesprächsforschung. Online-Zeitschrift zur verbalen Interaktion 2: 115–140. Blum-Kulka, S., J. House and G. Kasper (eds.) 1989 Cross-Cultural Pragmatics: Requests and Apologies. Norwood, NJ: Ablex. Brünner, G. 2000 Wirtschaftskommunikation. Linguistische Analyse ihrer mündlichen Formen. Tübingen: Max Niemeyer. Bruzzi, S. 2001 Observational (‘fly-on-the-wall’) documentary. In The Television Genre Book, Vol. XI, G. Creeber, T. Miller and J. Tulloch (eds.), 129– 131 London: British Film Institute. Clyne, M., M. Ball and D. Neil 1991 Intercultural Communication at Work in Australia: Complaints and Apologies in Turns. Journal of Cross-Cultural and Interlanguage Communication 10 (3): 251–274. DeCapua, A. 1989 An Analysis of Pragmatic Transfer in the Speech Act of Complaints as Produced by Native Speakers of German in English. Unpublished doctoral dissertation, Columbia University. 1998 The Transfer of Native Language Speech Behavior into a Second Language: A Basis for Cultural Stereotypes? Issues in Applied Linguistics 9 (1): 21–35.

114 Ronald Geluykens and Bettina Kraft Drew, P. and J. Heritage 1992 Analyzing Talk at Work: An Introduction. In Talk at Work. Interaction in Institutional Settings, P. Drew and J. Heritage (eds.), 3–65. Cambridge: Cambridge University Press. DuBois, J. W. et al. 2000–05 Santa Barbara Corpus of Spoken American English (Parts 1–4). Philadelphia: Linguistic Data Consortium. Eisenstein, M. and J. Bodman 1986 ‘I very appreciate’: Expressions of gratitude by native and non-native speakers of American English. Applied Linguistics 42: 167–185. Forsyth, P. 1999 Communicating with Customers. London: Orion. Geluykens, R. 2007 On methodology in cross-cultural pragmatics. In Cross-Cultural Pragmatics and Interlanguage English, B. Kraft and R. Geluykens (eds.), 21–72. München: Lincom Europa. Geluykens, R. and B. Kraft 2003 Sociocultural variation in native and interlanguage complaints. In Meanings in Contrast: The Cambridge Papers, Vol. 2, K. M. Jaszcolt and K. Turner (eds.), 251–261. Amsterdam: Benjamins. 2007 Gender variation in native and interlanguage complaints. In CrossCultural Pragmatics and Interlanguage English, B. Kraft and R. Geluykens (eds.), 143–158. München: Lincom Europa. Goffman, E. 1974 Frame Analysis. An Essay on the Organization of Experience. Boston: Northeastern University Press. Golato, A. 2003 Studying compliment responses: A comparison of DCTs and recordings of naturally occurring talk. Applied Linguistics 24 (1): 90–121. Granger, S. 1993 The international corpus of learner English. In English Language Corpora: Design, Analysis and Exploitation, J. Aarts, P. de Haan and N. Oostdijk (eds.), 57–69. Amsterdam: Rodopi. Granger. S., E. Dagneaux and F. Meunier 2002 The International Corpus of Learner English. Handbook and CDROM. Louvain-la-Neuve: Presses Universitaires de Louvain. 24 Greenbaum, S. (ed.) 1996 Comparing English Worldwide: The International Corpus of English. Oxford: Clarendon Press. Hartford, B. S. and K. Bardovi-Harlig 1992 Experimental and observational data in the study of interlanguage pragmatics. In Pragmatics and Language Learning Monograph Series, Vol 3., L. F. Bouton and Y. Kachru (eds.), 33–52. Urbana, IL: University of Illinois at Urbana-Champaign.

The use(fulness) of corpus research in cross-cultural pragmatics Hinkel, E. 1997

115

Appropriateness of advice: DCT and multiple choice data. Applied Linguistics 18: 1–26.

Iacobucci, C. 1990 Accounts, Formulations and Goal Attainment Strategies in Service Encounters. In Multiple Goals in Discourse, K. Tracy and N. Coupland (eds.), 85–99. Clevedon / Philadelphia: Multilingual Matters. Kasper, G. 2000 Data collection in pragmatics research. In Culturally Speaking. Managing Rapport through Talk across Cultures, H. Spencer-Oatey (ed.), 317–341. London /New York: Continuum. Kasper, G. and M. Dahl 1991 Research methods in interlanguage pragmatics. Studies in Second Language Acquisition 13: 215–247. Kelley, S. W. and M. A. Davis 1994 Antecedents to customer expectations for service recovery. Journal of the Academy of Marketing Science 22 (1): 52–61. Kilborn, R., M. Hibberd and R. Boyle 2001 The rise of the docusoap: The case of vets in practice. Screen 42 (4): 382–395. Kraft, B. 2007 Conflict Management and Complaints in Service Encounters. Mimeo, University of Southampton. Kraft, B. and R. Geluykens 2002 Complaining in French L1 and L2: A cross-linguistic investigation. In EUROSLA Yearbook, Vol. 2, S. Foster-Cohen, T. Ruthenberg and M.-L. Poschen (eds.), 227–242. Amsterdam: Benjamins. 2004 Beschwerdeverhalten in Muttersprache und Lernersprache: Eine kontrastive Analyse der Sprachen Deutsch und Französisch. In Beiträge zu Sprache und Sprachen 4. Vorträge der Bochumer Linguistik-Tage, K. Pittner, R. J. Pittner and J. Schütter (eds.), 257–265. München: Lincom. 2007 Defining cross-cultural and interlanguage pragmatics. In Cross-Cultural Pragmatics and Interlanguage English, B. Kraft and R. Geluykens (eds.), 3–20. München: Lincom Europa. 2008 Complaint sequences in cross-cultural service encounters. In Institutional Discourse in Cross-Cultural Contexts, R. Geluykens and B. Kraft (eds.), 221–242. München: Lincom Europa. Kryk-Kastovsky, B. 1999 On data collection and More: Problems in synchronic and diachronic pragmatics. In Form, Function and Variation in English. Studies in Honour of Klaus Hansen, U. Carls and P. Lucko (eds.), 361–372. Frankfurt a. M.: Peter Lang.

116 Ronald Geluykens and Bettina Kraft Labov, W. 1972

Sociolinguistic Patterns. Philadelphia: University of Pennsylvania Press. Margalef-Boada, T. 1994 Research methods in interlanguage pragmatics: an inquiry into data collection procedures. Dissertation Abstracts International (DAI), 55 (2). Mattson, J., J. Lemmink and R. McColl 2004 The effect of verbalized emotions on loyalty in written complaints. Total Quality Management 15 (7): 941–958. Maxham III, J. G. and R. G. Netemeyer 2002 Modelling customer perceptions of complaint handling over time: The effects of perceived justice on satisfaction and intent. Journal of Retailing 78: 239–252. Merritt, M. 1976 On questions following questions (in service encounters). Language in Society 5: 315–357. Morales-López, E., G. Prego-Vázquez and L. Domínguez-Seco 2005 Interviews between employees and customers during a company restructuring process. Discourse Studies 16 (2): 225–268. Nakabachi, K. 1996 Pragmatic transfer in complaints: Strategies of complaining in English and Japanese by Japanese EFL speakers. JACET Journal 27: 127–142. Rintell, E. and C. J. Mitchell 1989 Studying requests and apologies: An inquiry into method. In CrossCultural Pragmatics, S. Blum-Kulka, J. House and G. Kasper (eds.), 248–272. Norwood, NJ: Ablex. Rose, K. R. 1992a Speech acts and questionnaires: The effect of hearer response. Journal of Pragmatics 17: 49–62. 1992b Experimental and observational data in the study of interlanguage pragmatics. In Pragmatics and Language Learning 3, L. Bouton and Y. Kachru (eds.), 35–52. Urbana, IL: University of Illinois. Rose, K. 1994 On the validity of discourse completion tests in non-western cultures. Applied Linguistics 15: 1–14. Rose. K. 1996 Film in Interlanguage Pragmatics Research. Paper presented at the Second Language Research Forum (SLRF), University of Arizona, Tucson. Rose, K. and R. Ono 1995 Eliciting speech act data in Japanese: The effect of questionnaire type. Language Learning 45: 191–223.

The use(fulness) of corpus research in cross-cultural pragmatics Sasaki, M. 1998

117

Investigating EFL students’ production of speech acts: A comparison of production questionnaires and role plays. Journal of Pragmatics 30: 457–484.

Schaefer, E. 1982 An analysis of the discourse and syntax of oral complaints in English. Unpublished M.A. Dissertation, University of California, Los Angeles. Shaw, P. 2001 The intercultural validity of customer-complaint handling routines. Document-Design: Journal of Research and Problem-Solving in Organizational Communication (DoDe) 2 (2): 180–193. Svartvik, J. and R. Quirk (eds.) 1980 A Corpus of English Conversation. Lund: Gleerup. Tamanaha, M. 2003 Interlanguage Speech Act Realization of Apologies and Complaints: The Performances of Japanese L2 Speakers in Comparison with Japanese L1 and English L1 Speakers. Teachers College. Thornborrow, J. 2002 Power Talk. Language and Interaction in Institutional Discourse. London: Longman. Trenchs, M. 1995 Pragmatic strategies in Catalan and English complaints: A comparative study of native and EFL speakers. The Language Quarterly (LangQ) 33 (3–4): 160–182. Trosborg, A. 1995 Interlanguage Pragmatics. Berlin: Mouton de Gruyter. 1998 “Sorry” does not pay my bills. The handling of complaints in everyday interaction and cross-cultural business interaction. Hermes. Journal of Linguistics 21: 67–94. Yuan, Y. 2001 An inquiry into empirical pragmatics data-gathering methods: Written DCTs, oral DCTs, field notes, and natural conversations. Journal of Pragmatics 33: 271–292. Zabava Ford, W. S. 2001 Customer expectations for interactions with service providers: Relationship versus encounter orientation and personalized service communication. Journal of Applied Communication Research 29 (1): 1–29.

Hesitation markers among EFL learners: Pragmatic deficiency or difference? Gaëtanelle Gilquin

I guess what I’m trying to say is, I don’t think you can measure life in terms of years. I think longevity doesn’t necessarily have anything to do with happiness. I mean, happiness comes from facing challenges and going out on a limb and taking risks. If you’re not willing to take a risk for something you really care about, you might as well be dead. Diane Frolov and Andrew Schneider, Northern Exposure, Northern Lights, 1993

1. Introduction A couple of years ago, during a course on “Language and Thought” taught by Professor Herbert Clark at Stanford University, the audience was asked to transcribe a short sentence extracted from a recorded conversation. The sentence was simple enough: “I think that the situation that you describe has gone completely out of hands”. At least, this is how several people (myself included) transcribed the sentence. In reality, the recorded sentence included vocalisations and other (parts of) words which had escaped our notice. What happened was that we concentrated on the semantic content of the sentence, but did not pay attention to what we considered mere interference, namely the markers of hesitation present in the sentence. This account will probably sound familiar to anyone who has had experience with verbatim transcriptions of spontaneous speech. Since vocalisations, false starts, repetitions and other “smallwords”1 such as well or I mean “do not contribute essentially to the message itself” (Hasselgren 2002: 150), they tend to be disregarded. Not only are they disregarded by hearers, in fact, but also, until quite recently, by specialists of language (for an exception to this fact, see Maclay & Osgood’s study, dating from 1959). Talking about smallwords, Hasselgren (2002: 168) refers to an “essential but hitherto largely neglected body of language”. Spoken corpora, by giving access to 1

The term “smallword” is borrowed from Hasselgren (2002).

120 Gaëtanelle Gilquin detailed transcriptions of authentic speech, have made it possible to study hesitation phenomena with a precision and reliability that were practically unattainable before. Taking advantage of the availability of spoken corpora, and more precisely of a newcomer to the field, the spoken learner corpus, which contains samples of non-native speech (NNS), this paper sets out to investigate the function of hesitation among EFL (English as a Foreign Language) learners with French as a mother tongue, using as a baseline the way the function is performed in native speech (NS). The paper is structured as follows. First the function of hesitation in speech is briefly introduced. Next, the two corpora on which the study is based (LOCNEC and LINDSEI-FR) and the three categories of hesitation phenomena investigated (pauses, smallwords and other devices) are described. The following two sections present the main results of the corpus-based analysis and discuss these results in the light of Foreign Language Teaching (FLT) and English as a Lingua Franca (ELF). Section 7 concludes the paper.

2. The pragmatic function of hesitation in speech Spoken language, in its purest form, is unrehearsed and spontaneous (see e.g. Burns & Joyce 1997). Interactants build speech as they go along, in a process of on-line planning. As a result, there are times in a conversation when the speaker is inevitably hesitant and does not know what to say next (or how to express it). Such hesitation may manifest itself in various ways. Wiese (1984), for example, mentions filled pauses (e.g. uh, mhm), repetitions, corrections and drawls. But the literature also recognises a number of smallwords which, among other functions, allow the speaker to “buy time” (e.g. well, I mean or vague words such as stuff or things like that). Here, I will adopt a very broad definition of hesitation, which covers (silent and filled) pauses, drawls, truncated words, repetitions, as well as a representative selection of smallwords of hesitation which commonly appear in the literature.2 The function of hesitation is crucial as a conversational strategy. Since speech is dialogic in nature (Burns & Joyce 1997: 13), it is important that a speaker should indicate that s/he needs a moment’s reflection, but is still 2

Corrections (or “repairs”) will only be dealt with as a special case of word truncation, when the speaker starts a word and then goes on to produce another word (e.g. to a few p= bars). Other types of corrections (e.g. when the speaker starts with a particular structure and then produces a different one) are not discussed.

Hesitation markers among EFL learners

121

“in control” of his/her turn. Hesitation markers, by signalling a small delay, ensure that the speaker can keep his/her turn in the conversation and is not interrupted by the other participants. As Larcombe (1995) nicely puts it, “[i]n order to keep the floor while we hesitate, we place dummy words in the empty spaces between our words, much as we might drape our coats on a seat at the cinema to prevent others from taking it”. Even silent pauses have been demonstrated to play a part in the structure of the message and to contribute to its internal cohesion (Romero-Trillo 1994). For foreign (or second) language learners, hesitation is even more crucial. In their search for a formulation which is acceptable in the foreign language, they are likely to experience many planning problems and, therefore, need techniques that enable them to gain time while they are trying to solve these problems.3 Paradoxically, while hesitation is an inherent characteristic of spontaneous conversations and fulfils an important pragmatic function, it goes largely unnoticed in speech (at least native speech).4 Talking about the transcription of recorded spoken data, McCarthy (1998: 13) notes that even the best audio typists “often simply ‘miss’ relevant details” (he gives the example of repetitions) and that they can be “deaf to the presence of discourse markers and other ‘little’ words which become important the moment one starts to analyse the work they do in the creation of interaction” (see Lindsay & O’Connell 1995 for a clear demonstration of this). Watts (1989), considering the opposite perspective, that of the speaker him-/herself, shows that one may be unaware of the extent to which one uses such little words. This lack of salience explains why corpora are an ideal resource for the investigation of hesitation markers. Provided they have been transcribed carefully and with some detail, spoken corpora give access to a record of authentic language, including various types of disfluency phenomena, which can be queried automatically and analysed quantitatively and qualitatively. An analysis based on intuition, by contrast, would very probably underestimate the role of hesitation in speech. In the next section, I describe the two corpora that were used in this study.

3

4

De Cock’s (2003, 2004) study of recurrent sequences of (two or more) words in native and learner speech has highlighted the particularly large proportion of sequences containing hesitation markers among learners. See Chambers (1997) or Temple (2000) on the inequality between native and non-native speakers in this respect, and the fact that disfluency phenomena are more easily stigmatised in non-native speech.

122 Gaëtanelle Gilquin 3. The corpora In order to investigate the use of pragmatic markers of hesitation by EFL learners, I used a corpus of interviews with advanced French-speaking learners of English, and a comparable corpus of native speech which served as a baseline for the study of the NNS data. The NNS corpus, LINDSEI-FR, is the French component of LINDSEI (Louvain International Database of Spoken English Interlanguage). The LINDSEI project, launched in 1995 at the University of Louvain (UCL) and resulting from the collaboration between a number of universities internationally, aims at gathering oral data produced by advanced learners of English. To date, 11 mother tongue backgrounds are represented: Bulgarian, Chinese, Dutch, French, German, Greek, Italian, Japanese, Polish, Spanish and Swedish. All the components follow the same format so as to make the data comparable. They consist in transcriptions of interviews between a non-native advanced 5 learner of English and a (native or non-native) interviewer, recorded with the consent of the participants. Each interview lasts about 15 minutes and takes place in three stages. First, the subject is asked to talk for a few minutes about a topic which s/he has chosen6 and has had some time to think about before the interview starts (without taking any notes). The conversation then continues informally, with the interviewer asking questions related to the topic chosen by the subject and questions about more general topics (hobbies, life at university, travels abroad, etc). Finally, the subject is presented with four pictures making up a story and is required to recount what happens. Subjects also have to fill in a learner profile questionnaire with information such as gender, number of years of English or knowledge of other foreign languages. The interviews are transcribed orthographically but include indications of pauses (short, medium or long), fillers (eh, er, erm), truncated words, syllable lengthening, overlapping speech, etc.7 Each component of LINDSEI contains the transcription of 50 5

6

7

The proficiency level is established on the basis of an external criterion, namely the number of years of English at university (subjects were in their third or fourth year). The topic is chosen among the following three: (i) an experience you have had which has taught you an important lesson; (ii) a country you have visited which has impressed you; (iii) a film/play you have seen which you thought was particularly good/bad. The transcription, however, has its limitations when it comes to the (automatic) retrieval of certain hesitation phenomena. While simple programs can be written

Hesitation markers among EFL learners

123

interviews, for a total of over 100,000 words per component. The corpus has been marked up in such a way that the sentences uttered by the learner (B turns) can be queried independently of those uttered by the interviewer (A turns), and this is how the corpus will be used here. At this stage, the corpus has not been annotated, but there are plans to have it POS-tagged. In parallel with LINDSEI, a comparable corpus of native English, LOCNEC (Louvain Corpus of Native English Conversation), has been compiled (see De Cock 2003).8 Like the LINDSEI components, it contains 50 interviews made up of three parts (set topic, free discussion, picture description), which were transcribed using the same conventions. The interviewees were all native speakers of British English and most of them were undergraduate students in linguistics or English language. In this study, LOCNEC was used as a control native corpus. By comparing the LINDSEIFR data with data produced by native speakers, it was possible to identify cases where French-speaking learners’ use of hesitation markers diverges from that of native speakers, either quantitatively or qualitatively. Table 1 shows the number of interviews found in each corpus, LOCNEC and LINDSEI-FR, as well as the total number of words for interviewer’s and interviewee’s turns (A and B turns) and for interviewee’s turns only (B turns only). The results presented in this paper are limited to the B turns. Table 1. Number of interviews and words in LOCNEC and LINDSEI-FR Nr. of interviews

Nr. of words (A and B turns)

Nr. of words (B turns only)

LOCNEC

50

170,533

125,226

LINDSEI-FR

50

149,127

94,406

8

to extract exact repetitions (e.g. he knows he knows), cases of retracing with modification (e.g. he know he knows…) are not amenable to such treatment. See Osborne (2006) for an oral learner corpus, the PAROLE corpus, annotated using the CHILDES system (MacWhinney 2000) and encoded with several hesitation phenomena, including retracings. At the moment, access to LINDSEI and LOCNEC is restricted to the project collaborators. However, we expect to release a first version of LINDSEI on CDROM in 2009.

124 Gaëtanelle Gilquin 4. Three categories of hesitation markers and their extraction For the purposes of the analysis, the function of hesitation was divided into three main categories, namely (silent and filled) pauses, smallwords and a miscellaneous category. All markers, except repetitions, were extracted automatically using the program WordSmith Tools (Scott 1999). When necessary, the hits were manually disambiguated. Silent pauses, which are defined as gaps in the utterance, are probably the most basic way of dealing with problems of formulation. Not knowing what to say, the speaker just remains silent. As pointed out by Fillmore (1979), silent pauses are multifunctional, since they both have a rhetorical function and serve as a marker of disfluency. Because it is almost impossible to identify with any certainty cases where the pause merely has a rhetorical function, however, all silent pauses were taken into account in the analysis. Following the LINDSEI transcriptions, a distinction was made between short pauses (under one second), indicated in LINDSEI by means of one dot, medium pauses (one to three seconds), indicated by two dots, and long pauses (over three seconds), indicated by three dots. An example is provided in (1). (1)

I was there for ages and . again I thought right I can’t do this so I went out of Breda again

Alternatively, pauses can be filled by vocalisations, sounds such as er or erm, as shown in (2).9 Discussing uh and um (the American spelling variants of er and erm, respectively), Clark & Fox Tree (2002: 75) note that they are “characteristically associated with planning problems”, being used by speakers to announce a delay in speaking. In the LINDSEI transcriptions, the following “fillers” are recognised: eh (for a brief sound), em, er, erm and mm. Each of them was extracted automatically from the corpora. (2)

yeah I went to see erm another one Jane Eyre and that was really bad it was a a real amateur production

A number of smallwords can also be used to signal hesitation, as exemplified in (3) with kind of. Other examples include well, defined by Fuller (2003: 187) as “a delay device when the speaker is not sure how to respond”, I mean, 9

White (1997) counts as filled pauses cases of syllable lengthening (drawls). Here, however, drawls will be considered as a category of their own and dealt with as one the miscellaneous devices (see below).

Hesitation markers among EFL learners

125

which can be used “when pausing to think about what you are going to say next” (Longman Dictionary of Contemporary English, Summers 1995: 886), and vague words such as stuff and or something, which can be used to fill knowledge gaps or lexical gaps (Drave 2002: 26; see also Channell 1994). Most of these words are multifunctional (see Schiffrin 1987: 61 or Aijmer 2002: 19ff). You know, for instance, may be used “when you need to keep someone’s attention, but cannot think of what to say next” (Summers 1995: 781), but it may also function as “a speaker appeal for hearer cooperation in a discourse task” (Schiffrin 1987: 63). Often, several pragmatic functions are performed simultaneously by one and the same word and it may prove extremely difficult to disentangle them, despite the help of the surrounding context. Therefore, no attempt was made to disambiguate such cases in the corpus data. By contrast, some smallwords have, next to their pragmatic function(s), a non-pragmatic meaning which can clearly be identified. Kind of, for example, illustrated in (3) in its pragmatic function, may also be used with a non-pragmatic function, as a synonym of “type of”, cf. (4). Similarly, well has both pragmatic and non-pragmatic functions. Consider the sentence in (5). While the first two occurrences signal hesitation, the third one does not, functioning as an adverb together with as. Whenever a smallword was used with a non-pragmatic meaning, hence ruling out the function of hesitation, it was discarded manually from the hits returned by the automatic extraction. (3)

as I say during the year I kind of changed my mind about what I wanted to do

(4)

it produces just the most fantastic kind of sweets

(5)

yeah well I do enjoy it so I just thought well I might as well do something I enjoy

In the miscellaneous category, finally, I included drawls (i.e. syllable lengthening), truncated words and repetitions, which are all signals that the speaker is hesitating (see for example Fox Tree & Clark 1997 on drawls, Temple 2000 on truncations and Wiese 1984 on repetitions). Drawls and truncated words were extracted from the corpora by searching for the symbols used in the transcriptions, namely a colon for drawls and an equals sign for truncated words. Repetitions were identified automatically by means of a program written specifically for this purpose.10 Examples of drawl, truncated word and repetition are given in (6) to (8). 10

I thank Marie-Catherine de Marneffe for her help in this matter.

126 Gaëtanelle Gilquin (6)

they were brilliant so we went skiing and we went to: various parts of France

(7)

they’ve had to rethink and they’re coming back in their f= forties even in their fifties

(8)

but the last time I got organised and went to Spar first and bought some some supper to eat during the film

Once the different hesitation markers were extracted from the corpora and, when necessary, disambiguated, their frequency in the learner corpus was compared with that in the native corpus. It was thus possible to identify phenomena of overuse and underuse, i.e. cases where learners use significantly more or significantly less of a particular item than native speakers. Statistical significance was tested by means of the chi-square test and probability values of less than 0.05 were considered significant (but most of the results presented are actually significant at the 0.001 level). This quantitative approach was supplemented by a more qualitative analysis, which highlighted differences in use between native and learner English.

5.

The hesitation function in native and learner speech

This section presents the main results of the analysis for the three categories of hesitation markers outlined above, namely (silent and filled) pauses, smallwords and a miscellaneous category.

5.1. Pauses Silent pauses are very frequent, as appears from Table 2, especially short pauses (under one second), which represent the most common device for signalling hesitation, both among native speakers and learners. Medium pauses (one to three seconds) are slightly less frequent and long pauses (over three seconds) are comparatively rare. Learners, however, make heavier use of silent pauses than native speakers, in a way which is statistically significant. This is true of the three lengths of pauses. The words after which pauses typically occur are relatively similar in NS and NNS. In the top four list we find filled pauses (er and erm), yes (or yeah in NS) and and. This is illustrated in (9) to (11) for native speech.

Hesitation markers among EFL learners

(9)

127

he m= he was erm .. a lecturer at I think it’s Oxford it might have been Cambridge one of one of the[i:] Oxbridge universities

(10) yes yeah . you you don’t really think about anything else you’re just sort of sitting there thinking oh gosh (11) I think she acted very well and . again you really felt for her and erm .. you could kind of imagine yourself in her shoes Table 2. Relative frequency of silent pauses in native and learner speech (per 100,000 words)11 LOCNEC

LINDSEI-FR

X²

+/–

Short pause (.) Medium pause (..) Long pause (…)

2087.43 1636.24 54.30

4391.67 3153.40 295.53

958.04 554.10 198.57

+ + +

Total

3777.97

7840.60

1702.24

+

Like silent pauses, the category of filled pauses is, generally speaking, more characteristic of learner speech than of native speech, as shown in Table 3. The individual fillers follow the same tendency, except for erm, which is underused by learners. It is interesting to note that, of all the items listed in Table 3, learners prefer eh and er, which come closer to the normal filled pause in French, euh.12 The fillers involving “m”, by contrast, are less frequent in learner speech.

11

12

In this table and the following, a plus sign signals a case of significant overuse, while a minus sign signals a case of significant underuse. Sometimes, the values were too small to perform a chi-square test; this is indicated by means of “n.a.” (“non-applicable”). In fact, going back to the original sound files, it turns out that the pronunciation of these fillers is often closer to French euh than to English eh or er. This is in line with Clark & Fox Tree (2002: 93), who claim that “[s]peakers of English as a second language often import the fillers from their first language”. They add that this is “one reason they continue to be heard as non-native speakers”.

128 Gaëtanelle Gilquin Table 3. Relative frequency of filled pauses in native and learner speech (per 100,000 words) LOCNEC

LINDSEI-FR

X²

+/–

eh em er erm mm

81.45 42.32 754.64 1266.51 216.41

1326.19 381.33 5260.26 498.91 414.17

1361.18 329.60 4172.45 341.84 70.05

+ + + – +

Total

2361.33

7880.86

3636.22

+

5.2. Smallwords While French-speaking learners tend to overuse silent and filled pauses to express hesitation, they do not exploit the full range of smallwords that may be used to perform this function, as is clear from Tables 4 and 5, which give the relative frequency of a number of markers of hesitation regularly discussed in the literature. Table 4 lists the smallwords that are shared by native speakers and learners. For some of them the difference in frequency is not statistically significant between the two groups, but in the majority of cases (12 out of 18), there is a significant underuse among the learners. 13 Particularly striking is the underuse of like, illustrated by (12), which is extremely common in native speech (527.05 occurrences per 100,000 words) but is hardly ever found in learner speech (6.36 occurrences per 100,000 words). (12) I don’t wanna swim any more people come out with like bruises all over their legs where they’ve hit rocks at the bottom

13

Influence of the context in which the interview took place may not be totally excluded. In LINDSEI-FR the interviewer was one of the participants’ teachers, whereas in some of the LOCNEC interviews the interviewer was a fellow student. This difference may have resulted in a more relaxed atmosphere in the latter case, which in turn may have led to a heavier use of some smallwords (see Stubbe & Holmes 1995: 66). However, the number of interviews by a fellow student in LOCNEC is too small to be solely responsible for the observed differences.

Hesitation markers among EFL learners

129

Table 4. Relative frequency of smallwords of hesitation shared by native speakers and learners (per 100,000 words) LOCNEC all right all that anyway I mean in a way just kind of like or so or something right something like sort of stuff thing things like well you know Total

LINDSEI-FR

X²

+/– –

12.78 3.99 45.52 352.16 15.97 746.65 86.24 527.05 3.19 43.92 99.02 27.95 456.77 67.88 142.14 55.90 415.25 479.13

3.18 2.12 33.90 152.53 3.18 274.35 67.79 6.36 7.41 13.77 5.30 40.25 34.96 5.30 49.78 23.30 1076.20 190.67

5.73 0.59 1.79 80.75 8.41 222.07 2.34 482.73 1.91 15.81 80.55 2.45 348.65 51.47 44.86 13.66 338.60 126.57

– – – – + –

3581.52

1990.34

484.37

–

– – – – – –

A notable exception to learners’ tendency to underuse smallwords of hesitation is well, which, as Table 4 shows, is significantly overused by learners (sentences (13) and (14) are just a couple of illustrations taken from LINDSEI-FR). To paraphrase Hasselgren (1994), one could say that well is a “pragmatic teddy bear” for learners, who cling to it because it is familiar, 14 safe and widely usable. This over-reliance on well probably explains why learners do not feel the need to use other smallwords of hesitation, which results in an overall underuse of this category of hesitation markers.

14

As underlined by Mukherjee & Rohrbach (2006: 216), well is one of the only discourse markers not to be underrepresented in ELT (English Language Teaching) textbooks and materials.

130 Gaëtanelle Gilquin (13) but er well when I’ve seen the number of my room . well I I noticed that it wasn’t the case (14) er . with the school .. the first time I was well about fifteen or sixteen . secondary school .. er . the second time . well in fact .. er I’ve been studying in a: teachers training college before I was here Table 5 lists smallwords which occur in LOCNEC but are never found in LINDSEI-FR (at least not with the function of hesitation). Most of these belong to what Stubbe & Holmes (1995) call “set marking tags”, e.g. and things, or anything: (15) and so we spend the days doing things like climbing mountains and going swimming in lakes and things and the nights at the pub drinking you know (16) it wasn’t like a test or anything it was just like to see what they got right . how much they understand of it French-speaking learners’ failure to use common expressions including the word thing (and things, all that kind of thing) and their underuse of the word thing and the expression things like (see Table 4) are quite surprising in view of the fact that, in writing, they (like learners from most mother tongue backgrounds) overuse the word thing, which, because of its vague character, tends to be avoided by native writers.15 In fact, it appears that French-speaking learners use the word thing almost as often in writing as in speech (37.92 vs. 49.78 per 100,000 words, X² = 1.95, non-significant), which is in stark contrast to the situation among native speakers, who use it much more often in speech than in writing (113.69 vs. 11.01 per 100,000 words, X² = 12509.11, p