AN INTRODUCTION TO DISCOURSE EWU AAI Malcolm eedAPPLIED LINGUISTICS AND LANGUAGE STUDY Genera Editor Profesor Chriopar N. Candin, Macpasie Univenity Error Analysis Porpectives on sond Language cision JACK C, RICHARDS (ED) Solita the Teaching of Literature HENRY WIDDOWSON Language Tests at School A pragmatic appro JOHIN W. OLLER JNK Coietative Aealpis CARL JAMES, Language and Communication JACK B. RICHARDS AND RICHARD W, SCHMIDT (EDS) Lemming to: Write: Fist Language/, Second Language AVIVA FREDMAN, IAN PRINGLE AND JANIC ALDEN (EDS) Strategies in Interkimewage ‘Communication ‘CLAUS FAERCH AND, CGAMRIELE KASPER (EDS) Reading ina Foreign Language J, CHARLES ALDERSON AND AL URQUHART (EDS) An Introduction to Discoure Analysis New edition MALCOLM COULTHARD (Computers in English Language Teaching and Research GHOBFREY LEECH AND. (Crus TOPHIEK . CANDLIN (EDS) Bilingualism in Education Aspcs of thor, reach end rate JIM CUMMINS AND MERRILL SWAIN Second Language Grammar: Lesming ard Teaching ‘WILLIAM E, RUTHERFORD “The Clasiteom and the Language Learner Edhunrapy nd cond ‘deseo reanh LEO VAN LIER uae Vocablary and Langeage Teaching RONALD CARTER AND MICHAEL McCARTHY (EDS) Observation in the Language Clasoom DICK ALLWRIGHT Listening wo Spoken English Second Edition GILLIAN BROWN Listening in Langage Leaming MICHAEL ROST ‘Introduction to Second Language ‘Acquistion Research DIANE LARSEN-FREEMAN AND MICHAEL H. LONG Language and Discrimination A Sindy of comico aultvetinc worplaces CELIA ROBERTS, TOM JUPP AND EVELYN DAVIES Translation and Trandaing “Theory and Practice ROGER 'T. BELL Proces and Experience inthe Lingoage Classroom MICHAEL LEGUTKE AND HOWARD THOMAS An Introduction to Discourse Analysis New edition Malcolm Coulthard ‘np ot Pearson EducationPearson Education Limited Edinburgh Gate, Harlow Essex CM20 208, England ‘and Associsted Companies throughout the world Visit us onthe World Wide Web at: ‘wor pearsomed cok (© Loogman Group UK Limited 1977, 1985, All sights reserved; no pat ofthis publication may be reproduced, stored in a retrieval system or transmitted ia any form or by any ‘means, electronic, mechanical, photocopying, recording or otherwise, without either the price written permission ofthe Publishers or licence permitting restricted copying inthe United Kingdom issued by the Copyright Liceosing Agency Lid, 90 Tottenham Court Road, Loodon, WIT LP First published 197 ‘Second edition 1985, ISBN 978-0-582-55970-8 British Library Cataloguing in Publication Data Coulthard, Maleokm ‘An introduction to discourse analysis ‘New ed. ~ (Applied linguistics and language study) 1. Discourse analysis 1 Tile I Series 41s 302 Library of Congress Cataloging in-Publication Data (Coulthard, Malcolm ‘An introduction to discoarse analysis (Applied linguistics and language study) Bibliography: p. Includes index 1. Discourse analysis. Tite. I Series P202.C68" 1985 40I"41 85-4262 ‘Mlostration on page 113 reproduced by kind permission of ‘Academic Press, New York Setin 10/12p¢ Linotrom 202 Exhardt Printed in Malaysia, PP 2322 08 o7 Contents Preface Author's preface 1 Introduction 2 Speech acts and conversational maxims 3 The ethnography of speaking 4 Conversational analysis 5 Intonation 6 A linguistic approach 7 Discourse analysis and language teaching 8 The acquisition of discourse 9 The analysis of literary discourse Further reading Bibliography Index 13 3B 59 96 120 146 160 179 194 195 21Preface From the response received to the first edition of Malcolm Coult- hard’s pioneering Au Intraduction t9 Discourse Analysis, i is clear that ‘our hope that it would ‘put discourse analysis on the map’ has been amply fulfilled. ‘The references to the book in the linguistic literature would be evidence enough to this were it not also for the cortobor- ative impact it has made on a range of applications: language teaching and acquisition, stylists, reading and writing studies, speech pathology and many others. It would not be an exaggeration to say that it stands even now as a key work of reference for many teachers and students throughout the world Why then a sccond edition? Principally, because if you take the broad view in making connections between language form and language use, bringing together as Coulthard docs appropriate work from a range of disciplines: linguistics, social anthropology, philos ophy, psychology, there is inevitably the nced afier eight years to review judgements and to bring to a new generation of students and scholars something of the excitement of more recent research. The ‘time-bomb' of meaning, referred to in the first edition, has long since exploded and its effect on the narrowness of earlier models of linguis- ties has been fundamental. Taking the wider focus is now a way of life, Not that this development has been unproblematic; there remain the issues of deseription and analysis in discourse and pragmatics to ‘which much of the first edition was dedicated. What has changed is a general perception that the bold integration of different disciplines, subject 10 overriding principles of descriptive and explanatory adequacy, offers the most promising avenues for attack. It would be gratuitous to list here the additions and deletions, the changes in organizational structure which practice with the book hi dictated: perhaps most noteworthy among these, however, are the entirely new Chapter 5 on Intonation and Chapter 6 4 Linguistic Approach which bring. up-to-date the highly influential work of the ‘author and his colleagues at the University of Birmingham. ll sections of the book have, however, been revised, and wherevili Preface necessary and appropriate, made current and more clearly intercon- nected. As an aid to the reader, however, it may be valuable to provide @ summary view of current positions on the analysis of discourse; a way in to the richness of the reference provided here. Any approach to discourse analysis and pragmatics has, presumably, to represent two distinguishable but related discourse worlds in the pursuit of its objective, namely the characterization of speaker/writer meaning and its explanation in the context of use. On the one hand, more nomothetically, discourse analysis must portray the structure of ‘suprasentential text or Social transaction by imposing some framework. upon the data, explicitly or implicitly. On the other hand, more hermeneutically, discourse analysis should offer us a characterization of how, in the context of negotiation, participants go about the process of interpreting meaning (whether this is reciprocal as in conversation or non-reciprocal as in reading or writing need not detain us here, suffice that the process is interact In their structure-portraying role discourse analysis and text linguisties have much in common, as in fact does the ethnography of speaking, concerned as it is with the display of sequenced episodes in some social encounter. In its interpretation-characterizing role, on the other hand, discourse analysis is involved in the assessment of the communicative function of momentary messages, drawing upon general and specific background knowledge in the process of making inference. The object of the first type of discourse analysis is the determination of interactive acts, siting them within some larger inter- actional frame; the objective of the second type of discourse analysis, on the other hand, is more the capturing of illocutionary force, drawing upon general pragmatic principle, an understanding of ‘contextual expectations in the activity type under discussion, together ‘with knowledge of how information may generally be structured, and procedures of natural analogy. All this is an attempt to display coherence. So far, then, we have two different approaches to discourse analysis, one concerned with sequential relations interpretation; the one working for ‘rules’ whic alizations about intersential structure wherein the ‘function’ or of the utterances is in a sense taken for granted, the other working for ‘procedures’ where ‘function’ or ‘value’ is not a product based (on intuitive understanding of the utterances in question, but a matter of negotiative process among a variety of contextual factors all of which taken together lead to the establishment (or the revelation) of specitic social relationships berween the interlocutors, themselves, of Preface ix course, powerful sources of clue to illocutionary value, Rather than providing rules to account for relationships between product and form, in this second view procedures are introduced for the tracing of the negotitive process. One approach to discourse analysis is thus ‘emphasizing organization and mapping, the other emphasizing social relationships and interaction, Here, however, lies the paradox in our account: the organizational and the interactive (the structural and the procedural) each implies the other and cannot easily be abstracted from each other in any effective study of the discourse process, and for two reasons. Firstly, following studies in conversational analysis (amply documented by Malcolm Coulthard in Chapter 4), discoursal ‘place’ provides. an orientation for participants in their evaluation of illocutionary force. Secondly, taking such an integrated view enables us to see language forms as the surface realization of those communicative strategies involved in the interactive procedures working amongst those various social, contextual and epistemological factors we have identified as ‘crucial to the process of communicative inference and coherence. An example of such an integrated view exists in the work of Brown and Levinson (1978) where strategies of message construction are a key locus for an understanding of the interconnection between discourse structure and social structure. Their strategies of face redress, for example, act as a mediator between communicative intent and the circumstances or social relationships holding between the'interlocu- tors. It is via these strategies that the degree of modification of the impact of communicative intent on the addressee is negotiated between the intentions themselves and the social relationships, traced then in the appropriately chosen form. Itwould seem, then, as this book makes abundantly clear, the charac- terization of utterance funetion cannot be left to the tender mercies of linguistic form. How, then, can we map action to utterance? Following the suggestions outlined in Coulthard’s Fnimduction, we ‘can turn to a range of resources, all of which, we must acknowledge, are to be hedged around by the natural processes of contextual nego tiation, participants’ history and naturalized ideologies which makes ‘easy identification and labelling very difficult. We can marshal our knowledge of speech events, themselves of course culturally-specitie ‘constructs, and apply to them (if we are aware of them as outsider) the specific inferencing procedures relative to the event in question, using our framework of expectations about the nature of the speech events to which they contribute. We can make use of our know of social role which is itself a negotiable ‘good! if one takes a criticalx Preface ‘view of discourse, Furthermore, we can apply our understanding of the maxims attaching to various pragmatic principles and examine with care the placement of the utterance in question in the ofien qui lengthy, sometimes discontinuous and certainly very complex patterns of conversational structure. One could go on ... the point is that we are dealing with an immensely complex inferential process that makes use, as this book amply shows, of information of many kinds. If we do not take this into account, if we underestimate the quantity of text needed to make judgements of value, if we fall into the trap of failing to acknowledge culturally-biased presupposition, if we fail to embed utterances in the context of speech events, if we fail to make the conneetion between the formations of discourse and the forma- tions of society, then we will take a too simplistic view of the subject- matter of this book, What is more, if we do not review our methodologies and the reasons why we undertake the research then we shall neither have access to adequate data nor have any social warrant for their collection or their analysis. 4 In explicitly acknowledging Malcolm Coulthard’s contribution to an awareness of the caveats presented here, | can do no better than repeat the final appreciation of the first edition; ‘the crucial matter has been to have scen the connections between disciplines concerned with describing and explaining human communication and to have suggested a synthesis’. Christopher N. Candlin Lancaster General Editor 1985 Author’s preface In some ways itis more dificult to rewrite than to write — a second edition is necessarily constrained by the first, I have tried to retain the organization and as much as possible of the content of the first edition and thus, with the exception of Chapters 5 and 6 which have been totally rewritten, most of the new material occurs in the second halves of chapters. “The intention behind the book remains the sime — to introduce those interested in the analysis of verbal interaction to relevant re- search in a variety of fields. This of course means that few of those whose work is presented here would regard themselves as Discourse Analysts and that for purposes of presentation I may have linked together researchers in what they and their followers feel are totally inappropriate ways. The prime example is Labor, brilliant but un- classifiable, who has worked in a whole series of areas: in the last edition he appeared with Ethnographers of Communication and Con versational Analysts — I have rectified this, but now he appears with Speech Act Philosophers! “Textbooks cannot be writen in a vacuum; most of the excisions and additions I have made result from teaching Discourse Analysis to many soups of students sadly too numerous to be named. It is, howew possible and appropriate to acknowledge my debt to Dave Willis and Ken Hyland, whose theses I supervised and from whom and which I learned more than they will ever believe, In final place because he knows its real significance, Mike Hoey, a stimulating colleague and a true friend, without whom both content and form would have been more flawed. Birmingham August 19841 Introduction [Although it is now many years since J. R. Firth urged linguists to study conversation, for there ‘we shall find the key toa beter understanding fof what language is and how it works’ (1933), the serious study-of spoken discourse is only just beginning and currently much of the ‘Work is being undertaken not by linguists but by sociologists, anthro- ‘pologists and philosophers. ‘The explanation is not hard to find. While” all linguists woul agree that hurian communication must be described in terms of at least three levels — meaning, form and substance, or discourse, lesico-grammar and phonology — there are disagreements over the boundaries of linguistic. Firth (1951) asserted that ‘the main concer of descriptive linguis- tics is to make statements of meaning’. Part of the meaning of an ut- terance isthe result of contrasts in the levels of phonology and syntax, and Firth accepted that in order to isolate meaningful contrasts in these levels “we make regular use of nonsense in phonetics and gram- mar’, but, he argued, language is fundamentally ‘a way of behaving land making others behave’ and therefore ultimately the Hinguist must concern himself with the ‘verbal process in the context of situation’ For Firth language was only meaningful in its context of situation; he asserted that the descriptive process must begin with the collection of a set of contextually defined homogeneous texts and the aim of description is to explain how the sentences or utterances are mean ingful in their contexts Firth himself did not in fact explore the relation between form and meaning and his exhortations to others were ignored, because Bloom- field led linguistics away from any consideration of meaning to a con- centration on form and substance, by observing that linguists ‘cannot define meanings, but must appeal for this to students of other sciences ‘or to common knowledge’ (1933). The utterance ‘I'm hungry’ could be used by a starving beggar to request food or by a petulant child to delay going to bed; Bloomfield argued that linguistics is only con- cerned with those phonological, lexical and syntactic features which the utterances share — he felt it was no concer. of linguistics to2 Aw Introduction to Discourne Analysis explain how identical utterances can have different functions in dif- ferent situations, nor how listeners correctly decode the intended message. For a generation American linguists concentrated massively and highly successfully on problems within phonology and morphology — on the existence of the phoneme and the validity of unique pho- nnemic descriptions; on discovery procedures for isolating phonemes and morphemes in languages not previously deseribed; on the mech~ anical identification of morpheme boundaries and word classes. ‘When Chomsky redirected linguistics towards the study of sentence structure, the concerns were stil pre-eminently with the formal fea~ tures of language: ‘the fundamental aim in the linguistic analysis of a language Lis to separate the grammatical sequences which are sen~ tences of L. from the rgrammatical sequences which are not sentences of Land to study the structure of the grammatical sequences’ (1957). In arguing the independence of grammaticality from meaningfulness (Chomsky produced the most famous example of ‘nonsense’ in linguis- ties — ‘colourless green ideas sleep furiously” Earlier linguists, while concentrating on formal aspeets of language, had used collections of speech or writing as a source of examples. Chomsky suggested that not only was a corpus unnecessary, it was actually counterproductive, No corpus, however large, can be ad- ‘equate because it will never contain examples of all possible structures and will actually contain misleading data, performance errors, caused bby ‘such grammatically irelevant conditions as memory limitations, distractions, shifts of attention and interest and errors (random or characteristic) in applying knowledge of the language in actual per- formance’. The prime concern of linguistic theory, Chomsky argued, is with the underlying knowledge, the competence of the ideal speaker- hearer. The underlying competence is the same for all native speakers and therefore can be studied in the productions of any one individual, usualy the linguist himself, who proceeds by introspection, checking potential sentences for grammaticality against his intuitions. ‘The insights achieved by transformational grammarians were enor- ‘mous, but as time passed the problems became more serious, Ithecame evident that there was not in fact a uniform native speaker com- pretence; it became necessary to talk of degrees of grammaticality or scceptability; crucial examples were attacked as ungrammatical and defended as ‘acceptable in my idiolect’. Meanwhile the timebomb reaning was ticking away: in the late 1960s Ross, MeCawley and G. Lakoff began arguing that one cannot in faet describe grammar in isolation from meaning, that powerful syntactic generalizations can be Introduction 3 tchieved by making lexical insertions at an early stage in the gencr- on of a sentence. By 1972 Robin Lakoff was arguing that “in order lo predict correctly the applicability of many rues one must be able to refer to assumptions about the social context of an utterance, a well as to other implicit assumptions made by the participants in a Uiscourse’. ‘Thus the results of empirical investigation have forced ‘many transformational linguists to recognize the importance of context and to join a series of disciplines converging on the study of situated speech. Mrhere is as yet, however, no singe dscptne which concems iself with the study of interaction; in writing an introduction to discourse analysis Lam not, paradoxically, describing only the work of research= cers who consider themselves discourse analysts — many of those men= tioned here would be bemused or annoyed by the label. Rather, what Thave tried to do is draw together in the first six chapters research from many disciplines — philosophy, psychology, sociology, sociolin- szistics, conversational analysis, anthropology, ethnography of speak ing, phonetics and linguistics — which is useful to anyone interested in the analysis of situated speech or spoken discourse. Labels are always Thave chosen to maintain a distinction between spoken dis~ ‘oure and writen text, but this is by no means a universally accepted distinction; many German writers use ‘text’ to refer to speech a well, ‘while Hoey (1983) and Widdowson (passin) use ‘discourse’ to refer to writing, and to complicate matters further ‘pragmatics’ as defined by Leech (1983) and Levinson (1983) overlaps substantially with dis- ‘course analysis as I conceive it Early attempts at discourse analysis Although Firth urged linguists 10 study the total verbal process in its ‘context of situation he did not do so himself, choosing rather to con~ centrate on phonology. In the period up to the late 60s there were ‘only to isolated attempts to study suprasentential structure, one by Harris (1952), the other by Mitchell (1957). Harris's article, although it has the promising title ‘Discourse is’, is in fact disappointing. Working within the Bloomficldian ition Ihe sets out to produce a formal method “for the analysis of connected speech or writing’ which ‘does not depend on the analyst's ‘knowledge of the particular meaning of each morpheme’. He observes that in grammar it is possible to set up word classes distributionally a class of adjectives A which occur before a class of Jh a statement captures a powerful generalization, even4 Am Introduction to Discourse Analysis itis possible to show that a particular member ofthe class A, 7, may never occur before a particular member of the class N, ‘subjugation’. Harris suggests that a distributional analysis ean be successfully applied to a whole text to discover structuring above the rank of sen tence. As an example he creates @ text containing the following four ‘The tees tur here about the middle of autumn, The trees turn here about the end of October. The fist frost comes after the middle of autumn. We start eating afer the end of October. ‘The aim of the analysis is to isolate units of text which are distri- butionally equivalent though not necessarily similar in meaning; that is equivalences which have validity for that text alone. From the first two sentences above one establishes the equivalence of ‘the middle of autumn’ and ‘the end of October’, not because they are similar in ‘meaning but because they share an identical environment, ‘the trees turn here’. The next step is to carry over the equivalences derived fom the first two sentences into the next two and this allows us to ‘equate ‘the first frost comes’ with ‘we start heating’ and of coutse both with ‘the trees tum here’ which provided the original context. Thus, in terms of equivalence classes, all four sentences have identical struc~ ture, class X followed by class Y. ‘The analyst progresses in this way through the text creating a chain of equivalences and occasionally, as required, introducing a new class until the whole of the text has ‘been divided into units assigned to one or other of the classes. Harris points out that in evaluating his approach the only relevant questions are ‘whether the method is usable and whether it leads to valid and interesting results. In the thirty years since the article was, published no one has adapted or developed his method for the analy~ sis of discourse, though the idea of ‘transformation’, introduced to handle the equivalence relations, became, in a modified form, a central feature in Chomsky’s Generative Grammar. It may well be, of course, that any purely formal analysis of structure above the sentence is impossible. In marked contrast, Mitchell's ‘Buying and selling in Cyrenaica’ presents a semantically motivated analysis. Working in the Firthian| tradition he specifies the relevant participants and elements of ation in detail and divides the buying-selling process into stages purely oon content criteria, admitting that ‘stage is an abstract category and the numbering of stages does not necessarily imply sequence in time’. He describes three major categories of transaction — market auctions; Introduction 5 ‘other market transactions; shop transactions — although the second and third are distinguished mainly by situation because they share the following five stages: salutation, enquiry as to the object of sale investigation of the object of sale ‘anganing This is an ideal structure: sometimes stages 1 and 2 do not oceur and stages 3 and 5 may be realized non-verbally. ‘The following is an ex ample of a shop transaction: Personality Translation ever: Have you a bed to sell SELLER: [ve got one bur it’s rather expensive. nuvi: Let me have a look at it then rusk: Gensainly Ifyou want it for yourself Twill make you a reduction nuver: How muct SHUR: £4. never: What's your last pric shite: Believe me if t were anyone but you I'd ask him five even: [ll make you a firm offer of £3.50, stutek: Impossible, let it stay where iis. ryan Listen. Il'come this afternoon, pay you £3.70 and take i (Buyer crosses threshold of shop on his way out) - seize: Ie sil wants some repairs 5 2 2 2 While this analysis captures the structure of the transaction it is ar- guable that it is not a linguistic analysis at all — the stages are defined and recognized by the activity that oceurs within them rather than by characteristic linguistic features and, with the possible exception of stage 4, which when opened by the buyer apparently begins with the formula ‘How much”, there are no linguistic markers of transitions berween stages. However, once the stages have been isolated non~ inguistically Mitchell then characterizes them lingui ing examples of the kinds of phrases and clauses, often ‘occur within them, Discourse units and discourse function “The fact that Mitchell did not offer any internal structure for his stages must not be taken to imply that spoken discourse has no struc ture and consists simply of a string of grammatically well-formed ut- terances. The following examples from Labov (1970), the first fom6 Aw botrnduction to Dieonree Amelie an interview with a schizophrenic patient, the other fabricated, are grammatically unexceptional yet noticeably odd: as Whats your name? te Welle ay sou might have shoght you had something fom before, but you haven't got it anymore. 7 ss Pim going teal you Dean 4: Lee hot today te No. In both examples v's contribution obviously breaks rules for the pro- duction of coherent discourse, and one of the major aims of discourse analysis is to discover these rules and to describe the conversational structures they generate. Obviously an initial and fundamental que tion is the nature of the units whose structure and occurrence the sequencing ules will describe Harris (1952) observes that traditionally grammatical description has taken the sentence as its upper limit, and it is instructive to discuss the reasons for this. A grammatical description provides the struc~ ture(s) of a given unit in terms of allowable combinations of smaller units and an essential feature of any grammatical description is the specification not only of what structures ean occur but also of those structures which cannot oceur. ‘Thus a grammar of English would allow the following sentences: 1 bought these chairs yesterday Yesterday U bought these chairs. 'Phese chairs I bought yesterday but not: Yesterday these chairs bought 1 “These chairs bought yesterday [these chairs yesterday bought and a speaker's decision about which of the possible grammatical op. tions to select on a particular occasion will then depend on cohesive and stylistic considerations. Once one comes to look at choices above the sentence, however, there are no parallel restrictions on combi- nations of units and all the decisions a speaker or writer makes are stylistic ones — there is no way of describing paragraph structure in terms of allowable combinations of simple, complex or compound sen- tences because any collection of sentence types in any sequence can ‘constitute a paragraph and ‘rules’ about paragraph writing therefore take the form of advice about “topic sentences’ and the alternation of long and short sentences. Introduction 7 However, while it appears that structure deseribable in terms of formal grammatical units ends at the sentence, we can explain the Labov examples in terms of patterning of fictional units which cer- tainly does occur above the sentence and across utterance boundaries. In the following example itis impossible to deseribe or even contem- plate constraints on @’s utterance in grammatical terms but in func~ tional terms his options are highly restricted: ss Where's the typewriter? ve In the cupboard. Is iv in the cupboard? [Look in the cupboard Tthink i’ in the cupboard. In other words, whereas we cannot provide @ meaningful structural description of a conversation in terms of ‘declarative followed by movdless clause’, or ‘interrogative followed by declarative’ it is poss ‘ble to provide a meaningful structure in terms of Question and Answer, Challenge and Response, Invitation and Acceptance. Thus Labov (1972a) argues that the first and most important step is to dis~ tinguish ‘What is said from what és done’, and stresses that the unit of analysis is not the grammatically defined clause or sentence but a functional unit, which may of course be realized by a single clause or sentence. ‘Any attempt to characterize discourse structure in terms of func sional units must confront the problem of grammatical realization — how do the four major clause types, ‘declarative’, ‘interrogative’, ‘imperative’ and ‘moodless’, realize a multiplicity of different functions, and how can a hearer correctly interpret which function is intended? Labov (1970, 1972a), taking as given that question-answer is a basic interactive structure, focuses on ansmers and sketches out a se- ries of interpretive rales to explain how a second utterance comes to be heard as an answer to a question. ‘The simplest relationship is between @ question and an elliptical answer: A: Are you going to work tomorrow: ie Yes Here a simple rule can account for the relation: 1f utters a question ofthe form (Q(S,) and » responds with an existential E (including ye, no, prohaly, maybe ete.) then w 1s heard as answering with a statement (E) 8 A more complex relationship holds between the following pair of utterances:8 Am Intraduetion to Discourse Analysis: Are vou going to work tomorrow? se T'm on jury dy Grice (1975) argues that there is an underlying constraint on all on versationalists to ‘be relevant’ and for this reason 4 will assume, at least initially, that there is a proposition known to both which con- nects fs response to his question, i. ‘if someone is on jury duty he cannot go to work’. To account for this type of relationship Labov proposes the following rule I. makes a request of wofthe form Q (S)), and w responds witha statement Sand there exists no rule of ellipsis which would expand S; 10 include SS then 1 is heard as assering that there exists a proposition known 10 both 4 and 8, of the form IPSs, chen ( where’ (F) is an existential operator, and from this proposition there is inferred an answer to 4° request: () S1. ‘This rule makes clear the crucial importance of shared Anomledge in conversation; not simply shared rules forthe interpretation of inguis- tic items, but shared knowledge of the world, to which a speaker cean allude or appeal. Labow notes that this rule is invariant: 4. must inspect ns utterance to sce if he can detect an underlying linking proposition and ‘failure to locate such a proposition may reflect a real incompetence’. Younger members of a social group may not be able to find the proposition being asserted: lusts: Do you want to play with me, Violet vwourr: You're younger than me. (its the door) lusts: (pizeled) She didn’t answer my question. (Labov 19724) “There are two possible ‘solutions’ to this joke — one is that Linus, unlike the reader, is unable to derive the underlying proposition ‘If you are younger than me ‘THEN NOT I want to play with you’; the ‘other, more subtly, is that his previous experience or self-esteem lead him to conclude that the underlying proposition is nor in fact eoberent and that therefore Violet has not provided an answer. In either case the question had set up the next utterance as a potential answer and the questioner had used inferring strategies to evaluate the utterance as a possible meaningful answer. These first rales are concerned with explaining how statements fol- lowing questions come to be heard and interpreted as answers, but in these instances there is an expectation that an answer will follow; ‘much more difficult to explain is how some utterances, declarative form, come to be heard as questions. Labov presents the following extract from a therapy session, Introduction — 9 ‘nner Oh soe lyon riaansr: Ske did sy for yeu. tena No ‘east: And it never occured to her to prepare dinner. raviest: No. " and observes that it consists of a series of pairs where ‘the first ttterance is a statement and the second is “yes” or “no”, and it seems that a statement is functioning as equivalent to 2 yes—no question’ Labov suggests that the statements in the therapy extract are acting as requests for confirmation and have the same compelling force as re quests made in question form; but how ‘is it that we regularly and reliably recognize these as requests and not as assertions’, because is certainly not the case that any statement ean be followed by ‘yes? 1 Tdon’e like the way you said that & "Yes. se [feel hot today ‘No, ‘The proposed interpretive rule depends again on shared knowl- ‘edge. Whenever there are two participants, 4 and #, in a conversation, Labov observes, one can distinguish ‘a-events’, things that 4 alone knows about, ‘s-events’, things that m alone knows about, and ‘A events’, things that are known to both. Using this terminology he sates a simple interpretive rule: If A makes a statement about a n-ovent it is heard as a request for confirmation ‘The interpretation of such utterances as requests for confirmation depends crucially of course on speaker s’s assumptions about speaker H'sknowledge being correct. [n the following example the assumptions were wrong and 4's intended request for confirmation was heard as a statement of new information. 4: ‘There's no playgroup next week then © Oh, isnt there? This brief discussion of a part of Labov's suggestive work has raised some of the questions which discourse analysis sets out t0 answer —how does one characterize and label the basic unit of inter- action; how many different functions are there; how are these functions realized lexico-grammatically and what structures do these basic units combine to form? Succeeding chapters present a variety10 An Innroduction 0 Discourve Analysis of answers and in evaluating them itis useful to bear in mind the four criteria to which Sinclair (1973) claims all inguistic descriptions must conform: 1. ‘The descriptive apparatus should be finite or else one is no saying amy thing av all, and may be merely eating the illusion of classification, 2. ‘The whole of the data should be deseribable; the descriptive system should be comprehen Thi na ish rien met cause it is shvays posible 1 have a“raghag category into which go ems nt posively clsied by other eaten. [Of cours] we Bnd that 95% ofthe text goes into the ragbag we would reject the deserition 2s imal. While making apparently innocuous demands — that the system should have a demonstrably finite number of items and be able to handle the whole of a specified corpus — these two criteria cannot ‘be applied to all of the descriptions presented in succeeding chapters Indeed, as Laboy (1972a) notes, Sacks and his colleagues believe it is inappropriate to attempt to describe all the data at this stage, and Laboy himself only attempts to handle fragments. 3. There must be at least one impossible com “This is the basie notion of linguistic structure to wh paling in our earlier discussion of the status of paragraphs, and is ‘one which Levinson (1983, pp. 291-4) uses to criticize the descrip- ‘TABLE LL Categories of interaction analysis Hiymes ks ‘Scheflen Sinclair event ‘conversation presentation interaction topic positon transaction sequence sequence point exchange pair Introduction V1 tion proposed in Sinclair and Coulthard (1975). Of course this rule presupposes an interest in the structure of interaction — speech act analysis for instance is concerned only with the functional meaning of individual utterances. ‘Those descriptions that are concerned with structure vary in the number of analytic units they propose from two (Hymes 1972a) to six (Sinclair and Brazil 1982). Table 1.1 matches roughly in terms of size the category labels from four different descriptions, Scheflen’s being based on non-verbal aspects of interaction, 4. The symbols in the deseripsive apparatus should be precisely relatable to their exponents inthe data... ifwe call some phenomenon a ‘noun’ os ‘repair statesy’ or a “threat” we must establish exactly what constitutes the class with that label. TThe problem can be approached from either end: one can write re- ‘lization rules to show how functions are related to their lexico- {grammatical realizations, or interpretive rules to show how particular siretches of specch are understood as having particular significances. Subsequent chapters have examples of both approaches but it looks as if interpretive rules will ultimately be preferred “The other major concern of discourse analysis, which the Labov examples do not highlight, is the relationship between the discourse and the speakers and hearers by and for whom itis produced — a concer with how speakers take and relinquish the role of speaker, how social res affect discourse options in terms of who speaks when and what they can talk about, how non-verbal signalling. works and how the actual form of utterances is conditioned by the social re- lationships between the participants The research reported in the succeeding chapters comes from a wide range of disciplines with differing ideas on what constitutes rel- evant and acceptable data. As we saw above Firth argued for a text~ based description, Chomsky for a total reliance on intuition. Lyons (1968) suggests that there are in fact three degrees of idealization between raw data and the idealized sentences of Chomsky’s competence. ‘The first stage is regularization in which the analyst ignores such phenomena as slips of the tongue, hesitations, repetitions, self-editing and so on. ‘The second stage is standardization in which one ignores variation and treats whatever data one is examining as homogeneous — thus at the phonemic level, different pronunciations of the same ‘word? are treated as if they were the same; at the level of discourse, sariants of a muisappretiension sequence are all regarded as occurrences, of the same unit. This is an essential step in any classificatory system,12 Am Introduction 10 Discourse Analyse for in the final analysis all utterances can be shown to be unique However, there are currently disagreements among linguists over the degree of standardization and the amount of variation which can be successfully described (SankotT 1974). The third stage of idealization involves decontextuatization, which separates sentences from their con- teats of use oF oceurrence and treats them as self-contained and iso lated units. Much of the work described in the following chapters is based on transcripts which are in an unregularized form, but in fact analyses the data as if it were both regularized and standardized. The work by philosophers on speech acts, however, is based entirely on decon- textualized fabricated data and all approaches make some use of fabri- cated examples to make points and arguments clearer. ARETE oF RCNP cently they had come to realize that this was not always the case. 2. Speech acts and conversational maxims While linguisties restricted itself for a generation to a concentration on form, shestudy of meaning was left to Linguistic philosophe:s, who sented otters ond ingens Wf sete and parts of sentences, In J. L. “Austin Observed that while it nad long been The assumption of philosophers that ‘the business of “statement” can only be to deseribe some state of ais orto “state Some fact”, which 1 must Ge-=RHEE TE [MOTE ‘There are sentences which look like statements, or as Austin prefers , ‘5 al trom engi, Un a potted w veer! oc amar ie 7 formation about fact: some, for Sampler Hic The King of France is 2th—j bald” are strictly nonsense, despite unexceptional grammatical form; heey Bid promi are ‘perhaps intended, solely or partly, to in ot oF peter cony,o tact in Se ways". Austin focuses on a third group of sentences which he labels Erialiony in whlch te soriog of the wood Soempes gs ng Me Ee SEs Moe Sea Sea, etre “name his ship the Quen cael! ~ ered when smashing the tone spaet he ser "To" (ike tt woman tobe my lw weed wie) — as werd inthe Sore deg eronoey STpre and buch omy brother — es ocuing ina wil 8) In saying ‘T name this ship the Queen Elisateth the speaker is not describing what he is doing, nor stating that he is doing it, but actually performing the action of naming the ship; from that moment the ship is named. A confirmation that itis in saving the words that one per- forms the action is that very frequently one can insert the word “here~ by’ — 4 herey name this ship ...”. The uttering of the words alone is, however, not sufficient — while the performative utterance is “usually a, or even sf leading incident’ in the performing of the acts of naming, marrying or hequeathing, itis rarely fever the ‘sole thing necessary ifthe act is to be deemed to have been performed’. Austin stresses the conventions! nature of the performative act and the fact Jobs14 An Introduction to Dikcourse Anelytis that an agreed procedure must be followed. There are four conditions which must be satisfied if the performative act is not 1 misfire (there must exist am accepted conventional procedure, having @ ceria conventional effect, that procedurg tO nce the uttering of certain wyordfbsieeriin person(in erain-Greimstances. p26) By this condition Austin draws attention to the fact that there limited number of performative acts and{ one eannot arbitrarily a.procedurc in order to perform what appears 10 be a similar act = ‘there is a procedure for christening babies But not dogs, for naminig ships but not houses. For some acts procedures differ in different ‘countries — no one, whatever Kis religion, ca divorce Bis wife i England by saying ‘I divorce you; I divorce you; I divorce you’, while some aets are possible in one language community but notin another — there is no formal procedure in Modern English for insulting someone, to match that used by German students to initiate duels in the inter-war years. This is not, of course, to say that one cannot insult someone in English, but simply that one eannot insult them by saying ‘I insult you’ ive passitr pesos and cteumsinces a » pen. cae mus be P tomoprate forthe mnocn of ie atiular eens loka he ——— ‘This condition emphasizes the fact thatthe uttering of the correct and appropriate words is insuficient to achieve the succesful perform asice of the act: the words must be uttered by the appropriate person the blacksmith in Gretna Green may read the marrage service as well as any parson, but the ceremony is still invalid; while, 2 the -nproprate person cannot utr the appropriate words in pria rcumatancey ‘one of the umpires in the Test match when TCeonard-Hrton scored his record 364 claimed later that Hutton was technically out Ibw at 332, but, as no one on the fielding side ap- pealed, the umpire was unable to pronounce him out 2 Bela er amine Gad ind (0. 35-6) These conditions cover misfires which occur despite the existence of 4 conventional procedure and the presence of the appropriate patici- pants in the appropriate circumstances. ‘The problems may be verbal or non-verbal. The marriage ceremony includes yevno questions, ‘Do you rake this woman ...” but ‘yes’ is not an aceeptable answer and the ceremony has a fixed point forthe ring to be placed on the finger — filure to produce the ring or placing the ring om the finger at a Speech acts and comversational maxims — 15 different n the ceremony would again cause the act to misfire, [So far we have seen that the uttering of certain words by appro. priate people in appropriate circumstances can constitute the pi forming of certain conventional acts amr obvious-next-question is.W ‘Tormal features mark utterances as performative? All Austin’s initial “a ms Tove sh spl reset ate fm with a first_ person singular subject, e.g{Liinie this ship’, and this is apparently ‘Fomifcan since nether am naming this ship, naz-he nameviamed this ship’, nor ‘this ship is named by me’ is an acceptable substitute. However, it soon becomes apparent that there are some performative umerances with te verb inthe panve — ‘pamenger are sexusred “to retumrto-theirseat voters fave no subject or verb y prauvusect Wie neaiacat sy ar “abr on umpire —and- Austin is for clude reluctanty that there are in fact no linguistic features which reliably and unambiguously dis ~The achievement 50 far as been to isolate “a class-of utterances, Tinguistically quite heterogeneous, which have in common that, in vr= tue of non-linguistic conventions, to issue them (happily) counis as abing this or that’ (Warnock 1973). In one important sense these per formative utterances are idioms — the meanings of the individual swords are not of grcatimporsance and synonyms cannot be substituted 's the uttering of preetommined words in a fixed sequence in 9 few highly conventionalized and at times ritual situations, which con- KP stinues the performing of the action. If performative utterances were restricted to such situations th but not particularly existence would be an interesting jgnificant fact about language use. However, Austin noticed that the concept of the performative loing Something by saying something, -general application? for romedhing Uy sayin in saying ‘I promise’, ‘I apologize’, ‘I war you’, one actually performs the acts of promising, apologizing and warning. Thus these utter= ances also are performative, but are crucially distinet from the first sroup in that there are no rule-governed conventions restricting their tise — anyone can make a promise to anyone in any place at any time. This extension of the concept to ordinary language situations very exciting but it raises enormous descriptive problems because, although the performative utterance may Sexplic for example I warn you at py his much more lke w eBay — there are, as Strawson (1964) points out, ST ae he word “the ice over there is thin” to a skater isto issue a warning ... with= out its being the case that there is a statable convention at all, such that the speaker's aet can be said to be an act conforming to that dass16 An Introduction to Discourse Analysis convention.’ This, of course, raises the crucial question of how one recognizes a given utterance as performative. There are even problems ‘with utterances containing explicit performative verbs: [promise only when I intend to keep my word. (On p. 49 I protest against the verdict. but these can usually be sorted out with the ‘hereby’ test discussed above. Much more problematic are the utterances without a perfor- mative verb. Austin suggests that the problem is not, in fact, too dif= ficult, because any iterance whichis in act performative shouldbe reducible oF expund- aban es fo. Form inthe fs BET singular: ent indlaive ace «- This “out e equivalent to T declare, pronounce {r call you out’, ‘guilt’ is equisalent to ‘I find, pronounce, deem you to be gil’ (62) The discussion has now come full circle. We first established that there was a set of utterances of the form ‘I + present simple active verb’ which were performatives; then it became apparent that not only were there constative utterances with the same grammatical form, but also performatives with other grammatical forms which often did not even include the performative verb. We then suggested, however, that those utterances which were actually performative, but did not have the form ‘I + present simple active verb? were ‘reducible, expandable or analysable’ into that form. This revives the question of how one decides whether an original or transformed utterance of the form ‘I + present simple active’ is performative or constative. In the following table the first column contains explicit performatives and the third ‘column constatves, while the status of those in the middle column is doubtful. sama Pefmmatices ? __ Constativs Thank you Tamgratefol I eel grateful 1 id you welcome 1 weleame you T apologize Jam sory Lrepent (Austin 1962, p. 79) the Austin suggests four tests for deciding which way utterances middle column are being used: Speech acts and conversational maxims, \7 Covi He kee fae oF 1. Does the saying of the words enstitute the performing of an act? This were cans fev lie il Fea oar gerne 9 ok ac ae Se ase one me rae ahve ies Rance eee a De ee ey ie a nena eee Srooran heater te inion att oonne ey eae: BEE, pend tate oneness EEE Pint cso tnd wiout wring the word? One cn be "CAS Be Soe ee ee Lemon Soto an mac rec ompery acsenrad Onecare iia toby brat ete ates a cteton cra thane eae ba eu tives which can only be bagpy or unhappy-Reespite saying ‘Lam sorry it_ 7 need not be woe tat one i SonY- sf one ss apologie, however, i annot be fase that one has apologized — the apology may be insincere Sand the speaker may have abwed the procedure but tha is anther Using these ceria it i pease to assign uneranes ofthe om PsN Single aE eID te csr of exp performatives SF of eonstnVes is one being subject to «tea Tappinss the ther ‘0 a test of truth. However, yet again in the argument, having taken two steps forward wwe must take one back. Austin a which he labe of expastves phere “the main body ofthe utterance has generally or offen the straightforward form of a statement’, and which-are therefore subjeet-to. test of truth. However, prefacing the statcment is a ferb_phrase (ay carpietcnclate/cotivadai peer which in Gt SAGER al WE on 1 antag that hee so backside othe moon, and it doesn’t take long to realize that even ‘I state’ satisfies the per formative test. This is initially very disconcerting because the whole drift of the argument so far has been concerned with distinguishing | 4+. Can he anc hs italy ase? usin ses this sa eral distin ion between eonsatives a = Ihe true or false_and performs peviornaties fram-consae, bat I now erdent ar OERTSE | {nees previously labelled constative, even those with the grammatical | “~ Gent proce sinple are we arte fir primary sexes |p thes wii we "epeeltle or TRS ne ¢ oem Cee 70 —Thara mow an slegines ter MievdaecHpron-—Tnpwead oF | Claiming two classes of utterance, one performative and the other constative, Austin now asseref hat i saying angTMINE pne is perform- gion iad eae ee | Mme Lhe patel elem lalenkepy fe bedA Locwreneny —crekod seus sercsely 4 fell ae ~ ad polers Sergiy Get ak len Type pedtoe , te Disk Andhais Z assert at pel: Benen Of cer nettl ¥f Se Having demolloated that infact all uterances ae pefornane ‘sterner th sees fx Hic Mo eyvemaing maybe 02 fa smelly cones hn "rng ay caer esis "can perioem thres’ scr stinltenndesi: x leatiosars set which fe tbe tet yng someting Inthe fll pee ‘ay an Mla 9ct stich an et performed i saying someting thes Tendo by Ibe excl petermsties and Pacis sacperined Bipe a tak pag, THES Ti Cl if ald tome Shon he? meaning by boo!” sno and refriog by pa Aa 0 hein Ts rr ttre) et Aa Gu Pen ids pana ow ie ep. i) Zi nebo \ ein ig tae It is not Austin’s intention to suggest that in speaking one has the ‘option of performing one oF other of these acts; one er all three simultaneously, but it i useful for analytic purpose tinguish them, Austin first distinguishes locutionary and illocutionary acts. While “wo perform a locutionary act i in general, we may say, also a to perform an illocutionary act’ (p. 98), the interpretation of ccutionary act is concerned with meaning, the interpretation of the lgcutionary 2G wife Nh ne ustin glosses ‘meaning’ unhelpfully as the use of language with ‘a certain more or less definite “sense” and a more or less definite “reference”, but Strawson (1973) clarifies things by asking what a listener would need to know, so that he could be said to know ‘the meaning of precisely what was said’ on a given occasion, He points out that @ complete mastery ofthe linguistic sys~ tem, syntav ies, is almost always insufficient: any stranger listening Toa tape-recording of the uiterance John will get here in sw ours Tom now, would Enow either the person rl johm?-nor the tine and place designated by 4 ust Be seen and is the anna speaker Tiere, Thee sera in WBE he listener may, however, not Fave understood ‘how what was said was meant, that is whether the illocutionary fore of the locution con ceming John was assertion, prediction oF warning ‘The locutionary/illocutionary distinction is not an easy one. It could be argued that in explicit performative utterances like ‘I warn you ‘eres a ll in tha el (@ Knin dhe meaning of the lcutonay Spi a andcmroitinat int Wal a actalready co knowih illocutionary force; and Cohen ( Gates “in what way does the illocutionary force of 5 fom that part ofits meaning which belongs to it in virwe of its per- formative prefix. going on to argue tha ilocutionary forces do not in fact exist{ Strawson (1964) accepts that in Forse’ while Searle (1969), in a similar vein eyGohem) argues that ~) ibere « ceria fore i part of the meaning, ere tbe teasing SOV onigiely determines pala fos, hese tre not two-ifleen 9 (= acs, but two different labels for the same act’, and reaches the cine oy “lusion that there are only iliocutionary acts. These criticisms are i fact unhelpful and appear to pun on the meaning of ‘meaning’ for 2s Forguson (1973) observes, ‘even if there are cases in which mean= ing completly determin force i isn the sae thing a force’ in_himself expected the distinct illocutionary and ke perlocutionary acts to give more trouble, Basically andifocutionary ae ‘is a linguisiic act performed in uttering certain words in_a given ci Ww, The OU maa pc re pats Vee TORsequence forming the locutionary and illocutionary et Teved Through the uttering of certain words is potentially under the complete control ofthe speaker: provided he uses the correct explicit performative i ‘an be certain that the act will BSchappy’ — no one can prevent ‘Soneane Tan Warning oF alng them excep by refusing to liste, ‘The associated perlocutionary act, however, isthe causing of a change in the mind or behaviour of the listener, so that he becomes ‘alarmed’, ‘convinced’, ‘deterred’. Thus the act is the effect of the utterance on the listener, but because this is not an effect governed by convention — there is no conventional or reliable way of linguistically “convine- ing’ or ‘deterring’ someone — I may warn you hoping to deter you ‘but in fact succeed only in encouraging or even inciting you, For this reason, Austin feels it necessary to distinguish between perlocutionary objec, basically the intended result of the illocutionary fact, and perlocutionary sequel, an unintended or secondary result. It is in this way that Austin solves the problem raised earlier of ac~ ‘counting for those actions like ‘insult’ for which there is no perfor~ mative verb, therefore no illocutionary act and therefore no perlocutionary abject; some nerlocutionary acts can only be sequels: ‘thus T may surprise you or upset you or humiliate you by a locy though there is no illocutionary formula I surprise you by . upset you by ...”, “T humiliate you by...” (p.117). Unfortunately20 An Introduction to Discourse Analysis ‘Austin did not pursue the investigation of perlocutionary objects and sequels, but such a study could reveal persuasive and oratorical tech niques and form the substance of a companjon volume How to achieve things through words. , Qlzoely 2xv2& rom the discussion so far it will be evident that Austin ataches considerable importance to speaker's intention — he argues in fact that ifa listener misinterprets an utterance the speaker should be re his position ereates (wo major problems: First, the unstated as- sumption i that each Jacution has only one illocutionary force; bu, as Searle (1965) argued persuasively, primary performatives a ‘only potentially ambiguous but often deliberately so: suppose at a party my wife says It’s really quite late’ That utterance may be at one level a statement of fet; to her interlocutor, who has just re~ marked on how early it was, it may be (and be intended as) an objection; to her husband it may be (and be intended as) 2 suggestion or even a quest (Let's go home.) as well as a yarning (You'll feel rotten in the moming if we don't). Second, there is thg problem of discovering what the speaker's in- tention was, something literary cries have Tong rogarded asa Tress éndeavour, and of deciding what in fact has happened if no illocy- iimyy act Bag eer peta However, a eee ana cage fn use have discovered, there i, fortunately, no real need to concern ‘oneself with the speaker's intemtigg because interaction proceeds ac ‘eng to the ia SA Gilerpretatis of the Toree oF an UTerance. imondson actually suggests “hearer-knows-best” principle according to which, Is interpretation of S's behaviour may be said to determine what S's be~ Jhaviour counts as at that point of time in the ongoing conversation: this allows of the possibilty ofcourse that S may self-correct — i. the hearer= Inows-best principle may be applied sequentially. (p. 30) Subsequent developments Austin’s theory is suggestive, but he died before he was able 10 de- velop it. One significant gap is that whereas he proposed four con- ditions governing the ‘happy’ production of ritual or archetypal performatives, he suggested no conditions or rules for other perfor- atives. Searle (1965) attempts 4 detailed discussion of one conventional Mfocutionary act, /fromise’) to garded not as having (accidentally) produced a different ilocutionary et but ay Taving produc€ no act at all“the performance of an i ToeaigRAPy ac avoWes the securing of pie, that sated opal, TE “Thy be playing football, though i Specce acts and conversational maxim — 21 explicate the notion of illocutionary act by stating a set of necessary and sicient conto tr he perocmanee oC atic Knd Leu ey-acr-and enthacting from it a set of semantic rthe use of the expression (or device) as an illocu- He chooses not to separate an utterance into locutionary and illo- cutionary acts, preferring to see it as consisting of two (not necessarily separate or even separable) pars: p-propeition, and afimeion indicating which marks the illocution® its ‘a miscellaneous group concerned with attitudes and_ social Behaviour — apologize, cfitici7é, Bless, challenge; and Sxporititey, which clarify how utterances-Gi into ongoing discourse, or how they afe being used — argue, postulate, affirm, concede. —— “Fiore, Tae ar problems vib ir caniealon, es Seale (1976) points out: ‘there is no clear or consistent principle or set of _ principles on the basis of hich the-sanomy-is construct ‘and ereTore a very large number of verbs find themselves smack in the middle of two competing “categories” — for example, Austin lists ‘describe’ as both a verdictive and an expositive. However, the fun- damental weakness of Austin’s classification of illocutionary verbs is that it is just that, a classification of illocutionary verbs, As Searle comments, ring those Tor which there is no lexical label. This means that on one taned"T ofder you to", T request you to’, ] beg you to’, ‘I entreat you to’ are necessarily regarded as different though “all eould be expansions of the sanfe-primary performative ‘put down that gun’ uttered by speakers with Wiffemng status relative to their addressee; while on the other hand, whereas one can report both “| complimented her’ and insulted her’, only the former can be classi- fied as. report of im Mocutionary a> |.) yearly) argues That iti MOTE reasonable to think of speakers as performing a limited number of Mocuttonary aets and To see the oe Tomaprexes carrying other TalormaliTh Srenher 6 hee — a eae fee ox cece dterencesTT TE relative status of speaker and hearer, suggest/ propose/insist with variation in the strength with which the illoc- tutionary point is presented, and oast/ament and congratulate) console with ‘differences in’ the way the utterance relates to the interests of the speaker and hearer’.24 An Introduction to Discourse Analysis ationally unimpc “ipled explanation for ‘Searle argues that there are three major ways in which speech acts that p° and ordering ‘wanting that expansion, but onl Using these three dimensions, Searle proposes five macro-classes of iMlocutionary act: representatives, directives, commissves, expressives and declarations, Eek pret the point or purpose isto ‘comm speaker to something being the case’ — in other words, tis an nee h the speaker fits his words to the world and which ~ orporaies his BELIEF that p™. The degree of belief can obviously vary between swear’, Suggest" and ‘hypothesize’ and affective fea- tures can be incorporated as in ‘boast’ and ‘complain’ Piratixes are all attempts by the speaker to get the hearer to do thing = this-class the speaker k WANTING to achieve a “ature sion In Which-the world will marctrtis Words and thus this a a-simply ‘order andtequest but, more Sine vie’s“aare” and “challenge” Commisiogsa category taken over intact. foe Hos are Tike dzecves concerned with altering he wast vo suai ar WORE but ie tie pols camel she queasy sell oacting and it necsssacily involes INTENTION. no dynamic relationship between swords ane orn ane PTE primitive psychological verb. Instead ‘the Wlocutionany point of Tis class is express the psychological state specified in-the sincerity Speech acts and conversational maxims 25 about a state of affairs specified in the propositional content’, As ex- apes be os vgs meee OER Rk cles, daaies,covmias of act which in their uring tier the world smuSincludes iim of thase which Ansta Tet cone “eared as perfomaates. They piel eequre an exe-Engusié Aston which pve AUST ei wee Ga SS rahe Fe pel Cae of deskions ome ered Wi ge use lisetft-define;-abbreviate, Tame, call or ech (1983, p- 180), however, ares that there are good reasons for regarding most of these as not illocutionary acts at all Decaute "hey are conoeatonal rather than common act the linguistic parts of val Any attempt to 1s immediately throws up problems. First”s Willis (1983) points out, at the begin —throws_up problem Tings and’ ends of nfomy interactions and at strategic points during {Fe partpants produce utterances which are baskally Gare [ould one see hello” as a directive requiring a second ‘hello” (but friiBlst cave how does-one categorize the xecond one), ox'as'an ex pressive (but expressing what)? Rather, it is a propositionally empty deus wi a nego af cheated raisins madng 2 Vagiolag Secon) many representatives art ackmowled ye by’ the: hearer ss We raining again, wi: Yeah, as an overt signal of uptake’. Third, and most problematicaly, Searle regards questions as es but it is difficult t see which words the addressee is being asked to change the world to match — it seems much more insightful to see questions as a separate category which functions interactively to elicit instances of the other four major classes: and in these cases oe utterance appears to function merely 2) What time isi? Tes four o'clock, mre 'b) What can I do to help? Open the door. ose 6) Can you help us? Mil give a cake. cox 4) How do you feel? miso happy. #80 It is interesting that Lee Gaya his development of Searle's pro- posals decided to set up 4eategon6f gates tyd thus to separate Out questioning items from directive Searls proposals are obviously a frst but very suggestive attempt at casifation — Leech, as we have seen, adds one category; Wills Proposes three more and suggests more detailed sub-classii- at secondary delicacy while Stiles (1981), using a if26 Au Introduction 19 Discourse Analysis {or classification, also proposes a_division into cight categories. All these proposals represent a significant step Torva free speech act analysis from dependence on acts that happen to have been led ty speakers, and open the way to 8 more: soumly-ased indly-base ‘conirastive analjsis and thus to applications in the Field of language teaching and Tanguage acquisition. Indirect speech acts Searle's classification of illocutionary acts suggests that the problem of interpreting primary performatives may not be as great as had orig- inally been thought, but it certainly doesn’t solve the problem. (The liscussion at this point is complicated by the fact that different de- scripive categories are used — even for Searle (1975) the directive category doesn’t include questions, and representatives and commis sives are apparently subsumed under statements.) Searle formulates the problem of primary performatives “When the grammatical moods declarat 1, Can you pass the salt? 2. Would you pass the salt? 3. Pd like the salt 4. You ought to pass her the salt. Sentences 1-3 are representative of a large set of utterances which Sadock (1974, 1975) has maintained are in fact primary performative versions of ‘I request you...” — an analysis he justifies by the fact that they can all co-occur with ‘please. Sadock argues that in such terrogative-oF dedlarative item shot ‘be broken down but treated uunarralysect-as ome Conventfonal way of conveying w request. Certainty here“ someimanational support for this — as we point out on .13L, in many occurrences of such utterances the inital phrase is ‘marked intonationally as uninformative by being unstressed or not prominent (see pp. 102-4 for an extended discussion of the sig- nificance of prominence) However, her ae nor probs wh an tempt Ga Hteciness In terms of om. Firs, i cannot cope wih all the data np A shar Be requested action i= imple oF hinted 3 mt be eli ter was. Becontree ‘yinaly Tong. Phird)as Levinson (1983, -270) points out, “idioms are— ‘ampasitional and are therefore likely to be idio- ‘Speech acts and conversational maxims — 27 syncratic to speech communities . Chowever) most of the basic... structures fanslae_aeross Languages. Finally) and as Searle himself points out, most importanily, the addreSsce can respond vo both the) surface form and the underlying force: ure. (passes the sll) Va, Can you pass the sa ‘These examples are easy to instance and accept, but Searle goes on to generalize, suggesting that ‘the man who says “I want you to do this” literally means he wants you to do it’ this leads him to argue that when such sentences are uttered the literal ilocutiomary act is also performed, and thus he talks of indirect speech ats, that is, speech acts performed inditectly through the performance of another speech act As supporting evidence for his claim of simultaneous performance, Searle cites speakers’ reports of utterances, observing that ‘Can you pass the sat” can be reported by an utterance of ‘he asked me whether ?, but ths is an unreliable criterion because mothers ean frequently be heard complaining ‘I've asked you three times to ...’ when they have been heard by all present to use straight imperatives. Also, the following teacher directives don’t seem to admit verbal responses easily: ts How many times have [told you to vf Seven, sir ‘1 Who's talking now 2 Me, sir +: Can I hear someone whistling Yes, sir ‘The debate continues, but in fact one doesn’t need to accept the claim of simultaneous performance to appreciate Searle’s analysis of the options available for indirectly directing. He suggests that the possible realizations can be grouped into six categories: 1, Sentences concerning hearer’s ability; Can you pass the salt? 2. Sentences concerning hearer’s future action; Are you going to pass the salt? 3. Sentences concerning speaker's wish or want; 1 would like (you to pass) the salt 4. Sentences concerning hhearer’s desire or ‘Would you mind passing the salt? ingness; I might help if you passed the salt 5. Sentences concerning reasons for action; 6. Sentences embedding either one of the above Can I ask you to pass the salt? can explicit performative 1 don’t think you salted the potatoes.28 Aw Introduction 19 Discourse Analysis Searle observes tht the fing€TEED> types refer to the thre fl conditions on directive iMocutfonary HIS Which he proposed in 1969, ‘Fespectively preparatory, concerned with the hearer’s ability, Bropo™ atonal conieha, Concerned with the Futurity ofthe action; and since, Comterned withthe speaker wanting the hearer to perfarm NE wetion He combines groups 4 and 5 arguing, Got entrcconvincingy that “both concern reasons for doing A >. since something is a reason par cxccllence for doing it’. He is then able to. Show thar speaker cam make an indirect decve by 1. cither asking whether or stating that a preparatory condition con- cerning Hs ability to do A obtains; 2. either asking whether or stating that the propositional content con- ato obtain: 3. stating that the sincerity condition obtains though not by asking tthether it obtains Giterestinghy, though, questioning the sineriy condition can function to request the hearer to dest: “Do you think T enjoy isteing to you whiting”) 4, either stating that or asking whether there are good or overriding reasons for doing A, except where the reason is that H wants or wishes, etc. to do A, in which case he can only ask whether H wang or wishes, et do A ‘These penerliatisrepreseat pomecfal desetthn of it data hey eteeetze exh eons ese ways OF ctor ta indreer deere — but thre is mo atempt to aplate why theae ace ong al mal y optic atr haar hee a lve ttterance lke ‘Can you pase tne sl ct abou deciding whether the spasker onde Yeas a segues or'aquestoe.loceed, Rese ais tag Saban "ett ane way dog ect ec ery js just a question about his abilities and when it is @ request’, but_ cbserte unhelpful that tls at this point ht the general pringigles St eomeration (gether with factual background information) come inp iis interesting. at this pint to ‘compare: Seaile’s description: wit lint propoved oma diferent pepecive by Labov and Fanghel (1977). They begin first by characterizing the prerequisites for an bmeranee mperive In form wo be head at vald resi orinthels sea gue ac IFA addresses 19 B an imperative specifying an action X ata time Ty and B believes thar A believes that 1, a) X should be done for a purpose Y (ne forthe action) 1b) B would not do X in the absence of the request (nend for the reques!) 2 Bhas the ability w do X Speech acts and conversational maxims — 29 3. Bhas the obligation to do X or is willing o do it 4. Abhas the right to tll B to do X, then A is heard as making a valid request for action. (p. 78) Imperative urterances which fail to satisfy one or more of these pre- ‘conditions are, in. Austin’s terms, infeliitous, and may be variously interpreted as cheeky, insulting, joking or simply irrelevant. The rule so far only covers those utterances in which there is a close fit between intended function and formal realization, tha is im- perative directives; but as Labov and Fanshel observe, these are the minority of eases. They therefore offer a rule for indirect requests | TRA makes to B a request for information or an assertion about 4) the existential status of am action X tobe performed by B 3) the consequences of performing an action X 1) the time Ty that an ation X might be performed by B {any ofthe pre-condition for ald request for Xap give inthe Rul and STORET pe codons ae infect then Ae ea x mang ald request of B forthe action X. (p. 82) “They cite as an example an utterance from a therapy session, ‘well ‘youknow, w'dy'mind takin’ thedustrag an’ just dustaround?, and ex- plain that this is interpreted as an indirect request through being recognized as a request for information about the third pre-condition for valid requests, B's willingness. They go on to observe that al- though they are in the main concerned with ‘the text as it actually occurs their discourse rules represent ‘a general grammar of possible specch actions and possible ways of executing them’. Thus while the indireet request to ‘dustaround actually occurred inthe form quoted above, there are many alternative ways in which it could have been realized by questioning or asserting other pre-condition: 8) Bisental status Have you dusted ye? You don't seem t0 have dusted this room yet by Conseyuences How would it look if you were to dust this ‘This room would look alot better if you dusted, ©) Time referents When do you plan to dust? TTimagine you will be dusting this evening. 4) Prevanditons 1a, need for the action: Don’t you think the dust is prety thick? ‘This place really is dusty 1b, need for the request: Are you planning to dust this room? T don't have to remind you to dust this room.30 An Introduction 19 Discourse Anabsis 2. ability Can you grab 2 dust rag and just dust arou You have time enough to dust before you go, 3a, willingness: Would you mind picking up a dust rag? I'm sure you wouldn't mind picking up a dust rig and just dusting around, 3b, obligation: Isn't it your turn to dust? You ought to do your par in keeping this place dean 4. rights: Didn't you ask me to remind you to dust this ace? T'm supposed ro look after this place, but not do all the work. (83) ously the examples above are just a few of the large number of direct formulations ofthis particular request: as Labov and Fanshel ‘observe, there is an ‘unlimited number of ways in which we can refer to the pre-conditions and this poses a serious problem if we want to make firm connections between these discourse rules and actual sen- tence production’ (p. 84). Of course, a given indirect request can be made in an ‘unlimited number of ways" only if tis eonsidered in iso- lation;-in-reatiny, the constraints of the preceding discourse, the cur- rent-topic; the facts of the situationand’ the current speaker's ‘intentions for the progres of The succeeding discourse willall reduce the choice enormously, — - ‘The situation is not quite as simple as Labov and Fanshel suggest — the following is a counter-example to their claim that in producing an utterance following their rules, ‘A will be heard as making a valid request of B': : Maleolm, can you open this for me. Me: Tdon't know © No, I was making a request. o In fact, the Labov and Fanshel analysis has the same problem as Searle's: both need, as Levinson Gibid.) points out, an associated in- ference theory to explain how a listener comes to reject the direct in terpretation and select the indircet one — the most convincing pro- posals are those of Grice (1975) Conversational maxims As Garfinkel (1967) observed, it is never possible to say what one rmeans in ‘so many words? — speakers require hearers t0 ‘work’ to. greater or lesser extent to derive their message from the words uttered Grice (1975) offers an unremarkable example: ‘Speech acts and conversational maxims — 31 Suppose that A and B are wking about a mutual friend, C, who is now ‘working in a bank. A asks B how C is geting on in his job and B replies (Oh quite well, I think; he likes his colleagues, and fe hasn't been to pris ye Grice observes that in addition to what B has said he has implicated ‘something else — he has provided information from which A can de- duce extra infosmaat Make your conversational contribution such as is required, at the tage at which it occurs, by the accepted purpose or direction ofthe tlk exchange in which you are engaged. ‘This principle implies decisions in four major areas, relation, quality, quantity and, manner, and their significance is spelled out by maxims: 1 raion be vn ¥ 2. quality a) do not say what you believe to be false 1b) do not say that for which you lack adequate evid 3. quantiy 2) make your contribution as informative as is required (for | the eurrent purposes of the exchange) +) do not make your contribution more informative than is required 4. manner 2) avoid obscurity of expression ') avoid ambiguity ) be brief @ be orderly Tes important to realize that these maxims do not represent a de- scriptive statement of how conversational contributions are: firstly, there will always be occasions when a speaker decides to ‘quietly and ‘unostentatiously VIOLATE a maxim’ — he may lie, he may not give as much of the relevant information as he could, or he may, like the Delphic oracle, offer utterances which are only later seen t0 be am= biguous; secondly, and much more importantly, there will be occasions ‘when a speaker is seen to break # maxim either because he has been faced with a CLASH between two maxims making it impossible, for instance, for him 1o be as specific as he ought to be and still t say nothing for which hie lacks adequate evidence, or because he has chosen to FLOUT 2 maxim, ‘that is to say he may blatantly fail to fulfil it, In such instances the conversational maxims provide a basis for the ite wht eigenen jn =32 Ave Introduction to Discourse Analysis. Grice exemplifies the process with reference to his example about not having been to prison yet: Jina suitable setting A might reason as fellows: 1. B has apparently violated the maxim ‘Be relevant’ and so may be regarded as having flouted one of the maxims conjoining perspicuty, yet I have no reason to suppose that he {s opting out from the operation of the CP; 2. given the eircumatances 1 can regard his irrelevance 2s only apparent if and only iT suppose him to think that C is potentially dishonest; 3. B knows that I am capable of ‘working out step (2), So B implicates that C: is potentially dishonest. ‘Thus what is being claimed is a two-stage process — firstly recog nition of the apparent irrelevance, inadequacy of inappropriateness of the utterance, which secondly ‘riggers’ (Levinson 1983) the subse- quent infereneing, ‘This theory obviously supplements the descriptions proposed by Searle and by Labov and Fanshel and allows for occasions when the trigger fails and the listener takes the utterance at its face value, but it still has two crucial weaknesses. Firstly, there is no attempt to ex- plain why a speaker might choose one form of flouting rather than another, given that there is an infinite set of possibilities though Brown and Levinson (1978) and Leech (1983) make interesting attempts to explain some of the selections in terms of extra maxims concerned with politeness; secondly, and more worryingly, although Grice is centrally concerned with hearers’ inferencing, his examples are always explanations of one interpretation rather than a discussion ff how an utterance with a series of potential implicatures comes to have in the context only one. As Sadock observes: the Co-operative Principle has been very believably invoked, eg. by Searle (1973), to account for the fact that an utterance of ‘i's eold in here’ can convey a request to close a door. But it can also convey 2 request to open 4 door oF to bring a blanket or to pay a gas bill. In fat itis difficult to think of a request that the utterance could NOT convey in the right context. ‘Thus we are left with the conviction that only a theory of inference ccan cope with the way in which speakers derive meaning from indirect. utterances, but also with the knowledge that Grice has only taken a first step towards the solution. a eee hs BHMIMENET UAT aT epee phe th Le walt ott fo my poids 3. The ethnography of speaking set the goal of linguistic theory as the description of the hearer S{eompetence, his Knowledge of grammaticality, shi yetween ‘he hit me’, ‘it was me that he hit it was him that hit _me’, wighut attempting to explain why one and nor nother might be appropriate to a particular situation. Hymes (1971) argues that Chom- sky's dcfinition of competence is too narrow — linguistics ought to concer itself with comminicattoe competence, the speaker's ability 10 produce iate utterances not grammatical Sentences fe suggests that “an adequate approach Ht in vestigat€ four aspects of competened: systematic potential, apprapriate- ness; occurrence; feasibility By systematic potential he refers to ‘whether Se ee rating tt io eluet at eager PE “iris to this that Chomsky in eect reduces competence’. Appropn-— — ateness includes ‘whether and to what extent something Context suitable, eTfective oF the ike’. These two feanures can vaty “M2 independently: schizophrenic’s talk is often marked by grammatical but Za as in this example already quoted” Rom Labor (1970), 2 What is your name? 1 Well, let's say you might have thought you had something from before, ‘but you haven't got it any more. ‘Tm going to call you Dean, while Albert (1972) reports that among ¢he Burundi appropriate but Jungrammatical utterances occur frequenity in certain sitations — ifferences in rank require a peasant-farmer to make ‘a rhetorical fool {Tims when his adversary ia prince o herder along at oher fimes he ‘may show himself an able speaker’. =A speaker's competence also includes knowledge about occurence, — Cihether and to what extent something is done’, This theoretical di in some AyperprskThe ethnography of speaking 35 34 Am Introduction wo Discourse Analysis particular language or dialect; he angus that it is possible for speakers mension provides for the fact that members of a speech community to share formal linguistic features, phonology, grammar, Iexis, but still are aware of the commonness, rarity, previous oecurrence or novelty (abr Geble IioISPaCGIT cach ners messages. Forename, f many features of speech, and that this knowledge enters into their Labor (1972s), in a discussion of aspects of language use among ado- definitions and evaluations of ways of speaking’. The final dimension lescent New York negroes, presents utterances like ‘your momma's & jeasibily +s concerned with ‘whether and to what extent something peanut man’, oF our mother's a duck’, which are superficially intel- {s possible’. Hymes (19728) refers to the experience of]. R. Fox work- Tigibte but whose real significance as ritual inl is not avaiable 10 ': pesstebing among the Cochiti of New Mexico, who was unable to elicit the most English speakers. first person singular possessive form of ‘wings’ on the grounds that ‘Speakers who apparently share the same language may also have the speaker, not being a bird, could not say ‘my wings’ — ‘only 10 different ‘norms as to greetings, acceptable topes, what i said next ‘become the frst person able to say it in Cochit, on the grounds that ina conversation’, how speaking turns are distributed and so on. ‘your name is Robin”. example, Sacks, Schegloff and Jefferson (1974) suggest that for Any utterance, oF extended piece of discourse, can be described “American English there isa conversational rule that only one speaker | in terms ofthese four dimensions. Thus Hymes suggests that Lcontes’ speaks at a time, whereas Reisman (1974) observes that in Antigua speech in Act of The Winters Tal is ungrammatcal, appropriate, “the start of a new vice is notin itself signal forthe voice speaking ‘individual and difficult, while the bumbling speech of the Burundi to stop or to institute a process which will decide who is to have the | Peasants is ungrammatical, appropriate, common and awkward. Even floor’, Any group which shares both linguistic resources and rules for if the scope of | were expanded to cover these four aspects interaction and interpretation is defined as a speech eommunity and its oar | of Compson He dls it would still be to0 narrow and in his Seri ere or dissatisfaction TeeTs the need to propose ‘a second descriptive science i ce limiting eriterion of a speech community isthe sharing be ! of language’, the ethncgraply of speaking, concerned not simply with of one linguistic variety, most communities have several between which language some uw ng ta Wh "rules. of speiting they switch. Blom and Gumperz (1972) report an investigation in the «the ways in which speakers associate particular modes of speaking, Norwegian village of Hemnesberget where all the residents speak both topics, or message forms, with particular settings and activities’ the standard language, Bokmal, and the local dialect Ranamal. Bokmal {W9Taa) ay desepion oF ays of speaking” wil need vo provide iia tenginge al eda tii tayetooe, fon so abe along four interrelated dimensions: the mass media, but the local dialect still enjoys great rests, a eee by “identfving himself as a dialect speaker both at home and abroad, ect ae tan a ee eee amember-symbelizs his pride in his community’. In any interaction (Hcprs-sentential sirueturing — how many differently structured ti speakers have a choice of two varieties; Blom and Gumperz. were in- ‘guise events, like trials, religious ceremonies, debates, songs, are terested in the factors which influenced the use of one rather = athe nilec oF the other. After close observation and analysis of tape recordings of ares eee renege eect free speech they concluded that locals would typically use local dialect fe norms which govern different pes of interaction. except in situations ‘defined with respect tothe superimposed national Nonvegian system’. Even then, inthe community eministration ofice, Obviously, any attempt to produce a description in these terms would where the standard language prevailed, clerks were observed to switch bbe an enormous and perhaps impossible undertaking, and thus all the depending on topic and, ‘when residents step up to a clerk’s desk, work so far attempted within this framework is necessarily parti greetings and inquiries about family affairs tend to be exchanged in the dialect, while the business part of the transaction is carried on Si in the standard”, Ln_other words, it appeared that topic could only The speech community peskert to svisch femr standard ty Saleet =" whereap ara a ‘The dial ass the is dard language situation talk about family alfsirs might be conducted 10 produce ‘rules of F-\Hymes (1972a) stresses that it ‘in dialect, in a gathering of friends and kin speaker ‘never is not adequate 0 define a group as-all those who have access to a a36 An Introduction 19 Discourse Anabysis To test their hypothesis Blom and Gumpere arranged to tape- cord the conversation of two groups of local residents, both self recruited and consisting of close friends and relatives. On both asions the investigators first stimulated discussion among the group land then as the conversation progressed interjected questions and Comments fecling that the greater the range of topics the greater the chance of a switch to the standard language. In fact, as predicted, ‘in several hours of conversation ... marked by many ehanges in topic, [they] found a number of lexical borrowings but not a clear instance of phonological or grammatical switching’. At this point it would seem possible to write rules for the speech community of Hemnesberget to predict the choice of one or other speech variety for one type of speech event, conversation, However, Blom and Gumperz recorded conversations among two more groups, i ‘ne composed of members of a formerly active peer group who had ] spent the past few years away at university, returning only in the sum- met, the other comprising three speakers from families who tended to dissociate themselves from the local community. The students claimed to be pure dialect speakers, but for them topic mas a significant vari- | able: non-local topies evoked ‘a tendency to switch towards standard phonology while preserving some morphophonemic and lexical fea~ tures of the dialect. For the other group the local dialect was only used for local anecdotes, humour, and attempts to provide local colour and the standard language was the normal speech style. ‘Thus, what appeared to be one speech community sharing two dialects was now seen disturbingly to be three, distinguished by different selection rules. Dorian (1982) raises a different question about membership of a srech commoniy — wha ihe sas of senses nea passive bilinguals? In the community she investigated, whereas they said very little and produced grammatically deviant unterances, they wese _culturally fluent: ‘unlike the linguist-guest [they] were never uninten- Tionally rude. They knew when it was appropriate to speak and when not; when a qfiestion would show interest and when it would con- Siitute an interruption; when an offer of food was mere verbal routine ” Hymes (19822) appears to accept that such are members of the speech community and thus to go along with Corder's (1973) dete nition of a speech community which Dorian quotes: ‘people who regard themselves as speaking the same language’ (p. 53). Laboy (1972a) argues for a different definition; ‘The ethnography of speaking — 37 the speech community s not defined by any marked agreement in the use Of language cements so rach as by participation ina set of shared norms Sen ayes cbc in ree npe of elute oshaviut sh by the unformiy of attract patterns of variation which ae invariant in espect to parculr Ievels of usage. pp. 120-1) ‘This then allows him to sce the whole of New York City’s population as-2 single speech community because they react in similar ways to phonological variation despite the acknowledged major diferences in grammatical usage and norms of interpretation which he highlights elsewhere (1969, 19726). "As these three examples make clear, the speech community, though sen well and poner ene canes. et peskens 7 do not fll neatly into categories, but just as we can say very useful things about languages and dileets even though isoglsses don’t fall neatly one on top of another, so we can make useful generalizations about language use in speech communities, and just as some linguists have restricted themselves to statements about their own idiolects, so some ethnographer of speaking may eventually produce detailed rules for two-member speech communities. Speech styles Any ethnography of speaking must describe the linguistic options ‘open to the speech community. As we have seen, the residents of “Tlemnesberget had a choice of two major varieties, the local dialect and the standard language. Ferguson (1959) suggests that speakers of Swiss German, Arabic, Greek and Tamil are faced with a similar choice, this time between two standard languages — a high form typi cally used in sermons, speeches, lectures, news broadcasts, and a Jom variety used in conversations, political and academic discussion, ‘folk’ Titerature. By contrast Americans, according to Joos (1967), havea choice not between major varieties but betwcencfive different degrees ‘of formality within the one standard language; Labov (1968) provides Supparang evidence, Hom the diferenial occurrence of post-vealic Wot fom dees offomiy. Geertz (1968) reports that Javanese has three major styles which, unlike those suggested by Joos and Labov for English, are recognized and named by speakers of the language — ‘krama’, ‘madya", and ‘ngoke’, high, mid and low. These svles share some linguistic features with other levels, but also have unique lexical items and grammatical constructions. In addition there isa set of honorifics, mostly referring to ‘people, their parts, possessions and actions, which occur inde-38 An Introduction 19 Discourse Analysis TTARLE 34 = ~ x . . {2 {3 | ay i Bi: [a Fr) GL]E | fF gles gle |? Ub) Aue |e | glez| Bes] 28/83 S[S% | 23 (22 |23| 33 3 : 5 2 3 3 5 2 z z 54 a if Poe | dls g i Fl & (Geertz 1968, p. 284) The ethnography of speaking 39 pendently of the first style-defining set of choices and raise the style ‘half a notch’. These honorifics can only occur with the low and high styles, at least in the usage of the educated townsman, who thus has ie recognizably distinct varieties to choose between. Thus the ques- tion ‘are you going to eat rice and cassava now? could have any of the five realizations on page 38 depending on the context and relative status and familiarity of the interlocutor — a townsman would use low With a friend, mid with a non-intimate and high to a high offical from whom he would in turn receive low. Halliday, McIntosh and Strevens (1964) suggest that it is possible to make much finer distinctions than this and argue that a speaker's linguistic repertoire consists of a large number of varieties, or registers, epi, Tor speech communities within the same language area could is they had avaiable(bak a sports commentary, a church serv ‘bviously linguistically distinet’, and suggest that there is 4 resister ‘appropriate to each. This raises the question of how different regis ters can be recognized and isolated and they suggest that while there will be grammatical differences the major distinctions between regis- ters will be lexical. However, the claim that some lenca items sufi almost by themselves to identify a certain register: “cleanse” puts usin the Language of adversing, “probe” of newspapers, especially headlines, tablespoon” of recipes prescriptions, “nec fof fashion reporting or dressmaking instructions, is worryingly naive. In fact there are no restrictions on the concept: 1 register can apparently vary in size and importance from that of dressmaking to that of scientific English (Huddleston etal. 1968); and a eee the language used in dressmaking. patterns Ts the register of dressmaking and the register of dressmaking is that used in dressmaking patterns. Gis) (1974) suguests tha iis - ‘munity as comprising a set-of spies, where(siylé is used in the neutral sense of ‘a way or mode of de somethii Vhereas style has often IeEWUSEA as a ConCepE to account for variation according to autho setting or topic it has never been used as the general basis of d Scription, This, Flymes suggests, is now possible if one exploits the long recognized fact that a li Se al Tor eanngle, St hn mtom pap te doy, OE aE Te40 An Introduction to Discourse Analysis things a description must dois account for the other items which could ‘occur as paradigmatic choices instead of ‘the’; the other thing a de- scription must do is to characterize the set of items which can occur following ‘the’. Drawing og Envin-Tripp (1972) Hymiz> suggests that the cancept of symtagmatic relations can be generalized of items over lar low one to ‘characters speech in so of the rules of co-occurrence among them”. ‘The concept of paradig- frat choice cn Be simile. asistalice aan aor a cope wih The Cicice between styles. “The © Concept of spe may scem very close to that of it there is 4 crucial difference: registers are mainly defined and recug~ nized’ hy Topic and_context-specific lexis — the gagister of sermons Pade is The Tanguage used in giving sermons; styles, Gowever) as the rules, of alternation emphasize, re not mechanically connected a patcular situations — speakers may ch ng stvles and their choices hav al msinine ne of the most reliable ways of making, people laugh ow a t0 adopt a strle inappropriate to a particular content or message. I is of course one thing to define a siyle a oices it a set of co-occurring another to isolate different styles- As Hymies observes, "the relevant speech sles of SoRMMURy oon be arrived at mech- anicaly, for one could note an infinite number of differences and putative co-occurrences’. he aim, therefore, isto isolatesignfcnt speech styles, that is ones that speakers can distinguish and use. iymes accepts that some stylistic Features may be present in piece iscourse without defining a sign 7 ‘simply convey a “tinge or character’, but not be an organizing principle fed tit dose Dene Aco of sty Te. However Fis jas recognizes two kinds of groupings of stylisieTeatures which do con. stitute organized use — those which colour or accompany the rest of what is done; ziplistic modes) and which can be said to define fecurrent formscstvistie struc. aulisie modes set of modifications entailed in consistent use of the voice in a in toning, chanting, declaiming, As an ex- importance of mode Hymes refers to the basic distinction among the Wolof of Senegal between ‘restrained? and ‘unrestrained’ speech, distinguished principally by paralinguistic features; restrained speech being characterized as low pitched, hreathy, slow, soft with final pitch nucleus, unrestrained as high pitched, clear, fast, loud with I pitch nucleus. (Silistc struaipes, as the name implies, are verbal forms organized The ethnography of speaking 41 in terms of defining principles of devel 1 One kindrot is the organization_of sentences and utterances imo larger units such as ‘greetings’, ‘farewells’, ‘prayers’; the other is the systematic exploitation of arbitrary Tinguistic features which Sinclair (1971) callers panering — at the rank of word posts frequently. __use such features as inital consonant, final syllable, positioning of stress to add an entra layer of patterning which we recognize as sl- Frannie and mete reactive Repettions at regular inter. TFaT Tice poner erent suuctures we call erst YOR Tipmes calls these strictures elementary ox minimal genes, and ob serves that both kinds of groupings of features, modes and structures, complex groupings called _gompier zenres,, A church re, Containing the-el- service would be an es ‘complex geri “The work of Bricker (1974) on Mayan provides a useful exempli- fication of Hymes’ concept of style. She notes an initial division into formal and informal genres, the formal comprising ‘myth’, ‘p ‘song’, ‘contemplation’, ‘planning’, ‘war’ ‘argument’ and ‘frivolous talk’, the informal three other pes of ‘rivolous talk’, ‘gossip’ and ‘discussion’. All the formal genres ‘are structurally alike: they are ex- pressed as semantic couplets; the informal genres have no common structure, but specifically avoid couplets. The following prayer illus- trates the couplet structure, Well grandfather, Lord: How long have you been waiting here for my ea? How long have you been waiting here for my mud? am gathering together here; 1am meeting here I see the house of poverty; I see the house of wealth OF His Labourer, OF His tibute-payer Holy Esquipulas, thou art my father; ‘Thou art my mather Obviously the organizing principle is the highly marked semantic and syntactic parallelism. These couplets happen to be from a prayer but apparently could just as easily come from @ song, for songs closely42. An Insraduction to Discourse Analysts resemble prayers in their context, content and function — the distinction istic mode, ‘prayers are simply recited, w' songs are sung to a musical accompaniment’. Naturally some genres will have a more rigid and overt structure than others — indeed until recently many considered that conversation had no identifiable struc ture at all. Hymes suggests that for conversation the distinctive modes and structures are simply more difficult to identify, and the work of Sacks, Schegloff and Jefferson, described in detail in Chapter 4, provides growing evidence of a high degree of structuring in conversation Speech events Hymes stresses that it is essential to distinguish unique combination of static sinicture ofa peo in order to emphasize the distinction between gen Performance, a distinction frequently obscured by users of a language, who often employ the same label for both, Hymes suggests the categ- ries of speech event and speech act to parallel complex and elementary genres. All genres have contexts or situations to which they are fitted Tin which they are typically found. Some genres, lke ‘conve sation’, can occur appropriately in a wide range of situations, some, like ‘prayer’, are highly restricted; however, itis a defining criterion of 4 genre that it is a recognizable style and therefore can be used in inappropriate situations. The cultural implications of an inappropriate aa See eee pect imoledone fe bapeomae Spee aE ee aochain events and speech acts that one writes formal rules for their occur- can dicoverHgusie stucturand arenot necessarily uF whi a i ‘coterminous wilh the situation; several speech events can OCUr sic= cessively or even simultaneously in the same situation, as for instance with distinet conversations at a party. The relationship between speech events and speech acts is hierarchical; ‘an event may consist of a single speech act, but will often comprise several. imate aim of the ethnography of speaking is an exhaustive ist of the speech acts and speech events of a particular speech com- ‘munity, though the descriptive framework is currently “heuristic” and. ‘quite preliminary’. However, any researchér atiempting a description in these terms faces several major probleis, Firstly) unlike Austin and Searle, Hymes doesn’t offer a list of speech acts, even for Engis althougti “request, ‘command’, ‘statement’ and ‘joke’ are used infor it wee an remit (oeiennae sctaserae SAE bt The ethnography of speaking 43 ‘oom interaction and admiing that general ene of speech ae would appear to be central’ (p. 32), does no more than discuss ert ‘cally more recent formulations of Searle (1976) and Dore (1979). 6 ye the absence of any clear direction those working within the framework of the Ethnography of Speaking appear to develop their own cat- o2* egories ad hoc. Saville-Troike (1982, p. 146) quotes a student dis- sertation which analyses the opening of Japancse door-to-door sae encounters as a sequence of si acts: ‘greting’~‘acknowledgement’= ‘identification’-‘question about purpose’-‘information about purpose’~ ‘exprossion-of disinterest/interest’. A second linked problem)is that Hymes, although warning B— are fot identifiable witay single portion of other levels of grammar’, realize them; thus, when Savile-Troike reports another study in gee which-an event opens with the sequence ‘greeting’~“acceptance of wiIpfs greeting’ (p. 156) the reader does not know whether this is equivalent 4°". to ‘grecting™‘acknowledgement’ or significantly different. In other words, unless one knows the set of analytic catcgories and how the) relate t3 The dala one isi Tact creating the Musion of classiicaio (Sinclair 1973). However, to be fair, no one else has proposed a nor contentious list of acts, though Chapters 2, 4 and 6 discuss partial solutions from different perspectives. « “third difficulty arises from the fact that whereas @he Neserip- sions eae ead de® units of analysis, Ciymes roposes only two Sh eae a massive Consensis AONE Fecardiors dor ne a ‘of event contain utterances which typi TS occur in pairs wih the fist constraining the occurrence ofthe fee _sctond, Bat altiough Hymes gives erecting’ s an example of «speech (i) ‘act he has no way of showing that greetings typically consist of (wo & paired utterances — indeed it is not at all clear whether he would regard the two utterances as a composite realization of the act ‘greet~ ing’ or whether each utterance itself is a Components of speech events So far the discussion of speech acts and speech events has concen- trated on stylistic mode and structure and for many acts and events these are the defining criteria — a song is a song whoever sings it; at least in our culture. However, some genres are performed for speci-44 An Introduction t0 Diane Analysis fic purposes in specified plies with particular participants, An An- Bicen baptism tational ase aoe poste ne seven essential participants — the parson, the unbaptized baby, the parents and three godparents — anj the definition and description of the speech event requites participants and situation, as well as style, to be specified, - tk Jn fact, Hymes recommends that for_every speech event the cth- fos ne ene cm care ei per a ‘poses, Tey_tapte channel Gpoken, written, whistled, drummed) and vitor, raccate form, 50 that knowing the possible parameters one can check re whether an apparently irelevant one isin fact irrelevant. Hymes re~ ports that Arewa and Dundes (1964), investigating the uses of lan- ‘guage among the Yoruba, chserved that proverbs were used only by adults and were always spoken, but pressing the point discovered they could also be drummed, though in a slightly altered form, and used by children as part of a formulaic apology. In other words, by being aware of the passible parameters the ethnographer can more easily and | successfully discover the constraints on the performance of genres, | and the defining criteria of particular speech events. y Senin Al speech events occur of necessity in time and space — sometimes je itis defi ria_of an event that it occurs at a specific mamsop time ot in.a specific place Foster (1974) describes a series of fificen noes agrcatural fesivals which the Iroquois eeebraw ae ceeaae | pf Points ‘during the year. At two of the festivals two speech events con- } fs, cemed with asking the Creator for successful crops occur, the To- bacco Invocation and the Skin Dance, while a the other thisteen, the ‘major speech event is the Thanksgiving Address, Salmond (1974), by contrast, reports a speech event among the Maori, the Encounter Ritual, which can occur at any time bur only in a ‘marae’, a complex conssin of aed meeting house anda courant for orators Closer to home we also have speech events tied toa particular time — special church services for Easter, or the Queen's Christmas messages or 10 4 particular place — there is a very restricted number of places where ‘marriages can be solermnized or jtigation occur. Even when a speech event is not restricted to a particular setting, the setting may affect cither the stylistic mode — people tend to speak in hushed tones in church; or the stylistic structure — Geertz reports that the Javanese ‘would be likely to use @ higher level to the same individual at a wed ding than in the street’. Hlymes stresses that the ethnographer must also take note of the The ethnography of speaking — 45 poycholgical setting of an event — the cultural definition of an oc~ ‘casion as formal or informal, serious or festive. Prake (1972) compares litigation among the Subanun and the Yakan. For both litigation is ‘an integral speech event concerned with scttling disputes by means of a ruling formulated by neutral judges’; the major difference is nos in-the-cventitself-but in its place jn the overall structure of the cul ‘ture, The Subanun divide activities sharply into festive and non- Testive; litigation is festive behaviour and often accompanied by eating, drinking and merrymaking. The festive nature of the ovcasion con- ditions the choice of style — both litigants and judges employ esoteric legal language, often arranged into verse form and sung to of drinking songs. Yakan litigation on the other hand occurs in a very informal atmosphere and the process is initially indistinguishable from ‘a group of people taking together’. ‘The underlying structure of both speech events is very similar, but the psychological setting and r suing style very different. Perinat Tea nll ypenct ee ber dest a rs parans a speaker who transmits sage ct 4 over, aii ibe male of numces the pera who tsaseke Jags ans the addrsor or he suo of the ‘sentient hat are being ‘expressed and the words in which they are encoded’ (Goffman 1979), Shr Mlle “Spolearar and ‘porta fe wonce het thee a ie, ‘Similarly, altfough there Se ae gol Fels Wee ‘lena oe "pa A eps ‘the Soviet ambassador was summoned to the Foren Othe to hear the views of ----/ cognive the roe Fortis reason (lua argues that ther arc atleast four participant roles, addressor, speaker, addressee and hearer or audience, and that while Earn ai a Teauitciony an adiremor eal ensetdreeee ote speech acts require different configurations. Labov (1972a) reports that ‘rl Seah rogue tree pardlpent roles onc beng a aence “Shone TUNE Galante each conrbuion, When one considers ens nore Guaples, Serr (1974) describes x fpeal event anon, the Cums caled ‘chanting’ in which two che(s perform a ualzed interaction in front of an audience — one chief chants, and at the af ask Verge eer resale; wT reaper fate chic mcs cate con "Then, when Auchan, is over a third participant, the chief's spokesman, addresses the audi- noe diet) and Interprets for thee# se non-hum a clap of juhunder, an old Tndian_asked his wife if she had, 46 Am Introduction to Discourse Analysis ‘There are some speech events which have only one human parti pant — for instance in our culture some forms of prayer. Sherzer (ibid) describes discase-curing events among the Cuna where the Participants are the curer and a group of wooden dolls, ‘stick babies’, which are considered to carry out the actual business of curing once Joerg hey ve been tld as aldresses, what to do, Uymes pois out that ans can also be taken as addressors, citing arr occasion When ny comprehensive description of a speech community must in- ‘clude data on who and what can fill the participant roles, and in what Rpeech events and speech acts. Some speech events simply req that csr parkipa roles be filed anyone er seas ules “Wa Fly OF MAT TT omer evens Require parcipan rs par sex specific; while in most Maori tribes only male elders can deliver ‘eular age, sex, Kinship relation, suits, role or profession —only Ke tina Chiefs can chant; initiation or puberty rites are almost invariably speeches on the ‘marae’. In other events turns to speak are regulated by relations between particular participants — Albert (1972) repores that among the Burundi turns to contribute to a debate are strictly controlled by relative status, with the most important speaking first, the Icast important last; while the Wolof have a rule that in greetings the lower status speaker begins fist (Irvine 1974). What has been said so far may have implied that assigning the roles iddressor, speaker, addressee, hearer, audience to participants is un~ rroblematicg-this would be a mistakes Two major problems arise: jere_are many situations in which participants change roles Ore ee Ses pi go and rapidly; and-secondi the definitions ofthe roles are ‘ot emirdy deny, For insta, usta ahem beedeeed or del ‘egated to express sentiments that are not his own, or can he select himeelf thee are radio interviews when the intensewer reads gues- tions sent ‘in by Tisteners, evidently acting a8 a spokesman, but then sand Teeturer ay pres lecture about Manx wi exensie Toadies fron the works —is he a spokesman dering the whole af he loca, during the parts when be ie dcedy. quote, or he, beemee hel teaching bout Man rather than teaching Margin fat an addvessor iv his own ight? Apu, i lecturer choowes trend = pre-prepared cna kate fara meeeen oe Reape eradaeenoe beara tae tame tet could be read it he were, by someone he ~~ and irhe niakesimprompt loses on his own text docs he then fever to being an addressor because, interestingly, spokesmen and readers The ehnography of speaking 47 of other people’s lectures can't interpret, they can only convey the information in the tex. ‘The problems-with addressee and hearer/audience are even mor com Goffmap \ibid ) poins out, in any conversation beiweeh ‘more than two participants there will be times when not all the ‘ratified hearers’ are being directly addressed and the movement in and out of the addressee role can be rapid and short term, One can also dif ferentiate berween specch events according to whether the ratified hearers have the option to switch into the addressor role — there are certain events in which they don’t, when they are generally regarded by native speakers as an ‘audience’, but in fact there are two kinds of audience, which the label obscures — one which is directly ad- dressed and another which in some sense overhears, this being one of the features which distinguishes a lecture from @ play. But then again, actors ean ‘break frame’, as at the end of Murde Ca shed Tgemmiicncedireau Yet again there are seudoraddress 7 political interview the politician can treat his interviewer as hearer and address the voters directly, at times even attempting to make eye-contact with them. (Phe whole question of the role of cye-contact and other non-verbal signals in participant role definition isa fascinating one and is treated at some length in Gosling (in preparation),) eee Teompltenes: Gafan anus for the recogiton of “hy standers’, unratfied hearers who nevertheless hear, and he subdivides them into ‘overhearers’, who acknowledge they are listening in, and ‘eavesdroppers', who don’t. The transition from eavesdropper through ovetheater to ratified particjpant is onc we are all familiar, par- ly at parties: ‘I couldn't help (overhearing what you were just Saying Purpose All speech events and speech acts have ven if occasionally WTis-onkk phatie. Sometimes several events share the same style a are distinguished only by purpose and participants or setting. Hymes notes that among the Wai Wai of Veneauela the same genre, the ‘oho chant’, is used for a series of speech events which are distinguished according to their function in marriage contracts, trade, communal work tasks and invitations to feasts. Frake (1972) reports four speech events among the Yakan, dis- tinguished by purpose — mit “discussion, gisum ‘conference’ man>~ ‘pabkat ‘registration’, and hulum: ‘tigation’. Initially, to the outsider there is no difference — no special seting, clothes or paraphernalia,48 An Introduction to Discourse Analysis no activity other than talk. Mfitin is the most general and apparently refers to unfocused, purposeless conversation in which all. partic pants have equal speaking rights. Qfsum is a discussion with a pur- pose; some issue such as when to plant rice has to be decided and 1 all participants have equal speaking rights, but this time the ‘event has a recognizable end when a decision is reached. Marepabkat is a negotiation over a disagreement; its purpose is to reach a settle ment, and now the participants are divided into two protagonistic sides. Finally, hukun is concerned with a disagreement arising aver an offence; the purpose is to reach a legal ruling based on precedent and this requires additionally a coure comprising a set of neutral judges. Hymes observes that ‘the purpose of an event from a community standpoint may not be identical to the purposes of those engaged in it At every level of language individuals can exploit the system for personal or social reasons or artistic effects. vine (1974) describes «speech event among the Wolof, the ‘greeting’, whieh “is a necessary opening to every encounter, and can in fact be used as a definition of when an encounter occurs’. Relative rank determines who greets whom — its customary for the lower ranking party to greet the higher and there is a proverb ‘when two persons greet each other, one has shame, the other has glon’. However, individuals do not always wish to take the higher status position because along with prestige goes the obligation to contribute to the support of ow status persons. For this reason a higher status person may indulge in ‘self-lowering’ by adopting the opening role, and Irvine observes that no one ever asked her for a gift if they had not first managed to take the lower status role in the interaction, Key Within Aey Hymes handles the iin which an act oF event is performed. He suggests that acts otherwise identical ‘won mock ad serio, pefuncon_andpansaing. Sachs as ob- served that the first question one must ask of any utterance is whether itis intended seriously and Hymes emphasizes the significance of key by observing that when itis in conflict with the overt content of an act, it often over-rides it. Thus ‘how marvellous’ uttered with a 'sar- castic’ tone is taken 10 mean the exact opposite “The signalling of key may be non-verbal, by wink, smile, gesture or posture, but may equally well be achieved by conventional unity ‘The ethnography of speaking — 49 in English. Tic ae srcetind) discussed above has paralinguistic features associa Tole classifiable on the dimensions of ‘stress? and ‘tempo/quantiy’ a Sires Teno) Quani 2 Noble $ (= high, — loud) 1 (~ rapid, ~ verbose) Griot S (+ high, + loud) T+ rapid, + verbose) ‘Thus the opening greeting normally has the associated paralinguistic features ST, the response st. However, if speaker wishes to indicate that the status assigned $y his role isCat.variance with hisSteue status ‘he dec ts by using an inappropriate stress patern — a speech ale ST wil sometimes be used By a noble who has taken the role of in- itiator but wants to indicate that (he knows) he is being polite. Fle is showing deference (initiator role and T) even though he doesn’t have to (9). Channels Under channel the description concerns itself with the ‘choice of oral, ‘writen, telegraphic, semaphore, or other mediums of transmission of speech’, Mogt genres are associated with only ate fem to wse a diferent channel, a whe cae of te rtimmning a “Yoruba provers, necessitates some changes. The development of radio and television Tas GEE a siwvation in which some speech ‘vents have enormous unseen and unheard audiences, which subiy affect the character of the event. What is superficially round discuss de chat-can_in Tact be-an-appc ‘attempt, indirectly, to sway a.nation’s opinions. The channel itself has gre lond th Seton of te seek ems he prs commen tary and the quiz show, with their own highly distinctive stylist rode and structure, prescribed participants, typieal seting and key., Message content Hymes suggests that ‘content enters analysis fitst of all perhaps as a question of ti, and change of topic’. For many events and acts topic ig fully predetermined and invariable, though farathers, particularly which they borrow from flan (1976) and define as “something that is emotionally invested (2A and that can be lost, maintained or enhanced’. They suggest that many interactive acts constitute a threat to face and that many aspRcis of Daly 7 imerance form can be explained in terms of speakers atemptin fuse or mitigate a Face Threatening Act (FTA) Pacethey argue, +4 ‘5 San usefully be seen as con 2 ng of two aspects: ‘the positive, consistent slf-image .. crugally inclu at Fee and mgitive fags>the ... claim to ~Treedom of action and freedom from imposition’, Thus an act may hréaten positive face by belittling and/or negative face by imposing Bsyeerand Levins predispose us to think office a a propery of ‘the hearer bythe way they discuss it and categorize the options open to speakeys, but it is important to realizé that such acts as “con- fessions’, ‘apologies’, ‘offers’ and “invitaions’ threaten the speaker's Positive oF negative Face On any occasion when he decides to make a(FTA)’ speaker has four major options: he-may do it fdirec}) or “off record’, so thet iF challenged he canCdens rowever imlaunibhy that he meant it. cmp: Can you fix this needle? sour: Pm busy ‘up: Tyust wanted to know if you ean fix it. (Sacks MS) Alternatively, a speaker may perform the ae explicid, or ‘on record’, oF on record The exhnography of speaking 51 either with some Siac 0 mpi fase poste, face, ot ally ior ata Brown and Levinson see these opons ax ordered Tom max Ye iol’, though politeness we they ae i is SOEH DORE Wan sr usual ordlnry language Ril we tnd te ‘ones which are in strict conformity wit! CHEV tating ce pp. 30-32 for a discussion ofthe mat- ims), and in which there is;no linguistic concession to fice Such 4 are not frequent; as Brown and Levinson observe, ‘the ‘majority of ‘natural conversations do not proceed in this brusque fashion at alr. Archetypa cxamples are direct imperatives, ‘sit down and shut up’, cos eonngs our kacrean iy dripping’ etieranes used when 2. ame feats uniaportt or overridden by other considerations ae UIpEHGy PURUORERTSS WHER CB The speakers Taee Thats ‘Ba THPEToNeD ar with an fntaon, the significance can be re- ‘versed — in_most circumstances ‘the firmer the invitation the more le i door in/aga, "Wave aT eae E rc “options for the speaker to take account of the hearers fienitre trop his nee Wo Tes appreciated, Brown and Levinson die ‘cuss fifteen strategics, only some of which can be touched on here. (One eto strategies ino inked 1 individual FTAS at all, but eather Roncerned-more geherally with creating a better emotional environ-_ SHUNT Tor Toure PTAs trough Compmening Uc ere, "what SEate garden jourhave’ (p00), eliming shared interests and needs, ‘you must be hungry, i's a long time since breakfast’ (p. 108), and indulging in gossip and small talk to show that the hearer is valued for himself and not just his usefulness ~ At individual act level Brown and Levinson draw attention to the preference to emphasize agreement and to play down disagreement, 4 That's where you live, Florida? t& That's where [ was born. (p.119) which may sete far as “oie His’ =—> ing from ‘bald and record to ‘off record, but as Leech (1983, p. 109) 7 ‘observes it is quite insufficient vo not ‘correlation benween inditect- ness and politeness: we must be able to say not only@om polite an. ‘uteranee is but why a particular device ofindireciness contributes 1, licular locutionary goal’. On the first point, Brown and Levin- s0n note that ‘some simple compounding of hedges and indies increases the relative expressions’, and the following is not an unusual combination: Pm sory to trouble you again but could you possibly lond me a tiny cup of sugar.” Leech himself argues that the following examples b Answer the phone. [want you to answer the phone. Will you answer the phone? Can you answer the phone? Would you mind answering the phone? Could you possibly answer the phone? (p. 108) Brown and Levinson suggest that suiting a particular fo to aspect ocasonseque s cSIGUnaT ie myhngs othe ETS and this depends not simply on the seriousness of the threat or jo- orition but also-on th nd of speaks is, however, one thing to recognize these as relevant rallies sof quince eg en peed Sona specific utterance form] One complicating problem for forcigners is that anne Doe eal era ol poe er seas sizes negative fact Using ‘please’ and ‘thank you’ extravagantly, even between intimates, andthe Culture enshrines the py between Tntimates,and-the-culrure enshrines the practice in stories Tor children Tike The Bed Baby whose crime was that "he never once said please’54 Am Introduction to Discourse Analysis ‘Gon and Levinson 2onfes analysis ‘a sasisfactory ‘Sgt lineata Sting as es mt pees Gu tures concerned with avoiding FTTAs, Invitations, for example, expose Ser ee a as hres chu op ie ale ee eran enone emai 2. are you busy later I'm going out how exelting 3. are you busy later Tim afraid Tam that's OK ~ Rule breaking A successful ethnography of speaking will describe the normative artis oli eco al Sone ter ‘munity by detailing for each act and event the necessary configuration cays al apache stared to and each community ha mn rules for interpreting rule-t ing. studens dicusion w other tabene her Go hee thatthe pecker were menus of he oe speck ees then, ven hey ecopaed te wis come dee ‘artificial speech’. On this oceasion the rules were apparently not broken deheracy seme ofthe paripane vee ee prised at their own code-switching — but very frequently rule. breaking is detiverate andr specific list = Salmond (974) suggests that the main justification for writing a detailed description is that ‘only when the rules are laid down as economically as possible and all the options are made clear can an outsider appreciate the manipulation that people practise’. She reports that at the Maori ‘rituals of encounter’ deliberate rule-breaking re- sults in a great los of or gain in prestige — the unsuccessful contend ce leaves the ‘marae’ in utter humiliation, the successful is greatly honoured having proved himself above the constraints that bind ordinary people ‘On one famous occasion a Maori group from a part of the country where women were allowed to speak was visiting another where ‘women were not. The hosts opened the oratory but when it came to the guests” turn there was a problem, for the most senior in rank was an old chieftainess. After a moment's hesitation she began to speak Immediately there was a protest from the hosts but the chieftainess calmly ignored them, continued her speech to the end and then said ‘The ethnography of speaking 55 ‘ov Arawa men, you tell me to sit down because [am a woman, of you would be in the world if t wasn't for your mothers, This iswhere your lering and your grey hairs come from!”; then turning her back on them she bent over and flipped up her skits “in the su- prem gesture of contempr’. Most rule-breaking is less lamboyant and less risky than this. Tn the 16th century, English, like many modern European Jan= wages, distinguished two second person singular pronouns, Now: as eustonary for aol owe Yo repay, w recive "you' from their inferiors but to address them as thaw’. If speaker i ie “rules, the rule-breaking was meaningful a Se Er gee Edward Cok Tnsult Sir Waker Raleigh at his trial by addressing ‘that he did at thy instigation, thou viper for I thou Evin Tripp (1972) presents 2 simitar insu roucesins: What's your name, hoy? octox: Doctor Pousszint. ma physician oucras: What's your first pare, boy? boctox: Alin, and observes that the policeman insulted the doctor three times. Firstly, he employed a social selector for race, in addressing him as ‘poy’; secondly, he treated the reply as a failure to answer, a non- ‘name; thirdly, he repeated the term ‘boy’ emphasizing the irrelevance of the name Dr Poussaint. So, Ervin-Tripp points out, ‘communication had been perfect in this interchange; both were familiar with an ad- dress system which contained a selector for race available to both blacks and whites for insult, condescension or deference, as needed. Only because they shared these norms could the policeman’s act have its unequivocal impact.” Norms of interaction All communities have an underlying set of non-linguistic rules which govern hen how and how af seh osu Ths the Aran value speech highly and the young are trained in the arts of speech, while for the Wolof, speech, especially in quantity, is dangerous and demeaning. French children are encouraged to be silent when visitors are present af dinner, Russian children are encouraged to talk. Among “the Arucanian there are different expectations of men and women, ‘men being encouraged to talk on all occasions, women to be silent, indeed a new wife is not permitted to speak for several months.36 Am Introduction 19 Discourse Analysis Even within North-Westem Europe there are surprising differ- ences. One ethnographer reports how, when he was researching in Iceland, neighbouring Eskimos would visit once a day for an hour to check that all was well, During the hour there would be no more than half a dozen exchanges, and all the rest of the time was spent in si- lence. Another ethnographer describes staying with in-laws in Den- mark and being joined by an American friend who, despite warnings, insisted on talking with American intensity until ‘at 9 o'clock my in Jaws retired to bed; they just couldn't stand it any more’. Other norms govern the physical distance at which speech events, particularly conversations, take place. Watson and Graves (1966) re port that compared with Americans, Arab students confront each other ‘more dircedly when conversing sit closer, are more likely to touch each other and speak more loudly, behaviour which is often interpreted as aggressive or over-friendly by Americans. There are differing norms for tum-taking; ‘Tannen (1982) notes that whereas a typical feature of New York Jewish style is for speakers to over! a “a way of showing enthusiasm and interest, this same phenomenon sof interpreted by members of erent groupe Eula difrencs inte ways WHEN palo ea oth ead “titteutes, occasions Wher Tey ave Toroed to sop midwas teeuge grammatealsructureFymes (1O7IR) suggests tr white middle lass Americans the 1 lis Arvercans Te normal bestaion behaviour Kea eee ol for : Meanse (ometimes move Thay GPG] Reawling Saisie soe, times evident in children’s speech but may be interpreted as a defect i ‘The etashes of norms described so far may produce some personal scorers ens” and even unused censure, but ‘sory the ‘norm-breaking is accepted as the performance of someone who doesn’t _ “ipant-asume that Ted share the same Rare Papa) discovered that Mesquaki Fox children imerpreted thé normal loud- ness of voice and directness of American English teachers as ‘mean- ness’ and “getting mad’, and even more serious has been the ‘misinterpretation of the behaviour of thousands of Negro children in New York schools. These children were observed to be failing at school — on all standard tests of reading and verbal and non-verbal intelligence they were retarded while in individual interviews with ‘educational psychologists they said very litle and many of them The ethnography of peaking 57 appeared to be ‘linguistically deprived'( Bereiter and Engelmann (1966) working with four-year-olds in U municated by gestures, ‘single words’ and ‘a series of badly Connected words oF phrases” such as “they min’ or ‘me got juice’ —_ they observed thatthe Negro children could not ask questions and” ‘without eaggerating --. these four-year-olds could make no state ments of any kind’. From observations like these Bereiter and Engel- mann conchuded that one mast teat these children 2 if hey had no language at all, and therefore they devised a pre-school programme using formal nguage dri to teach the children English. Even when the teaching was underway, they observed that some children, havir been taught “two plus one equals three... would continually lapse into amalgamations, “two pluh wick three. Ie was claimed that having done this the children were no longer abe wo substitute ther mumbers for the “one”, it having become fosed wth the besining sound of “equals bal dei hhas no basis in socal reality — itis a nonsense cre= ed by edlertonar paetorogis who Tbe very Ile SEH guage and even less about Negro children’, He observes that to say {he children have no language or even that they are linguistically de~ prived is a complete misunderstanding — they come from a culture ‘where linguistic ability is highly valued, as is evident from the impor- tance of ‘sounding’ and the faet that verbal skill i a prerequisite for peer group or gang leaders. The truth is that the children donot Os their abilitics at school because ot ones, the school values and beeause the school is ‘Negro children faced with a Targe, though friendly, coloured inter- viewer, let alone a white interviewer as used in Bereiter and Engel- ‘mann’s tests, choose to produce monosyllabie, non-committal answers, whereas white middle-class children are wiling to chatter away in the ‘ame’ station, Ii eens the ution o rer he spec js assumed to be the sume, with the same norms of interaction, that ‘aychologists and sociologists alike feel able to compare the perform- ance of different class and ethnic groups — only when one realizes that the norms are not the same can one perceive the uselessness of the wei. Tina striking demonstration thatthe psychologists were wrongly in- fering from what the children said to what they were able to sa Labov took a rabbit into a classroom where young Negro children were dutifully “learning English’rnalegee 58 Am Intraduction to Disenure Analysis atl bop Dosrerrinn Ards | ‘1: This isa book, What is it? P: Iisa book. > What colour is the book? Iisa red book tele Unfpenne ae je explained to ae ire habe ad «rabbit that vas ve aty 4 Conversational analysis Ti the next room sa —_— ld be quite happy played grammatial illed to their fell He nce hee lien apy ds SSTiation lar in excess ofthe structures bing is next door, A Hymes (1972) stresses that it is heuristically important for the eth- ography of speaking “to proceed as though all speech has formal characteristics of some sort as manifestation of genres’, though he adit that the very notion of care unmarked speech points to the fact that there is_a ‘great rans Ssriong genres in the number auLex- ‘snes fatal marc Pt fr hs en he eth ‘nographers of speaking who have provided detailed structural descriptions ave focused on well defined and often rivalled events. — greetings (Irvine 1974), ritual encounters (Salmond 1974), chant- ing (Sherzer 1974) — while those who have looked at conversation have examined not its structure, but factors affecting the choice of code or style (Blom and Gumperz 1972; Gumperz 1964; Geertz 1960). Until very recently most of the advar nvers hhad been made by three sociologists, > \ ‘sho originally sirssed that they work with conversational isterls ‘not because r it . they saw conversational analy as a fist sep towards achieving a “naire observational ine itealwith details of social interaction in “a rigorous, empirical and formal way, Latterly, However, they and Those who f6I6W inthe tadlton ave seen thei main concem asthe nly eave cial cre al pee so hee i ings are useful to and usable by anyone interested in the structure hooener ten Turn-taking ) —Gne-of the basi GE of copsersation is thatthe roles of speaker a lbeener change; and WisCeeeget a cematabfy tle etary ‘gpeechrand remarkably few silences. Sacks (MS) suggests that there Ts an_underlying rule in American English conversation and not more than ong Faryyalks ata time’ This ts not am empIFIC fact, because there are obviously, many instances of short pauses and _short overlsbut rather a nomiative oF “ObseRaBTTorfented 1 Tea60 An Jutraduton to Discourse Analysis ture of conversation — in other words, it is a rule used by conver- sationalists themselves. If more than or less than one party is talking itis ‘noticeable’ and participants set out to ‘remedy’ the situation and return toa state of one and only one speaker. Ifthe problem is moze than one speaker one of the participants usualy yields the floor quickly: ome But that xz — Then you wentuh Fre:ds sis Were left — we kt tex: LNo. That the time we let Feds (Sacks MS) Gone problem is filence other speakers begin speaking, of in- ate their intention t0-spcak by nofses Tike ‘er or “mnt. In other words uns apeak pial cur sucesily Wit craps or gaps between them. Overlapping is dealt with By OnE speaker ending his tr gull, as bemeen tres by snuher speaker hesnnng his turn or simply indicating that his Tarn has begun and incorporating the silence ito it ‘A second feature of conversation is that speaker change recurs, and this presents problems for the participants — how can they achieve change of speaker while maintaining a situation in which at least, but ‘ot more than, one speaker speaks ata time? Sacks suggests thar@lcurrent speaker can exerci ees ily he ean select which participant Lcompdtte e ket ae will speak next, ¢fthep ing him or by alluding to him with a descriptive phrase, ‘the Right Honorable Meniber Tor Bealey South ee ee ee are the, at nce-by producing the fir udiacency, Fifer belt» ) Eg Fin the selected speaker to product appropiate atone rene vocra His Mie oe rere: Hoge ae Fea ere oes Hale rhe current speaker'xsGcond option is simply to constrain the next -utterange, but not sclect the next speaker, while the third option is to select nether and leave it1o one of the other participants to co ‘hue the conversation by selecting himsel— — ——S Sacks emphasizes that these options ae in an ordered relationship = the frst over-tides the second and the second over-rides the third If the current speaker selees the nest speaker he alone should tlk next Sacks notes that when an unselected speaker takes a tara ale errr Conversational analysis 61 _ready assigned 10 a selected one, the right of the selected speaker to speak next Usual Preserved ‘(to Q: Tell us about yourself so we ean find something bad about you te Yeah hurry up Importantly, these selection techniques operate only utterance by ut, terance: there i mechani ation Speaker an selec the = choice of the next Speaker is aTvays The prerogative of the current speaker if he chooses ‘to exercise it. In more formal speech situations — feet Sane rooms, formal discussions — it is, of course, quite possible for one speaker, whose role assigns hin@estra author, to select the speakers or several SUCCES Uiterances “While these speaker aptions-explain how the next speaker comes to be selected, or to select himself, they do not na foe i ew aker knows when the current speaker has finishedand therefore Gian iecan begin, although this s pbsiousy. essential fhe istoavoid overlap of silence. Sacks) suggests that in fact next speakers are not See Mf and eypnee be concerned wth actualy completed ‘erances, because one can never be sure that an utterance is complete sible to add mo n apparently complete utter= frefore, next speakers are Tin developing this idea, Sacks observes that turns consist of one or more Sfenc3, with a sen- tence being defined'as a unit which has ‘ts completion recognized on its completion, and that itis not complete r=), cognizable by partiipants aso it can be monitored, fom is beginning, 4 Em begin ta he for cpl Be rose BGG 'S Nay tin on is completion iv completion” may be) + nope = ‘sacks MSY Speaker change takes place aie end omen the next speaker or next action has been selected, the next speaker will take over at the end of the sentence during which the selecting was done; if the current speaker has not selected, any participant may self-slect atthe tend of any sentence, Thus, a speaker is vulnerable at every sentence ‘completion whether he selects next speaker or action or not, and even if he gets past one sentence completion he is equally vulnerable at the end of the nest sentence. Turns to speak arCvalued?snd sought. and thus the majority of turns im-any_conrersation_consist of only a ‘Sale sentence, unless fermissiog has been sought fora longer turn, perhaps to tell a@tory of a iake > - “The argument so far is that conversation is made up of units which62 Am fntraduction 19 Discourse Analysis 2 coil a the incomplete pussy complete and that speakers can begin as 50 rent speaker has reached a porsible completion. This fat Sacks Sugeest explains the lo ine cidence of overlap and silence(Floweversthe ability to come in as soon a. spear Das renee possBle completion requires a high degree (TSH on he par of-paripanis — they nesd tobe ble not only analyse and understand an ongoing sentence in order to recognize ‘when itis possibly complete, butalso to produce immediately a rel gyant next utterance. Do speakers have this abilit 7 ea (973) argues that the recipient of an ongoing utterance ‘ ws te technic capacit-tn select a precise spot to srt his own talk ‘no later” than the. fe moment’. gives three kinds luce a completion to a prior speaker's otherwise complete utterance: ext An’ there er, sr — because a'that there we at least fen mies of trafic bumper tuh bum= Much more impressive ae instances of recipi the right moment wih thei ox proposed completion of an as ee uncompleted sentence: PO St ee ‘oust: No Soshe i someone whois 4 eb i sx: 1 who carbon copy ofthe en [ain Ppa, SE end A variant of his is when the re ' en the repent sable to prec the end ofthe sentence and stems to sy the same thing tthe ane es ts The gu who docs’ oes 7 “Thus speakers demonstrghh i — nit, but 'e hat the ailiy to plac thir em recision. theles3, ualntentional overlaps still ecu, fequeny eased Tre current speaker has thot selected a neat speaker, a self-sclecting speaker, beginning at a possible completion, may well overlap with the current speaker who us decided cominus; or with second selecting speaker. The Broblem is usualy ‘remedied’ quickly by one of the speakers yielding th floor when the onerip the result of to el-selecting speakers re appears to be a rule that the ‘first starter shee ap le that the ‘fist starter’ has the right to On occasions when both speakers cease 5 x speakers cease speaking simultancously, fs in the last example above, there is the question of who has = % iF one speaker was producing a completion of the other’ utter. Conversational anabsis 63 lance, as Roger is, the speaker whose tur was completed takes the hext turn. If, in other circumstances, an overlap is ended by one speaker stopping first, if only by a syllable, he takes the next turn: x: ‘The hat, 3) sc: En the hat, one bcs us sag 4 problem and parspants feel that ajsilence is attributable, “esuall)) to_some intended next speaker, Schegloff and Sacks (1973) quote a report ofa silen He hadtuh come out tuh San Francisco, So he called ih frm thir plac, fout here to the professors, en setup, the time, and hh asked him to bb, oN Fthey’d mabe a reservation for him which they did cuz they paid for biz room en etcetera en he asked them ths: make a reservation fori par= ens, Ein there was a deep silence, she said, atthe other end ’e se2 Oh well they'l pay for their own uh-hh- room and secommodations “They observe that the silence was noted by the speaker as not his, but that he then transformed it into a pause by continuing his turn “There is a very low tolerance of silence between tums and if the— intended _next speaker docs JImost at once the previous. eaker is likely to product * past completo which is-either a question, noticing the silence such as Didn't you hear me’,or a marked repeat of his utter This, speakers who have not yet formulated ‘what they want to say tend to indicate their intention to speak by ‘erm’, ‘um’, ‘mm’, or an audible intake of breath and thereby incorporate the silence into their turn that atleast and not S aslo We have seen that the basis of conversation more than one party speaks ata time, and that the system for alloting Tens works only one Tur a atime Sacks ofa} (1974) distingsh “Efferen spech exchange sysieng according to the organization of turn-, Uh taking They observe that whereas in conversation TORTS are alTocared yeh singly, in some systems, debates or law suits for example, there is a Se high degree of pre-allocation of tums, The most extreme form is the Eto system governing public discussion among the Burundi, where ‘the order in which individuals speak in a group is strictly determined by seniority of rank’ (Albert 1972). Turns come in a fixed order and only ‘when everyone has had a turn and the cycle has recommenced can i eagareae tear apm oo Gi) sungent hat different tu-taking systems gn0- fifferently structured turns. In a pre-allocated system there areote 64 An tdci te Disoune Anal Se ‘no interruption pressures, turns tend to be longer and for these rea © ‘sons consist of a series of linked sentences. In a turn-by-turn allo- \ cation sstem there are stong_presuss untae peng TEcniguts, open to the paler the wishes to nina speaking past a parcular ‘posal completion. The simplest sc ep stat Sassen ine these are items such aku Cand) Choweve ant other Cae on netors, whose importanes in soiesition that they umm a porns tially complete sentence into an incomplete one. This, of cours etnigue, Because a seleselecing Tr the ats Completion’, may have sleady planned or even begun his tin — Ferguson (1075) in am cain, ion of eleven hours of conversation, ote ta 289 of intemapions ‘urred after conjumctions. Therefore the successful Hoorhokter roles uteraness Wc, ‘whe they could perfectly well becom. posed of ak that, in terms of “posible sentence completion would Grave oe, oF mare than one, such occurrence within them af bat in sucha way ast not have possible compietions within tea” (Sace CO fis One technique is to begin with an incompleton marker, if’, ‘since’, or any other subordinator, which informs the other participants that there ill he atleast swo_clauses before the first possible completion, A speaker can also,prazsirlicture a fairly large unit of speech by such devices as Vd like tormake-two points’ or simply “frsth’, which ex- plicily indicate that there i more Yo come after what could otherwise have been regarded as a possible completion point. n the following extract a skilled politician operates the system beautifully to guarantee himself time to answer the question fully: yw U think one can see seceral major areas ... there's fst the question ow the send big area of course is the question of how you handle incomes and I myself very strongly believe that we have to establish in Britain are fundamental principles. Fist ofall (Denis Healey: Analysis, BBC Radio 4) None of these devices can guarantee that the speaker keeps the oor uthey de foreejthe other speaker into a position where he must tions, They Choose not to veld the loot bo speaking mae Taal “More quickly and in 4 higher pitch; often the surface grammar uF ‘Phowomgy breaks down, and frequently there i a reference to the Conversational analysis 65 sun they ave at thei dpa enormous asesfand their policy ms ie can Fuse come non hat as yar mel ee Ines i ae you'll know what the point is (Money at Work, BBC2) mn-speaker who wishes to speak, but is unable to find a suitable quently heard as rudeness, trdown tomy parcolar sta, NL you talk ike the chairman of Bish Rail . inces his desire or of indicating by repeated short, single-tonic, uttera for the floor. This is a technique which children master early as the following example of a two-year-old shows. storm no are what sh’ een seing re [ Pourcehat Hetaal ty i tos: Lah ah ah ah ah sworn: ob well at feast wll ned to know when she “Tc foe om [; ah ah ab sto, rors: amy con sorry darling ron: stop talking [At the other extreme a non-speaker who is offered the floor but doen wat muy spy vem slent etl the speaker conus (Kendon 1967; de Long 1974) or produce a minimal response to con- fn, grec oer interes or sete -l of his um to produce 1a possible pre-closing ‘alright (‘okay" Swell, (see below p. 90), anc thereby indicate that he fas nothing further to add and is willing to close the topic ; “STE-erMdence adduced so far to describe and explain speaker change has been almost entirely grammatical and semantic. We have seen how grammatical markers and considerations of meaning tum certain points in utterances into ‘possible completions” and it has been suggested that at each ‘possible completion’ «speaker is vulnera There is, however, evidence to suggest that speakers signal paralin~ suistcally and kinesially to the other participants at which possible rbreblih Gace = hyve 66 An Iniraduction to Discourse Analysis completions they are actually willing to relinguish the floo De Long (1974) reports a deuiled analysis of a series of conver- sations between four- and five-year-old pre-school children, which shows a marked correlation beiween certain body movements and change of speaker. The transcription noted eight basic movements, including ‘up’, ‘down’, ‘lef, ‘right’, “forward” and “backward for «ight parts ofthe body, the head, the trunk and the left and right arms, hands, and finger. Anas shone ta wo movements co-ctaed cither simultaneously or in rapid succession to signal a termination ‘The fist was aleward movement ofthe head, the second a dann, ward movement by the head, arms or hands individually ot in an combination, 7 However, we must evaluate this evidence with great care, becaus 4 Levinson (1985, p. 302) points out such signals can only be mp portve and not the basis ofthe turn-taking system, a turn-taking can ‘be managed perfectly well in telephone conversations with no visual cues at all, with ‘ess gap and shorter overlap and apparently no com | Pensitory ‘special prosodic or intonatonal patterns De Long stresses that ‘to say thatthe intention to terminate ver- % balization, willingness to yield the floor, is signalled by downward and leftward movements does not mean that every time a left in the head is accompanied by a down in the head or in other parts ofthe body ina speaker intends to terminate’. In fact it is only when such sig- nalling occurs at possible completions that termination is signalled. To <7 support his analysis de Long describes two apparent exceptions: Jong gap between two utteranees, and the only recorded interruption, ‘The gap occurred when one child ended ‘Til show you what's brown’, ting his head to the left but not making any downward sig- nals. During the neat few seconds he made intermittent downward and leftward movements while the other child remained silent then ‘after no less than an eight second delay we find left and downward suc- cessive signalling in the head’. Immediately, the other child began speaking. As confirmation of the significance of this signal, the only interruption is found to oceur immediatly following series of dow ward movements by several parts of the body and a sucession of lowns and lefis in the head which the interrupter apparently took as speaker change signals Dierappasey fos Kendon (1967) suggests that one important fat mooth chany as Adooks at and away front PAGE more equal “Periods, Focusing on the ends of wrterances ofmore than five seconds, Conversational analysis 67 Kendon notes that ‘usually the person who is bringing a long utter- ance to an end does so by assuming a characteristic head posture and ‘by looking steadily at the auditor before he actually finishes speaking The auditor, who spends most of his time TooKing atthe speaker, sC“t~ + able to pick up these signals and tends to respond by looking away j.. 7 just before or just as he begins to talk. By this time of course the le ae Speaker is looking atthe auditor and can pick up the signal that he has accepted the offer ofthe floor. ‘Such changes in behaviour which precede the utterance itself clearly make it possible for each partici= pant to anticipate how the other is going to deal with the actual point Of change of speaker role, perhaps facilitating the achievement of Smoother ... change overs,’ Kendon notes that fewer than a third of the utterances which ended with an extended gaze were followed by lence or delayed response, as compared with almost three- quarters of those that ended miziut the speaker looking up. In larger groups the situation is more complicated, but Weisbrod (1965), study- fg a seven person discussion group, found thatthe person whom the current speaker 1&t lookedhat before endim mo Weto-sp newt — “Most conversations, particularly two-party of, have periods when se i nr ett Godwin TBI, p 100) ha. ineserting r= “ations He notes that the ‘boundary between full engagement and mutual disengagement snot srtured av ¢ Sharp Sear aI rather, ‘a= ticipants are afforded a space within which tt organize their bodies. He gives an example of a listener withdrawing gaze just as the speaker was concluding and then holding her head “facing just to the side of the speaker’ who was thus put in the position of being able but not obliged to continue. In other words, in this position one artisans. ale iano the thet through peripheral vision and su San Tagh oS re-attend if there is tom This observation rings true and ‘most people will acknowledge having used iin embarrassing situations iene comrsaton fad deed up_ and Jigs je-conaek was too ‘sressful. When there are three or more participants the ‘gazing past” posturé can lead smoothly to complete disengagement as skilled cock~ tal paryy circulators know. ‘As we observed above, ums to speak are valued and speakers com. re for them. One of the points at which a speaker is vulnerable is {hen Te pauses within a phrase — Ferguson (1973) discovered that almost a third of interruptions occur following ‘filles? such as ‘um’, ‘er, and ‘y'know’. Kendon observed that speakers tend to avoid guz~ ing during hesitant speech — they spend only 20% of the time gazing a contin te spake.68 An Introduction to Discourse Anabssic during hesitant speech as compared with 50% of the time during fluent speech — probably as a defence against interruption. In Ken- ddon’s data interruptions were rare but in those cases where a smell battle for the turn occurred both speakers stared fully until the con- fle 1973, 1974) suggests that the cues for speaker change can (Cor ‘or any combination of all three, A listener may claim the speaking turn when the current speaker {Ves a tum signal, defined as the display, at the end of a phonemic clause, of at least one of a set of six cues. ‘The cues are 1. Intonation: the use of any pitch level/crminal juncture combi- nation, éther than 22.7 2. Paralanguage: drawl on the final syllable or on the stressed syllable of the phonemic clause. . Body motion: the termination of any hand gesticulation or the re laxation of a tensed hand postion Sociocentrie sequences: the appearance of one of several stereo ‘yped expressions, such as ‘but uh’, ‘or something’, ‘you know’, labelied sociocentrie by Bernstein (1962) (| 5 Paralanguage: a drop in paralinguistic pitch and/or loudness, in & conjunction with a sociocentric sequence. {6 Symx: the competion of «grammatical clause involving a subject predicate combination Duncan (1974) observes that the occurrence of a speaker turn signal does ot necessarily conn change of apesker he tees aoe hhas the right to decline(urthe more cues displayed simultaneously the greater the likelihood that the listener will take over r. Duncan emphasizes the importance of the speaker turn signal by showing that in his data every smooth exchange of the spcaker role followed a speaker tum signal, while every attempt by an auditor to Claim the tun while no cuts were Being daplayed rsced in aa ‘lemon ~~ Duncan (1973) suggests that the speaker turn signal can be over- flo*7idden by acclsin-suppressing signal) wiiGh Const of Fand gees Aiculations, and is apparently almost total 2 “of 88 auditor sot cory eam Fesponse To ham sgl oh ts sch sane oe ey aceon ota sto i ene 4 Once the auditor takes over the speaking role he_tpically displays - aker-state signal, which consists of at least one of four cues: the ‘ering of gave noted by Kendon, the nitatea sf ene, yi F Conversational analysis 69 > audible intake of breath; and paralinguistic overloudness. ‘any description of turn-taking there is the problem of what con— -es.amum, and while most analysts accept that nods of agreement murmurs of assent do not count, there are some important dif- § ices of opinion, Duncan uses the tren Rk chant haviour wo og ‘contributions which do not constitute aTarT bar which provide peaker with useful information as his turn progresses. Under thi ing Duncan inchudes ‘sentence completions’, ‘request on’ and ‘brief restaiémemts), all of which for Sacks would be ccompletémums — indeed Dunean and Niederche (1974) express some disquiet that ‘for some of the longer back channels, particularly the brief restatements, the boundary between back channels and speaking ‘tums became uncertain. On an intuitive basis, some of These Tonge longer back channels appearéd to take on the quality of a tur.’ Duncan's main concern is to describe speaker change, and for this ‘Purpose it-may-be sufficient to categorize all utterances as either ‘back, channels’ or ‘turns’; for those interested in describing the structural relations Berween utterances, however, this is not sufficient — the options open following a ‘request for clarification’ arc very different from those following a “brief restatement’ Conversational structure Sacks (MS) obserses that a conversation is string of dt east rx sums)Some turns are more closely relat ‘others and he isolates class of sequences of turns called adjaceney paitwhich have the following features: they are no ufferances long; the utterances are_ ‘produced successively by different speakers; the utterances are or ered — the first must belong to the clas of first pair pars. the second [to the class of sewmnd pair pats; the utterancg are related? &F second pair can Toll its ft pair par( aa} onlyan appropriate one; SS Ragu pat Oe a me tor ant ahoee es e scant Ws ses up 8 taniion tant and expects ik Be ocegs of eee "i «wisn Fp ro amma ge ee ah answer-wittfottow-{Sacks 1967). jere a class of First pair parts which includes Questions, Greet= figs, Challenges, Offers, Requests, Complaints, Invitations, An- naroceaments Ter some Erst pels part the vend pale Part fe _gecproca (Greeting Greeting), for some theres only anc eppreprits Sw second (Question-Answer), for some more than one. (Complaint- 0? Apology/Jusifcation)| 70 Am Introduction to Discourse Awalysis Adjacency pairs are the basic structural units in conversation. They are used for opening and closing conversations, Hi there Hallo Bye then Bye 36°%and are very important during conversations both for operating the ‘o°_ tum-taking system by enabling a speaker to select next action and next “speaker, and also for enabling the next speaker to avoid both gap and Maar A It is, however, no difficult matter to discover a question not fol- V7 Piowed yan ane nd anes 2 question shu the eo he pair. Sacks argues that, whereas the absence of a particular item in ey rsdon has nial no inprtance becase thee ae a number PS ngs ate sry sense cs fs sce at “3° the first part provides specifically for the second and therefore the 3 absence of the second is noticeable and noticed. He observes that PX, , people regularly complain “You didn't answer my question’ or ‘I said yes hallo, and she just walked past’ x €” Preference organization* Work by Pomerantz (1978, 1984), Atkinson and 1979) and Levinson (1983) has developed the notion steecaey oP into a very powerful concept. As we saw, some first pair parts allowed for alternative seconds; however, we can now demonstrate that some op- tions are preferred and some dispreferred — a distinction which may have a psychological basis and explanation but also has linguistic realiz~ ations: ‘preferred seconds are unmarked — they occur as structurally simpler turns; in contrast dispreferred seconds are marked by various, nds of structural complexity’ (Levinsoitbid., p. 307). Invitations naturally prefer acceptances arid we can see the differences in real- ization between the preferred acceptance and the dispreferred rejec~ tion in these two examples from Atkinson and Drew (ibid, p. $8): 4. Why don't you come up and see me some ies ® would like to "This section draws heavily on Levinson’s (1983) 30-page discussion which readers are recommended to cons Comoenational analysis 71 © Uh if you'd care to come and visit alittle while this morning I'l give you a cup of affee by hehh Well that’s aufully sweet of you (DELAY) (MARKER) (APPRECIATION) don’ think I ean make it this morning hh uhm I'm running (REFUSAL OF DECLINATION) an ad in the paper and uh I have to stay near the phone (ACCOUNT) Levinson observes that dispreferred seconds are distinguished by in= ‘corporating a ‘substantial number of the following features’: 9 ded by pause before delivers; (i) by the use ofa preface (ee b); Gi) ty dgplacement over & numberof tums via use of par fiat, ion sequences » G) he use of markers or announcers of dispreferreds like Uh ‘and Wal (ji) the production of ken agreements before disagreements, {ip the tse of appreciations if Flevan (or offers, ivtatons, sugges Sons, advice) Gn HE we OT apoToges if relyant (or request, imwta- tions ete); (the use of qualifiers (eg J dom’ inom for sur, bat =~); (ie) hesitation in various forms, inchading se-editing ©) Sours iy formulated explanations for why the dispreferred act is Being done d)_deciration compoTy: of 3 form suited to the nature ofthe fist part of the pair, but charactersticlly indirect or mitigated (pp. 334~3) Preference is a very powerful concept and once it has been estab- lished with examples like those above it can be used to explain the ‘occurrence of quite a number of other conversational phenomena as the results of speakers trying to avoid having to perform dispreferred seconds. Thus Schegloff et a. (1977) argue that conversationalists prefer the speaker to correct his own mistakes rather than to have to correct them for him and that they therefore use a series of repair initiator devices ranging from pausing, to return question, to actual, frequently mitigated, correction: 1: But know single beds’ awfully thin to sleep on ts But ymow single besa hint sleep Single beds, // The’re~ . Yimean narrow? 1 They're awfully narrow yeah We can also now see a general explanation for pre-invitations, pre- requests and. pre-arrangements — they are psychologically motivated structures to avoid loss of face for one or both participants resulting from a dispreferred second having to be performed: ssck: Say what ya doin? sur: Well, wee going out. Why? sack: Oh, i was just gonna sty come out and come aver here and talk to the people Pre-incitation72 Am Introduction to Discarrse Analysis Participants can recognize pre-sequences fairly easily and indi in their replies whether the invitation is likely to be accepted: 2 Say what you doin’ tonight? 1: Ob, I'm just ‘The very recognizability of the first part of a pre-sequence sometimes results in ‘collapsing’, in which the next speaker produces what is really an answer to the first part of an anticipated next sequence. In the following example utterance 1 was produced by a five-year-old and the actual reply was utterance 4: utterances 2 and 3 represent ‘one form Sacks suggests an uncollapsed version might have taken: 1, Can you fix this needle? Precrequst 3: G20 3 Will your Rewer busy. The ellipsis in fact allowed the five-year-old to protest that an i tended question had been misanalysed as a pre-request and to respond: 5. 1 just wanted to know if you can fx it Of course, an alternative analysis is that utterance 1 is an indirect rect Gn 29-30 wih lato ec peconon on requcass Wait Ws reply Pm buys then heed ata refs because denis the third pe-condin, his lings Hower Talasn psmatndyrgecr or tak tr ss sts is nisguided — he sts that one_of_ vantages of a is fads by’making the request-at-all In some instances the addressee may respond with an offer: .& Do you have any pecan Danish today? Yes we do, Would you like one of those? cc Yes please: (Merritt 1976) or even proceed directly to the action: s: Have you got Embassy Gold please? vs Yes dear. (paid) Asi Initially, of course, the indirect speech act analyst would argue that the first utterance in the above example is an indirect request, but then, asks Levinson, what does one do with the example below — surely one can't classify one as a question and the other as an indirect request 1976) Comvercational analysis 73 Hi Do you have wh size € flashlight bateries? Pre request we Yes si. (Go aid) Dithave four pease. (eqns) Tums to get Rayon (Meret 1976) Inserted sequences ‘The structures described so far have been linear, one pair followed by ater: thee area aes of embedding, of one pair occurring Schegloft (1972) lls these edivedded pais etn sequences. S Sueahapciber habs ‘he doesn’t understand, or be- cause he doesn't Want to commit himself until he knows more, or be- ‘cause he's simply stalling, a next speaker produces not a second pair part but another fist pair part. The suggestion is ‘if you answer this ane, I will answer yours 4 Ldon't know where the — wh — this address //i. Q Insertion |: Well where do — which part oftown do youve? Qt sequence) 4 Hive four ten East Lowden, Ai 1 Well you don’ live very far from me. A One question which immediately arises is in what sense is the pair QUAI inserted into the pair QA; surely this is treating converss as an accomplished product rather than a developing process, because A may never occur. Schegtoff, however, argues that the Q utterance makes an A utterance conditionally relevant. The action the Q does (here, direction asking) makes some other action sequentially relevant (here, giving directions by answering the Q). Which is to say, afier the Q, the next speaker has that action specifically chosen for him to do and can show attention to, and grasp of, the preceding utterance by doing the chosen action then and there. If he does not, that will be a notable In other words, during the inserted sequence the original question retains its transition relevance, and if the second speaker does not then produce an answer itis noticeably absent in exactly the same way as it would be if shere-wese na intervening sequence, and the-ques- tioner can complain about the lack of answer in exactly the same way. aiaconcy pais are normative structures, the second part ought to occur, and thus the other sequences are inserted between the fst pair part that has occurred and the second pair part that is anticipated.74 Au butroduction ww Discourse Analysis Methodological considerations Its appropriate at this point to discuss the nature of the statements being made by the conversational analysts and their descriptive aims in making thegrc Schenlein (19784), ty his introduction to collection of conversgtfonal analyses, obse that ‘the descriptions presented here offer ‘movement towards an empirically based grammar of natural conversation’. For readers ith linguist perspective this is slighty misleading because, a Levinson points out, the aim o versational analysts is not simply to show that ‘some aspect of co seston ge viewed’ ete smucued in «parca wa, but also ‘that t-actually is so conceived by the pardctpamy producing ir, ‘@p. 318-19). In other words, the turn-taking mechanisms, the tran- sition relevance set up by first pair parts and the existence of pre- ferred and dispreferred second pair parts are significant because they are demonstrably ‘oriented to’ by conversationalists. Paradoxically, the evidence showing thatthe structures are nor just an artefact of the analysis comes from points where structures break dawn — then we can observe that participants ‘either try io repair the hitch or ... draw strong inferences of a quite specific kind from the absence of the expected behaviour’ (p. 319). Thus in the following example there is a two-second pause between two contributions by participant c ©: So I was wondering would you be in your office on Monday () by any chance? (pause 20) ©: Probably not (iid, p. 320) and the addressee’s silence iseGhvioushpSnterpreted as a negative spoiise to the question — T= Tee The strength of the conversational analysts°appibach is that thee’ structures they have isolated do not, lke Grice’ maxims, predict sim fy ply that inferencing will take place but also predict the result of the \ fnferencing. ‘Thus to spell out the steps ir an-interpretaion of the ““%] aaTIpTE above: by the rules of the turn-taking syste c)has validly /eq_ selected the next speaker-and thus the pause is assigned to The ad? gy. yp dresse; the adjacency pa stem makes a yes/no answer tansiionally relevant and the preference system isolates delay as one of the markers M4") of dispreferred seconds, inthis case ‘no’. Thus the system predicts, “A = seiker than singly accounts for, the: fect‘thar silence wil befintar: T° peed tn implicating a eee Teis examples like this which justify Levinson’s observation that the research methods of conversational analssis are to be recommended Convercational analysis 75 because they ‘offer us a way of avoiding the indefinitely extendable. and unverifiable categorization and speculation about actors’ intent tf J so typical of Discourse Analysis-style analysis! (p. 319). However, all conversational analyses are s0 rigorous. IT the aim is, as Schenkein udod ? suggests, to produce a grammar of conversation we must bear in mind, a ee Rlowever, too often categories are devised to cope with cach new pjece Seiatty-“artractive ‘transparent’ categories like “misapprehension se- nero “Sarifcaon and ‘pus’. We need w ext he problem some detail “Jeterson (1972) proposes an embedded sequence different from Schegloft’s insertion sequence and labelled side sequence. She observes that the general drift of a conversation is sometimes halted at an un- predictable point by a request for clarification and then the conver- sation picks up again where it left off. The following example is children preparing for a game of ‘tag’ A. srevex: one, ovo, three, (pause) four, five, six, (pause) eleven, eight, susan: Elever cight, - B 359 = sacs ce cage, Th tres Sp savers Elev te, aprwh—l. sri Se ei net. = * Ea Tne, scsi: That's better, bee molt Whereupon the game resumes. The side Sequence begins with a gues-—2 pi Cioning repeat an interrogative iter indicating that there ia problem im what has just been said, ‘and its work is to generate further talk~€-7 directed to remedying the problem’. Questioning repeats occur typi- cally after the questioned utterance has been completed, because only then can one be sure that the speaker is not going to correct himself or explain the unclear item; an interrupting questioning repeat is liable to attract nota clarification but a complaint, if you'd just let me finish’. Jeticrson suggests initially that the misapprehension sequence has a three-part structure, consisting of ‘a statement of sorts, a misappre~ hension of sorts, and a clarification of sorts). The exampte-above is “in fact more Complex consisting oT statement followed by two mis apprehension and clarification pairs. So far the side sequence looks rather like Scheglof's insertion sequence. There are, however, wo major differences: firstly, the frst item, the statement i nota first pair part, the other items are in no sense inserted and there is no expec- tation of who should speak at the end of the sequence or of what type16 An Introduction to Discourse Analysis of utterance should follow; secondly, while the sequence misappre- hension-clarification looks like a pair, there is actually a compulsory third element in the sequence, an indication by the misapprehender that he now understands and that the sequence is now terminated —"That's better’ inthe example above, or ‘yeah’ in the example below. 2. Statement: If Perey goes with — Nivon Pd sure like that. Misapprehension: Who? Clarification: Perey. ‘That young fella that uh — his daughter was murdered. (1.0) ‘Terminator: Oh yeash. Yeah. Because the first item, the statement, is nota first pair part, the con- versation cannot resume with the second pair partas after an insertion sequence, and there remains the problem of a return. Jefferson ob- serves that ‘it is not merely that there occurs a return to the on-going sequence, but that to return to the on-going sequence .... is a task performed by participants’. She suggests that the return can be ef- fected either as a rewmption or as a continuation — a resumption is achieved by attention getters such as ‘listen’ or ‘hey you know’, which mark that there is a problem in accomplishing a return, while contin- tuations, attempted by ‘so’ or ‘and’, are directed to ‘covering-up’ the problem, to proposing that there is no trouble, sometimes where that may not be the case. Thus the full system is: 3 statement: 1: And a goodlooking gel comes to you and asks you, y'know, misapprehension: 6: Gi (hh) rl asks you to — ‘Side ig | clarification” c: Wella its happened a Jota times, sequence | Sermination Okay okay go ahead. continuation: ®: So he says ‘no’ In trying to understand and use the descriptive categories outlined above, the intending analyst has all sorts of problems. Pair isthe only technical term which is defined; sequence appears to be a structurally coherent collection of not necessarily successive utterances oF ut ance parts, up to four in number, and pairs are also referred to as sequences; misapprehension sequence is apparently a subclass of side sequence, But we have no idea what other types of side sequence there From the way the authors describe and exemplify their categories it would appear that the real difference between Schegloff’s insertion _ sequence. and Jefferson's side sequence is that the forimer tas a ready- jade réiura, the second pair part of the question-answer pair, while Conversational analysis 77 forthe latter it has to be ‘worked at”. However, one could surely insert 4 misapprehension sequence inside ScheglofI’s Question Answer pits would it, could it, then be classified as an insertion sequence? s Tdon’t know where the — wh — this address // i. Q te Which one? Misapprehension ss The one you just gave me. Clarification Oh yeah, yeah. Termination '& Well you don't live very fir from me. A Pethaps itis a mistake to assume that insertion and side sequences necessarily have different distibutions; perhaps the main difference between them is the fact that they have different internal structures. As it is difficult to see how misapprehension and clarification differ in any fundamental way from question and answer respectively, one must assume any structural difference lies in the termination elemen swhich completes the side sequence. However, there Seems be nd reason why Scheglofl's insertion sequence couldn't also have a termination. ss Idon’t know where the — wh — this address // is Q Which par of town do yom fie a x: Llive four ten East Lowden. Al we Ah yeah, Termination Well you don't lve very far from me. A ‘Thus it appears that in fact these two sequences have different labels. only because they have been labelled from different viewpoints — insertion sequence is a structural label, misapprehension sequence a se~ mantic label attempting to capture the relationship between the first item in the sequence and the preceding utterance. “There is a similar confusion in the labelling of the constituent units of the misapprehension sequence. Following an item labelled clarification one might expect an item which indicates that the addressee now understands, the apparent function of ‘oh yeah, yeah in example 2 above, and therefore labelled something like acknox- edgement, Ip fact itis labelle mination, a struc emantic label and one which leads the re -why-the first item is sa ntl spear or fa LT TIRE caro Took. for misapprehension sequences in his own. dans the der has ony the obseraon tnt they begin with "ni ee ctesvlonat aan tod thee exiles -work feet pecan trace eter wenigle te waleeprelensiony the imei ceamples 1 and 3 lod a if they would be more satisfactory Ualed a sallng,Tolowed by a erection and ution re IU | _stt78 An Introduction 12 Discourse Analysis spectively. Confusion is increased by the data and analyses presented in Jefferson and Schenkein (1978): rarny: Oh Pd say he’s about what five three enna half = a rar: = Arenichu Ronald a Rox: Five four a perms: Fie fot, a parm: En 'e weighs about 2 hunnerd’n thiny five cs pounds «6 ovatp: = Aauggh! Whadds ca arn: Well how — = cs rar: = Ovright? How much dyou weigh a. Roxalo: One iment fixe C10 parm: Oh one ten five cu ‘They comment: Relying on Ronald's overhearing of her exchange with Gene (C1), Patty initiates an encounter with Ronald by soliciting a carrection from him on her estimate of his weight [sc] (C2); Ronald offers the correction (C3), Paty acknowledges the correction (C4), and returns at once to her exchange with Gene (C5}. Thus the proposed analysis is Correction Solicitor (C2) Correction (3) Acknowledgement (CA) ‘This seems perverse because, to this outsider at least, it appears that the last thing Patty is looking for is a correction; rather she is looking for confirmation or support. However, this labelling allows Jefferson and Schenkein to treat CS~C11 in the same terms, when this time we can agree that in C9 Patty certainly is looking for a correction fol- lowing Ronald’s outburst in C7. However, Ronald's C7 utterance is also labelled correction solicitor which wrongly implics that the ut- terances are similar — they differ crucially in terms of who holds the information and therefore in terms of what ean occur in the third slot following the correction — if, asin this ease, the person. supplying the correction ‘knows’, ic. itis in Labov’s term{ a B-event, ihe third slot will typically be filled by acknowledging ite and CH; however, had Patty responded to Ronald's C7 the third slot would ‘ypically have included something like ‘that’s better’, at which point We realize it would have been a sequence remarkably similar to that analysed in example 1, p. 75 as a misapprehensiof sequence. ‘The situation is further complicated by Schénkein (19; a Sequence which he labels Pyééle, Pass, Solution, Comment: Conversational analysis. — 79 11x: Fine just fine thank you 'cept for this fucking infection brn: Infection? HLLES: [can’t seem to get rid of this fucking, ub urin parr: They're impossible I know all about it deary bel track infection in which the second and third utterances have more than passing similarity t0 a Schegioff QA sequence and a Jefferson smisapprehension-clarificaton, Te 168) observes that initially he did mot attempt to describe cause it seemed to be ‘the sort of thing in which direct con- tent considerations were necessary and .... {he} couldn't proceed in {his] usual fashion... to extract relatively formal procedures’, Now he considers structure to be massively present. An intial Yaestion iw ga and do form topics in conversation? Some topics are not relevant to particular conver sé eit is a generat rute about convetsation that it is your b ‘well people What you carr suppose they know" (1971), and the suitabiliy OF other topics i PEFSOR one is talking to. We experience, see, hear about evemts all the time: some are ‘tell- able’, some aren't, and of those that are tellable some are tellable to everyone, some have a restricted audience, some must be told im= mediately, some can wait and still retain their interest. For instance, if one’s sister becomes engaged, some relatives must be told immedi~ ately, others on a first mecting after the event, whereas some of one’s friends might not know the sister or even that one has a sister, and for them the event has no import ven interest. *hy that now and to me’, and someone who con- i newowortines Sisal grees ola oi ceweentiy eseted to - aie Geom on poop to nom rele news, ve Sneed bythe existence of the telephone one no longer needs to wait unt te mees fiends or reaver nor doer ‘one weed to mae spel or difficult journeys to pass on information. Sacks has a good example of this in a tape of a series of telephone conversations. 4 and w are friends; w works at a local department store; A was driving past the store in the morning when she noticed an incident outside involving police cars (part of the transcript is reproduced on p. 88); knowing80 An Introduction to Discourse Analysis it was ss day off she rings up to tell her the news, thus ful- filling her obligations of monitoring the world for her friend and re- porting relevant events. However, 8 has a second friend, who also works at the store, who didn't ring up to tell her about it and whom she then rings to discover what all the commotion was about. This puts ¢ in a difficult position; she has been caught out failing to keep her friend informed and she takes the only possible way out, that of denying that what happened was newsworthy ©: Tt was nothing, wh — in faet I da'c even say ansthing to Willy about it In other words, if she didn’t consider it worth telling her husband to whom virtually everthing is newsworthy, it must have been an in- significant evel _/~. Thus we see there are certain things which one must say to pa G, sicular people and cern things which ae tllable if one happens to meet them. This leads on to the idea of reason fer a cal or visit isa basic assumption of all except chance encounters that the person ‘who initiates the encounter has some reason for so doing, and if there is no such reason people regulary feel the need to state this, “I was just passing’, ‘I just felt like giving you a calP. Conversations tend 0 xin with the topic which isthe reason for the-encounter and-ihen— ‘move on To otker tops; hough, ofcourse, the association of "reason for call wit ypic’ can be exploited by producing a false reason for the call and introducing the real reason as just another topic later in the conversation, ‘opic change Sacks (1971) bbserves that in a conversation which is progressing well ey rceptibly from one topic to anothgr, and suggests that The relative frequency of marked topie introduction is some measure aL ually Ft saeryion “Turns must display ‘why that n ‘and “Speakers specifically place almost all their utterances, and the most usual way of doing this is hy tying grammatically and topically to what has gone before. However, as Sacks (1968) stresses, talking topically and talking about some-topic chosen by another speaker is not the same thing at all, One can perfectly well have a sequence in which successive speakers talk in a way topically coherent with the last utterance, but in which each speaker talls on a different topic ‘That is to sey, ‘talking topically” docsn’t consist of blocks of talk about ‘a ‘opie’. And when one presents a topic, except under rather special cir- Consersational analysis 81 cumstances, one may be assured that others will ty to talk topically what you've talked about, but you can’t be assured that the topic you it ended was the topic they will talk to Speakers are aware of this as a problem and have ways of formu- lating a topic to make it more likely that other speakers will talk to ‘Sacks exemplifies with a hypothetical speaker who wants to talk about surfing. He could introduce the topic by saying ‘I went surfing yesterday’, but that allows topically coherent utterances of the form {did X yesterday’; that is, ‘surfing’ has been presented as one of 3 class of activites and a topically coherent next utterance ean focus fon another of that class. A better opening he suggests would be ‘I ‘went to Malibu yesterday’. Malibu is also one of a class, this time @ class of places, but it is a known feature of the place that people surf there, and predisposes the next speaker to at least begin with a ref- ‘erence to surfing, if only to say that they don't like it. Thus a possible sequence would be Las at Mali yee Yea us a Coy Fore ate (Sads M5) E \ c ) Topic conflict 4-—— “The phenomenon Loic eit jan be Frustrating at times for con- versationalists. Everyone Tas had the experience of failing to get in at the right time with a good story or experience, and then seeing it wasted because the opportunity never recurs. At times, within con- Cerstins, here is compettve talkwhen two speakers want ode ‘clop the tpi in ferent nays oth ht Benue thes know there sie no farther opportunity tos what they want 10. In the sample Koger nants "New Fike 10 be the tpi, Jim, Tent the New Pke depressing? Hh The Phe? EGA On the place i isqsing Any dy ofthe weck [Tin ur BOP [ieresng ns — But you go — You go — take — sine Those guys are losing money {yim go down — dow - down to the New Pike there's people oh: and they're old and they're pretending they're hasing Fan , but theyre really not kes: How e'n you tell’ Mm?82. An Introduction to Discourse Analysis xocte: They're — they'e ting make ting, but the plac ison the decline, 's like a dergencratc place 7 = wm so ROP. focen: Yknow? ns POP. is just — sooet: Yeah its one of thse per joints y'know? ne 15a flo! ehh (Sacks 1967) As we saw earlier, utterances normally relate back to the previous utterance — here Roger and Jim compete by skip-connecting, relating back to the last-but-one utterance, their own. Each time one of them gets a tum he declines to-tak about the previous speaker's topic and reasserts his owik Skip-connecting ig not an uncommon phenomenon, but apparently speakers-only-skip-connect over one utterance and thus, Ken's entry with what is a challenging question ‘How c'n you tellin fact preserves Roger's topic. Jim in his next turn is forced to produce @ normally connected utterance, but stil is able to use it to assert P.O.P. as a possible topic, ‘So's P.O.P. it appears that Jim, seeing he loses out, making a gesture of acquiescence sie holding oto P.O. pe fom Roger am acquiscenc in cudng OP, inte el ht in ope’ nen Yeah a ne of se pe 1 connects to Jim's as. (Sacks 1971) Once this competition has been resolved the conversation moves forward again. tories, 2 \ (Zor Sacks. store is any report of an_event — it may be only one (Sentenvé tong, buts usually longer, and therefore presents a problem speaker for any intending teller tending story-teller therefore nee turn-taking “to station this morning’. The function of such utterances is to select not next speaker as is usual in the turn-taking system, but next speaker but one, and to guarantee this speaker a stretch of uninterrupted talk in which to accomplish the story. Usually anything but a direct refusal is sufficient to allow the speaker to begin: Ws Las at the pie station this morning. ve Big deal, "Big deal’, yeah, somebody stole all my radio equipment outta my car (Sacks 1967) To seek a suspension of the usual and one ay of aching his. ya foo yrefiach, ‘Wanna Hear a joke? or ‘I was at the police” Conversational analysis 83 eefferson (1978) looks at how stories are fitted into conversation. She notes that one problem facing storytellers is ‘to display a relation ship between te sory and prior tal the following extract we carr-see-Om-OT The ways i Which tH is done: ex: . ther’s a place up in Mulholland where they've — where ihe’ bldg those housing pres ae nooee ‘oh have you ever taken. them Molhollan time tials? Hére'we sce two: dence: dtice tiene 10. dates wiarler, TORS, which shovs that what is to come isn't directly related to what just been said, and then an embedded repetition, ‘Mulholland’ which marks ‘he element of prior alk which triggered the story’; here they do, but ‘the wo devices need not occur in combination’ Azabov and W 966) in an ana f raat that the-stori res typically begin With an-abstractSan item which Carre the narrative is intended t0 i- lustrate’, and thus enables the listener to see the relevance of ini vidual narrative events: 1. Now, [think I i the sight thiog (Labo and Fanshel 1977, p. 105) 2 [talked a man out of — Old Doe Simon, I talked him out of pulling the tigger (Labor 19726, p. 363) As they point out, ‘one of the most important problems to solve in delivering a narrative is how to finish it (p. 109) — the listener needs to know when the account is over and when itis appropriate for him to respond; the abstract sometimes gives the listener some idca of what is required for the story to be complete, but also speakers have a series of ways in which they can bring the listener ‘back to the present time and so let him know that the narrative is completed’ and I see that man now and again ‘whem they see me now they say and ever sinee that time (Labor and Waletshy 1967, p. 10%) normally. Jefferson (1978) focuses on an instance when no one ap- produced five possible completions before solving the problem by pro- odd erie con re i eta84 Am Introduction to Discourse Analysis soos: through circumstances (gi) * beyond thr cony tol oe "[oeyond their entrol “While re-engagemen of turn-by-turn talk may be the primary sue upon a ston’s completion’ there are, as Jeflerson points ot, “ther maters to which a story-teller may be oriented” — specifically what other participants have made of the story, what its significance is taken to be. Obsiously the preferred reception is listener agreement with the proposed significance — in Labov and Fanshel’s example the therapist responds to the patient's evaluation with ‘yes I think you did too' — but as Ryave (1978) points out participants can disagree about what the real significance of a story is. Ryave confronts the question ‘how do conversational participants 0 about orienting to a present story in such a way as to transform the results ofthat orientation into the work of constructing their own succeeding story, so as to assure the constitution of acres of His data is two closely-linked stories. He notes, first of all, the obvious similarities of content and specifatons: both are concemed with the ‘faulty operation of . .. amusement park rides’, both have the ‘storyteller... impliated as a principal character’ and both are con. structed in such a manner as to make ‘a moral point’ (p. 121). In other words the second story can be secn as parallel tothe fist. However, it is not merely similarity of content that participants are sensitive to when constructing a series of stores: ston-tllers concer themsclss ‘not simply with telling their stories but with expressing the ‘i 2 general procedure employable by a suceeding stry-tlr fr construct ing a story that obscrablydspiys a sercyotstodes relationship with preceding story ito organize the stor in terms ofa spice atest Sich algo serves to formule preceding sry. 127) ‘BiB, sec ns cso ist en ante ox Hien naam abou Resear eal Shee : understang ; HIppOr. Spaliae and er WN We Teo ae ‘overt significance statement, second stories typically don't and are tac sen afer suppoe exeumcecie oie cea cance sutemeat. Hoveweyaseod peers ct ppnee aleeans iterprctations of previous stores bettas pata cet eho in's toy bu sa 'pupen’ phenometon' ard in te ree Ryave discusses the second speaker's story and significance statement are concerned with ‘altering and undermining the implications’ as- Conversational analysis 85 i ands with 9 signifi- serted by the first speaker — in other words he enc cance statement which fits his story and subtly re-interprets - fi story thereby sung her params with new sense of what the preceding story is actually about mt of ous, ot sors tat ar nee w mio fr inte ion, Sharrock and Turner (1978) Took at the ways 11 whi Pept ake omg to epee They observe tht complain TEITSTS Can correspondingly appear inCihis wersion as victi Ie) Ta aye come eg a st they haven't complained at the first opportunity, haven't specifically looked for the evidence oF incident about which they are complaining ‘and don’t necessarily want to implicate people by name — all of these devices to ensure that the police accept that ‘their complaints are what they seem t0 be. Topical coherence ‘The referential and descriptive items within any story are lated in iy complex ways, agLeach occurrence serves to reinforce and 1s asize The topic. ks (ISTRY presents some techniques for ana- psig topical eoherence By focusing on the ‘stor” ‘The baby cried. The mommy picked it up. Initially he suggests that most people wil ‘heat the story in the same way and wl agree wih the flowing as Fis, alhough there is no genitive)in the story, the mommy who picks up the baby is the baby’s 'y; secondly, that the two events occur sequentially; and thirdly, that the second eveni*occurs because of the first event. He sets out to produce a descriptive apparatus that will account for these facts, and, because it i ‘overbull, for similar facts in other stores. Ss Tiroduces the concept of prombeshipcalgprzation dvi iy ¢ D>) handle a descriptive category, for example ‘ex’, which comprises one SF ROE suOTaTMATE comets. eaegerey for example "mateahd enables oie t yulation mer arizaion devices and, because ‘some words av ma Fea SORE cRegores wil be members of more hun one device ‘Thus, while ‘baby’ along with ‘mommy’ and ‘daddy’ is a member of the device ‘family’, it is also a member of the device ‘stage of life’ along: with adul and ‘adolescent onIo rule. The economy rule states that “if'a member uses a single category \ eZ] ‘ ‘86, An Introduction to Discourse Analysis Sects then theducen ro rules the grag rile arith se ‘ie onp device thea aan ey recone es he ag eae a jose pS WS Soalomny vale once Ua ee Oe REE Bf deta i bon weed "tis concoct heme illo a) used To categorize further members of the population’. From the Coney eile Er deter a ecoliay, WAT sole ® becert mathtl Weve otsasle Caegriet cromelio ciepurke wear mre {, Members of some population and those categories cane heard as categories from the same colle then: Hear them thai way.” Thus ‘momnmy’ and ‘baby are heard asbeingeo-tiemver OTHE dete fam ily’, but how do we hear that this particular mommy and baby are-re~ tated? ‘Sacks fhe BoTeS—MuE-many-TonceS ae duplicative iplicatively organized, tha i, the population is seen to consist of a series of such “Tevices and is analysed as such — the populatio a large number of families, not a large number of unrelated mothers, fathers, children, babies, and-each device has a number of each. [tis for this” ‘Feason that The mommy is heard as the mommy of the baby, and for exactly the same reason the story ‘The first haseman looked round, the third baseman seratched himself will be heard as implying that the basemen are in the same team. Sacks next introduces the concept of categor-bound activites. He suggests that some activities are appropriate t, or done by, members of certain devices — thus “eying” TS amwetvity bound to the category "aby" when iris a member of the ‘stage of life’ device. Some devices he notes, incidentally, are organized in a positional way, that is there is an oniéted relationship like baby, child, adolescent, adult, and i this case category-bound activities ean he instanced to praise or de- grade. For babies and young children, erying, in certain circum- stances, may be the norm andl an absence of crying isa cause for praise, ‘what a big boy you are’, while alternatively an older child who cries may be told not to ‘bea baby’. Because, inthis story the baby is crying and this is an activity bound to the category ‘baby’ in the ‘stage of life’ device, both meanings of baby are simultaneously present. ‘The idea_of membership outlined in this article is developed in exloft (1972)Scheglotf points out that any speaker wishing to refer 10 @ place or location has a relatively large number of possible formulations — as he writes he could describe the location of his notes as ‘right in front of me, next to the telephone, on the desk, in ry office, in the office... in Manhattan, in New York City... While all these ‘correctly’ describe the location of the notes, on any eeeasion Conversational analysis 87 actual_use not all of them are ‘correct’. The problem Schegloft ‘poses is ‘how is it that on a particular occasion of use some term farily on whom one is talking to and partly on the topic. ‘Whatever the topic of the conversation, the speaker must member~ echer trges. Each time topic changes the listener must be re-member- Shipped, and during a comsersation the same person may be member. Shed! as a doctor, a rugby player, a Hera a gurdencr, a bridge plaver Speakers usually membership thei frends corres but may miake mistakes wth strangers, and shoppers membershippcd a shop- assistants can become very annoyed. Sacks (1968) reports an exchange aan oeropline xssrxcrr: Do you have a cigarette? Sin anorss: No we don't. They don't provide that service any more. “The stewardess assumes quite naturally that she has been membe shipped as a stewardess and that the question is addressed to her in that role. She indicates ths in her use of we’, and replies, on behalf of the organization, that cigarettes aren't available any more, Had she taken it that the passenger was membershipping her asa stranger not a stewardess, Sacks argues, she might have offered him one of her own, cigarettes. "Thus membershipping is not simply a feature recov hat is stid, iis also a vitally necessary determine cone says. As ‘Schegloff points ‘out, before a speaker can produce even a location term he must membership his hearer, and if he gets it wrong one gets sequences like: ts 1 just came back from Ireuapa we Where’ that When about how to membership someone speakers play safe and use a pre-sequence: © Diyou know where the ‘Tiboro Bridge is? we: Yeah, fs Well you make a right there fact that there is a diversity of possible formulations for the same person, place oF event allows for much greater topical coficrence SORRE ee ese car chor categories om The same deve Sacks (1968) and Schegio HATE THE Tolfowing piece ofa tele- phone conversation:88 An Introduction to Discourse Analysis srrize: Well I just thought I'd — re — better report to you what's happen’ at Bullocks today. Well I-s-got outta my car at ive- thi. I drove aroun’ an’ at frst Thad go by the from door a” the store, seanerre: Eyed, sities An’ there was two p’leece cars arse the ste, andeh — colored lady wan'wh go in the main entrance there where the ster is an sl the gifts an rhings, seaverre: Yeah, tstriie: And they, woulda’ let ‘er goin and he adda gum ke was holding 4a gun in is hand a great big long gun seawerre: Yeh, PSL: An’nen eer om the other side, L mean to the righ of there, where ‘he rmpayes come out there was a whole oh musta been tenuh eight ten employees stanning there Its immediately obvious that once Estelle has given the name of the store where the incident happened, all other places are described in relation tit. The police ears were ‘across the street’ the coloured lady wanted 10 go ‘in the main entrance’, ‘where the silvers the eight or ten employees were standing ‘oser om the other side’, “to the right of there’. Bullocks is the tic and all the locations are formulated to emphasize this. Had the teller, for instance, been coming out of the store across the road, the police cars would have been parked in front of the store and the incident would have been ‘acros the street” ‘The choice of location terms follows the consistency rule, Bullocks fs the topic and the way in which the places are formulated emphasizes its centrality. As Sacks observes, ‘the phenomenon of being ‘parked across the street from” is obviously one sort of characterization which turns on not onky where you are but what itis that is being talked about, and where shat is The analysis of conversation ‘The techniques described above are designed to handle parts of con- versations, short sequences, topics, stories. Sacks (MS) asks whether fone can usc-enmtenaling as an analytic unit. The Basie qUeSTOn is a THEFT features w conversations sharey-or-svhrether CoMETSaTONS consist of a random collection of Sater units in no fixed sequence. He suggests that greetings are close {Being universal in conversation and although they sometimes don't ‘cur, on some of these occasions their absence is noticeable, which suggests. that conversationaliss feel they are an almost invariant feature There are two important features about grectings: firstly, they oceur Comersational analysis 89 ; gorta ime myths ‘at the very beginning of a conversation, and cannot ies ee etucrntn, secondly the allow all the speakers 9 tm, wt at the beginning of the conversation: Pe fiw ae Hi Hi there 1 i Jon on which a conversation does There are two major types of occasion not open with a greeting. Firstly, conversations between people who do not consider themselves co-conversationalists, for example age ing terms and therefore do not ex- strangers. ‘They are not on greeting te e Change # greeting. The speaker who opens must demonstrate in.b first utterance why he is beginning the conversation: rst utterance why he is DEpINning Whe COMETS re: Slap Tine aruse me. Can you tell me the way 10 Hey. Youve dropped your book. “The other conversations which typically don’t open with a greting are telephone conversations, Sefegior (1968)-angues_that-althowh the person who answers the telephone may say hello’ this isn ; re ie rr rm heal eo the ringing ofthe telephone. Poll r chan in ane ihn) gating sequence io bein the-comer- simp ansers th Sation proper, although sometimes, if the answerer Peg ree ht a cheling sequence to make sure the caller i iki o he night penn, Sanmins ‘Telephone ings fore ete Greeting | 1: Goodmorni sequence \ x2 Oh bi, Tone more coi hough canal 25 Schegolf and Sars the it pi Pe datums mts when he clr Ps Ren hd90 Am Jntroduction to Discourse Anabysie uses his second turn not simply to reply to the greeting but to initiate the first topic: Where you been all day, Ive been trying to get hold of you? Even if the called doesn’t initiate, the first topic may still not be the ‘reason for the call. We mentioned above that sometimes a caller ‘may not want the real reason to occur in the distinetive first topic slot and may therefore substitute another. At other times conversationalists ‘may not feel they have anything sufficiently important to be preserved as the ‘reason for the conversation’ and there are ways of talking past the first topic slot. © Hello there hs Hello, ss What's new with you? : Not much, and you? &: Nothing (Sacis M5) The endings of conversations are also things that have to be achieved — speakers don’t just stop speaking. Conversations virtually always end with a casing pair, composed of ‘goodbye’, ‘goodnight’ “see you’, and so on, However, the elosing sequence can only-oecut when a topic has been cnded and other speakers have agecd.not 10 introduce any new Topics. Arsising-at-a-pointwhere a closing se- (quence can Hegin Fequires a certain amount of work. As we noted earlier, topics frequently merge one into another. ‘There are, however, ways of bmunding topics to produce a clean end- ing. One way involves one party producing a proverbial or aphoristic summary or comment on the topic which the other party can agree with. posts Uh — you know, it's just lke bringin the — blood up. ‘nuns: Yeah well. THINGS UH ALWAYS WORK OUT FOR THED BEST. porrsse: Ob certainly (Schegloff and Sacks 1973) Another technique is fr the speaker to indicate that he has nothing further to add to the topic by using hs turn wo preduce simply al righ, “okay, ‘so, well’ often lengthened and ith a falling Sinton. ation contour. In daing this the speaker ‘pases. This allo the next spcaker the choice of cither introducing an entirely new topi, be- cause the constrains of topical coherence have been ited, or of alsa passing and turning the fist speakers offered pesble pre-dsing into 4 pre-closing sequence. Then, a5 neither speaker has raised + new topic they can move into a closing sequence and end the conversation: Conversational analysis 91 fe [Lew Seqmee {Tam Goodnih vit) In this example both participants agreed thatthe conversation had ‘gone on Jong enough; however, sometimes one speaker wants to en fut for some reason is unable to achieve a topic bounding sequence and is then forced into a different type of pre-closing: either a state iment which presents a reason for stopping: 1 gotta go, baby’s ening. ‘or an offer to allow the other speaker to stop: Wa, a mae op re i easing 3 a of mone / Oe ih py tack to wach your Dla ia) e-dlos especially the later Again, these are only possible pre-closings and especially IIR nay not be accepted the other speaker may den that he wants to get aviay, though if he does accept they can then move straight into the closing sequence’ thas ae 1 init but hs Ben ld going out vo dinner rts ae Yeah. Well get on your clothes and get out and collect Prectnng |" some of tate ood and we ake sme ok ine rete’ | judy hen 7 ce Okay Jack. ae US the be (Scheglof and Sacks 1973) “These examples contain only the essentials of a closing, an achieved preccag acne and sing pi re-clnngs my nde making arrangements, re-emphasizing arrangements made calier,re- stating the reason for the call, as well as many repetitions, and may Continue for many utterances: Schegloff and Sacks quote a ‘modest sample of a closing containing twelve utterances. The slot after the ‘possible pre-closing’ is the one provided for introducing any topic which has not yet received mention, but new topies ean be introduced afier a pre-closing or even afer a closing, provided they are marked as being misplaced92 An Introduction to Discourse Anabssis cues: Okay, thank you. cawsonti: Okay dear. cana: OH BY TH WAY I'd just ike to... (bia) Any items inserted during the closing after eal 3 closing after earlier opportunities have been passed up have the status of afterthoughts and this position ean be exploited in order to take away the importance of a piece of news = one doesn't normally forgct really important items. ‘Schegloff and Sacks Gbid.) quote the following example which oc- curs, at the end ofa fairy long telephone conversation, following the pre-closing, when both speakers have apparently indieated they have no more topics. ae -— ah Tdi wanna tel you an I ida? wanna tell you uh; Un because you had entert — uh, compen, Li had sonehiag tele el 30, Soph » omany EET had soning ® hove terre isi? x Uh, tuh as warse it could be or) we Wey? mean Ada? x Uh yah 1 Whatshe do, die? a: Mamhm, Stylistic features of conversation Ina fascinating series of lectures given during (970 Sacks suggested that conversation has much ofthe additional phonological, grammatical find thematic panering which usually thought wo pps wars of “Tira discussion of an extended version of the “skip-connecting’ passage (i volume pp. 81-2) he points out marked phonological atterning. There is a series of words, “depressing, ‘disgusting’, ‘le feces ecine ‘dese wt he fen eo shone a identical, displaying: eersc hy ‘There are also words and phrases which echo each other Because they share phonemes — ‘de- gencrate’, and ‘pier joint’; ‘walkin aroun drunk’, and ‘all kindsa fun’; “alcoholics’_and ‘all kindsa things’. 7 chs (que) that features such as these cannot be reiccted as st erase They occur oo Tequentt and any rejection on the grounds of implausibly jznores the fact that speakers ae all the ime achieving effectsof similar sophistication and complexity, These Phonological echoes are evidence of how closely attentive speakers 16 1o-cach other 1 speaker's choice of one formulation rather than another is partly determined by the phonological patterning of the Concersational anahsis 93 previous text and the alternative Formulations. Texts also display marked lexical patterning. This same fragment has a large number of marked contrast rerms, for example, ‘these’ and ‘those’; “go to" and ‘come from’; ‘in’ and ‘out; ‘you' and ‘they’ ‘men’ and ‘ladies’; ‘new’ and ‘old’; ‘ever’ and ‘never’; ‘pretending’ and ‘teal’; ‘depressing? and ‘fun’ The use of such contrasting terms is particularly appropriate at this point in the conversation because of the topic conflict which we discussed above Tn the following example from a group therapy session Roger is complaining about the way in which he has to describe his preferred ‘occupation — he wants 10 describe it as life-pervasve: ocx: When I say I wanna be something, it’s not just that I wanna be ths, its just 1-1 just — that’s the only thing It people that T wanna be an artist. I's really a whole way of ile « You visualise yourself ul living a certain way 1 see it as a whole picture 1 don’t see it that way at all 1 — How am I gonna live, what am T gonna do for a living, and the whole scene And ub since most people don’t think along these lines (Gxcis 1971) ‘One must remember that this is unplanned conversation and that while” there is avast number of ways to describe his problem he consistent’, uses specially visual, ats-appropriate, tems. "You visualise you Selly see it asa whole picture’ T don't se’ ‘the whole scheme’ ‘along these lines’. a a uses of lang ‘resents another extract froma group therapy seasion where the patients have been talking round the subject x without actually mentioning it. The therapist observes, “Well so far, all of you skirted around the subject. That sce(hh)ms to b(b)e pre- dominantly uh on your minds at any rate’. The problem facing the therapist in producing this utterance isto indicate that he knows the topie they have been talking around, without actually introducing i himself He dacs this by choosing an expression, skirt around’, which both means “to allude to’ and ako stsef alludes, by a pun on ‘shit to the hidden topic. Stories in conversation tend to be created anew for each retelling;94 An Introduction to Disease Analysis jokes usually have a fixed form and can be ruined by slight akerations, Sacks focuses on the structural complesities of the following ‘dirty joke’: ‘ses: You wanna hear muh — ch my sister told me a story lastnight Phere — There was these three girls. And they were all witer, And they'd just got married to thr So first ofall, that night, theya're — on their honeymoon the — uh mother-in-law, says (to "em) well why don'tcha all spend the night here an then you en 20 on yer honeymoon in the moming. First night, th’ mother walks Up # the first door an” she hears this wuuuuuuuwuhht Second door is HHOOOHH! "Third door there is nothin. She stands there fer about reunny five minutes waitin fer sumpra happen — nuthin, Next ‘moming she talks 1 the fist daughter an she sz— wh how come ya — how come y'went YEEEFEEAAHAGGHB last night, daughter sez woll it teed mommy, ‘n second git how come ya sereamed. ‘mommy it hurts. hh third gel, walks up t” her — why didn ya say anything lat night, Whew told me it was always impolite talk with ‘my mouth full. (Gacks 1978) Sacks points out thatthe joke has a simple temporal ardering — the events are told in the order in which They are said to have hap- ened. There is also a more complex sequent organization by which each of the events depends for its significance on its position with relation to other events — ‘next morning’ has its relevance because there was a ‘last night’ and in order to understand what is being im- plied by the reference 1 ‘second door’ one uses that there was a “first door’. The joke has two major units, the “rst night” and the “next ‘morning’; the ‘frst night” poses a problem which is solved the ‘next morning’, and the solution is nicely placed a the punch line which closes the second sequence ahd the joke itself. The mother in the joke acts as a guide — the listener sees the events through her eyes. The fact that she moves from first door to Second and second door to third as soon as she hears a noise shovis that itis noise which is important and enables the listener to see that lack of noise i the important event at the third door. Her questions next morning elicit explanations of why the noises were different and the meaning of the silence. As Sacks notes, the economy and organization ofthe events in each sequence is admirable. The joke depends on the fact that silence in the third room is unexpected, but the joke doesn't actualy say the ‘mother was surprised. In order to show that something is not normal els a least but fot more than three TaSGANCES, TWO ormal and — ne“Hon-normal, The events could Be presented in any Order “Teh I WOUTT BE necessary to comment on the unexpectedness of one Conversational analysis 95 of them, The joke presents to “normal” evens fst and einforess this by the fact thatthe mother moves from one dor to the nent the third events sen 8 erent rom the fist wo an the mser was The seeond sequence, ‘next morning’ preseres the sequenee ard alos the esplnation ofthe uneypected ee ecu asa to ‘normal’ explanations can predispose the listener to hear the p Ee anck neo his ke ha an extra dimension The ok was lo Ken yin 13a sister, and emis adolescent til obectons to parental roles and interference, The cleverness of the punchline that it uses one rule to explain the breach of “nother and incidentally complains that sometimes itis impossible satisfy one's parents.5 Intonation 1s suprising if not starting wo realize that although sgl nh pei capes tos ft ore we analsis of speech, there hasbeen virtually no attempt to accoun for th sic fans inthe mor hare pec mens of supra-segmentals: paralinguistic features of voice quality and prosodic features such as pitch, pitch movement, loudness and length. Voice aia embracing ih caterers “eth “a i ea ato “oa sa NC of but, although Crystal (ibid.) and Laver (1980) fer po de ssp i itt ach specie meanings the shies yond the aber tht tinction might well be w the adonal emphasis or pong to an aide andy present i utterance’ (Crystal +» p- 137). - “= __ Austin (1962) refers on several occasions to the importance of es shes cadence, emphasis’ and ‘intonation’ but in fact his a ‘eS NO account of these features; Searle (1965) includes intonation contour among his “ancion indicating. device, bat never expands: while of the ‘ethnography of speaking vestigation reported above, only Irvine's (1974) discussion of Wolof cectngs I of ps ees — so! omen —o and ol Jefferson ho (1974 kes tay no atcout the ea able prosodic and pars-linguisic information, though much of the ‘conversational analysts’ data is quoted in the detailed Jefferson trans- cripton which shows hestation phenomena, lengh, sess and some ilomaon aot pick and ich meme at a the ala scene so far can be applied very successfully to ordinary ortho- azaphictansrpts with no accompanying tape, which confi that appeal to intonation ts spasmodic, if not haphazard, and. ec when differences are perceived for which there can be het explanation. — Tntonation, the syste patterning of prosodic etre, is of couse Intonation — 97 also a problem area — whereas native speakers have no difficulty g the system communicatively, they find it very difficult to intro- pect about the significance of the choices they make, and even to Jroduce ciation forms reliably and corrcels. Sadly, when discourse valyets coming from varying backgrounds look to linguistics for hep, they find very little — in the main American linguists have ignored iimonation, Trager and Smith (1951) produced a simple partial de~ scription, ‘while. Transformational Grammar saw intonation as a feature to be added later, once all the major grammatical decisions dnd lexical insertions had been made. Thus all the work reported here will be on British English. ‘Theoretical preliminaries Digcourse analysts need to be able to describe the meaning of supra~ segmental choices and this requires a two-stage process, first as- Signing the data to categories, and then assigning meaning to the Categories. For the first stage, we need to draw on three traditional principles of phonetic and linguistic description. First, features that are coustically on a continuum must be analysed as realizations of a small thumber of discrete units that ‘form a closed set defined by their mu- tual oppositions’ (Labov and Fanshel 197, p. 42) — just as we accept that sven phoneme conflates a large munber of acoustically distinet Sounds, so a falling tone, for example, will asa category include large variations in length and steepness of the fal. “The second principle is that there is no comstant relationship be- tween particular acoustic phenomena and particular analytic cat- tories itis contrasts and not absolute values which are important ‘This principle is not novel and creates no problems theoretically or fs analysts of tone languages discovered long ag0 tone languages have @ major characteristic in common: i is dhe relative JReght of their tonemes, aot their actual pitch whieh is pertinent 10 thei Frastsie anasis the important feature i the relative beight of» sit Iitfe in relation to preceding and following syllables. A toneme is ‘high’ Ui is higher tran its neighbours inthe sentence, not i frequency fof vibrations i high (ike 1948, 4) practically, |A third principle is that there is no necessary one-to-one relation- ship between a given suprasegmental choice and a meaning: on the fone hand, as Bolinger’s (1964) ‘wave’ and ‘swell’ metaphor suggests, a given pitch choice can at the very least be simultancously carrying both general information about emotional state and a specific local98 An Introduction to Discourse Analysis meaning of the kind described in detail below, pp. 104-110; on the ‘ther hand, certain interactionally significant signals, like for instance 4 request for back-channel support, may be carried by the co- ‘eccurrenee of a particular pitch choice and a particular kinesic one, each of which singly has a different significance (Gosling, in preparation). Descriptions of intonation ‘There is in fact a large measure of agreement about the basic facts of English intonation, though descriptions differ markedly in the sig- nificance they attach to choices. O'Connor and Arnold (1959) sug- gested that ‘the contribution that intonation makes is to express in addition to and beyond the bare words and grammatical constructions used, the speaker's attinde to the situation in which he is placed (.2). However, their attempts to provide valid generalizations about the attitudinal meanings of the tunes they isolated serve only to dem- onstrate the difficulties. For instance, they suggested that a speaker uses a low falling intonation with a statement to indicate that it is definite and complete in the sense that it is a ‘separate item o terest’; but that in addition the intonation conveys a ‘eool, calm, phleg- atic, detached, reserved, dispassionate, dull, possibly grim or surly attitude on the part of the speaker’. Their own examples demonstrate the problems admirably You've got dipstick on your collar again, Ws geting te, Itis dificult to hear the first utterance as ‘coo!’ ‘calm’ or ‘detached, and the second as ‘grim’ or ‘surly’. In fact, even though O'Connor and Ammold offer a hundred different ‘meanings’ including such fine distinctions as ‘mild surprise but acceptance of the listener's premises’, ‘critical surprise’ and ‘affronted surprise’, these meanings seem to depend, as Brazil (1975) observes, far too much on ‘the co-occurrence of particular lexical items’, Could it be that O'Connor and Arnold's intuition that intonation carries attitudinal meaning is correct but that the attitudinal labels they assign are badly selected? Apparently not; Crystal (1969) reports an experiment which demonstrated that native speakers find it virtually possible to agree when matching attitudinal labels with intonation contours, and this confirms feclings that individual intonation choices «do not in fact carry specine attitudinal information. Crystal himsel? Proposed a very detailed analysis of the intonation choices available in English, which has met with general agreement, but while accepting we Intonation — 99 prosodic contrasts... must in ht ‘statements of the meaning of the pro ams in Mec be dinar al of he ng (282) Be cones i= fel unable to make any valid generalizations — and by 1975 he was Angin that ‘the vast major of ones in conneted speech car m2 meaning’, although accepting that a few carry attidinal opt ‘absence of emotional involvement. ; ; Dy conrst aly (1970 ase tat ‘he importance of itn r hati a means of saying ferent things. Ifyou change ation is... that it is a means of say yom hanes the intonation ofa sentence you change its meaning’ He suggest that Jnonaton choices camry two kinds af information: firey the rela importance of diferent pars of the message determines and therefore iseanveved by decisions about when and where 0 make major pitch mmements; secon the cies of one Bich cmt rather thn Anorer Flats to grammatical ‘mood kinds of statement, question Cie), modaly (assessment ofthe possibiiy, probaly vay, re Chance, ete, of what is being said) and key (speaker's attitu politeness, assertiveness, indifference, et)" (P. 22) — llidy (1967) suggests that she rumber of significant choices i only fie (Table 5.1), When he comes to attach meanings to intonation choices he suggests an even more powerful generalization, tha there So Ee ee agnor indicates whether polarity is certain or uncertain (polarity is the Choe Heteen poste and EAE aaa pay rer or query noc aang ol ard ve wih fas and then res Won which invokes oF entails some quer. fas and then Hes a pS which rss and then flys dsmsed Tone 22 Goartcrcd by an assertion, ‘Tone 3 avoids a decison; as an SESPlon is at bes confirmatory, contingent or material ‘rants: §.1 Significant intonation choices Tore Symbol Tovicmveen Terminal pth tend . ling 8 ° i flinging oe 3 sng nad $n ting ing mi 5 ting iin io (Halliday 1967)100 An Introduction to Discourse Analysis Wecan tempi the way these gener meanings tegen spe ea etude Senst bremialobigadecates tmmnates Ver dairies wid raise ogee te accetne nace shee al alee noel we eg timed cece T9e oat lines // indicate the boundaries of the tone group, the number at the beginning the tone, while the syllable underlined is the one on which. So sat norm oi I.1- {saw him yesterday f V2 U saw him pesterday ny Cohallenging’ “oppressive “defensive, indignant ete) I Cdisengage’, “uncon : | cemed,“discouraging) If 4. saw him yesterday fteseration (theres 3 “but” about) 1/5. saw him Sesterday / committed Ciwohed, ‘asertne’,super- ior, “encouraging 1/3. Lsaw him yesterday “Thus, the speaker is seen as having four major options which allow him to indicate his degree of involvement with the information and his imerlocutor. It is evident that this kind of analysis has great potential for dis- course analysis — it has a small and therefore powerful set of categor- ies and a linked set of general meanings which, eombined with ini vidual clauses, generate a more delicate meaning in context. Sinclair al. (1972) is one early attempt to employ this system, which also showed up some problems. - Brazil Ina series of publications Brazil (1973, 1975, 1978, 1983) has mod- ified, refined and gradually diverged from the original Hallidayan ‘model, adapting it to fit with the model of discourse structure presented in Sinclair and Coulthard (1975). Brazil's model keeps the major distinction between rising and falling tones, though he attributes different significance to the opposition, but radically moves away from tying the description to the grammatical clause, arguing that the in- tonational divisions speakers make in their utterances are motivated by the need to add moment-by-moment situationlly-specific mean- ings to particular words or groups of words. It isto the still develop- ing description, available in a more detailed form in Brazil er al (1980) and Brazil (1985), that the rest of this chapter will be devoted. * This dscasion devs ay on Boa’ conerbuton to Br Jo (1980) and Coulthard and Brazil (1982). Sas Intonation 101 Like the other descriptions already discussed, Brazil's is expressed in terms of pitch choices, though this is almost certainly a simplifi- cation. Intensity and durational features regularly co-oceur with the pitch choices and it may well turn out that the choices described as being realized by pitch phenomena are being identified by hearers at east some of the time through associated intensity and durational phenomena — Lieberman's (1960) experiments on stress urge eaution in this area Brazil suggests that there are four sets of options associated with the tone unit — prominence, tone, key, and termination — each of which adds a different kind of information, The tone unit itself has the fl- lowing structure. (Proctitic segment) Tonic segment (Enelitic segment) As this structure implies, tone units may consist simply of a foie seg- ‘ment, and many do; indeed a considerable number consist of no more than a tonic syllable, ie. the syllable on which there is a major pitch movement: {J GOOD /[/ YES /,! ME /, / JOHN // Most tone units, of course, do consist of more than the tonie segment, and here the question of segmentation arises. With the syllables fol- Towing the tonic there is in fact no analytic problem: even though the pitch macement of the tone may be continued over succeeding syl- lables, for reasons which will be explained later the tonic segment is considered to end with the tonic lable, ‘Thus Enclitie segment ness knows // Tonic segment ny’s coming // However, while the final boundary of the tonic segment is by defi- nition unproblematic, recognizing where the tonic segment begins is ‘a more difficult matter and depends on an understanding of the con~ ccopt of prominent syllable, which will necessitate a short digression Tt is not always easy to be sure what significance writers attach to such terms as ‘stress’, ‘accent’ and ‘prominence’. For Brazil, accent ‘means the attribute which invariably distinguishes the marked from the tunmarked syllables in words like ‘aur tain, con‘iain, re'la thon, and dis- tinguishes the lexical items from the others in a sentence like “Tom is the best bay in the class. Thus, when we say “Tom is the best boy102 Aw Introduction tn Discourse Analysis in the class’ we are not acenting ‘is’, we are making it pitch prominent (Readers will notice, when performing the example, that ‘is’ high- lighted by having a higher pitch than ‘Tom’,) Prominence is thus the name given to a property that is not inherent, like accent, but only associated with a word by virtue of its function as a constituent part of a tone unit. ‘To return, itis now possible to define the scope of the tonic seg- ‘ment: it begins with the first prominent syllable, henceforth called the onset, and ends with the last prominent syllable, the donic, which has the additional feature of tone or pitch movement, whose significance will be discussed at length below. There are thus, by definition, no prominent syllables in the proclitic and eneltic segments. All promi- rent sjlables will now be distinguished by capitals and tonic syllables will in addition be underlined. Prolite segment Toni segment Endive segment he was GOing to GO that’s a VERy TALL STOR it was a WED nesday If-we expand the first example to {fe was GOing 10 GO again ‘we now have four classes of syllable: unaccented, ‘he’, ‘was’, ‘to’, ‘a-’; accented, ‘-gain’; prominent, ‘GO'; and tonic ‘GO’. It is interesting to speculate how far these are in fact the four degrees of stress which Trager and Smith (1951) proposed, and which others have had great difficulty in recognizing. Prominence, then, isa linguistic choice available to the speaker in- dependent both of the grammatical structure of his utterance and of word-accent, What then is its significance? Let us begin with the following question response pairs 1. a: Which card did you pla n // the QUEEN of HEARTS // 2. us Which queen did you pay? ve // the queen of HEARTS // 3. a: Which hear did you play? xe // the QUEEN of hearts // 1 fl discussion of the fandamena frequency characteristic of prominemt svlables wil be found in Bra 978 and blr ir mre cee Sacion in Bra, Coulthard snd Jonns 1980, Intonation 103 ‘The three responses differ only in terms ofthe assignment of promi- hence and we can see already that prominence has something to do ‘ith informativeness. In exploring this concept further let us begin With the word ‘of It is easy to see that ‘of? is totally predictable: it is the only word that could occupy the place between ‘queen’ and ‘hearts’. If we think of each word in the phrase as representing a selection from a set or paradigm of words available, then atthe place filled by ‘o? there is a set of one. In this respect it can be compared with the places filed by ‘queen’ and ‘hearts’ where the options are seater: a hears r ‘a clubs een 7 diamonds King spades In the contest of the question ‘which card did you play?" the speaker has a choice of thirteen possibilities at the first place and of four at the second, but this time the limitation has nothing to do with the working of the language system: as Brazil points out, there is no fing Tie reason why the response should not have been ‘the prince of forks’ or ‘the seventeen of rubies’, or any of an enormous number of combinations. What imposes the limitation this time is an extra~ linguistic factor, the conventional composition of the pack of playing: cards, “This is @ crucial distinction and it makes clear the need to dis- tinguish the existential paradigm, that set of possibilities that a speaker can regard as actually available in a given situation, from the general paradigm which is inherent in the language system. Obviously at the place occupied by ‘of” the two paradigms coincide: there can be no possibility of selection in the existential paradigm because there is none in the general paradigm, whereas at the place occupied by ‘queen’ the general paradigm of thousands of nouns has been re~ duced to an cxistential paradigm of thirteen. We can now explain the prominences in T and 3 as marking a selection from an ‘existential paradigm — when ‘queen’ occurs in the question as in 2 there is no choice in the response, ic. no existential paradigm at the position occupied by ‘queen’, and the word is therefore non- prominent. We can support this assertion by imagining a context in ‘which ‘of’ can be seen as a choice from an existential paradigm — for example, when a foreigner makes a mistake in his choice of preposition then the correction contains a prominent ‘of":