Questions and Responses in English Conversation 91-40-05042-4

694 152 10MB

English Pages 312 Year 1984

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Questions and Responses in English Conversation
 91-40-05042-4

Table of contents :
Front......Page 1
Contents......Page 9
Preface......Page 13
1.1 QUESTION, RESPONSE, AND FOLLOW-UP......Page 15
1.2 AIM AND METHOD......Page 17
1.3 THE CORPUS......Page 18
1.4 DIALOGUE TYPES......Page 20
1.5 CLASSIFICATION OF TEXTS BASED ON Q PURPOSE......Page 22
2.2 TURN-TAKING......Page 25
2.2.1 The turn......Page 26
2.2.2 Backchannels......Page 27
2.3 CONVERSATIONAL CUES......Page 29
2.3.1 The interactive aspect......Page 31
2.3.2 The communicative aspect......Page 33
2.4 COOPERATION......Page 35
3.1 PAIRS IN DIALOGUE......Page 36
3.2.1 Problems of definition......Page 38
3.2.2 Context of situation and common ground......Page 39
INTONATION......Page 42
SYNTAX......Page 45
LEXIS......Page 46
3.2.4 Features of R......Page 47
3.2.5 Predictability......Page 48
3.2.6 Cohesion and coherence......Page 49
3.3.1 Discourse actions......Page 52
3.3.2 The single/multiple aspect......Page 55
3.3.3 Constraints on questioning......Page 58
3.3.4 Elicitative force......Page 59
3.3.5 Conduciveness......Page 61
CONDUCIVE DEVICES......Page 62
VARIANTS OF CONDUCIVE Q......Page 67
RESPONSES TO CONDUCIVE Qs......Page 70
3.4 HOW R FUNCTIONS......Page 71
3.4.1 R sets......Page 72
3.4.2 Comments on R......Page 80
3.5 HOW F FUNCTIONS......Page 81
4 Model of Analysis......Page 85
4.1 ACTS......Page 87
COMMENTS ON THE ANALYSIS OF ACTS......Page 96
4.2 MOVES......Page 97
4.3 EXCHANGES......Page 100
COMMENTS ON THE ANALYSIS OF EXCHANGES......Page 102
4.4 SEQUENCES......Page 103
4.5 TRANSACTIONS......Page 104
COMMENTS ON THE ANALYSIS OF TRANSACTIONS......Page 107
5.1 MIXED AND UNMIXED TRANSACTIONS......Page 108
5.2 TRANSACTION BOUNDARIES......Page 109
5.3 TRANSACTIONS IN A SUBCORPUS......Page 114
TEXT 11.1: A COURTROOM EXAMINATION......Page 115
TEXT 6.3: A POLITICAL INTERVIEW......Page 117
TEXT 7.2: TELEPHONE CALLS......Page 118
TEXT 3.4: A VISCUSSION BETWEEN AN ADMINISTRATOR AND ACADEMIC STAFF......Page 119
TEXT 3.3: A DISCUSSION BETWEEN AN ADMINISTRATOR AND STUDENTS......Page 120
TEXT 4.1. A DIALOGUE BETWEEN HUSBAND AND WIFE......Page 121
5.4 COMMENTS ON TRANSACTIONS AND SEQUENCES......Page 123
6.1 EXCHANGE STRUCTURE......Page 125
6.1.1 The eliciting part of the exchange structure......Page 127
6.1.2 The responding part of the exchange structure......Page 133
6.2 LINKAGE......Page 135
6.3 EXCHANGE OPENINGS......Page 136
6.3.1 Opening strategies......Page 138
INITIAL OPENINGS......Page 139
NON-INITIAL OPENINGS......Page 148
6.3.2 Initiators......Page 152
PREVIOUS APPROACHES......Page 153
DEFINITIONS OF INITIATORS......Page 154
[ə:m] AND well COMPARED......Page 155
DISTRIBUTION OF INITIATORS......Page 157
COMMENTS ON INITIATORS......Page 160
6.4 COMMENTS ON EXCHANGES......Page 162
7 The Q move......Page 163
7.1.1 Primary acts......Page 165
......Page 169
and ......Page 170
......Page 171
Alternative Qs (ALT Qs)......Page 172
MULTIPLE FUNCTION......Page 173
......Page 175
......Page 180
......Page 184
COMMENTS ON SECONDARY ACTS......Page 185
7.2 SUBORDINATE MOVES......Page 186
7.2.1 [Checking]......Page 187
7.2.2 [Re-opening]......Page 189
7.3 SUMMING UP THE Q MOVE......Page 190
8.1 PRIMARY ACTS......Page 193
8.1.1 R to ......Page 199
CHOICE OF WH-WORD......Page 200
THE SPEECH SITUATION......Page 203
......Page 204
......Page 205
......Page 206
......Page 207
......Page 208
......Page 209
YES/NO-Rs......Page 211
......Page 215
......Page 216
......Page 219
......Page 220
Choice of terms......Page 221
......Page 226
8.1.4 R to ......Page 227
and ......Page 231
......Page 234
, , and ......Page 235
8.1.5 R to ......Page 237
8.2 SECONDARY ACTS: Elaboration......Page 243
8.2.1 What elaborations do......Page 248
8.3 SUMMING UP THE R MOVE......Page 249
9.1 OPTIONS......Page 251
9.2 THE [FOLLOW-UP]......Page 253
9.3 THE [RE-OPENER]......Page 254
9.4.1 Distribution......Page 255
9.4.2 Pitch direction......Page 259
10.1 GENERAL CHARACTERISTICS......Page 262
10.2 MULTI-PARTY DIALOGUE......Page 263
10.3 TELEPHONE DIALOGUE......Page 265
10.3.1 Openings......Page 266
10.3.3 Closings......Page 268
10.3.4 Comments on telephone dialogue......Page 269
10.4 COURT DIALOGUE......Page 270
10.4. 1 A civil case......Page 271
10.5.1 A political radio interview......Page 273
11 Conclusion......Page 276
SEQUENCES......Page 277
EXCHANGES......Page 278
THE Q MOVE......Page 279
THE R MOVE......Page 280
ELICITATIVE FORCE......Page 282
STRATEGIES IN DIFFERENT TYPES OF DIALOGUE......Page 283
APPENDIX I: TEXT DESCRIPTION......Page 285
APPENDIX II: INFORMATION ABOUT SPEAKERS AND TOPICS......Page 286
APPENDIX III: SIGNS AND SYMBOLS......Page 296
APPENDIX IV: EXCHANGE PATTERNS......Page 297
References......Page 298
Index......Page 303

Citation preview

LUND STUDIES IN·ENGLISH 68 EDITORS CLAES

SCHAAR AND JAN SVARTVIK

QUESTIONS , AND RESPONSES In English Conversation

BY

ANNA-BRITA STENSTROM

J

CWKGLEERUP

GAVE FRA STATSilIBLIOTEKET I ARHUS

INS'fITUT FOR LINGVISTIK Aarhus Universitet

QUESTIONS AND RESPONSES IN

ENGLISH CONVERSATION

BY ANNA-BRITA STENSTR~M Fil.mag., Krsd

DOCTORAL DISSERTATION to be publicly discussed in English in lecture room 239 at the Department of English on 17 March 1984, at 10.15 a.m. for the degree of Doctor of Philosophy.

Document name DOCTORAL DISSERTATION

Organization LUND UNIVERSl1Y

Department of English Helgonabacken 14 S-223 62 Lund Sweden

Date of issue

February 1984 CODEN:

LUHSDF/(HSEN-1009)/X+298pp (1984) Sponsoring organization

Author(s)

\

Anna-Brita Stenstrom Title and subtitle

Questions and Responses in English Conversation Abstract

The strategies for asking and responding to questions hn genuine conversation arc described by means of a hierarchical discourse model and accounted [or statistically in 25 face-to-face and telephone conversations rrom the London-Lund Corpus of Spoken English. Utterances arc analyzed on a pragmatic/functional basis as moves and acts in exchanges, which form higher units, sequences and transactions. The basic i.ntcractional unit, the exchange, is realized minimally as Q(ucstion), R(csponse), and an optional F(ollow-up) move. In order to cover the whole range from most to least Q-like, four different types of Q have been identified. These are matched by a corresponding set of Rs, which are further subclassified according to the adequacy of the information given. The data confirms that an utterance can function as Q regardless of in· tonation contour and syntactic form. The determining factor is the role of the utterance in a particular context, where lexical markers, intona~ tion, and syntactic form often serve as functional clues. Due to the speakers' cooperativeness, Qs are nearly always responded t1 gen~rally by an R answering Q. Minimal Rs are unexpectedly common, eg yes/no answering yes/no-Qs and syntactically incomplete utterances answe: ing WH-Qs. 'Real' Qs gen~rally elicit the information wanted and are £re• quently follow~d by a confirming F move on the questioner's part, as opposed to statement-like Qs, which only receive acknowledgement. This di£· ference illustrates the degree of elicitative force in Qs.

1

Keywords

Question, response, follow-up, conversation, interaction, elicitation, elicitative force, information, cooperation, exchange, move, act Classification system and/or index terms (if any) Lang(!age

Supplementary bibliographical information

English

ISSN and key title Price

Recipient's notes Security classification D1stnbuhon by (name and address)

Liber Forlag, S-205 10 Malmo, Sweden I, the undersigned, being the copyright owner of the abstract of the above-mentioned dissertation, hereby grant to all reference sources permis;/on t o ~ n d di1sei2e abstract of the above-mentioned dissertaH~ 1

Signature

d"f=-:,-

/::J.---

Date

CJ fe ~ ~ I °I I)if

\"· //

,

LUND STUDIES IN ENGLISH 68 EDl1DRS

CLAES SCHAAR AND JAN SVARTVIK

\

// ,,/ /

QUESTIONS AND RESPONSES In English Conversation

BY

ANNA-BRITA STENSTROM x

I

CWKGLEERUP

LUND STUDIES IN ENGLISH Editors: Claes Schaar and Jan Svartvik Publishers: Uber Forlag Malmo

\

CWK Gleerup is the imprint for the scientific and scholarly publications of Uber Forlag Malmo Orders for publications in Lund Studies in English should be addressed to any international booksellers or directly to the publishers: Libcr Forlag ·· S-20510 Malmo, Sweden

SPROGLABORATORIET AARHUS UNIVERSITET Bygning 327, Ndr. n;nggri.de 80CO /\ !r~ r;

© Anna-Brita Stenstrom Wallin & Dalholm Boktryckeri AB, Lund 1984

ISBN 91-40-05042-4

\,,

/

/

V

Contents

INTRODUCTION 1 1. 1 Question, Response, and Follow-up 1 1. 2 Aim and method 3 4 l. 3 The corpus 6 1. 4 Dialogue types 1 . 5 Classification of texts based on Q purpose

8

2

CHARACTERISTICS OF CONVERSATION 11 2.1 Conversation defined 11 2.2 Turn-taking 11 2.2.1 The turn 12 2.2.2 Backchannels 13 2.2.3 Speaker shifts 15 2.3 Conversational cues 15 2.3.1 The interactive aspect 17 2.2.2 The communicative aspect 19 2.4 Cooperation 21

3

CHARACTERISTICS OF Q, R, AND F 22 3.1 Pairs in dialogue 22 3.2 How Q and Rare identified 24 3.2.1 Problems of definition 24 3.2.2 Context of situation and common ground 3.2.3 Features of Q 28 Intonation 28 Syntax 31 Lexis 32 3. 2. 4 Features of R 33 3. 2. 5 Predictability 34 3.2.6 Cohesion and coherence 35 3.3 How Q functions 38 3.3.1 Discourse actions 38 3.3.2 The single/multiple aspect 41 3.3.3 Constraints on questioning 44 3.3.4 Elicitative force 45

25

I

,. VI

3.3.5

3.4

How R 3.4.1 3. 4. 2

3.5

llow F

Conduciveness 47 Definitions 48 Conducive devices 48 Variants of conducive Q 53 Responses to conducive Qs 56 functions 57 R sets 58 Comments on R 66 functions 67

4

MODEL OF ANALYSIS 71 4. 1 Acts 73 4. 2 Moves 83 4.3 Exchanges 86 4.4 Sequences 89 4. 5 Transactions 90

5

TRANSACTIONS AND SEQUENCES 94 5.1 Mixed and unmixed transactions 94 5.2 Transaction boundaries 95 5.3 Transactions in a subcorpus 100 5.4 Comments on transactions and sequences

6

\

109

EXCHANGES 111 6.1 Exchange structure. 111 6.1.1 The eliciting·part of the exchange structure 113 6.1.2 The responding part of the exchange structure 119 6.2 Linkage 121 6.3 Exchange openings 122 6.3.1 Opening strategies 124 Initial openings · 125 Non-initial openings 134 6.3.2 Initiators 138 Previous approaches 139 Definitions of initiators 140 [a:m] and well compared 141 Distribution of initiators 143 Comments on initiators 146 6.4 Comments on exchanges 148

/

/

VII

7

8

THE Q MOVE 149 7.1 Superordinate moves 151 7 .1.1 Primary acts 151 155 and 156 157 Alternati':e Qs (ALT Qs) 158 Multiple function 159 7.1.2 Secondary acts 161 161 166 170 Comments on secondary acts 171 7.2 Subordinate moves 172 7.2.1 [Checking] 173 7.2.2 [Re-opening] 175 7.3 Summing up the Q move 176 THE R MOVE 179 8.1 Primary acts 179 8.1.1 R to 185 Choice of WH-word 186 The speech situation 189 190

8. 1. 2

8.1.3

191

192 193 194 R to 195 195 Yes/no Rs 197 201 202 205 206 R to alternative Qs 207 Choice of terms 207 212 213

I

l VIII

R to 213 and 217 220 , , ~nd 8.1.5 R to 223 Secondary acts: elaboration 229 8.2.1 What elaborations do 234 Summing up the R move 235 8.1.4

8.2

8.3 9

TIIIJ F MOVTI

9.1 9.2 9.3 9.4

10

11

221

23 7

Option 237 The [Follow-up] 239 The [Re-opener] 240 Summing up the F move 241 9. 4.1 Distribution 241 9.4.2 Pitch direction 245

Q/R STRATEGIES IN DIFFERENT TYPES OF DIALOGUE 248 10.1 General characteristics 248 10.2 Multi-party dialogue 249 10.3 Telephone dialogue 251· 10.3.1 Openings 252 10.3.2 Message 254 10.3.3 Closings 254 10.3.4 Comments on telephone dialogue 255 10.4 Court dialogue· 256 10.4.1 A civil case 257 10.5 Interview dialogue 259 10.5.1 A political radio interview 259 CONCLUSION APPENDICES REFERENCES INDEX 289

262 271 284

\

""

//

J?reface

The availability of a large corpus of transcribed, recorded speech, The London-Lund Corpus of Spoken English, and ac.cess to the original tape-recordings at the Survey of English Usage, University College London, made it possible to carry out this study. A source of inspiration was Kay Wikberg's study of questions and answers in Shakespeare. In the first place, I wish to express my thanks to Jan Svartvik not only for supervising my work but also for many excellent suggestions of improvement. I also wish to thank Randolph Quirk for kindly giving me access to the tape-recorded material whenever needed, David Brazil for his generous advice, Bengt Altenberg for his accurate and very helpful criticism, Muriel SpaldingLarsson for her witty remarks when correcting my English, Oscar Forsheden for his careful proof-reading, and my daughter Heleena for drawing the figures. I am especially indebted to Ami Gayle. Without her friendly tolerance and skilful typing my manuscript would never have been transformed into a book. Finally, I wish to thank my large family for coping with my sometimes complete absorption in scholarly work. Lund, February 1984 Anna-Brita Stenstrom

I

/ /

/

/

/

·'/

1 Introduction r· 1.1

QUESTION, RESPONSE, AND FOLLOW-UP

Primarily, que',stions are used as requests for information. Questioner A asks for information that he does not already possess but believes that addressee B possesses and is willing to impart. Questions seeking information are generally perceived as real and genuine unlike, for instance, rhetorical questions, which do not require an answer, and classroom questions, to which the teacher already knows the answer and only asks in order to check whether ·the pupils know. In conversation, however, questions have various other uses, often with the main function of keeping the conversation going; they may, for instance, introduce a new topic or link up with a topic already dealt with; they may ask for clarification or reflect B's understanding of information to be confirmed by A; or they may serve a merely phatic purpose. In order to cover all these and other uses of questions in casual conversation, as well as in more formal and/or rule-bound dialogue types·, questions and responses need to be considered in a very broad sense. A preliminary definition might read as follows: Given that the speakers A and B cooperate, a question (Q) is any utterance by A that may elicit a response (R) from B; R is consequently an utterance elicited by Q. The close interdependence between Q and R leads inevitably tl a circular definition; it seems indisputable that Q depends on a potential R for its definition and that R presupposes a Q. A definitive definition will be presented in section 3.2, where the identification of Q and R will be discussed in detail. It will be made clear that 'elicit' is a loaded term, which does not involve command function, and that not just anything that B says , immediately after Q qualifies as an 'appropriate' R. Although the definition permits non-verbal.as well as verbal Rs, the study.will be restricted to linguistic Q/R sequences which, in principle, excludes non-verbal utterances.

*

2

I have adopted an interactive/pragmatic approach, attempting to describe utterances on the basis of what they DO in discourse. Consequently, the form of the utterance is not immediately taken as a criterion of its function. Q and Rare dealt with in terms of moves and exchanges, the exchange being the minimal sequence of discourse interaction. A typical Q/R exchange is illustrated in (1) (where the prosodic notation is simplified compared to ,subsequent illustrations; for explanations of prosodic symbols and other typographical conventions, see Appendix III): (1)

A: B: A:

Q how did you get down to WINDLEBURY R how oh by TRAIN F by TRAIN YES

5.9:1190

This exchange consists of a request for information (the Q move), a direct informative answer (the R move), and a confirmative Follow-up (the F move). Exchanges consisting of Q, R, and F occur frequently in casual ~onversation; Q/R exchanges will therefore be describ~d minimally as QR (F), with the F move as an optional constituent. Exchanges form different patterns in different situations and as a result of indivi~ual speaker strategies. The identification of Q is often complicated by a clash between form and function; what looks like a Q may not function as one in a particular context, while an utterance which does not have an interrogative form may perfectly well function as Q, as is illustrated in the following genuine, but simplified examples: V

(2)

B: A:

(3)

B: A: B: A:

and he was in quite a bad MOOD because he was VERY UNWiLLING to Gb when it came fo the P5INT then WHY does he ACCEPT the assignment I MEAN this is the THiNG 1 .6:728 this YEAR YES this YEAR this is over [5i:] renewal for two YEARS over YES the RENEWAL for two YEARS

.

1.2:146

In (2), A's utterance, although interrogative in form, is not used to ask for information but to object; it means something like he shouldn't have accepted the assignment. In. (3), a declarative utterance with falling intonation acts as a Q asking for a confirm-· ing R. \"" Q, R, and F will be described at three different levels: interactive, pragmatic, and formal. It will be seen that utterances /

3

differ in their degree of 'questionness'; there is a scale from 'most Q-like' to 'least Q-like', which is reflected not only in the type of R move but also in the presence or absence of an F move. 1.2

AIM AND METHOD I

The purpose of the thesis ii to describe spoken interaction as it occurs in Q/R sequences in genuine English talk. The strategic and formal realizations of Q, R, and F will be described, and the extent to which syntactic, lexical, and prosodic features correlate with certain exchange types will be demonstrated. My point of departure is WHAT the speaker intends to achieve at a particular point in the discourse. HOW he does it is my second concern. The study is based on a corpus of spoken English (specified in section 1.3) consisting of twenty-five 'conversations' with two or more speakers and includes both face-to-face and telephone ' talk, and, for the sake of comparison, a radio interview and a courtroom examination. The reason why I include a variety of conversational types instead of simply keeping to two-party face-toface dialogue is that I want to give an overall picture of Q/R sequences in verbal interaction by pointing out the main similarities and differences in the various categories of talk. An adequate description of utterances in conversation requires an appropriately designed model of analysis, not least to avoid subjective or arbitrary judgements. For this purpose I adopted the model of Sinclair & Coulthard (1975). Although designed to handle classroom interaction, this model was found to be applicable, with certain alterations and elaborations, also to my data. After all, the basic pattern of classroom interaction, 'Initiation-Response-Follow-up' is similar to that of Q/R exchanges in conversation. The model of analysis is hierarchical, which means that a unit at one rank is realized and describable by units 1 at the rank just below. This type of model has considerable advantages. Not only does it make clear how conversation is built up, the way smaller items form part of larger items and finally form a complete whole; it also makes it fairly easy to add categories when the need arises. The present modifications of Sinclair & Coulthard's original .

./~

4

model have been prompted mainly by the subsequent work of the Birmingham group, as described in Coulthard, Montgomery & Brazil (1981) but also by what has been achieved by other scholars investigating verbal interaction, especially Goffman (1971, 1974, 1976, 1978), Labov & Fanshel (1977), Schegloff (1972a, 1972b, 1978, 1979a, 1982), and Schegloff & Sacks (1973). A detailed a,ccount of the coding scheme will be given in chapter 4. Using a model alone is not enough, however; it presupposes that the analyst is familiar with the context of situation, ie 'all relevant factors in the environment, social conventions, and the shared experience of the participants' (Sinclair & Coulthard 1975:28). The more of this background knowledge he possesses, the more adequate his interpretation is bound to be. However, for an analyst approaching spoken ~ata 'after the fact', and without access to, for example, video-recordings, a classification providing a water~tight description of conversation is difficult to achieve; unforeseen and ambiguous utterances are bound to occur. Qs have already been widely discussed in the literature: by philosophers and logicians, psycholinguists and sociolinguists, grammarians and phoneticians. Rs have received considerably less attention, whereas pairs of Q and R have, for instance, been dealt with in terms of 'adjacency pairs' in natural conversation (cf section 3.1) and as Q/R systems in. man-machine communication. Since my purpose is to describe Q and R in terms of conversational moves from a pragmatic and functional point of view, I shall limit my discussion of previous work to the field of discourse (mainly in chapters 2 and 3) before entering on my own analysis of Q, R, and Fin the corpus. 1.3

THE CORPUS

The Q/R corpus which has yielded the data for the present study consists of 25 conversations from the London-Lund Corpus of Spoken English (LLC), which is the computerized version of the corpus of spoken English collected at the Survey of English Usage (SEU), University College London. The SEU corpus consists of surreptitious and non-surreptitious tape-recordings of monologues \ and dialogues representing various speech situations rangi~g from prepared oration, as represented for instance by political speeches, to spontaneous conversation, such as casual chats be/

5

/

•/~

tween friends.The dialogues are either face-to-face or telephone conversations. Each conversation or 'text' consists of approximately 5 OOO running words and is analysed in terms of tone units. The auditory analysis of the texts has resulted in a detailed orthographic transcription with a full description of prosodic and paralinguistic characteristici (see Crystal & Quirk 1964 and Crystal 1969 for details). Symbols used in this book are listed in Appendix III. The speakers are identified as to sex, age, and occupation or education. The relation between the speakers is described in terms of intimate/distant on the personal level and equal/nonequal on the level of social and professional status. Occasionally, the information is more precise and indicates for example that the speakers are mother and daughter or husband and wife (cf Appendix II). Information about the speech situation is usually not given, with the exception of texts belonging to the category of spontaneous oration where there are such general indications as 'horse racing', 'case in court', and 'dinner speech'. (For a more detailed description of the SEU corpus see Svartvik & Quirk 1980.) The London-Lund Corpus (LLC) differs from the SEU version in that only the basic prosodic distinctions have been retained, whereas indications of tempo, loudness, modifications in voice quality, voice qualifiers, and voice qualifications have been omitted (see Svartvik et al 1982). Although I have used the LLC version of the material, I have also had access to the original SEU slips for more details, and on several occasions I have consulted the actual tape-recordings at the Survey of English Usage. My main concern is the study of Q/R sequences in genuine faceto-face interaction, and consequently most of the texts represent this category. They will be compared with Q/R sequences in telephone dialogues and with questioning and responding in radio interviews and courtroom sessions (see chapter 10). It should be observed that each of the telephone 'texts' consists of a number of short conversations, varying from nine to sixteen in this selection. For the sake of comparison I have included a number of multi-party conversations (see Appendix I); there is evidence to show that the structure of Q/R sequences differs a great deal when more than two speakers are involved.

6

Since it is generally assumed that the speakers' knowledge of being recorded affects their speech production, the investigation covers surreptitious and non-surreptitious as well as partly surreptitious recordings, where at least one of the speakers is aware of the microphone in his role of prompter with the task of ~eeping the conversation going. Prompters were marked by lower case' letI ters in the original SEU version- (a, b, c, etc); in this book, all speakers arc indicated by capital letters (A, B, C, etc). Note that utterances produced by non-surreptitious speakers have not been marked for prosody (refers only to text categories 1-3). In the large majority of the conversations, the speakers know each other well and have a similar social or professional status. This appears from the labels 'intimate' and 'equal' in Appendix I. They are either personal. friends, close colleagues, or related by kinship. Since most of the speakers are linked to the academic world in one way or another, the topics of conversation are very often related to university matters. Quite a number of texts are labelled 'academic'. The term 'general' stands for anything outside the academic world which cannot be labelled either 'domestic', referring to domestic life, or 'specific', referring to a very specific topic; 'mixed' is used when both academic and other matters are discussed. The telephone conve-rsations, which consist of a number of short talks, are unspecified for topic. Judging by Appendix I, there seems to be a predominance of malemale conversations in the corpus, but it turns out that the telephone conversations, which are not specified for 'sex'' and 'topic', contain more female than male speakers. Although much remains to be done in the investigation of differences between male and female language, I shall not be concerned with potential sex differences in the domain of Q and R in this study. 1.4

DIALOGUE TYPES

The different types of dialogue in the Q/R corpus are shown in Fig 1:1. One disctinctive feature of telephone conversation is that there are generally only two participants, one speaker and on€\ listener, whereas the number of people in face-to-face conversation varies from two to many. A second and more distinctive feature is that

~

7

telephone

casual

DIALOGUE

two-party {

-[ :::::/n terv iew

face-to-face

7.

[ multi-party

Fig 1:1

5

11

-{

court examination TOTAL

25

Dialogue types

face-to-face interactants can always see each other, while telephone speakers cannot. This, in turn, ought to imply that speakers should be more explicit on the telephone, since facial expressions and gestures, which accompany speech and even replace speech, play a very important role in face-to-face interaction; a head movement can for instance. substitute for a yes or a no, and a widening of the eyes may e'xpress astonishment. I shall not deal with body language at all, however, since this corpus is purely linguistically analysed. Speaker switch occurs in the same way in telephone and face-toface conversation, but there is an obvious difference in the distrib~tion of the opening and closing turns. Telephone conversation is strictly rule-bound in this respect; 'the distribution rule for first utterance' (cf Schegloff 1972a) says that the answerer speaks first. A second rule says that he also speaks last, since it is normally the caller who initiates the closing of the conversation (see further section 10.3). Different restrictions regulate the speaker's behaviour in face-to-face interaction, eg restrictions depending on a particular speech-situation, the relationship between the speakers, etc. The interview situation, finally, particularly as observed in the courtroom text, obeys the strictest rules and dbes not generally permit any random turn-taking whatsoever (section 10.4). Although the number of interactants in face-to-face conversation varies, there is an upper limit to how many people can take part in a conversation .. It has been noticed, for instance, that even four-party conversations tend to develop into two two-party conversations. In this connection, Speier (1972) emphasizes that the conversational rules differ depending on whether the interactants are two or many. He remarks that 'There is, of course, a

8

set of contingent conditions bearing upon these numerical fea-\ turcs that emanates from the nature of the occasion and the setting, ic two-party talk is more structurally conducive in a busscat or a love-scat than at a large round table in a conference room (1972:399). Pinally, whereas the turn-taking in two-party conversation is regular and follows the J\BAB pattern, sequencing in multi-party dialogue is unpredictable and forms irregular patterns. The different types of dialogue introduced in general terms above will be dealt with in detail on the basis of corpus data in chapter 10. 1.5

CLASSIFICATION OF TEXTS BASED ON Q PURPOSE

Q/R sequences may be considered in relation to their primary function in terms of purpose in a certain text or a certain context cif situation. This function varies with the questioner's more or less conscious intention, so that Qs asked when two friends are chatting do not have the same ultimate aim as Qs in a courtroom; nor does a radio interviewer ask Qs for the same purpose as a professor interviewing a student for admission to college. In the first case the Q/R procedure seems to constitute an end in itself, with no further aim than keeping the conversation going and being sociable, whereas, in the second and third cases, the questioning is directed towards a specific goal which lies beyond R, namely to acquire the relevant information for the actions of the jury or admissions committee. The aim of the radio interviewer is to inform or entertain an audience, whereas the college professor tries to find out whether or not his interviewee is fit for academic studies. The ways Qs work with respect to aim form specific patterns along the following lines: A= first speaker [Q]; B = second speaker [R]; C = third person [outside the conversation]:

(a)

social interaction

(b)

information seeking ( confirmation seeking

"''

//

9

(c)

(b) (c)

(d)

(e)

(f)

Q~

A-action { inform C entertain C

(d)

Q~~

(e)

Q~~

C-action

(f)

Q~R-

B-action

Fig 1:2 (a)

~

goal directed

Aims of Qin speech

The closed ellipsis signifies that Q and Rare exchanged for a purely social purpose; talk .itself is in focus. The curved arrow indicates that the questioner requires information or confirmation with no further intention. The questioner has a further action in mind, indicated by the arrow curving forwards, but this action depends on R. His underlying intention is understood as something like: tell me X and I will act accordingly. Just as in (c) R is not an end in itself in this type of exchange. The speaker asks Qin order to achieve acertain effect. This is what characterizes Q/R exchanges in front of an audience, eg a radio programme, which may either be meant to inform or just to entertain. In this case the goal consists in getting a third person C to perform an action, but the action depends on the fulfilment of R. An appropriate example would be Q/R exchanges in a courtroom, when counsel asks Qs of the accused in order for the jury to form an opinion about the crime. This is an illustration of a straightforward request for action, such as when a caller asks the operator to connect him with somebody. It is an open request for action insofar as the speaker's intention is apparent, as opposed to the case in (e) above.

A tentative application of the system of patterns suggested above to the conversations in the corpus is reflected in Table 1: 1 (numbers refer to texts). The telephone conversations cannot be specified. The reason is that the short calls that each conversation consists of are so different that they cannot be grouped together under one heading,

I

10

Table 1:1

Relation between text and Q-purpose

SOCIAL INTERACTION

(b) INFORMATION/ CONFIRMATION SEEKING

(c) AACTION

(d) INFORM/ ENTERTAIN C

1. 3

1. 1

1. 8

6.3

1. 5

1. 2

1. 6 4. 1

1. 4 3.2

3.3 3.4

4.2 4.3 4.5 5.8 5.9 6.2

5. 1 0

(a)

\ (e)

c-

(f)

B-

ACTION

ACTION

11. 1

most telephone conversations

although the majority, belong under (f). (For information about speakers and topics, see Appendix II.) Table 1:1 is of course not intended to represent a final classification of the texts in the corpus but rather to be a rough description of what goes on in a conversation in terms of Q and R. It seems possible· to traie certain general tendencies although there may be various purposes for questioning in one and the same conversation. Some of the texts fall more or less automatically under one heading, however. The questioner's intention is fairly clear in texts 1.8, 3.3, and 3.4; in text 1~8 the questioner's aim is to have her personal opinion confirmed in order to facilitate the choice of pictures, and in texts 3.3 and 3.4 a university administrator requires information in order to improve conditions for students and staff. The general purpose of the Qs in the texts under (b) is less uniform, but on the whole, the majority of the Qs, which concern academic matters, seem to be asked solely for the purpose of getting information. Characteristically enough, most of the texts fall naturally under (a), where Qs have a social purpose in the first place.

/

/

11

2 Characteristics of Conversation

2.1

CONVERSATION DEFINED

The concept of 'conversation' is used in two different ways in this study. On the one hand, a conversation is equivalent to a 'text', and on the other, it stands for verbal interaction in general. A text has been delimited to contain approximately 5 OOO words, with the result that the beginning and end boundaries of a conversation are arbitrary. In the face-to-face category particularly, conversations seldom have a specific opening and closing section. In the telephone category, by contrast, where one text generally consists of a number of calls, the natural boundaries are often intact; a call begins with an initial hello and ends with a closing good-bye. Crystal & Davy (1975:86) define conversation in the general sense as 'any stretch of continuous speech between two or more people within audible range of each other who have the mutual intention to communicate, and bounded by the separation of all participants for an extended period'. This definition is broad enough to cover all the various types of talk represented in the present corpus: face-to-face as well as telephone conversation, two- as well as multi-party dialogue, although courtroom talk and radio interviewing are admittedly different in character from the rest. I

2. 2

TURN-TAKING

What basically distinguishes dialogue from monologue is what has been referred to as the 'chaining principle' (cf eg Good 1979), which implies that conversational partners take turns. A normal string in two-party conversation is ABAB: (1)

A: B: A: B:

/are we 'going to 'get a 'TELEVISION# R I /don't see 'any NE:CESSITY# to /get one NOW# Q /WHY# R well if /I'm going to WORK# and /go 'back to COLLEGE# 4.1:533

Q

.

12

The fundamental principles of interaction in terms of turn-taking were accentuated by Schegloff (1972a) and have been further elaborated, especially by Sacks, Schegloff & Jefferson (1974) i~. their model for the organization of turn-taking in conversation. 2.2.1

The turn

Strictly speaking, 'turn' is a technical term used to describe the segmentation of conversation into each speaker's continuous talk, but it may also be regarded as working at a pragmatic level of speech. On principle, scholars seem to agree on what makes a turn, although definitions vary slightly. According to Schegloff (1972a: 376), a turn is an utterance that 'may contain anything from a single mm (or less) to a string of complex sentences', not necessarily identical with any syntactic or grammatical unit or combination of units. Earlier, Fries (1952:23) defined an 'utterance unit' as a chunk of talk marked off by a shift of utterance boundaries. Goffman (1953:165) talks of 'natural message' in terms of the 'sign behaviour of a sender·during the whole period of time through which a focus of attention is continuously directed at him'. Markel (1975:190) defines a turn in terms of 'solo talking'. Stephen & Mishler (1952:600), finally, take a somewhat different approach and regard pauses as utterance boundaries. I take it that they regard an 'utterance' as equivalent to a turn. But a turn is not just a ·certain amount of talk. From a pragmatic point of view one should distinguish 'what is said' from 'what is done' (see eg Labov, 1972:121). This is what Schegloff seems to have in mind when he says that 'it ·is crucial to distinguish a single turn in which two activities are accomplished from two turns by the same party without an intervening turn of the other' (1972a:376). (2) is an illustration: (2). A: B:

she /won't be fed 'up with a 'chance with !DAVE# /WILL she# (laughs -) - - no /Mi.Uy 1,Jas :SCREAMING# Uke /MAD# 4. 1: 45 as we /LEFT# • /didn't you HEAR#

The first part of the addressee's turn responds to the previous Q, whereas the second part, didn't you hear constitutes another Q. /'

/

13

The meaning of an utterance and its function in relation to other utterances in the discourse have been adequately captured by Sinclair & Coulthard (1975) in their analysis of classroom interaction. They describe utterances in terms of 'moves', defined as the smallest free discourse units. According to their model, what is said in one turn may constitute one or more moves, which, in their turn, are realized by 'acts' (see further chapter 4) •

2.2.2

{ r I

Backchannels

What constitutes a speaker's continuous ·talk, or turn, does not include so-called 'backchannel' items, a term introduced by Yngve (1970). Such items are emitted by the listener as feedback signals and should not be mixed up with turn-claiming signals. Backchannels vary considerably in length, from short vocalizations like mm to fairly long expressions, such as I think you're right. There is no general agreement on the function of backchannels. According to Duncan (1974), so-called 'auditor backchannel signals' permit the listener to participate in the interaction by indicating whether he has received or understood the speaker's message during the course of the turn, Crystal & Davy (1975) argue that the primary function of backchannels is to reflect the listener's attention to what the speaker is saying. Good (1977) looks upon what he ca.lls 'informationally minimal' Ms and yeahs as realizations of the 'parity principle', which means that he sees them as confirming the listener's role as equal partner rather than just confirming his attention. Schegloff (1982:81) uses the term 'continuer' for items such as uhuh by which the I hearer claims understanding and/or attention but whose primary function is to indicate that he understands that the current speaker intends to continue speaking - by passing on the opportunity of taking the turn. And Faerch & Kasper (1982:74), referring to Edmondson (1977) and Edmondson & House (1981), define 'uptaking' in much the same way, in terms of 'receipt' and 'goon' signals. Another detailed description of backchannels is found in Coulthard, Montgomery & Brazil (1981:24-25). They employ the term 'supporting act', which has three subcategories forming a 'sernan-

14

tic cline': the 'acknowledge' indicates that B hears and accepts the message and expects A to go on speaking (eg yeah, uhuh, mm), the 'accept' implies minimal understanding of what is accepted (og okey, oh I aee), and the 'endorse' supports the, point made by the current speaker (eg you 're quite right ) . (3) is an example of a supporting act: (3)

B:

A:

(B:

"/well ( . coughs) YES# /well he :doesn't :MENTION a'bout this at ALL# h9 /just SAYS that# . there'll be /twenty-'five 'students 'in a CLASS# / which is 'rather* 'NICE# . / not too BIG#*,. * /pretty GOOD#* so I'd /be RE:SPONSIBLE# for /fifty 'students in ALL 'summed UP# 6.2:412

Note that pretty good is uttered within speaker B's turn and not after, which would have been the appropriate place. In this connection, Duncan (1974) has observed that certain speaker-listener interaction is going on at a lower level than that of the turn, the 'within-turi interaction level', and he suggests that an early production of backchanneis in a speaker turn indicates that the listener is not only following the speaker's message but that he is ahead of it. Although backchannels are not directly part of my study, since they do not constitute Rs, I shall have to deal with them to some extent in connection with Follow-up moves and Re-openers (see sections 9.2 and· 9.3). Backchannels occur in Supporting moves, the most characteristic features of which can be summed up as follows (for Supports in the London-Lund Corpus, see Orestrom 1983:1071 08) : I

• • • • • •

they are not turn-claiming they encourage the current speaker to go on speaking they signal the hearer's active participation they express agreement, acceptance, and understanding they can occur almost anywhere in the current speaker's turn they have a random distribution in a conversation and are both speaker and situation specific

15

2.2.3

Speaker-shifts

Speaker-shifts are mostly smooth and unproblematic procedures, as was illustrated in (1) and (2). Speaker A uses certain strategies to show when speaker Bis required or invited to take over, and speaker B may signal through verbal and non-verbal Supports that he is willing or ready to take the turn. Different techniques used, 'turn-allocation techniques', have been described by Sacks, Schegloff & Jefferson (1974) who distinguish between the 'current' speaker selects next' technique and 'self-selection'. Moreover, they state that speaker-shifts usually occur at possible completion points, which are projectable before they occur, so-called 'transition-relevance-places' (TRPs). Speaker change does not always occur without disturbances, however. In general, it seems that the more spontaneous and intimate a conversation is, the more often trouble arises. Orestrom (1983: 141) describes what he calls 'unsmooth' speaker-shift, including cut-off speech, overlapping speech, parallel speech, and simultaneous talking. One example from.the present data is (4): ( 4)

B: A: B: (A: B:

did you /know that !DORNHOPE# was /master of !HUSTLEFORD# /YES# ((and)) /did you "!HEAR# that +he [?]+ +((he))+ got ++/MARRIED# to to /Miss ( - - laughs) ++"MARRIED to# "/Mary MISTLE'TON# ( - laughs) /EMl# ++ 5.9:121

This small sample illustrates how cut-off and parallel speech cooccur in one short extract. Turn-length varies considerably. This may be due to the speaker's individual characteristics; it may have to do with speaker relations in terms of status and degree of intimacy, it may be the consequence of particular speech situations, and so on. 2.3

CONVERSATIONAL CUES

Speakers use various devices to claim, hold, and yield the turn. Some of these are used strategically in the ongoing planning and production process and act as 'fillers' for time and 'planners'. Others can be considered from a strictly technical point of view

16

as 'interactive cues', regulating the turn-taking system, while at the same time acting as 'communicative cues', reflecting thi:\ speaker's attitudes and intentions. These t~o functions are so closely related that they should rather be seen as two aspects of the same function, or expressed somewhat differently: the interactive function is the effect of the speaker's communicative intent. Verbal fillers have been discussed by Brown, who states that their main function is to 'fill the silence and maintain the speaker's right to speak while he organizes what he wants to say' (1977:109). And she adds that they contribute very little information. Brown includes such items as well, I think, obviously, and actually. 'Planners', like fillers, make it possible for the speaker to stop and think before he goes on to formulate what he intends to say. The difference between fillers and planners is that planners can partly be described in syntactic terms, whereas fillers are typically outside the syntactic structure of the utterance. The difference is illustrated in (5), (excerpt from extended corpus): (5)

.

'

and /I /I get ! really [am] - - ((you/know)) when ! [?] when I'm 'trying to COOK# and /people come and CHAT# 2.7:69 I /I get !terribly put OFF# -

Here I suggest that really serves as a planner while you know serves as a filler. Really points forward to a head later on in the utterance, being replaced in the modifying position by the intensifier terribly. The exact function of you know is difficult to determine, but judging by ,,the long pause, it. is obvious that the speaker has not ·yet made up his mind how to continue. Other terms used for typical discourse items are 'hedges', eg: sort of, kind of, really, 'softeners' for eg: you see, you know, I mean, and tags, and I suggest the'term 'initiators' for eg: anyway, however, now, and well. The term 'hedge' was introduced by G Lakoff (1982:195) for words 'whose job is to make things fuzzier or less fuzzy'. A similar description is given in Brown & Levinson (1978:150ff) who include such items as rather, I suppose, I wonder, and tags, divided into 'strengtheners' and 'weakeners'. My list of hedges is much more restricted. Th~ term 'softener' was introduced by Crystal &\D,vy for items which not only 'alter the stylistic force' of an utterance but

/

/

17

also 'maintain the continuity of discourse' (1975:91). Quirk et al (1972:778) employ the term 'comment clause', and Edmondson (1981:155) uses the term 'cajoler' for items through which 'the speaker seeks to increase the likelihood that the communicative act being performed will be acceptable to the hearer'. Initiators oc~ur in utterance-initial position where they perform specific duties, such as resuming a narrative after a short digression, pointing .forward and introducing new information or a new topic, or simply linking one part of the discourse to what preceded (see further section 6.3.2). 2.3.1

The interactive aspect

Interactive cues may be inherent in a specific structure or utterance type, or realized by strategically used lexical and/or prosodic markers. Some utterances have the inherent property of eliciting or inviting an R, which makes them 'automatic' turn-yielders. Here, I am not only referring to Qs with varying 'elicitative force' (see section 3.3.4) but also to 'adjacency-pairs', ie pairs consisting of a first pair part and a conditionally relevant second pair part (cf Schegloff & Sacks 1973 and section 3.1). But even a plain statement may expect some reaction: (6)

A: B:

I met a FRIEND of yours last night DID you

.

A non-reaction on the part of B would have been noticed and might have caused trouble for the continuation of the conversation. Sacks, Schegloff & Jefferson (1974) emphasize that the grammatical structure of a turn serves as ari important turn-yielding device. The next speaker generally takes over or tries to take over the turn at ' "possible completion points" of sentences, clauses, phrases, and one-word constructions, and multiples thereof'. Duncan (1973:37) suggests 'the display of at least one of a set of six clues', syntactic, prosodic and paralinguistic, as potential turn signals. OrestrHm (1982) shows that five factors correlate with turn-taking, namely prosodic, syntactic and semantic completion, loudness reduction and pause. He calls the point where these factors converge a 'hit', defined as a possible. completion of a turn. Edmondson points out that turn-yielding signals do not predict speaker shifts: 'that a turn is "avail-

I

18

able", given various clues, may be said to be "known" to a hearer: whether or not he takes the available turn however is a mat~er of subjective choice' (1981:41). What a particular item does in a discourse is related to its position. A good example of this is you know, which can occur in initial, medial, and final position in the utterance: (7a) (7b) (7c)

you know, that chap phoned me again last night that chap, you -know, phoned me again last night that chap phoned me again last night, you know

You know in (7a) is turn-taking, indicating that the speaker has something to tell, thereby arousing the hearer's interest; in (7b) it is turn-holding in its capacity of delaying the message, and in (7c) it is turn-yielding, signalling that the message is delivered and that the hearer's reaction is welcome. Sacks, Schegloff & Jefferson (1974) exemplify what they characterize as heavily used devices for turn-entry and turn-exit by 'appositional' beginnings such as well, but, and, and so, which serve as 'pre-staits', and tag-questions which serve as 'postcompleters'. Keller (1979) found that overt, ie verbal, turn-taking signals appear to be rare in informal discourse with only two parties involved, where the intention to take the turn is usually signalled non-verbally. He_ lists a number of conversational strategy signals, so-called 'gambits·,, consisting of semi-fixed expressions that are generally used as introductory and terminating devices. Wait a seaond is given as an example of a turn-taking signal, while that's about it is an example of a turn-yielding signal. Occasionally, he points out, gambits share the same purpose as fillers, in the sense of fillers for time, and therefore also serve as turn-holders. The turn-holding function is ~arried out by different strategic devi~es which tend to cluster in unplanned discourse, especially at the beginning, or near the beginning, of an utterance, where the speaker has not yet worked out how to put his message into words. · (For a similar observation, see Brown, 19 7 7: 1 24) . (8) is a good illustration (excerpt from extended corpus): (8)

and /ALL this was DONE [a:]# - -/ by - - kind of letting - [a:] - - . /{WELL} REALLY by 'just [a,:] 'sort of [a]# -'/starting from !NOTHING# 2.3:115

""

// /' /

19

The following devices cooccur: • • • • •

hesitators, realized by [a:] long and short pauses (- .) hedges, such as kind of, sort of and really initiators, such as well reformulation: by kind of letting reformulated into by just s~rt of starting from nothing

The use of such devices reflects the speaker's ongoing planning procedure. Not only do they give him more time; they also prevent the hearer from taking the turn before the message is delivered. Brown (1977) includes repetitions among verbal fillers with a turn-holding effect, as in her example no+ not really+ no+ definitely not. As regards the turn-yielding function, Jefferson (1972:329) places you know among the standard completion signals, and Sacks, Schegloff & Jefferson (1974) state that it has the same function as a tag question and thus belongs to the category of 'postcompleters'. Crystal & Davy (1975:94) also compare the effect of you know in final position to that of a tag question, stating that if uttered in a high pitc~ range with a high rising tone it invites the listener to agree or at least expresses the speaker's assumption that the message has been understood. Another turnyielder suggested by Crystal & Davy is but er, as in their example so I said I'd see him on Tuesday but er ... ,which indicates that the speaker has nothing more to say and invites the hearer to take over (1975:102). Finally, pauses and rising pitch in general may act as turnyielders. The speaker may for instance make a pause longer than usual, thus giving the hearer a clear indication to take over, / or he may end an utterance with a rising pitch to indicate that the hearer is expected to respond. 2.3.2

The communicative aspect

Leaving aside the more technical aspect, what messages do the discourse items referred to in 2.3.1 convey? Let us first consider the turn-initial position (whe~e hedges and tags do not occur): (9)

A: B:

weZZ /can you 'not get CLOSER# /yes I :COULD have DONE#

4.1:81

20

(10) (11)

B: A:

well /what are we 'doing 'this WEEKEND# ,,, well I've /nothing DOWN anyway {at /ALL#}#

B:

I /meant to 'Joint it 1auT when I /READ it# to you# /WELL# it /wasn't a 'bad SUG!!GESTION# /REALLY# 4.1:801

A:

4.1:14

In (9), well links Q to a previous utterance in the dialogue. The meaning of well is something like OK but ... showing that the information is received but that speaker A has a better idea. In (10), well is equivalent to now in the same position. It points forward to something new, and there is no apparent linking to what preceded. The second well in (10) serves as an R-prefix, and as such characteristically introduces an indirect R, ie an R which is in some way insufficient; A cannot come up with a good suggestion, but instead of saying so straight out he lets B understand his meaning implicitly. Well in (11) introduces an evaluation and seems to have a strengthening effect. Note the cooccurrenc~ with the evaluater really. They both reflect the speaker's attitude to what was said in the previous'.utter~nce. (For a detailed analysis of well in conversation see Svartvik 1980.) Extract (12) illustrates fillers in medial position: 1

(12)

/this is the funny "thi,ng A! !BOUT 'aca'demics# that/ [?] if you're [no ?a:] you /KN6w# I /I've COME to it# /so LATE# , ((I .mean)) /I've had a !LIFETIME of EXPERIENCE# /rolling A:ROUND# 1.6:346

The extract speaks for itself; it is obvious that the speaker has some difficQlty in getting to the point. Items that typically occur as communicative, interpersonal cues in final position are you know, you see,and tags, as in (13) (from extended corpus): (13)

B:

A:

/there was !trouble A:FOOT# /you KN6w# /YEAH# - . /trouble A!LEG ANYWAY#

8.4:663

You know and other softeners in this position do not only soften the effect of a statement, they also appeal to the hearer to agree. Generally the rising tone is taken to indicate that the speaker is not completely certain of what he is stating but invites the hearer to confirm. There is a slight difference in meaning between you know and you see. Whereas you knoi,,.,gives the im-

21

pression that A presupposes that Bis already familiar with what he proposes, you see rather signals that the information is new (cf R to , section 8.1.5). Interestingly, if items which may serve as initiators and hedges occur in utterance-final position, they resume their syntactic function, as it were, and should rather be classified as adverbials. 2. 4

COOPERATION

According to Grice (1975), conversational partners are expected to preserve a general 'cooperative principle', expressed in his by now well-known maxims of Quantity, Quality, Relation, and Manner. In reality, however, ordinary successful conversation only partly works according to his principle. Speakers commit all sorts of violations and still manage to get their message across. Ostman (1981 :37) goes as far as suggesting that the communicative aspect of language is acquired not by following but by violating the Gricean maxims. Numerous characteristics of informal, relaxed conversation are listed by Crystal & Davy (1975): speakers get all mixed up and forget what they wanted to say; they make grammatical mistakes, use words vaguely, are interrupted, switch speech styles, manipulate the rules of the language, and they argue erratically and illogically, etc. Curiously enough, despite such irregularities, conversation may be judged successful; the people involved are not prevented from cooperating. Interruptions, which may result in reformulations and repetitions, may for instance be signs of cooperation and not the opposite. The extent to which disorderly features of this kind are acceptable, ie do not result in misunderstandings or a general breakdown of communication, depends in the first place on the amount of shared knowledge (see section 3.2.2) that the parties possess, and in the second, on the degree of tolerance between them. Certain things are taken for granted in conversation, and the more intimately the speakers know each other, the greater their shared knowledge; the speaker is understood without being explicit. Leech (1977:9) says that 'the principle of systematically conveying more or less than one "says" is continually and dynamically in operation in human interaction ... '.

22

3 Characteristics of Q, R, and F

3.1

PAIRS IN DIALOGUE

Turn-taking generally r'esul ts in a meaningful discourse; whatever one speaker says acts as a stimulus for the other participant(s) to· react, and the reaction is usually coherently related to the preceding utterance. This, according to Grice (1975), is a reflection of the speakers' willingness to cooperate. Sinclair (1980) proposes in this respect that discourse organization is 'prospective', which implies that one speaker's utterance not only anticipates a response from the other party, but also sets up expectations as to what type of R is appropriate; a question anticipates an answer, a statement an acknowledgement, etc. Re·sponses, on their part, are 'retrospective' and constrained by the nature of the initiating move.· Goffman (1976) argues that the dialogue format has to be extended in order to cover a whole range of utterance pairs and introduces such general terms as 'statement' for the initiating element and 'reply' for the responding element. But he goes further, maintaining.that th~ term 'statement' may not be adequate, since one Q can be answered by another Q. He therefore suggests the term 'reference' for all the things we can respond to and proposes that 'our basic model for talk perhaps ought not to be dialogic couplets and their chaining, but rather a sequence of response moves with each in the series carving out its own reference, ... ' (1976:293). Although this might be what dialogue comes down to in .the end, I find th·at a model based on this principle is not very illuminating until it has been considerably elaborated. The way a simple dialogue is structured has been described by, among others, Longacre (1976) and Halliday & Hasan (1976). Longacre discusses form and function in terms of surface and deep structure. In his model, a simple dialogue paragraph begins with an initiating surface structure utterance which encodes three units of deep , structure: question, proposal, and remark. A dialogue can be initiated in three different ways: 'by asking for / ',,

/

23

information, by proposing an action, or by submitting an observation. It concludes with a resolving utterance which encodes three units of deep structure: answer, response, and evaluation. In addition, there may be a terminating utterance which encodes two deep structure units: acquiescence and rejection (cf 1976: 169-170). The three levels may be represented in diagram form, as in Fig 3:1: INITIATING UTTERANCE Remark Question Proposal

Answer

RESOLVING UTTERANCE Response Evaluation

TERMINATING UTTERANCE Rejection Acquiescence Fig 3:1

Dialogue structure

The diagram makes clear that what Longacre suggests is a threepart interchange consisting of Initiation-Resolution-Termination, which is similar to the three-part exchange proposed by Sinclair & Coulthard (1975) and by Mishler (1975) (cf section 3.5), Halliday & Hasan describe two~part exchanges from the point of view of cohesion, manifested by the particular syntactic, lexical, and prosodic cohesion existing between an utterance made by a second speaker and an immediately preceding utterance made by a first speaker. Cohesively tied second utterances are called 'rejoinders', of which there are two main types: those following Q, 'responses' and, those following non-Q, 'other rejoinders' (for an overall survey of rejoinders, see 1976:207). Two- and three-part models of exchanges in terms of QR (F) will be discussed in detail in chapter 5 and following in connection with the analysis of my data and will not be dealt with any further at this pdint. Q and R const.itute a standard sequence in dialogue and have been considered the typical example of an 'adjacency pair' ever since the term was introduced by Sacks (1967). Adjacency pairs have the following characteristics: they consist of a 'first pair part' and a 'second pair part' produced by different speakers, with pair part one establishing a 'conditional relevance' on what follows in the next slot; consequently, a Q is not an-

24

swered until that slot is appropriately filled, According to the definition, not just any second utterance will do, only an appropriate one; a greeting is, for instance, followed by a greeting and an apology by a 'smooth-over'. As will become apparent, the\ notion 'adjacency pair' is relevant to my definition of Q and R. 3. 2 3.2.1

IIOW Q AND R ARE INDENTIFIED Problems of definition

Q and R can only be identified by what they do in the discourse - pragmatically and interactively. Therefore, I analyse utterances as Q and R by going from functidn to form, by asking such questions as: What does the speaker say?, Where does he say it?, and How does he say it? Goffman (1976) points to two crucial problems in defining Q and R. One is the problem of circularity, which I have already mentioned. If R depends for its definition on whether th~ preceding utterance was a Q or not, it follows that a Q is determined by reference to the sequence it establishes. The other way round would be to take the occurrence ~f Ras a criterion for Q: since there is an R, a Q must have preceded. Bolinger regards 'interrogative distribution' as one of four potential criteria for Q (cf section 3.2.3): 'Since most Qs elicit answers, an answer usually indicates that a Q has preceded' (1957:2-3), and so does Churchill (1978) who uses Bolinger's criteria for Q. Similarly, Brown et al (1980) identify '·what they thought was a Q' as a Q, if their informants responded to it. And Gunter (1974) finally, argues that there is no projecting things forward to R but the relevance of R depends on its antecedents. The second problem consists in knowing whether the utterance following Q was meant to be an answer or not. And this is where the analyst falls short; even if he overheard the conversation, he would not be able to read the ·minds of the speakers involved. The analyst has to keep to what he hears (and reads in the case of trans~ripts). Goffman poses the question: •.:. how could one show that what followed a particular question was in no way an answer to it?' (1976:293). I have defined Q as AN UTTERANCE THAT MAY ELICIT AN R, which implies that not all Qs are followed by an R. Consequently, Q cann9t be recognized merely by the presence of R. This may seem

/ /

/

25

a bold definition in the light of the above quotation from Goffman and involves a certain amount of subjectivity on the part of the analyst, but I think it makes sense, considering the data I have investigated. Although I basically agree with Goffman when he says that the initiator will want to know that his message has been taken in and that the recipient will want to make known that he got the message, I am not going to deal with R in the b~oad sense of 'reply to statement' but will confine myself to trying to single· . ' out what is characteristic of 'responses to questions'. As I see it, there are two main criteria for R; one is the placement of the utterance in a sequence of utterances produced by different speakers, and the second is appropriateness in relation to 'here-and-now', which involves 'context of situation' and, 'common ground' (see section 3.2.2). With Q defined as an utterance that may elicit an R, R is bound to be THE UTTERANCE ELICITED BY Q. But this does not signify that any utterance following Q qualifies as R. The term ELICIT does not simply mean 'cause the addressee to say something'; what he says should be coherently related to Q. In other words, the utterance following Q is an APPROPRIATE R only if this second condition is fulfilled. This is the type of R I am going to deal with. 3.2.2

Context of situation and common ground

What an utterance does at a certain point in the discourse should be seen as the effect of a ~ariety of correlating factors, linguistic as well as extralinguistic. The linguistic context, sometimes referred to as the 'co-text', is what immediately precedes and follows an utterance and helps to determine its meaning. This definition needs no further comment. The extralinguistic context is far more complex. Clark & Carlson (1981:7), give the following definition: context is information that is available to a particular person for interaction with a particular process on a particular occasion They further distinguish between 'intrinsic' context, which includes a person's knowledge of objects, facts, circumstances, etc, and his expectations related to the actual situation, and 'incidental' context, which includes thoughts and feelings not immediately connected with the situation but which might affect

26

understanding. The -intrinsic context is the 'common ground' that the hearer believes holds at that moment between speaker and hearer, defined as: the mutual knowledge, mutual beliefs, and mutual suppositions that exist between two people, Coulthard & Brazil (1979:24) define shared knowledge by 'what knowledge speakers (think they) share about the world, about each other's experience, attitudes and emotions'. Lyons (1977:572) regards 'all the factors which, by virtue of their influence upon the ~articipants in the language-event, systematically determine the form, the ap_Propriateness of the meaning of utterances' as 'contextual' factors. Six factors are listed under the heading of 'communicative competence': 1) role and status, 2) space and time, 3) degree of formality, 4) medium, 5) appropriateness to subject-matter, and 6) province and domain. Labov & Fanshel (1977:73) employ the term 'shared knowledge' for the particular knowledge that is shared by participants in a conversation and which is crucial for the formation and understanding of their rules of discourse. These rules are set up in order to bridge the gap between what is said and what is done and presuppose that the participants make certain assumptions about each other's mutual knowledge, known as: A-events: B-events: AB-events: 0-events: D-events:

known known known known known

to to to to to

A, but not to B B, but not to A both A and B everyone present be disputable

The identification of an utterance as an A-event or a B-event proved to be decisive for my distinction between 'informative statements and requests for confirmation (see section 3.3.4). Let us consider the importance of context and common ground for a plausible interpretation of interrogatives by looking at two extracts from the corpus: (1) concerning the purchase of a house, (2) referring to the removal of furniture: (1)

B:

A:

and /ANYWAY# for /God's SAKE# /how {QUICKLY} am :I LIKELY# to have "/two ahiZdren ((of)) the 'age of FIVE# - - I'm A/MAZED# at the /price of houses !OUT HERE#

4.2:508

/

27

(2)

A: B: A: B: (A: B:

but there /is a PROBLEM# */WHAT#* *un */LOADING# ** /how** **[a]** do I DO it# well /I shall 'just have to :RING you# and /you' 11 have to COME# 4.2:819

Both extracts contain an interrogative initiated by a WH-word, but only in (2) does the interrogative function as a Q. Why is (2) understood as a Q and not (1)? In what way does the context contribute to the understanding? What contextual clues are there? At least two factors speak against the interpretation of (1) as a Q. One has to do with the area of common ground, and the other with the linguistic context. First, it is obvious to both speakers that you do not get five-year-olds in no time. Consequently, B's utterance is not a Q to be answered but either an exclamation, an expression of mistrust, or a statement of facts. Second, the initiating anyway for God's sa~e is not likely to introduce an ordinary Q. Note that A continues without paying attention to what B said. in (2), which is clearly a The interrogative how do I do it request for information in this context, might have functioned as a rhetorical Qin a different context, meaning something like I .aan 't possib7,y do it. In this example, already the introductory but there is a prob1,em, which belongs to the linguistic context, points to a following Q, and moreover, both interlocutors are aware that help is required and that A needs information on how to get that help. But even greater demands are made on knowledge about the context when one is faced with potential Qs in declarative form with no lexical or intonational clues: (3)

C: A: C: A:

I take

I /don't 'think there's enough !MONEY in the department {to /SPEND#}#. ( - b·reathes) this is /money 'for the 'buying of BOOKS# YES# /YEAH# /YEAH# 3. 3: 404 this is money for the buying of books

as equivalent to

28

d'you mean money for the buying of books, interpreting it as a request for confirmation on the basis of the preceding context, where the shortage of books in the library is being discussed. The analyst can never know exactly what the speakers meant by what they said, but the more he learns about the speakers involved and their common ground, the situation, and other extralinguistic factors, the more likely he is to approximate to a 'correct' interpretation.

3.2.3

Features of Q

INTONATION In the earlier literature it is often assumed that there exists a typical rising interrogative intonation, and it fs also suggested that rising intonation may turn a statement in declarative form into a Q. Recent investigations indicate, however, that there is no such thing as a typical interrogative intonation in the sense of 'Q-determining'. According to Bolinger (1957:{-S), Q is a behaviour pattern which cannot be satisfactorily defined; it is fundamentally an 'attitude', which is emphasized by the importance of intonation and gesture for determining what is and what is not a Q. Bolinger lists four ingredients, any of which·'MAY serve to define a given Q' (1957:2), adding that no Q will lack all of them, namely 1) interrogative distribution, 2) interrogative syntax, 3) interrogative intonation, and 4) interrogative gesture. One of the criteria of Qin Quirk et al (1972:386) is·rising 'question intonati.on', as in their example: You wi"lZ speak to the BOSS? The subordinate role of intonation has,been emphasized by eg Crystal, who states that the intonation contour alone cannot decide the meaning.of an utterance but that the 'interpretation of an intonation contour can depend on the recognition of the relevance of other prosodic or paralinguistic features' (1969:283). He maintains that reference must be made to grammar and lexis. Referring to O'Connor & Arnold (1961), he points out that one may find that a 'meaning' of an utterance is 'attributed to an intonation pattern which in fact derives largely or wholly froi:n the attitudes implicit in the vocabulary of the utterance' (1969:284). Crystal emphasizes that 'so far as the evidence goes, the many

29

/

/

assertions concerning the r~s~ng intonation as the usual mark of yes-no questions in English have not been based on any adequate body of quantitative information' (1969:4). (For· results based on the present data, see section 7.3.) Halliday (1970:22-23) is of the opinion that each sentence has a 'most likely' intonation, which s~ould be used 'except for good reason'. A change of intonation from the most likely pattern is a change of meaning. He discusses falling and rising intonation contours in relation to known and unknown polarity; rising pitch, which means 'polarity unknown', is the neutral (or normal) tone for yes/no-Qs and indicates uncertainty, whereas falling tone indicates certainty with regard to yes and no. Falling tone, on the other hand, is the neutral tone for WH-Qs where there is no question of polarity. O'Connor & Arnold (1973:46) maintain that 'no sentence type always requires the use of one and only one t~ne group', but they regard some tone groups as 'normal' for fertain structures. Yes/ no-Qs are, for example, normally said with the · 1 1ow Bounce'. They try to describe the attitudes conveyed by the different tone groups (or intonation contours) in connection with various sentence types, but stress the assumption that part of the speaker's meaning is carried by the structure of the sentence, ie words and word order. Brown argues that it is not clear whether the pitch contours on the tonic syllable contribute to the meaning of the utterance and concludes: 'The most we can claim for the "meaning" of one tone group as against another is that it indicates whether a tone group is the final tone group in a sentence or not' (1977:102103). Furthermore, she does not find that high rise is typical of yes/no-Qs, as is generally claimed. This is rarely so in her data, and she declares that statements implying that 'a rising intonation can turn a statement into a Q' are most unfortunate. What rising ,intonation, and especially high rise, does is demand a response (ie not necessarily an answer to a Q, my remark). Brown et al (1980) found no consistent intonational clue to what is or is not a Q. But the context had a disambiguating effect in most cases. Surprisingly, it proved harder to decide what was and what was not a Q for certain utterances in context, which, /the authors explain, was due to the fact that the notion Q had not been clearly defined.

30

A somewhat different approach has been presented by Brazil (1975, 1978). Brazil's main ideas were summed up in C,oulthard (1977) and discussed in relation to exchange structure by Coulthard & Brazil (1979) (see also Coulthard, Montgomery & Brazil 1981). In this approach, relative pitch level, and not pitch direction, is regarded as decisive of utterance function. Coulthard & Brazil point out that the approach offers a solution to the problem of distinguishing Qs from non-Qs when it comes to utterances that are declarative in form or moodless~ namely going by the 'termination' of the utterance, ie the pitch choice made at the tonic syllable. And since it is te·rmination and not tone choice which carries the eliciting function of a move, there is no need to 'draw on assumptions about speaker's and hearer's knowledge of A-events and B-events' (1979:36). Coulthard & Brazil suggest that all moves set up expectations about what will follow and they have noticed 'a remarkable tendency for concord between the termination choice of the final tone unit of one mov~ and the initial key choice of the next' (1979: 32) (for key, see eg Brazil, 1975:9ff). This, they argue, indicates that the speaker predicts or asks for a particular key choice (implying a particular meaning) from the hearer. There ~re three choices: high, mid, and low. By choosing high or mid termination the speaker constrains a particular kind of R, whereas low termination is seen as a sequence closer (cf Brazil, 1978:24) and sets up no constraints. High termination is typical of Qs that expect a yes/no contrastive R, whereas mid termination expects agreement. In addition, there is the selection of tones: referring tone and proclaiming tone. Referring or end-rising tone marks what is said as common ground: proclaiming or end-falling tone marks the matter as new. I find ihe approach just outlined most interesting, and although I have not been able to adopt it for my data other than marginally, since the corpus is not consistently marked for high, mid, and low pitch, it has helped me to clarify a fair number of tricky cases. I agree with those who claim that pitch direction plays a subordinate role in the recognition of Q-function, but I would not go as far as to say that the significance of tone contou~s should be disregarded. On the contrary, the choice of falling or rising

31

tone reflects the speaker's attitude to what he is saying and may therefore affect the elicitative force of the utterance (see section 3.3.4). Moreover, there are cases where pitch direction alone distinguishes Q-function from other functions, as will be seen in the next subsection. SYNTAX My definition of Q implies that a pbtential R-eliciting utterance can take any form except the imperative; it is hard to imagine the imperative form not having command function, and 'Rs' following commands are not elicited but ordered. Let us consider the following isolated and prosodically unmarked examples: (4a) (4b) (4c)

what's your name your name tell me your name

(4a) and (4c) are unambiguously a Q and a command, respectively. But what about (4b)? There is nothing in its syntactic form that reveals its function. If tone is added,;on the other hand, the function becomes clearer: (5)

your NAME your NAME

In isolation, what determines the function is the direction of · the pitch: your NAME will be interpreted as a Qin the same way as (4a), whereas your NAME will undoubtedly be taken as a command. In context, by contrast, this is not necessarily so. Interrogative form, just like interrogative intonation (cf section 3.2.3), is often regarded as one of a set of possible criteria of Q-function, as in Bolinger (1957) and Churchill (1978). And Quirk et al (1972:386) mention 'the placing of the operator in front of the subject' and 'the initial positioning of an interrogative or WH-element'. Anthony (1974) points to a number of problems of definition; there is for example a considerable degree of syntactic and functional overlap·; negative interrogatives and imperatives can be realized by identical surface form, such as in Don't you drink any brandy, which can either function as a command or· as a Q (syntactic overlap), Who cares has a similar function to that generally associated with exclamations, whereas wiii you pass me the book bears a close _functional similarity to imperatives

32

(functional overlap). Confronted with the task of defining Q, Anthony states: 'A definition which attempts to cover utterances as syntactically and functionally disparate as those which we intuitively label questions necessarily reduces itself to near-vacuity' (1974:6). He concludes that the recognition of Q very often depends on nonlinguistic information. Lyons (1977) is of a similar opinion. He declares that the grammatical structure of a sentence has to be distinguished from the type of communicative act that is performed in a particular situation since declarative sentences can function as Qs, and interrogative sentences may serve as commands. Brazil (1975:3) goes a little further and states that 'In real, situated speech, it is the discourse function which determines the form the utterance takes'. And Coulthard (1977) points out that the linguistic form of utterances is almost irrelevant in discourse; what is structurally important is the linguistic function. Since I adopt a basically pragmatic/interactive approach, I agree entirely with those who do not trust form for the identification of utterance function. ks will appear, however, my subcategories of Qare described in syntactic ter~s (cf eg section 7.1.1). I wish to emphasize that the formal description is an end result, based on a functional approach, and not a starting-point. l

LEXIS The speaker's questioning strategy can involve the use of particular lexical items as Q-markers. Not much has been said on this topic in the literature, but Crystal (1969:284) has emphasized the important role of lexical Q-markers in determining the semantics of intonation, as have O'Connor & ~rnold (1973:.46). To my knowledge, specific lexical items functioning as Q-markers have only been discussed in any detail by Bolinger, in terms of 'imputations' and 'tentations'. Both types occur in utterances that are generally regarded as non-Qs as far as syntactic structure is concerned. Imputations may be realized by you say it's too late? or it's too late, you say? for instance, where the speaker 'ascribes words or opinions to someone else', and tentations by eg no doubt it's true? or I assume it's too late?, where the speaker 'employs an expression denoting hypothesis or\assuran~e' (1957:10).

33

Extracts from the data will make clear that lexical cues play an important role for an utterance functioning as Q, especially as regards utterances in declarative form. This matter will be dealt with in sections 7.1.1 and 7.3. Crystal & Davy (1975:103) touch upon the questioning effect of certain phrases when dealing with so-called 'softening connectives'. Some softeners, they sugges;-, especially you see and you know and also I suppoie and if you like, assume a literal meaning in utterance-final position, 'almost like a question-tag' (cf also Bolinger, 1957:61-62). 3.2.4

Features of R

If position alone were decisive, any next utterance after Q would do perfectly as R, whether or not it is adequate in other respects. A Q may, for instance, be followed by another Q and constitute a conscious and deliberate R strategy on B's part. Even a pause may serve as an R under certain coriditions and would be most revealing for example in a courtroom. Surprisingly, Halliday & Hasan (1976:206) make the following statement: 'Not all questions have an answer; but no less significant is the fact that not all answers have a question', although they define 'answer' as that type of direct response that follows Q; with that definition there can be no answer without a preceding Q as far as I can understand. Generally, Rs are considered to occur in connection with Q, and some scholars, among them Bolinger (1957), Gunter (1974), Chur·chill (1978), and Brown et al (1980) use the occurrence of Ras a criterion of Q. Sinclair & Coulthard (1975) describe responding moves in general as reciprocally related to soliciting moves, which consist of Qs, commands, imperatives, and requests. Rs following Qs, or elicits, are called replies, the function of which is to 'provide a linguistic response which is appropriate to the elicitation' (1975:42). In their extensive study of answers to WH-Qs, Robinson & Rackstraw (1972) describe Ras a verbal utterance as well as an act of behaviour. They define answers as that category of R which responds to Q, and questioning as an act of behaviour which is explicitly related to the acquisition of knowledge. In order to separate answers from non-answers they use a ·set of behavioural, formal, and contextual criteria: an answer should follow an inter-

34

rogatively posed Q, be given by the receiver of Q and be expressed in 'language'; as to form, it must consist of at least one declarative clause and have lexical continuity with Q; contextually, it should convey a statement, not consist of a refusal to answer, and function within th~ same referential category as Q. Pranck (1979) deals with what she calls 'linguistic acts' in terms of 'conversational moves' which are '.steps in the process of interaction'. One such move is R which gets its status as a result of a preceding Q move; whatever the addressee does when a Q is posed will be interpreted as a reaction to Q. She states that the function of moves is determined both by the conversational context and by the utterance itself, with the illocutionary force of the utterance as a potential contributing factor. But Rs, she says, do not generally coincide with a particular speech act class and are therefore only a conversational move type, although illocutionary and conversational qualities are intricately connected. Linguistic features of R will be dealt with in section 3.4.2. 3.2.5

Predictability

In accordance with the principles of adjacency pairs, the first pair part not only predicts the occurrence of the next pair part but also selects the next action, since only an appropriate second part can follow. Gunter (1974), among others, is strongly opposed to this way of reasoning. Although he basically believes in the cause-effect principle, he argues that its power to predict what will happen next in a dialogue is nothing but a powerful illusion. What we do when relating utterances to each other is to look backwards, since 'We are not wise enough to see how one speech can elicit another; bµt we do have the faculty, once the two speeches are made, to relate the second back to the first' (1974:87). Coulthard & Brazil (1980:100), on the other hand, claim that there are structure frames in dialogue which set up certain predictions, but the predictive power is not so strong that it can prevent a next speaker from coming up with a totally irrelevant next utterance. However, because of the predictive pow~r of the structure frame, a speaker who asks a Q will not treat the addressee's contribution as a non-R until he fails to see its relevanc~. And Sinclair (1980:114) finds that, even though liter0

/.

T

35

ally anything aan happen in a discourse, it is illuminating to regard each turn as 'a predictive classification of all possible next turns', but, he continues, 'each turn, although it must fit into the classification which has been devised for it, can choose which class it will realize - and the classification includes such options as challenging thJ predictions of the previous turn, and reopening after a conclusion' (1980:115). Utterances which do not fit any of the options prescribed will terminate the exchange by initiating a new exchange. The fact that anything can occur after an introductory move is also emphasized by Franck (1979), who describes all possible next moves as belonging to a 'set of continuation options', imposed by the initiating move. A Q for instance, is not necessarily followed by an answer but might also be followed by silence or by the introduction of a new topic. These are all possible reactions to Q, ie continuation options. Some continuations are, however, more socially acceptable than others - giving an answer is more acceptable than a refusal to answer. Franck therefore suggests that there exists a hierarchy of preference for continuation options, consisting of three levels: 1) acceptance or positive confirmation, 2) refusal or doubt, and 3) indeterminate, postponing reactions like check-ups. (See also Levinson 1983:307ff.) Sinclair & Coulthard (1975:133) suggest the following options for utterances ending in an open elicitation: 1) a minimal, totally fitting response, 2) something which satisfies the notional presuppositions of the elicitation but is structurally independent, 3) something which implies an adequate answer but principally sets up further presuppositions, and 4) something which challenges the terms of the question. The different sets of options that are available after a Q has been posed will be deali with under the heading R Sets in section 3.4.1. 3.2.6

Cohesion and coherence

The close relationship between Q and R is manifested in more than one way, through grammatical and lexical cohesion, but also through prosodic, semantic, and pragmatic agreement. As mentioned earlier (section 3.2.3) Coulthard & Brazil (1980) maintain that there exists a certain pitch concord between Q and R in that a Q ending in high or mid termination puts constraints

36

on what follows. High termination asks for a yes/flo contrastive R, whereas mid termination asks for a mid key agreement. If the predicted pitch concord is disregarded by the addressee, R is regarded as non-compliant behaviour. Similar observations are made by Brown et al (1980). It might be mentioned in this connection that, in American English, Pope (1976) has observed that when it is obvious what R is expected, as with tag Qs, an R that agrees with the expected R has usually 'mild' (slight fall) intonation, whereas a disagreeing R has 'sharp' (steep fall) intonation. Generally, cohesion and coherence are treated as different manifestations of the relationship between utterances in discourse, one of which does not necessarily presuppose the other. Widdowson (1978:31), for instance, describes cohesion as the 'overt, linguistically-signalled relationship between propositions' and coherence as 'the relatioriship,between illocutionary acts'. If cohesion is the formal relation between utterances, co~erence may then be characterized as the logical and pragmatic counterpart. Labov & Fanshel (1977), on the other hand, treat formal connections, such as ellipsis, under the common heading of coherence. Halliday & Hasan (1976:206ff) describe the cohesive relationship between·sentences in general in terms of grammar and lexis. When it comes to Q/R sequences, they state that there is a special type of cohesive relation which has its own characteristic grammatical properties. A direct R realized by a simple yes/no-R, as in: Are you coming? - Yes (for Yes, I am or Yes, I'm coming), acquires a cohesive effect through ellipsis; yes expresses polarity on its own, whereas the remainder.of -the clause is presupposed. Similarly, ·a direct R to a WH-Q in its simplest form, as in How much does it cost? - Five pounds, gives the fnformation that the questioner asked for, ie the gap is filled, while the rest is presupposed by ellipsis. Grammatical cohesion exists also between an indirect Rand its Q, as in Did you get the appZication form? - It's on my desk!, where R gives information other than that which is asked for; but the answer can be implied from this supplementary information (cf , section 8\J). This type of R presupposes the whole Q. Robinson & Rackstraw (1972) use a similar argument when describing lexical continuity; no R is an answer, they say, unless there is·lexical continuity between Rand Q. 'This ·continuity may be

/

37

explicit or presupposed, as in the case ot answers like Yes to confirmation/denial questions in which the whole lexical content of the question is presupposed' (1972:23). They further emphasize the grammatical and lexical impact on R which is caused by the social situation, suggesting, for example, that 'examination answers' will keep closer to the lexical collocates of Q than 'conversational answers', which in turn will probably consist of simpler and looser sentences with more pronominal references, sin~e the answer is expected to fall into line with Q. Discussing responses to requests for information, Labov & Fanshel argue that, in most cases, the grammatical form of Risso closely connected to the grammatical form of Q that there is no difficulty in recognizing the discourse rules that are in operation. Thus, in a case like Are you going home? - Yes, it is possible to expand R to supply what is/missing by recourse to the regular rules of ellipsis. In other cases, formal links are missing, and the form of R must be located ii an underlying proposition by a certain rule (see 1977:99) to make R coherent with Q. Their example is: (6) A: Are you going to class today? B: It's such a beautiful day We perceive this as a coherent sequence due to the underlying proposition which is known to both A and B; eg that B usually does not bother to attend classes when he would rather go swimming. Thus, it is perfectly possible for Q and R to be coherently related without cohesive ties. To what extent a discourse is perceived as coherent is related to the shared knowledge of the speakers. Coulthard & Brazil claim that 'The absence of a deterministic relationship between form and function makes it possible for virtually any rejoinder to have coherence given the shared background of understanding of the participants' (1979:16). Conversely, Widdowson (1978) points out that, if we are not familiar enough with a certain topic, we may not see the discourse as coherent in spite of cohesive links, since we do not recognize the illocutionary significance of ~he relationship. Gunter (1974) strongly emphasizes that we are not looking forward but backward when establishing coherence and that we do not predict but rather create coherence 'after the fact'. Thus coherence is a uniquely anaphoric relation, but cohesion may also

38

work in the opposite direction, as appears in the following example: ( 7)

A:

B:

What are the boys doing? They are swimming

There is an anaphoric, r&ferential link that connects they with boyo, but doing points forward and anticipates the verb for which it acts as a pro-verb and which answers Q. Similarly, there is a cataphoric referential link connecting the WH-word of a WH-Q with that part of R that provides the missing item, as in: (8)

A: B:

Where were you? I was at Covent Garden

where there is a link from where to Covent Garden. 3.3

HOW Q FUNCTIONS

A Qin discourse is not ,an object per se, recognizable on its own merits. It is, first and foremost, an interactive element which can only be interpreted by what it does 'here and now' in a particular conversation. Therefore, it is not possible to deal with Qs only in terms of illocutionary acts, which, using the formulation of Widdowson (1979:138), ari· 'essentially social activities which relate to the world outside the discourse ... '· The concept of 'interaction' is crucial, both from an organizational and a pragmatic point of view. 3.3.1

Discourse actions

Both Austin (1962) and Searle (1~69) regard illocutionary acts partly as products of context, but neither of them develops this notion. Austin began by isolating what he called 'performa.tives', ie such utterances where, in saying the words we also perform an action, from 'constatives', ie plain statements which are only intended to state or give facts and which, contrary to performatives, can be true or false. Later, he gave up the distinction performative/constative, realizing that all utter~nces are performances of illocutionary acts: 'Once we realize "'that what we . have to study is not the sentence but the issuing of an utterance/"/ in a speech situation, there can h~rdly be any longer a possibifity of not seeing that stating is performing an act' (1962: 1:fo).

39

Austin emphasized A's role, arguing that the happy outcome of the illocutionary force of an utterance lies in a successful realization of A's intention vis a vis B. Searle (1969), on the other hand, stresses B's interpretation of the utterance, analysing speech acts as illocutionary acts on. the basis of a set of universally valid conditions in terms of intentions and expectations on the part of A. He cl~ims that the production of speech acts is governed by 'constitutive rules', which have the effect that an utterance of a certain form will be taken as realizing a particular illocutionary act. An utterance will be taken as a Q under the following conditions (cf 1969:66-67): propositionai content ruie - any proposition or propositional function preparatory ruie - 1) S does not know 'the answer', ie does not know if the proposition is true, or, in the case of the propositional function, does not know the information needed to complete the proposition truly. 2) It is not obvious to both Sand H that H will provide the information at that time without being asked. sincerity ruie - S wants this information essentiai ruie - counts as an attempt to elicit this information from H

It should be added that Qin Searle's terminology is either a 'real question', ie a request for information, or an 'exam question'. Sociolinguists like Labov & Fanshel (1977), ethnomethodologists, especially Sacks, Schegloff & Jefferson (1974), and discourse analysts such as Sinclair & Coulthard (1975) look upon utterances, in the first place, as interactive elements. But whereas Sacks, Schegloff & Jefferson concentrate on the organization of utterances as turns in the system of spoken discourse, Labov & Fanshel and, even more efficiently, Sinclair & Coulthard handle both the interactive and the pragmatic functi'on. Labov & Fanshel define speech acts as 'actions', or 'interactions', which are 'carried out by means of speech' (1977:59). Actions are said to reflect what is really meant, ie what the speaker intended, an·d express the way in which A meant to affect B. But since what is said in communication is not always equivalent ~o what is done, they have formulated a number of 'rules of production and interpretation' to help B to understand what A intended. These rules are set up from both A's and B's point of

40

view and are based, primarily, on the shared knowledge of the interactants. Sinclair & Coulthard, who analyse utterances in terms of moves and acts (see further chapter 4), are interested in finding out not only what utterances mean but even more how they affect the continuation of the interaction; whether they initiate exchanges, constitute responses, or serve as exchange boundaries, etc. They explicitly point out (1975:14) that these acts are not identical with speech acts as defined by Austin (1962) and Searle (1969); the analysis of utterances as speech acts, Sinclair & Coulthard emphasize, is in terms of circumstances, intentions and expectations which must hold universally, whereas discourse acts realize what an utterance or part of an utterance does in one specific discourse. They stress the role of the 'situation' in their analysis by demonstrating that it is possible to predict what type of move, for example, an interrogative or a declarative realizes, by the use of certain interpretation rules which, in ~urn, depend on the answer to a set of questions about the situation (cf 1975: 1

30).

Similarly, Schegloff (1978:84) declares that what a certain form does depends on 'what it is attached to and wh~re that is placed'. He rejects the 'action determining' power, which implies. that the function of·an utterance depends on where it occurs in the sequential organization of· conversation. An utterance which is interactionally a Q asking for information, for example, lays constraints on the next slot in the discourse of a 'Q-A sort', as in: (9)

A: B:

What time is it? It's noon

where the answer giving the information required follows Q. However, another first pair part of an adjacency pair, like: (10)

A: B:

Why don't you come and see me sometime? I ,would like to

does not lay constraints of the same sort, since it does not function as a Q asking for information but as an invitation. My identification of Q-function bears traces of.all the ap- · proaches outlined above but, following Sinclair &C~ulthard (1975), I shall deal with utterances in terms of moves and acts with // ' moves representing what an utterance DOES in the discourse and./

.

/

41

acts telling us what it MEANS (cf Chapter 4). 3.3.2

The single/multiple aspect

In conversation, utterances often constitute more complex sequences than a simple seri~s of Q and R. First, the same utterance can be doing more than one thing from the point of view of discourse organization, such as responding to an immediately preceding Q and then serving as a new Q which is responded to in the ~ext speaker's utterance: (11)

A:

B: A:

Q R/Q R

when are you LEAVING I don't KNOW why do you.ASK I was just CURIOUS

or it may have two functions at the same time, one of which is ellipted and implied: (12)

will you get up at FIVE as usual B: (R)/Q wouid YOU A: R OK see you at EIGHT then

A:

Q

Second, an utterance may have more than one meaning, although, in the context, one of them predominates: a literal or direct meaning and one or more indirect meanings. There are at least two ways of answering the following Q: (13)

A:

Q

B:

R

when are you LEAVING at FIVE 01'

R

OK I'll not BOTHER you any more

Q can either be understood literally, as a request for information (which means that A wants to know when Bis leaving, no more and no less), or it can be understood indirectly, as a request for B to leave. The first of these aspects is briefly referred to in Schegloff (1972b:78) in his discussion of 'insertion sequences' where he remarks that although Q2 in a sequence where Q1 is followed by another Q instead of by R is not analysable as an 'answer', R is still not felt as 'absent'. Merritt (1976) 'discusses instances like (11), so-called 'coupling' where the person who asks the first Q goes on and asks the next Q, what Goffman (1976) calls the 'two-person sociable chain'; and 'elliptical coupling, where

I 42 \

the answer to Ql can be arrived at by implication from Q2, as in (12) (for a detailed discussion of chains in discourse, see section 6.2). McTear (1978) suggests that the best way of describing utterances with two move functions is to 'double-code' them. The double-coding device, he argues, makes it possible for us to mark continuity in discourse explicitly. I agree with McTear and would say that, for my analysis of Q/R exchanges, the double-coding device is absolutely vital (cf also Codlthard & Brazil, 1979:40). The multiple-meaning aspect has been the object of considerable interest. A recurring feature in the studies of speech acts is the setting up of rules for interpreting the illocutionary force. Sinclair & Coulthard (1975:32-33), for instance, demonstrate how it is possible to formulate rules based on responses to certain questions about the situation and predict when declaratives and interrogatives will realize functions other than statements and questions. And Labov & Fanshel (1977:65) stress the importance of rules of production and interpretation, based on the shared know' ledge of the interactants, for understanding indirect speech actions: the more indirect a speaker's mode of interaction is, they say, the more such rules are required. The most detailed set of rules, both for direct and indirect illocutionary acts, has been worked out by Searle (1969, 1975) who sets up the following hypothesis for indirect speech acts (1975:60-61): In indirect speech acts the speaker communicates to the hearer more than he actually says by way uf relying on their mutually shared background information, both linguistic and nonlinguistic, together with the general powers of rationality and inference on the part of the hearer The case he discusses is when A utters a sentence and means what he says but also means another illocution with a different propositional content. Searle restricts himself to directives, such as Can you reaah the saZt?. What A does is to utter a directive by way of asking a question, so that one illocutionary act is performed indirectly by way of performing another. He emphasizes that what is added is not an additional sentence meaning but an additional A meaning. In this connection he pointi,out that some sentences are conventionally used as indirect speech acts. ,,, The questipn is now: How does B understand the indirect or, a~~ /

/

43

Searle prefers to call it, the 'primary' meaning? He suggests a number of steps to derive the primary illocution from the literal act. The crucial step refers to B's inferential strategy, which presupposes that he infers that the primary 'illocutionary point' departs from the literal meaning, as well as what the primary illocut~onary point is. The theoretical appara~ tus which is needed to explain how B can understand the indirect meaning, Searle suggests, includes mutual background information, a theory of speech acts and certain general principles of cooperative conversation (cf 1975:61). Like Searle and others, Churchill (1978) regards Qs as requests and assumes that although all Qs are requests for information, in a technical sense, they are ~sed to carry out various kinds of interactional activities, so that their main purpose may not be to convey information at all but to request some other action 'beyond giving information' (cf 1978:65). He claims that all requests for action or, as he calls them, 'permission Qs', belong to the category of 'specific' 'qs (as opposed to 'general' Qs), which means that they all take yes or no as their literal answerset. He suggests that the general frame for all requests can be specified as Will you do X for me?. Such indirect acts consist of two parts: a first part which addresses B's right to agree or refuse to comply, and a second part which states what act A would like B to perform, eg: (14)

A: B:

Do you have the time? Yes/no or Yes, it's six

B's strategy for understanding and planning responses to indirect speech acts has be~n outlined by Clark (1979), who proposes that the literal meaning of an indirect speech act, such as Do you know the time?, can be intended to be taken seriously along with the in~irect meaning, or proforma, meaning I request you tell me the time. In the first case, Bis expected to respond to both meanings: Yes, I do - It's six, but, in the second case, only to the indirect meaning, It's six. In order to find out how B decides whether an utterance should be taken directly or indi~ectly and how he determines what the indirect meaning should be, Clark studied indirect speech acts and their Rs in a number of experiments which supported his pro-

44

posal that Buses certain properties of indirect acts as sources of information. The sources are of two kinds: one is connected with the linguistic characteristics of the utterance itself and the other with expectations B has built up, based on the circumstances in which the utterance occurred. These Q-properties are matched by a corresponding set of R-properties. Multiplicity of meaning in indirect speech acts corresponds, for instance, to multiplicity of moves in Rs: ,. (15)

Can you teZZ me the time? Meaning (literal) - Move 1: Meaning 2 (indirect) - Move 2:

Yee I aan It's six

The list of meanings and moves can be made longer (see eg Labov & Fanshel, 1977:166-167) .. The normal o~der of moves is identical to the logical order of the corresponding meanings of the indirect speech acts, which implies that Meaning 1 is logically prior to Meaning 2, but, for special purposes, moves can be twisted out of normal order. The logical contingency between Meaning 1 and Meaning 2 has a rational basis. For example, A must assume that he and B have mutual knowledge of certain background facts, and he must observe certain cooperative principles of conversation. Moreover, there is a certain conventionaiity as to what sentences can be used for what indirect speech acts (cf Searle above). Other fac~ors are. politeness and purposefulness. An expected· R to an indirect speech act, Clark.points out, is one that deals, explicitly or implicitly, with 'all the meanings of the act', and the only move that can ever do that by itself is the move that responds to the final meaning in the chain. I

3.3.3

Constraints on questioning

Levinson (1979) maintains that particular types of activity set up particular constraints on the way utterances function; first, they constrain what will count as an allowable contribution to each activity and, second, they help to determine the way an utterance will be taken. For instance, Qs are used and interpreted in different ways in the courtroom and in the classroom. Qs in a cross-examination are goal-directed; their function is to extract such answers from the witness that the jury\ get a 'natural' argument for their verdict. Qs in the classroom are used for a different purpose, to impart knowledge, organize know-

\

45

ledge and impart abilities. They should, however, be taken literally in both places, according to Levinson. Goody (1978), who studied the uses of Qin the West-African Ganja tribe, reports how the role of Qs can be a matter of social constraints. Her main purpose was to find out under what conditions genuine information Qs were possible, and she discovered that they were only possible~among status equals. She discusses two types of constraints on questioning: those related to activitr type and those connected with the social relations between the interlocutors. Questioning in Ganja is strongly attached to command function, which in turn is related to status relations, and has been institutionalized in a number of ways. Qs (especially rhetorical Qs) are used as indirect speech acts, functioning as 'joking challenges', as greetings, and in court cases. They might, however, be posed by a person in an inferior position, a junior asking a senior, but in that case the questioner is placed at a disadvantage since he displays ignorance of the power that lies in knowledge. If the roles of the interlocutors are defined, Goody states, the role expectations will bias the interpretation of any Q asked. 3.3.4

Elicitative force

According to my definition, not all Qs elicit an R. Moreover, I have emphasized that the form of an utterance cannot be relied on since interrogative form need not jndicate Q-function. Qs can take almost any form, from interrogative to plain declarative and even ellipsis. My basic assumption is therefore that the function of an utterance in discourse is determined in the first place by certain preconditions in the contextual situation, eg: that A does not know X, that A believes that B knows X, that A believes that Bis willing to impart X, etc which are preconditions for requests for information, and that the relationship between A and B, their common ground, and the context of situation have to be considered. Syntactic form, intonation, and lexical features are of secondary importance and serve to modify the function of the utterance. But I do not · think that it is possible to interpret what A says without considering B's reaction, or non-reaction, since Q and Rare so closely related.

\

46

Different types of Q have different 'elicitative force', which implies that whether or not R follows is more noticeable with some Qs than with others. If no R follows a request for information, such as What's the time?, the absence will be noticed, but if no R follows a request ~or acknowledgement, eg Thie I think io a ioveiy piature, isn't it?, it would hardly be noticed at all. I suggest that the degree of elicitat~ve force is related to the form of Q, which in turn is related to its specific function, and can be registered by the presence/absence and shape of R (see further section 7.1.1). Let us consider three types of Q: requests for information, requests for confirmation, and requests for acknowledgement (for subcategories of Q, see chapter 4). For the recognition of the first two types, I find that the rules in Labov & Fanshel are very helpful since they are set up from both the speaker's and the listener's standpoint. Their rule for requests for information (with a slight modification, cf 1977:89) reads as follows: If A addre$ses to Ban interrogative focusing on I, and B does not believe that A believes that a. A has I b. B does not have I then A is heard·as making a valid request for information. Their rule for request for confirmation reads (1977:100): If A makes a statement about B-events, then it is heard as a request for confirmation. I suggest the following rule for requests for acknowledgement on the same lines: If A expresses his personal .opinion or makes an assertion about an A-event either of these utterances may be heard as a request for B to acknowledge. Lexical indicators play a fairly important role for the interpretation of an utterance. Expressions like you mean, no doubt, probably, and I suppose indicate that the speaker is not certain and that he wants the hearer's confirmation. This subject will be developed in section 7.1.1. Preliminary conclusions regarding elicitative force in relation to R expectation and~ type are presented in Fig 3:11. / /

/

/

47

3.3.5

Conduciveness

Referring to the previous section, where I argued that the eliciting force of Q is related to its form, I shall now proceed to discuss a similar relation~hip, that of form and conduciveness. I want to emphasize, however, that what determines whether or not a Q is conducive, as well as the degree of its elicitative force, is neither form nor speaker intention on their own but the way B interprets a particular utterance in a particular situation. Although his interpretation does not necessarily have to agree with A's intention, it is likely to be influenced by the form of the utterance. A conducive Q is one where A is taken to show a preference for one R rather than another, as opposed to a straight Q where no such preference is manifested. Compare examples (16) and (17) with (18): (16)

where did JOHN go

(17)

did John go to LONDON or READING

(18)

didn't John go to LONDON

Both (16) and (17) are pure requests for information, with no preference for a particular R since there is no evidence for an underlying assumption about what that R should be. In (18), on the contrary, the negative form indicates A's assumption that John went to London and that A expects an affirmative R. All conducive Qs have in common that they are not pure requests for information but ask for some kind of confirmation of A's assumption. Although conduciveness occurs both in WH-Qs and yes/noQs, it has mainly been dealt with in its manifestation in the latter, where positive conduciveness implies that A expects agreement, and negative conduciveness implies :that he expects disagreement, as in: (19) (20)

did someone SEE him didn't anyone SEE him

Bolinger (1957:157) points to the crucial difference between WHQs and yes/no-Qs in this respect. For yes/no-Qs there is a choice between two possible Rs; it is therefore easy to suggeit an R in advance. But WH-Qs are answerable in a number of ways, which ·makes it difficult to anticipate a particular R. According to Bolinger, the most common interrogative word in WH-Qs is why,

48

which 'puts the hearer on the defensive' (see 1957:160). DIJPINITIONS What I refer to as 'conducive' Qs have been termed ahd defined differently by different linguists: 'conducive' by Bolinger (1957), Hudson (1975) and Bublitz (1980), 'biased, eg by Pope (1976) and positively vs negatively 'oriented' by Quirk et al (1972). According to Bolinger, a conducive Q is 'one that shows that a given answer is expect~d or desired' (1957:97). Bublitz, who finds this definition too vague, in that it does not distinguish between 'expected' and 'desired', defines a conducive Q as one in which A 'expresses his assumption regarding the kind of answer he expects' (1980:2), with speaker-assumption including both original and new assumptions. In the definition of Quirk et al, a Q has positive orientation if it uses assertive form in preference to non-assertive form, and'negative orientation. if it uses non-assertive form in preference to assertive form. Hudson uses a similar definition but states that the conduciveness lies in the 'act of uttering' and not in the sentence itself and, furthermore, that conduciveness ea~ be explained in terms of pragmatic constraints. Thus the conducive aspect is read into Q by B, who hears a sentence of a certain form and interprets what he hears on the basis of what he knows about A (cf Bolinger, 1957:10 and Bublitz, 19~0:1). Pope, finally, regards the bias of r~gular Qs as a 'supposition' which is identical with A's original belief, and not its expected R. I agree with Bolinger when he defines a conducive Q not only as one showing thit a given R is 'expecied; but also that it might be 'desired'. Here I am referring to two sub-types of conducive Qs, Rhetorical Qs and Suggestions, where I find that R has less to do with A's assumptions than with his wishes.

CONDUCIVE DEVICES There are various means available to A by which he can show both that he expects/desires B to respond in a certain way and\ how strong his assumptions/wishe~ are. He can use a particula~.Q~form and/or certain lexical markers, he can employ a specific tone pattern and, perhaps more efficiently, pronounce Qin the negative form~ which is generally a clear hint to B that he give his confirmation.

I

.I I

49

Let us compare the following syntactic realizations of yes/noQs and see how they reflect A's underlying assumption and also how they reflect the strength of his belief: (21)

Did John go to London?

(22) (23)

John went to London, didn't he? John went to London?

Weak assumption least conducive

I

Strong assumption most conducive

If only syntactic form is considered, I take it that inverted Q, which expresses A's weakest assumption, is also the least conducive type. A does not know whether or not the proposition is true, and Bis free to respond yes or no. Still, it cannot be doubted that A assumes that John went to London rather than to another place. Note the difference if he had asked Where did John go? or Did John go to London or Reading?'. The difference i~ terms of conduciveness between simple Qs and alternative Qs has been pointed out by Pope (1976:75): ' ... questions in full alternative form are ALWAYS more nearly or exclusively neutral than single term questions'. Example (22) is more conducive, since it consists of a statement part where A expresses his assumption and a tag part where he invites B's confirmation; (23) represents A's strongest assumption, being nearly an assertion, and is consequently the most conducive Q-form. To indicate whether he expects a positive or a negative R, A often uses verbal clues, both when Q has interrogative form, such as (24) and (25) (cf eg Quirk et al, 1972:388-390) and, especially when Q has declarative form, (26) and (27):

/

(24)

Has Mary gone already?

Positive expectation

(25)

Has Mary gone yet?

Negative e~pectation

(26)

John got something to drink?

Positive expectation

(27)

John got nothing to drink?

Negative expectation

Bolinger refers to declarative Qs with verbal clues like I assume, no doubt, I guess, etc as 'tentative assertions' in which A states something that he wants to have confirmed instead of asking a straightforward Q (see eg 1957:18, 77). He also suggests lexical markers used to make WH-Qs conducive, such as expressions indicating 'undesirability', eg Why bother? and Who believes sueh

50

nonsense? and expressions pointing to an 'absurdity', ,eg Who can know the unknowabte? In examples (21) - (23) I deliberately disregarded the possible effect of intonational contours. With reference to conducive intonation, Bolinger states that although certain intonational contours arc frequent, they are generally not conducive regardless of the components of the verbal chains where they occur. In his 1 opinion, the most significant factors are accent and proportion of , see section 7.1.2).

QZ

speaker B, the wife, forestalls the request, which she has implicitly understood, by asking her husband to do what he supposedly intended her to do.

100

Second alternative analysis: Q1

speaker A really wants to know what his wife is going to do and has no hidden purpose. Q1 is intended and heard as a .

QZ

speaker B suddenly remembers that the trousers will have to be picked up but realizes that she will not be able to do it, so she asks her husband if he can do it.

According to the first analysis, QZ is understood as I know what you intend to ask me but I'm going home.so ean you ... ; Q1 triggers QZ and R1 serves as a justification. According to the second analysis, QZ is understood as · (I just remembered) ean you sinee I'm going home; R1 triggers QZ and is the reason for asking. The analysis suggests that the pre-request (Q1), introduced by the mind, serves as t~e transaction-initial boundary marker. The only safe final boundary marker is the opening of a new transaction following the long pause after the F3 ;R 2 move. Briefly, discourse coherence can be perceived in at least three different ways. Blum-Kulka (fo~thcoming) describes coherence in terms of 'topical cohere·nce', which manifests itself in the relevance of the various parts of a sequence to a given topic, and 'presuppositional coherence', which is related to the topic of the discourse and its implications. To these might be added the type of coherence that is related to the pro- and re~rospectiveness of moves in discourse interaction. 5.3

TRANSACTIONS IN A SUBCORPUS

In order to verify a) to what extent exchanges could be described in terms of transactions in different types of conversation, b) whether transactions were generally mixed or unmixed, and c) how they were generally recognized, I selected seven conversations representing categories that are as different as possible (cf Appendices I and II): Text 11. 1 courtroom examination \ '· Text 6.3 radio interview Text 7. 2 telephone calls Text 3.4 multi-party discussion (distant/equal) Text 3.3 multi-party discussion (distant/disparate)

// ,/

/

101

Text 6.2

two-party face-to-face conversation (intimate/disparate) two-party face-to-face conversation (intimate/equal)

Text 4.1

The texts are in a descending order from most to least planned, with planning seen as a refl~ction of the situation. Text 11.1, I for instance, is constrained by the· procedure in the courtroom, where talk typically consists of series of Q and R, whereas the talk produced by the married couple in text 4.1 obeys no other constra1nts than those imposed by the turn-taking system.

TEXT 11.1:

A

COURTROOM EXAMINATION

Due to the specific conditions in the courtroom, everything in text 11.1 is dealt with in terms of Q and R. The questioning has a specific purpose, that of obtaining sufficient evidence for a final decision. Since the whole proceeding concerns the same issue, 'whether the mother of the defendant knew what she was doing when signing a couple of wills', the Q/R exchanges do not seem to be adequately accounted for as constituents of transactions. All the details connected with the signature are examined within this main topic, such as whether she had been drinking, whether the will was on a book or on a tray, whether she used a magnifying glass, etc. Thus, the examination consists of one long Q/R transaction which can be divided into a series of sequences, each dealing with one aspect of the main topic. The sequences in their turn consist of exchanges where the details of the new aspect are presented in the first exchange and dealt with in detail in the successive exchanges, as illustrated in (4) (Rs are omitted except in the first exchange): (4)

A:

[FRAME] Q

B: (i) (ii)

(iii) (iv) (v)

R

Qq Qq Qq Qq Qq

((NOW#)) - - /did !she - READ# /NORMALLY# or /did 'she re '.quire AS!SISTANCE {in /READING#}# . she /USED a MAGNI'FYING {/GLASS {to /READ#}#}# . /did she have !glasses as WELL# !use her MAGNI I FYING 'glass# and /did she !read /did you it ALOUD to her# the "/whole WILL# in /ONE 'go# or /how many TURNS# - 11.1:66-84

.\ 102

What is new with respect to the preceding discourse is how the will was read. The fact 'she read' is old information, and both ohe and read are anaphorically related to what preceded. (4) is an unmixed Q/R sequence which can be represented as: {PR Q R} {Q R} {Q R} {Q R} {Q R} {Q R}

Since I regard the whole proceeding as one,single transaction, (Frame] and [Focus] are taken as markers of sequence boundaries besides their 'normal' function of transaction-boundary markers. In general, the transition between sequences was marked. In cases where [Frame] and [Focus] were missing the transition was either marked prosodically by a long pause or intonationally by high pitch (indicated by 'H'), as illustrated in (5) - (8): (5)

[FRAME] Q

(6)

[FOCUS]

/WELL# /Captain and Mrs KAY# /lived in a . !FLAT# on their /OWN# 11 .1 :677 weU I'd /'like to 'know what you're ":SAYING# a'bout THIS#

(7)

Q

[?]

R

Q

/NO/# I sa~d /nothing like 'that to my MOTHER# NOTHING# /did you !know 'mother had been DRINKING#

Q

H

F

are you /SAYING#

11.1:331

11.1:141 (8)

/did you KNOW# that the doctor had TOLD# - - , , ,

11.1:617

These markers often occurred in combination. Significantly, [Frame] and/or [Focu~] had the additional function of marking emphasis, for instance when counsel wanted to make something absolutely clear before taking the next step, as in: (9)

[FRAME] [FOCUS] Q

/WELL# you /say 'several PEOPLE#

/who was 'in the HOUSE# a/part from 'you 11.1:664 and your WIFE#

Here counsel uses a preceding statement made by the wit~ess as a point of departure for further questioning, notably on the same topic. In other cases they were used to mark that counsel was ready do draw certain (implicative) conclusions from what the witness had said or make an emphatic :

/

/

/

103

(10)

A:

[FOCUS] Q

B: A:

R

[FRAME] Q

B:

R

A:

F

you /SEE# is :THIS#

/what I'm (

coughs) 'asking

/your suggestion IS# that /it was your "!wife's PHONE MESSAGE# that /caused the DOCTOf# to /bring DETAILS# of the /NURSING- 'homes# /YES# I /think it WAS# /WELL# /you'd PHONED# the /day BE"!FORE {A/BOUT it#}# /I'd 'phoned the 'day BEFORE# and the /doctor 'said it 'wasn't !NECESSARY# - - /YES# ( clears throat) - - - you SEE# /I SUG!GEST# /Mr POTTER# /quite PLAINLY# that . /your !MOTHER# /telephoned the :DOCTOR#. and sh~ /was in a 'state of IN!TOXI!CATION# and /it was a'bout !three o!CLOCK# 11.1:402

[Frames] and [F6cuses] with similar functions were also found in other types of discourse, for example discussion.

TEXT 6.3: A POLITICAL INTERVIEW Text 6.3, a political interview, is an example of planned discourse which is very clearly structured for the benefit of an audience. Interviews are by definition composed of Q/R exchanges. The 16 exchanges in this interview made up 11 Q/R transactions, all opening with a [Focus] move. The only [Frame] occurred at the very beginning, when the interviewer addressed the interviewee by name before presenting the first issue. All transactions but one were pure Q/R transactions, either consisting of a single exchange, {FOCUS QR}, or of two exchanges, {FOCUS QR} {QR}. The remaining transaction was mixed; only the first exchange was a Q/R exchange. The number of transactions in relation to the total number of exchanges indicates that the majority of the transactions consisted of one single Q/R exchange. On the other hand, the interview as a whole consisted of relatively few transactions, which shows that each of them was fairly long. The length of the transactions is partly a reflection of the

104

long introductory [Focus] moves where the new subject matter was presented, but especially of the very long and exhaustive R moves. Owing to the particular features of the interview, which was intended to be both enlightening and entertaining, the interviewer had to express himself very clearly and fully when presenting the issue, and the interviewee co~ld not possibly come up with short yea/no-Rs but had to develop his political views, motivate and justify actions taken under his leadershipn and so on.

TEXT 7.2: TELEPHONE CALLS Generally, telephone calls differ from casual face-to-face conversations by having a distinctly marked beginning and end (although this distinction is not valid for all the telephone calls in the data, due to the recording conditions). The particular features of openings and closings will be dealt with in section 10.3 and will therefore not be discussed at this point. Moreover, the telephone calls in the data are typically made to convey a message, while casual conversations are often mainly social activities. Another characteristic of telephone calls is that one of the speakers, the caller, is partly in command of the turn-taking mechanism, in particular when thi conversation is about to finish (cf section 10.3;3). Although this has certain impliiations for the general framework of telephone calls, it is of no great consequence for the structure of transactions. The important thing is rather a matter of who the speakers are and what type of message is· delivered. Calls where intimate speakers talk about personal matters tend to be structured differently from calls where distant speakers talk business. Text 7.2 consisted of 16 calls of varying length. The number of transactions.per call varied. Only one call consisted of a single transaction, probably because most of the calls involved speakers who knew each other fairly well and therefore took the opportunity of having a nice chat in addition to conveying their real message. A brief loo·k at more business-like calls in the remaining telephone conversations in my corpus showed that they generally consisted of only one transaction. Whenever unmix.ed tr.ansactions occurred, they were accompanied by one or more mixed transactions, so in that respect the telephone calls did not differ from casual face-to-face conversations. /

/ /~/

105

By contrast, explicit boundary markers seemed to be more important in a situation where the interactants could not see each other. Two thirds of the new transactions were introduced by [Frame] and/or [Focus], which may be compared to the general lack of explicit boundary marker~ in unplanned face-to-face conversation. High pitch, on the ot~er hand, did not mark transaction openings in the telephone calls. Whether this was only coincidental is difficult to say.

TEXT 3.4: A VISCUSSION BETWEEN AN AVMINISTRATOR ANV ACAVEMIC STAFF Text 3.4, which is a discussion between a university administrator and a group of academic staff, was much more difficult to account for in terms of hierarchical structure than both 11.1 and 6.3, where all communication was in the form of questioning and responding and where Q/R exchanges for~ed neat transactions and, in the case of 11.1, transactions consisting of fairly regular sequences. The discussion in text 3.4 consisted of only two transactions, each of which could be divided into a number of sequences dealing with aspects of the main topic. But transactions as well as sequences were mixed. Generally, the first exchange was a Q/R exchange, but what followed was an unordered series of informing and eliciting exchanges. The first transaction was opened by the chairman the administrator (note the irregular opening by Q): (11)

Q [FOCUS]

(a:m] - - . "/how about things GENERALLY# I /mean ((have !you[e])) /let's 'start with the AC'COMMODATION# "/{OBVIOUSLY} this is a PROBLEM# - - [:e] 3.4:29

and the second transaction by one of the participants: (12)

D:

[FRAME] Pre

A: D:

R [FOCUS] Q

/CHAIRMAN# /may I 'raise ((a *'com'pletely DiFFERENT))~ point# *"/YEAH#* - ((/TOTALLY different# /RIGHT#))

'''

/would it 'not be POSSIBLE#

'''

3.4:754

The way the introductory moves differ is an excellent illustration of differences in speakers' rights in discussions: while the

106

chairman in (11) offers an issue for discussion, the participant in (12) asks for permission to bring up a new issue. Note that his request is explicitly granted. The speaker roles also manifested themselves in the way sequences were opened. The participants always opened a new sequence by a , generally addressing the chairman: but I want here chairman to ask do you know chairman if chairman is it ever envisaged

Occasionally, were used by the chairman, but with a somewhat different effect: aan I say something about this may we have a word on this

There are two differences worth pointing out (which also refer to transaction openings). The participants' might be followed by the chairman's ; the opposite would hardly occur. The chairman's could be intended to invite the participants to express their opinion, as in may we have a word on this ; a similar would hardly be uttered by one of the participants. There were very few Q/R exchanges in all and consequently few sequences. All transactions and sequences were opened by a Q/R exchange which was always followed by a long discussion, consisting of [Informatives]. It is obvious that the speaker who managed to get the floor kept it long enough to develop his.views on a particular matter.

TEXT 3.3: A VISCUSSION BETWEEN AN AVMINISTRATOR ANV STUVENTS For comparison, I looked at the second discussion in the data, text 3.3, where the same administrator met a group of students, and found that the relationship between the speakers in terms of authority plays an important role. There were considerable differences in speaker behaviour which affected the structural organization as a whole. The administrator's role of discussion leader was much more pronounced in the student group. Only once, for example, did one of the students bring up a matte\.Jor discussion. The students' Rs were generally very short which forced the chairman to narrow down the problem by asking seve·ral more

\\

107

detailed Qs, to the effect that this discussion was more like an interview than a discussion. Transaction openings were recognized in the same way as in text 3.4. Openings were generally not signalled by a but mostly by a [Frame] (and [Focus]), occasionally by high pitch (indicated by 'H'). Note thatI the effect of [Frame] and [Focus] in (13) is similar to that pointed out in the courtroom talk: (13)

[FRAME] [FOCUS] Q

weU /NOW# ' TllEN# /if this is a FACT# [ i] /is it a BORE# to /go to some 'other 'part of the LIBRARY# 3.3:169

The [Frame] well now points forward to the new aspect whereas if this is a fact points back to and sums up what has already been said on the topic.

TEXT 6.2

A VIALOGUE BETWEEN A UNIVERSITY LECTURER ANV A STUVENT

Unplanned, casual conversation is characterized by topic drift and absence of explicit boundary markers. New topics are introduced in what seems to be a fairly irregular and random manner, a result of the ongoing process where utterances are planned and produced on the spot. Text 6.2, a non-surreptitious recording of a conversation between a male university lecturer and a female ex-student, was not characterized by topic drift. The number of topics was fairly low, and each topic was thoroughly discussed, which resulted in long transactions. Transaction boundaries were marked in one third of the cases. Text 6.2 differed from text 4.1 (see below) in very much the same way as the two discussions differed from each other. Just as in 3.3 where the administrator was the dominating party and suggested new issues, the person who introduced the new transactions in this dialogue was the ~ecturer, not the student. In text 4;1, with intimate and equal speakers, either of them could bring up a topic.

TEXT 4. 1. A VIALOGUE BETWEEN HUSBANV ANV WIFE Text 4.1 is a non-surreptitious recording of a married couple's chat at lunch-time. This conversation is marked both by topic drift and lack of boundary markers. The talk centered upon events

108

in the morning, plans for the afternoon, the evening and the weekend, and on typically domestic matters. The parties,were aware of the recording and knew that they were supposed to go on talking for a certain period of time, which partly explains why so many topics wore brought up. Transactions wore made up of exchanges, and there was no intermediary level. They varied cons id·erably in size, depending on whether they contained a narrative part or not. Only very short transactions were unmixed Q/R transactions, although two out of three were opened by a Q/R exchange. Only one fifth of all transaction openings were explicitly marked; the majority of the new transactions were opened without explicit markers: (14)

/are we 'going to 'get a 'TELEVISION# - -

4.1:533

(15)

I /wonder ((when)) [bena?] . 'when 'Berna'dette will be ARRIVING# .

4.1:884

/has 'Ivor gone HOME# -

4.1:1013

(16)

That these are new transaction openings is indicated only by the lack of anaphoric reference, which seems to indicate that there is not much need for explicit markers when intimate speakers are talking. Focusing did occur, however, but in two out of three cases the focusing was ac_hieved by 'fronting' '(cf section 6.3.1) and not by the use of regular [Focus] moves, eg: (17)

/I 'think I'll 'go to ((SLEEP d'you#))

4.1:422

New transactions were also opened by eliciting exchanges with reference to the 'outside world': (18)

/so if !you don't COME to'night# to /this RECEPTION# 4.1:441

The reception has not been mentioned earlier; such extralinguistic references are only possible when the speakers share a great deal of common ground. (19) illustrates a 'broken' transaction: (19)

are you /going to do some 'work on the 'car TOMORROW# - . 4. 1 : 54

which is resumed much later by: \

I /think the :NOISE# /is [5i] - !TAPPETS#

,L1:351

In the meantime a couple of completely different topics were dealt with.

109

5.4

COMMENTS ON TRANSACTIONS AND SEQUENCES

Coulthard, Montgomery & Brazil (1981) raised the question of how far transactions depend on the organization of speaker roles and on the purpose of the interaction. They felt that the need for prospective structuring and retrospective summarizing must be greater in some types of disco.arse than in others. Their observations indicated that the speaker who had control over the turntaking machinery, ie chairman and doctor in their data, also marked the boundaries. Prospective delimitation of the topic was typically found in the form of [Frames] marking end of topic in telephone calls. These observations agree entirely with my own. Prospective structuring was revealed in its most obvious form in the radio interview, where the topics of discussion had to be projected out of consideration to the listeners. In cas~al conversation, where the conditions are different, especially with intimate speakers, prospective structuring w~s almost nonexistent. Retrospective focusing was found to be a typical feature in the courtroom examination, as in the telephone calls. Distinct speaker roles could be observed in more than one type of discourse. The interviewer had complete control over the turntaking in the interview, and so did counsel in the courtroom. The chairman's role in the discussions varied depending on who the other participants were. His authority was much more obvious in the student group than in the staff group, and his control was therefore greater in the first group. Similarly, in the casual conversations, the speaker with more power, such as the lecturer in text 6.2, controlled the turn-taking, whereas in text 4.1, where the speakers were on equal terms, there was an equal distribution of turns. In the telephone calls, finally, the caller had control over opening and closing. It was confirmed in the data that the speaker in, control of the turn-taking also marks the boundaries in the discourse. The dialogues examined can be seen along a scale ranging from most to least structured. The features characteristic of each extreme type may be listed as in Fig 5:1. The label 'unmixed' Q/R hierarchy indicates that a transaction consists of nothing but unmixed Q/R sequences and Q/R exchanges. 'Mixed' signifies the opposite. Boundary markers are [Frames] and [~ocuses]. Note that 'no boundary markers' may involve high pitch and the type of

110

LEAST structured

MOST structured

SITUATIONAL FEATURES: unplanned private intimate/equal parties

SITUATIONAL FEATURES: planned public distant/disparate parties

:

li

INGUISTIC FEATURES re/retrospective focusing nmixed Q/R hierarchy xplicit boundary markers

Fig 5:1

{>

LINGUISTIC FEATURES no focusing mixed Q/R hierarchy no boundary markers

Scale of Q/R exchange structure

fronting which is achieved by dislocation of syntactic or structural elements in the Q move (cf section 6.3.1). It is obvious that the inore casual a conve.rsation is, the more difficult it is to account for it in terms of structural organization. The way exchanges formed into sequences, and sequences into transactions, varied depending on discourse type. The courtroom examination was found to consist of. only one topic-bound transaction which c~uld be divided into a number of aspect-bound sequences consisting of a series of exchanges where details were examined. In the radio interview, 11 of the transactions consisted oi one single Q/R exchange, and there were no ~equences. I

.,,>

/

111

6 Exchanges

6.1

EXCHANGE STRUCTURE

For Q/R exchanges, I shall adopt Coulthard & Brazil's definition of the exchange as the unit concerned with negotiating the transmission of information and its polarity (1981:101). But since my primary aim is to capture the unbroken coherence of series of Q and R, I will go beyond their description of exchange structure, which is I (R/I) (F) (F) (where 'I' stands for 'initiating'). My description is based mainly on pragmatic criteria. In casual conversation, the negotiation of information is more complex than is suggested in Coulthard & Brazil's model, and I find that a certain 'discontinuity' ,'(cf Stubbs 1981:111) must be allowed for, if by discontinuity is meant not a complete break in the negotiation but a momentary hold-up which is somehow involved in and sometimes necessary for the procedure of completing the proposition made in the opening Therefore, I regard embedded (subordinate) exchanges, consisting of a [Checking] move followed by a [Responding] move as 'part of' the main (superordinate) exchange, if the transmission of information which is negotiated in the main exchange depends for its completion on what is done in the embedded exchange, as in:

q:

(1)

A: B: A: B:

Would you like to come to the concert? What time is it? It begins at five. OK, I'll be there.

Another discontinuity which should be accounted for in a model of exchange structure is caused by the re-opening exchange, Qr R, which also constitutes a temporary hold-up before the final completion: (2)

A: B: A: B:

When did you arrive? At five. Did you really? Yes.

112

The frequent occurrences of [Checkings] and [Re-openings] in casual conversation justify their inclusion in a model oi maximal exchange structure. The following variants of move combinations occurred in the Q/R exchanges in the corpus, plus a few other irregular and more complex variants. Qc Rand Qr R like Fare optional. Fis recursive, indicated by Fn. Qc R may recur; this is not indicated in the formalized version of the exchange structure.·

INF INF

~ ~

{QR} {QR Fn} {Qi R} {Qi R Fn} {QR/QR} {QR/QR Fn} {Q Qc RR} {Q Qc R R Fn} {Q R Qr R} {Q Qc R R Qr R} {Q Qc R R Qr R Fn}

The minimal realization of a Q/R exchange is thus: {QR} For the maximal iealization of exchange structure I suggest the following model: { (FRAME) (FOCUS) Q

Qc) ( R/Q

R (Qr R) (Fn)}

The model is intended to cover all the variants of move combinations in Q/R exchanges in two-party conversation in the corpus (with the exception of occasional and randomly distributed [Informatives] and [Supports]). My assumption is that it is applicable to Q/R exchanges in two-party conversation in general, with the reservation that certain combinations occur more frequently in some types of conversation than in others, and, as will be demonstrated in the following pages, that the R/Q move may be less straightforward than it was in my data. The model is not intended to account for exchange structure in multi-party conversa~ion, which is often much more complex and will be dealt with se~arately (see section 10.2). The options available at Q and R, as illust~ated in the model, will be discussed stepwise, with special emphasis on the QR/QR series.

/

113

6.1.1

The eliciting part of the exchange structure

QG~Q)

R

There are three completely different strategies available at Q. The strategies have the followitig basic characteristics in the formalized version: Strategy 1:

the initial Q is responded to straight away

A: B: Strategy 2:

the initial Q is responded to on the basis of what is negotiated in the embedded exchange: A: B: A: B:

Strategy 3:

[~

~~c

the initial Q is responded to by an utterance with dual function:

A: B: A: The initial Q is responded to in all three cases; in the second move for strategies 1 and 3, but in the fourth move for strategy 2. Note that, if strategy 3 is chosen, the exchange will be terminated by A, the initiator of the exchange (if no F move occurs). In its simplest form, strategy 1, the Q/R exchange consists of a straightforward, unambiguous Q immediately followed by R which constitutes the direct answer, as in (3): (3)

C: A:

/do you LOCK your 'room Q which /LEAVE it#}# R /YES# /I DO#

{when you 1.8:431

But in ( 4) ' strategy 2, Q is responded to in a roundabout fashion: (4)

/

/

/

..

A: B: A:

Q /well /[wo e:]/what shall we do about [e) :THIS boy t~en# Qc DUVEEN# R /[Ml#

114

B:

R well "/I propose to :WRITE# [a] /saying . I'm : very sorry I ! CANNOT# - [a:] /teach . 'at the INSTITUTE# 1 .2:1133

(4) is not an example of an ill-formed exchange, despite the withholding effect of the [Checking] move (cf Burton's challenging move, 1980:150). Although the [Checking] move holds up the progress of the negotiation, it is only a momentary delay which does not prevent its completion. The answer to the: initial Q depends for its fulfilment on what is negotiated in the checking exchange which, consequently, should be seen as a necessary constituent without which the proposition would not be completed; Q would not be answered. Since my analysis allows for embedded exchanges within the main exchange it follows that a move can occur more than once in the same exchange. In the analysis by Coulthard & Brazil, by contrast, 'a second occurrence of any move makes a new exchange' (1981:101). The R/Q move, strategy 3, has been described by Coulthard & Brazil (1981 :98) as a move which 'faces both ways' in that it serves as an R to the preceding Q and at the same time as a Q ~ith respect to the following move. This is illustrated in (5): (5)

A:

Q

B:

R/Q

A:

R

B:

/do you !LIKE 'this 'work HERE# . in /this DEPARTMENT# /you were !HtRE# /{W~RE you} 6NCE# *((several sylls))* *'/{I} was an UNDER*"GRADUATE 'here# ((of)) /very !{RIPE} !YEARS# ,,, ( - giggles) 1.6:132

you were here which is followed by an R by B, seems to imply 'then you should know' and constitute an R to do you Zike it here. However, the dual function is more complex than it may appear at first sight, as is reflected in the following expanded version of the model of exchange structure (cf p 112): {Q R/Q

C; )

(F)}

My claim is that one of the functions predomina t.es, depen,ding on the type of initial Q:

115

Type 1 : Type 2: Type 3:

/

A A A A B A A A

does not know R thinks that B knows R wants to know R does not know R knows R wants to know R before he proceeds knows R wants to check/whether B knows R

At first sight, types 1 and 2 are similar in that both seem to constitute what are generally taken to be ordinary requests for information; they do, however, differ in terms of questioner intention. Type 1 involves Qs by which A asks for information with no apparent secondary purpose in mind (this is of course an overgeneralization), but in type 2 the request for information is used as a face-saving device, a preparatory step, which, ideally, will permit A to go on to his real message. Type 3 consists of test Qs and quiz Qs. The general assumption that whatever utterance follows Q will be taken as R 'if that is at all pos~ible' is crucial for the analysis. The assumption implies that, in terms of discourse interaction, the R/Q move is nearly always an R because of its position after Q, but from the pragmatic point of view, it does not automatically follow that it answers Q. The answering function is more obvious in some cases, the questioning function in others. Which of the functions prevails is a matter of who possesses the information, the real purpose of Q, and B's readiness to comply. The degree of significance of these factors for the continuation of the exchange varies depending on what type of Q initiates the exchange. For type 1 Qs, B's display of knowledge/want of knowledge decides whether R/Q counts as an answer or not. If the exchange is initiated by a Q of type 2, B's readiness to comply by asking for further information determines whether A will be able to proceed the way he intended. As to type 3 Qs, whatever B says will be interpreted as an Rowing to the particular situation. Which of the functions predominates is of consequence for the exchange structure. If the R function predominates, the third move is an optional F so that we get the structure {QR (F) (F)}, and if the Q function predominates, the third move will be R, optionally followed by F, and we get the series {Q (R)/Q R (F)}.· Note that the degree of optionality in the F move varies with the nature of the R/Q move.

116

Brazil's suggestions about interpreting the effe~ts of tone selection and especially choice of pitch level (see eg 1981:146ff and section 3.2.3) may serve as a guideline for the analysis of the R/Q move (r = referring checks the truth of an assumption; p = proclaiming introduces new matter). The examples to which I have applied Brazil's prosodic analysis are my own, as are the interpretations in the margin: Initial, Q type 1

(6)

A: B: A: B:

where did he go p DIDn' t he go to LONDON oh yes I see

Q R/Q F (F)

(7)

A: B: A: B:

where did he go r DIDn't he go to LONDON no he didn't oh

Q R/Q R . (F)

Assuming that Bin (6) chose the proclaiming tone (p) implying new information, and low termination which indicates completion and sets up no constraints on a next move, didn't he go to London will be equivalent to I think he went to London and answer Q. Note, however, that the F move is felt to be obligatory although the prosodic pattern in the R/Q move indicates the opposite. That this has to do with the interrogative form of the move becomes apparent if the interrogative form is replaced by the statement form, eg ~e went to London , where Fis not felt to be obligatory. If B instead chooses high termination in combination with referring tone (r), as in (7), the referring tone shows that he is dealing with old information, didn't he go to London as he said he wou1,d' whereas the high termination indicates that he anticipates a polar decision from A. In other words, it is evident that Bis not at all certain of what he suggests; consequently, A takes his utterance to be a Q and answers it. In this case, ·the F move, which is optional, constitutes the fourth move. Examples (8) and (9) illustrate non-compliant behaviour; B does not make an effort to answer. They have been coded diff~rently although I will assume that why don't you ask somebody etse is pronounced with proclaiming tone and low termination in both cases.

117

(8)

A:

B: A: B: (9)

A: B: A: B:

where did he go p WHY don't you ask somebody ELSE OK

Q

where did he go p WHY don't you ask somebody ELSE because I thought yoJ knew well no/oh

Q

R/Q F

R/Q R (F)

Since there is nothing in the form of the R/Q move which indicate? how it should be understood, the different interpretations must be related to the speakers' common ground and the relationship between them. A's reaction in (8) presupposes a rather unfriendly atmosphere which probably does not exist in (9). In the first case R/Q is taken as an R with command function, and in the second case it is taken literally as a Q and answered accordingly. But regardless of A's interpretation of the R/Q move the implicit R to the initial Q, I don't know , is conveyed in both cases. InitiaZ ( 1 0)

Q

A:

B:

A: B:

type 2

do you know what happened what my car broke down oh

Q

R/Q R F

By responding what B shows compliant behaviour, which is necessary for the continuation of the exchange. He does exactly what is expected of him and encourages A to go on the way he intended. what implies no I don't know, teii me. Thus, it answers the initial Q by implication while at the same time asking for information. In this case, the F move is obligatory; its absence would be noticed and would signify that B either did not hear or was not interested. (See also section 7 .1. 2 for this type of questioning strategy.) Type 3 Qs ( 11)

( 1 2)

A: B: A:

do you know the Spanish word for friend is i t amigo good

Q R/Q

A: B: A: B:

which is the hi,ghest mountain in the world is i t Mont Blanc no it isn't oh

Q

F

R/Q F (F)

118 \

Assuming that (11) is a classroom Q and that (12) is a quiz Q, which means that A knows R in both cases, my claim is that the R/Q move will be taken as an answer in both cases, regardless of tone selection and pitch level choice. Due to the particular situation, classroom and quiz, this type of utterance will be taken as an answer in disguise. The ~eacher's and quizmaster's F moves arc obligatory, predicted from the beginning, as a confirmation of right or wrong. A second F move is hardly to be expected in the classroom but is very plausible as the contestant's reaction in a quiz situation. The R/Q move has been dealt with by Coulthard & Brazil (1981: 96ff), Stubbs (1981:112ff) and Berry (1981:127ff), who do not seem to be in full agreement about whether the next move after R/Q is R or F, which of course reflects their view of the R/Q move itself. Coulthard & Brazil state that what distinguishes the R/I move (which corresponds to my R/Q move) from R is high termination and/or interrogative syntax. They give the following example to illustrate the way a pupil's response 'is looking for an evaluatory [Follow-up] from the teacher' (1981 :97): (13)

T: P: T:

can anyone tell me what this means does it mean d~nger men at work yes ...

The structure they suggest is thus I R/I F. Stubbs prefers to regard the third move as Rand suggests the structure I R/I R for the same example (1981:112). H~ argues that the syntactic form of the R/I move is irrelevant, interrogative, declarative and moodless forms being largely neutralized and equivalent as R/I moves; they all expect a following R. Berry, finally, in her interesting analysis of discourse structure in terms of three different layers (interpersonal, textual, and ideational), does not need to specify whether she regards the third move as R or F, but although she uses a different terminology, she seems to be in line with Coulthard & Brazil (for details see 1981:120ff). Depending on how the R/Q move operates we get the following exchange structures in relation to initial Q type (predominant func\ tion underlined):

/

/

119

Type 1 - Q request for information

A wants B to inform him (a) if B knows R (b) if B does not know R

{Q ~IQ

F (F)}

{Q R/.Q R (F)}

Fin (a) is predicted by the 1syntactic form of R Type 2 - Q pre

A wants to check B's knowledge/interest for an ulterior purpose

{QR/QR F (F)}

Fis predicted as a marker of compliant social behaviour Type 3 - Q:test/quiz

A wants to check B's knowledge (a) if B knows R (b) if B does not know R

{Q ~IQ F (F)} {Q R/.Q RF}

Fis predicted in the situation arid constitutes A's 'stamp of authority' (cf Berry 1981:126) 6.1.2

The responding part of the exchange structure R. (Qr R) R (Fn)

An exchange completed by R can be re-opened by an immediately following [Re-opening] move whereby the questioner expresses doubt about R for some reason, or in Berry's terminology, it queries 'the state of knowledge which the primary knower has indicated for himself' (1981:137). The [Re-opening] move is generally realized by words like really or a tag-Q, often with high pitch, which elicit a confirming R, which is optionally, but not frequently, followed by F. Since the information sought in the initial Q move is still being negotiated, the re-opening exchange does not constitute a new exchange but is part of the main exchange. ( 14) is a good example: (14)

A:

Q

B: (A: B:

R

but /what 'made you DE:CIDE# - /did you /did you 'go from UNIVERSITY# [s] to /SECRETARIAL *'school# [a]* */YES#* /AUTOMATICALLY# ((well)) it was a RE/{ACTION} against !FINALS# -

l 120

A: B:

Qr R

/WAS it# ./YES#

5.9:294

It is obvious that the [Re-opener] is in many cases a fairly weak query, often just a device indicating polite interest. Besides what has already been §aid about the F move, I would like to emphasize that, generally, there is just one F move, by which A expresses his attitude to R. But the,number of F moves is not predictable. In telephone conversation, where the simple tripartite structure {QR F} is more common than in face-to-face conversation, r moves tend to recur. A final example will illustrate how a Q/R exchange in a telephone dialogue is terminated by a series of F moves: ( 1 5)

/how /how was the :WEDDING# "/OH it was it was "/really GOOD# it was [a] it was a /lovely DAY# /YES# /and. it was a "!super PLACE# . to /HAVE it . of COURSE# /YES# /and we 'went and 'sat on "/sat in an :{ORCHARD at} !GRANTCHESTER# a:\').d /had a "!'{HUGE} {TEA} . *!.AFTERWARDS# (laughs -)* * (laughs - -)*

B: A: B: A: B: A:

B: A: B:

A: B: A:

+ [a:]+

F F F F

+it /DOES+. sound# /very 'niae INDEED# it /rea7,7,y WAS# /GOOD# /SMASHING# /YES#

7.3:1441

The model of exchange structure suggested in this section presupposes that no ~erminological distinction is made between A's [Follow-up] and B's [Follow-up]. They are both indicated by F for the same reason as [Question] and [Response] are indicated by Q and R irrespective of who the questioner and the respondent are:

~; ~~ A:

R

B:

R

For the sake of clarity I will distinguish between F and F , for 1 2 A's and· B's [Follow-up], however.

.1

121

6.2

LINKAGE

Following Merritt (1976), the way moves in exchanges and successive exchanges are related can be described in terms of four basic types of linkage: 'chaining', 'embedding', 'coupling' and 'elliptical coupling', as iilustrated in examples (16) - (19) (examples (16) and (17) are from the extended corpus): (16)

A: B: A: B:

CHAINING Q1 R1 Q2 [ R2 [

(17)

are you /NORMALLY 'free TUESDAYS# I'm /normally 'free TUESDAY# /and THURSDAY# and THURSDAY#

EMBEDDING

:: ~~~ A:

R2 R1

B:

( 18)

/can I 'speak 'to !JIM 'Johnstone PLEASE# /SENIOR# /YES# /YES#

Q1 [

(19)

A: B: A:

9. 9: 1

COUPLING

A:

B: A:

7. 1 : 18

~2

R2

/OH and# /by the !WAY# /you !didn't 'try 'using 'your :READING-'room TICKET# as a /STUDENT card# - - . /NO# - /COULD I 'have# /YES#,,, 7.2:771

ELLIPTICAL COUPLING well d'you /know 'what they GOT# icR1) /Q2 /WHAT# I I R2 they /didn't get RE!PLIES from. from 'most PEOPLE# - ,,, 4.1:790

rq1

CHAINING implies that A makes the initial move in the first exchange and goes on to make the initial move in the next one. The result is an eliciting exchange followed by a re-eliciting exchange. Merritt employs the term EMBEDDING in preference to 'insertion sequence' (Schegloff 1972b) and 'side sequence' (Jefferson 1972) since it 'captures the•important notions that the embedded Q-A sequence is relevant and subordinate to the host Q-A sequence' (1976:334). In such cases B needs R to Q2 before he is able to respond to Q1; in other words the 'conditional relevance' (cf Schegloff & Sacks 1973) of R to Q1 is 'deferred' until after the

\

122

\

completion of the Q2-R2 exchange (cf Merrit 1976:334). In my terminology the sequence consists of a superordinate eliciting or re-eliciting exchange with an embedded subordinate checking exchange. COUPLING, which Goffman (1976:259) refers to as the 'two-person sociable chain', implies that B responds to Q1 and goes on to ask Q2 in the same utterance. The result is an eliciting exchange followed by a re-eliciting exchange. Whereas coupling consists of two moves pronounced in one utterance, ELLIPTICAL COUPLING consists of one move (Q2), from which R to Q1 can be implicitly derived; in other words, R1 is ellipted. The result is an eliciting exchange initiated by A and a re-eliciting exchange initiated by B, just as with coupling but unlike chaining where A initiated both exchanges. Generally speaking, Q, being the first pair part of an adjacency pair, is expected tp be immediately followed by R, and the conditional relevance is 'satisfied' (see Merritt 1976:329). In the case of embedding the next utterance after Q1 was Q2, but still it cannot be said that the conditional relevance is dissatisfied since the Q2-R2 exchange is necessary for the fulfilment of R1, which is postponed until after the completion of the embedded exchange. Nor can it be said that R is 'officially absent' (cf Schegloff 1972b:76) in elliptical coupling, since the direct R to Q1 can be arrived at by implication. Example (19) can be expanded into: (19a)

A: Q1 well d'you /know 'what they GOT#. B: (R1)/Q2 \no\ /WHAT#

Q2 which fills the R slot is formally'not a proper answer to Q1 but it serves to provide the answer. In unplanned conversation, complex patterns consisting of more or less irregular combinations of the basic types of linkage are often met with rather than the regular patterns exemplified in this section. For complex linkage patterns, see Stenstrom (1979). 6.3

EXCHANGE OPENINGS

Openings of Q/R exchanges in conveisation are seldom ind~pendent of the previous discourse. The extent to which the opening move is related to the immediately preceding move depends on whether it open~ a superordinate (eliciting/re-eliciting) exchange or a

/

/

/

.1

· 123

subordinate (checking/re-opening) exchange. This is generally reflected in its syntactic and lexical properties. The superordinate/subordinate relationship can be further described in terms of initial and non-initial (for a discussion of !initial, see Stubbs 1981:107-119).

SUP ERO RD INA TE - [

. . . --======={ (FRAME) INF.._{Q (FOCUS) Q

initial

non-initial

INF,.__{Qi

SUBORDINATE ~~-non-initial --=-=-{INF -----{INF Fig 6:1

Openings (braces indicate

Qc Qr

exchange opening)

Superordinate openings are either initial or non-initial, depending on whether or not they have to be :expanded from the previous move in order to make sense. Subordinate openings are always noninitial. Examples (20) - (22) illustrate superordinate initial and non-initial openings: (20)

A:

Q

how's your father getting on?

(21)

A: B:

INF Q

I did the shopping why didn't you return the money?

( 2 2)

A: B:

INF Qi

I didn't return the money why?

(20) is independent not only of the immediately preceding move but also of the previous context for its understanding since it introduces a new topic. (21) and (22) are more intricate. In neither case is Q structurally predicted; what is expected after [INF] is F, for example realized by an such as oh or I see. I have analysed both (21) and (22) as realizing two exchanges, informing and eliciting, on the grounds that the F move can be regarded as ellipted and implied (cf ellipted coupling, section 4.3.2). However, Q. is initial with regard to the preceding move only in (21) and not in (22) where why has to be expanded from what A just said in order to make sense: (22a)

A: B:

INF (F)/Q

I didn't return the money !ohi why?

In (21) Q is syntactically and lexically complete whereas.Qin (22) is ellipted. (21) is consequently independent with regard to

---

!

124

the previous [INF] while (22), which presupposes an 'antecedent', must be regarded as dependent, ie non-initial. (23) and (24) are examples of subordinate non-initial openings: (23)

(24)

A: B: A:

R

13:

F

A: B:

INF Qr

A:

R

INP

Qc

your train is late sorry? your train is late oh your train is late really? yes

Both the checking exchange in (23) and the re-opening exchange in (24) are subordinate to an informing exchange. 6.3.1

Opening strategies

There are various strategies available to A by which he can draw B's attention to what he is doing or about to do. I shall discuss these strategies in terms of 'framing', 'focusing', and 'fronting'. Framing and focusing involve the use of particular moves; fronting, on the other. hand, 'implies certain tactical manipulations which affect the Q move with consequences for the [Focus] move. framing is achieved:





by using certain lexical items: 'attention-getters' such as Zook, hey, and listen, or less explicit markers like well, now, actually, and 'hesitators', such as [e:m]. These items realize [Frames] (cf initiators, section 6.3.2) by addressing the .hearer by name (initially), which is another type of [Frame]

focusing is achieved:



by using a [Focusing] move, which either contains explicit reference to the act of asking, , or indicates what Q will be about by a or a (cf fronting)

fronting is achieved:

• • •

by re-ordering acts in the Q move by dislocating syntactic elements in the Q act by pre-empting Q focus

I

125

INITIAL OPENINGS

Especially in informal, casual conversation new exchanges are mostly initiated by an [Elicit]: (25)

Q

/have you seen !G.A}WAY College# I

5. 9: 17

(26)

Q do you re/member GREENBRIDGE#

5.9:65

(27)

Q /how did you 'get down to WINDELBURY#

5.9:1190

In each of these examples the [Elicit] not only initiates a new exchange but also a new transaction. It is interesting to note that neither [Frame] nor [Focus] is needed to signal the transition from one transaction to the next. But in (25) a rise in pitch, signalled by the booster, indicates a new beginning. This is entirely in line with the observations of Coulthard, Montgomery & Brazil (1981) who noticed that although their doctor/ patient transactions were marked by TFrames], it is not the lexical items peP se that are important but the choice of pitch height, notably high pitch. Generally, in casual conversation, one utterance leads to the next so that new exchanges are either implicitly or explicitly related to what preceded. It follows that they are only rarely transaction-initial and that what is negotiated is mostly something that can be associated with the topic already being discussed. The parties engaged in the conversation referred to in (25) (27) were two friends, a·secretary and a male academic, who talked about ·their old college, their mutual friends and their present position; in other words it was a friendly chat where the parties brought up whatever came to mind. The fact that they were intimate friends implies that they shared a great deal of common . ground, which at least partly explains why no introductory [Focus] moves were needed. Fll.ambig

Framing is achieved by the use of [Frames] which mark a boundary in the discourse and signal a transition between two stages. The following (simplified) example occurs in Sinclair & Coulthard's classroom data (1975:99): (28)

Boundary Elicit

Now then Will this saw cut through this? A

126

Now then with falling intonation and followed by 'silent stress' (A) has been analysed as a [Framing] move which signals that one stage in the discourse is ended and another will follow. Similarly in casual conversation:

(29)

ll:

[FRAME] Q

"/HEY# [wiw] if /we [go?] if/ /I got this CHEST# /where'll we PUT it# - - - /where would 4.2:342 you 'have it in the HOUSE#

.

Hey attracts B's attention and signals that something new will be brought up, in this case the question of where to put the chest. Although [Frames] like now then and hey have nothing to do with the negotiation of the ~ew information, they are included in the structure of the exchange. I regard initiators, eg now, anyway, Zook etc (cf section 6.3.2) as [Frames] only if they constitute a separate TU. They are frequently, but not always, folldwed by a pause. The intonation is mostly falling, sometimes level. Initiators whic~ do not form separate TUs are regarded as constituents of the move they introduce. The initiator welt as part of an [Eliciting] move is exemplified in (30) (cf chapter 5: (3) .for a long version of this excerpt):

(30)

well /what are we 'doing 'this WEEKEND# -

4. 1 : 14

Weit has been analysed as a link which serves to connect Q with the previous utterance. In (31) and (32) initiators realize separate moves:

(31)

[FRAME] Q

/[a:m]# ((/was it /did you say you were going to :BURGOS#)) 1.1:544

[a:m] could easily be replaced by such boundary markers as now or welt. Occurrences in the data were rare, however; the framing function of hesitators was generally difficult to determine. The second type of framing, ie direct reference to the addressee by name, typically occurs in pre-planned, formal conversation, where it constitutes a polite way of signalling exchange boundaries besides indicating who is addressed: (32)

[FRAME] Q

/Mr POTTER#

/did YOU# - - AR/RIVE# a/bout !two o'CLOCK# /on [Oi] . !SUNDAY# . the /date the 'will was • SIGNED#. 11.1:1

,/

/

127

Example (32) is from the courtroom proceedings. Other similar examples are found in the radio interview, text 6.3. Foc.u1.,ing

Focusing implies the use of [Focus] moves which serve to indicate that new eliciting exchanges are being introduced in that they point forward to what is to come. They consist of an optional initiator and a , a or a (cf p 124), as in (33): (33)

[FOCUS] Q

wel,1, I'd /Zike to 'know what you're ":SAYING# a'bout THIS# [?] are you SAYING# that [a] • the /doctor's RECOLLECTION# was com/pletely WRONG# . or that /POSSIBLY# /he !got CON!FUSED# with an/other ! PATIENT# 11.1:331

The initiator well constitutes a link between the [Focus] and the preceding move, while the rest of the [Focus], realized by a , I'd like to know ... , points forward to the [Elicit]. As a contrast, consider the following example, where what looks like a [Focus] consisting of a referring to the act of asking constitutes nothing but the [Eliciting] move: (34)

A:

[FRAME] Q

B:

R

/WELL# . "/MAY I ask# /what goes !INTO that paper NOW# be/cause I !have to ADVISE# . ((a)) /couple of people who are !DOING [oi: a] well /what you :DO# /is to - - ,,, 1.1:8

is an indirect way of asking for information, ie may I ask a , and does not serve as an introductory move in this case, as is also reflected in the fact that it is immediately followed by R. The [Focus] move can also be realized by a where the questioner prepares the way for the following Q by referring to what he is going to ask about: (3S)

B:

[FRAME] [FOCUS]

[am] /LISTEN# /mother's :CLOCK# is /VERY is /running !quarter of an 'hour :SLOW#. I've /slightly 'moved it to a :faster PO:SITION#

128

Q

but /how do I move the HANDS#

7.2:265

The [Elicit] is preceded by both [Frame] and [Focus]. The present data indicated that similar sequences belong to planned rather than casual conversation and also that they are not very frequent. It also appears that [Frame] + Q ~ccurs more often in casual conversation whereas [Focus] + Q is typical of pre-planned, formal conversation. In the radio interview, for exa~ple, 11 out of 16 eliciting exchanges were introduced by a [Focusing] move (= transactions). So far I have picked out uncontroversial examples of [Focusing] moves, but as will be seen in the next section, there are complications, in that focusing is sometimes the result of 'fronting'.

F1tonting Fronting is a conversational device whereby a certain element is dislocated from the place it us~ally occupies in the Q move to front position. It constitutes a way of manipulating the structure of the exchange in order to draw B's attention to a certain point, remind him of something, give adequate background information, etc, before asking Q.. The effect of this procedure is increased emphasis on the fronted element. Fronting affects acts as well as single elements of the Q act and involves: 1) 2) 3)

RE-ORDERING OF ACTS DISLOCATION OF SYNTACTIC ELEMENTS PRE-EMPTION OF Q FOCUS

A general characterization of these procedures will be followed by a more detailed description. 1)

RE-ORDERING.OF ACTS:

The part of the Q move which generally occupies second position after the main Q act and serves as a occurs before the Q act, where it serves as a in the [Focus] move: (36)

Q

he's a good lecturer have you heard him

which corresponds to the unmarked sequence: (36a)

Q

have you heard him he's a good lecturer

_J

129 2)

DISLOCATION OF SYNTACTIC ELEMENTS:

An element in the Q act, ranging from a single item to a clause, occurs in front position and is replaced in the Q act by a pronoun or a prepositional phrase: I

(a) single element: (37) those seminars - are they conducted by him which corresponds to: (37a) are those seminars conducted by him (b) clauses: (38) she signed the will - did that tire her out paraphrasable by: (38a) did the signing of the will tire her out 3)

PRE-EMPTION OF Q FOCUS:

The Q act is realized by partial 1ellipsis, the recoverable elements of which occur in the : (39)

I haven't seen Gillian - have you

which corresponds to: (39a) have you seen Gillian There is a crucial difference between type 1) on the one hand and types 2) and 3) on the other. In types 2) and 3), the fronted element is syntactically 'part of' and can be inserted in or merged with the Q act, a~ is exemplified in (37a) - (39a). In type 1), there is no such 'part of' relation; the acts can occur in any order. The fronting of an element obviously implies focusing B's attention on what is to come, but it does not necessarily imply that the focused element is part of or realizes the [Focus] move, as I will try to demonstrate. 1)

RE-ORDERING OF ACTS

It is evident that a different order of the acts does not produce exactly the same effect. This is illustrated in (40) and (41): (40)

Q

(41)

Q

have you got a pen I'll leave a message we must do something ab6ut it so can you tell me

...

130

There is a causal relationship between the acts in both examples, beaauae I'll leave a message and sinae we must do something, but they differ in the placing of the emphasis. In (40) the reason for making the request comes second and the serves to provide B with background information. In (41), on the other hand, where Q is asked as a consequence of what is mentioned in the first act, the , the reason for making the request, given beforehand, is strongly emphasized~ An example from the corpus illustrates that the can be fairly long: (42)

A:

[FRAME] [FOCUS]

Q

/Prime MINISTER# /WE# - [?] read in the :NEWSPAPERS# that /Mr :Heath's 'putting on WEIGHT# . /and that. !you've been getting some of your !SUITS 'taken 'in# /you're !not - 'lying A:WAKE at night# /worrying a'bout the 'date of the :next E!LECTION ARE you# 6.3:1

A similar causal relationship exists between we read in the newspapers that ... and you're not lying awake ... as between the acts in (41) and (42) must be regarded as belonging to the same category of.strategies as (41), i~ a [Focusing] move followed by an [Elicit]. Briefly, the position of the act determines its discourse function. The length of the act is irrelevant. In (40) and (41) the acts are coherently related withi~ each example, and there are no restrictions on the possibility of reversing the order of the acts, other than those emanating· from a particular situation. With cohesively related acts this is possible only if Subject and Object are referred to by a pronoun in both acts, as in (36) he's a good lecturer - have you heard him. In (43), on the contrary, where the noun John in the [precursor] is replaced by the pronoun him in the Q act no fronting is possible: (43)

Q

(43a)

Q

John is a good lecturer have you heard him

*have you heard him John is a good lecturer

This is illustrated in (44): (44)

Q

your /MONEY has ARRIVED# do you /want me to 'put it in an envelope and !SEND it to you# or do you /want me to KEEP it# 7.2:181

where your money is placed in focus before Q is asked. Of course it would have been possible to say do you want me to put your money in an envelope - it has just arrived but then the emphasis on money would be lost. 2)

DISLOCATION OF SYNTACTIC ELEMENTS

The following types of 'syntactic fronting' occurred: + Q (a) the whole constitutes the complementation of Q; it is replaced in Q by a pronoJn, a prop word or 0. When merged with the Q act, the two acts form a main clause followed by a subordinate that-clause. The verbs involved in the Q act are verbs of perception, production,and opinion: but we're going to ... - have you seen that they'll probably be bored - do you not think so you didn't know - are you sure 0 I ve got the Lennox Berkley - did I tell you it (b)

the whole constitutes the Subject of Q; it is replaced in the Q act by a pronoun: she signed the will - did that tire her out

Cc)

the whole constitutes the adverbial of Q and is replaced in Q by a pro-form or a prepositional phrase: she hardly spoke she did not say

2

/ /

-

...

is she right or wrong there - is she wrong in that

Lifting of single element to fron~ position with replacement by a pronoun· or prepositional phrase or occasionally repeated in the Q act (a) in single position: but garden - you d6n't get garden ... do you

132

(b)

accompanied: time for tea - would you like some

Let us consider one example in its proper context: (45)

Q ((this)) /POLLY# . you /know !that GIRL# whom I've [?] I [m?] I [mm] pre/sented . ((a)) :rather ABSURD re'port in a 'way# ((tha~)) /genuinely :rcpre'sented what !I FELT# I /((said)) she !might FAIL#

'''

((do)) you RE/MEMBER#

'''

/do you "KNOW 'her#

1.4:887

The fronted item, Poiiy, has a long, parenthetic postmodification which identifies the girl in question. The length of the postmodification, however, does not alter the fact that the whole utterance constitutes'a single Q act. It would be perfectly possible to re-order the wording of this utterance so that we would get: do you remember this Poiiy ... which makes the Q act function fairly obvious. 3)

PRE-EMPTION OF Q FOCUS

This category differs from the preceding one in that the Q act is realized by ellipsis. Two types occurred: (a)

the Q act realized by AUX +Sis introduced by a in which the ellipted elements are recoverable: I think I'll go to sieep - do you.

(b)

the Q act is realized by an ellipted indirect interrogative. the ellipted elements of which are recoverable in the precursor: I meant to take the typewriter baak - do.you think I could

The effect of fronting is different in (a) and (b), it seems, although the same strategy is used. In the first example telling seems to be more important than asking. A will probably\go to sleep re~ardless of B's decision; but the utterance may ~f course be intended as an indirect speech act, meaning I think you shouid aome to bed. In the second example, asking is more important, since the returning of the typewriter depends on B's opinion, re-

133

fleeted in R. I have had some difficulty in deciding where some utterances in this category belong, whether they should be regarded as consisting of two acts making up two separate moves, [Focus] + [Elicit], or as two acts making up the Q move. Compare the following two examples: I (46)

(47)

((/I !haven't)) !SEEN 'Gillian for a LONG 'time# /have YOU#

1.4:1024

but we're /going to go 'on with it "next [a] - TIME# ((/see be'cause of)) the !PROVOST is ad'dressing us TOMORROW# ((/ISn't he#)) have you /SEEN that# 1.4:1099

Although (46) and (47) have been placed in separate categories because of their syntactic properties ((47) is in category b), they seem to function interactively in the same way, ie as [Focus] + [Elicit]. The fact that the Q act in (46) is realized by an ellipted clause and the one in (47) is not does not seem to be of any consequence. Focusing realized by a and focusing caused by fronting may occur in the same exchange: (48)

/let me !bring you . [kl !quickly 'back to EUROPE 'Prime MINISTER#

[ei] . the "/tone of the 'speech 'Mr CALLAGHAN 'made . {this /WEEK#}# .· /struck !many PEOPLE#/[?] as·. !notably 'more AC":COMMO'DATING than his FIRST# RENE'GOTIATION 'speech# on 'April the FIRST# does /that re'flect a. 'shift of INTENTION# . on /your 'government's PART# 6.3:878

Q

Similar sequences only occur in planned discourse, such as the radio interview, from which this extract was taken. It appears that fronting affects moves and acts in the following ways in discourse organization: • •

Acts in front position acquire function and realize [Focus] moves~ Fronted syntactic elements occur either in the [Focus] move or remain part of the Q move.

In addition to what was said earlier about the emphasizing ef-

134

feet of fronting (p 128), I wpuld like to suggest that the reason for fronting may not be to give emphasis at all. Since fronting typically occurs in conversation, especially unplanned conversation, it may just reflect the individual speaker's ongoing planning of the Q strategy. In,,other words, what comes to mind first is mentioned first. It will then be a process phenomenon rather than a stylistic device.

NON-INITIAL OPENINGS So far I have dealt with exchange openings which are captured by the formula: {(FRAME) (FOCUS) Q Such openings are defined as initial since they do not depend for their understanding on the preceding utterance, to which they bear no explicit relati6n. They initiate new exchanges and pre, diet the next move; [Frame]/[Focus] predicts Q and Q predicts R. But there are cases where an utterance that serves to initiate an eliciting exchange depends entirely on the previous speaker's utterance to be understood. This was exemplified in (22), here supplemented with an R: (22a) A: B: A:

[ INF] Qi R

I didn't return the money why? I had to buy a book

In (22a) Q does not make sense in isolation, since most of its surface structure is ellipted and has to be expanded from the preceding utterance which serves as an 'antecedent'. Similar exchanges may cause proble~s: how many exchanges are involved, and. how are the ~oves related? One way out of the dilemma might be to regard exchanges in which Q is non-initial and presupposes an antecedent as 'incomplete' in the sense that part of the information being negotiated is to be found in the preceding move:

INF {QR (F)} Another possibility might be to regard sequences of thi~ type as informing exchanges, initiated by an [Informative] with Q and R making up a subordinate exchange: ,/

{INF QR (F)}

/

But neither of these descriptions fits all cases. I found that similar sequences could be satisfactorily analysed with the help of the 'F test', which is based on the assumption that [Informatives] presuppose a [Follow-up] move, (implicit or explicit) in order to completr an informing exchange; this means' that a complete minimal informing exchange has the structure {INF (F)}. The F move is generally realized by an such as oh or I see. The F test implies that: if the [INF] move is or can be followed by a [Follow-up] in a sequence consisting of an [Informative] and a successive Q it constitutes the initiator of an informing exchange, and Q is the initiator of an eliciting exchange: {INF (F)} {Q ... 2

if a [Follow-up] is not possible after [INF], the whole sequence constitutes an informing exchange with Q initiating a subordinate exchange: (Qc . i . {INF \Qr ...

Examples (49) and (50) illustrate the first case: (49)

A: B:

[ INF] F Q

A:

R

so /there we !SAT# in the /grand ! CIRCLE# I [Ml# /not the 'royal :BOX# /NO [e] in /FACT# (laughs-) 7.3:1201

.

'''

(50)

B: A: B:

[ INF] Q

R

I'm "LUCKY ACTUALLY# I've /got some !rather 'rich RELATIONS# /WHERE!ABOUTS# 6.2:539 well there's "/one 'in

'''

The only difference is that, in (SO), no F move is expressed although it would have been quite possible: (50a) B: A:

[ INF] F

I

Q

I've got some rich relattons oh! whereabouts

F will therefore be considered ellipted and implicit in similar cases. (51) and ( 5 2) , which are not entirely identical, illustrate the second case: (51)

A: B:

[INF] Qc

/you're !HOPELESS# /SORRY#

136

A: B: (52)

A: B: A:

R F [ INF] Qr R

/you're !HOPELESS# /WELL# /no 'more than :YOU#

7.2:844

/oh I !got I /got a MASTER 'mind {as /WELL#}# /oh you ! !GOT one# . /YES# 7.2:129

In neither example is a [Follow-up] possible after [INF]. The expanded version of (51) indicates that the informing exchange, initiated by the [Informative], is not complete: ( 51 a) A: B:

A: B:

[ INF] F /Qc R F

you're hopeless oh sorry/ you're hopeless well , , ,

The completion of the exchange is momentarily held up by the subordinate checking exchange; the acknowledgement of the information given in the [Informative] does not occur until F confirming R, ie when the embedded exchange is completed. F cannot possibly occu·r after [INF] since the information given so far is perceived as incomplete. Notice that utterances such as oh you got one analysed as a [Re-opener] (Qr) in (52); function as a kind of F move (feedback) in much the ·same way as oh and I see would have done in the same place; the only difference is that oh and I see do not usually elicit an R. It may sometimes be difficult to determine whether a Q following [INF] should be analysed as a [Check] or a [Re-opener]. Qin (53) has been analysed as a [Re-opener]: (53)

B: A: B:

[ INF] Qr R

but /rumour 'has it that :HE's 'go#}# . "/KEN is# /YEAH# -

{A'/BOUT to

7.3:443

but compare ( 54) : (54)

A: B: A: B:

[ INF] Qc R F

.

!HAS been CANCELLED# /CP'T /for TO*MORROW#* */TOMORROW#* /YES# /GOOD#

7.3:20 /

137

If Q can be interpreted as oh you mean it has been analysed as Qr, if it implies do you mean as Qc. Once more I think that the potential occurrence of a [Follow-up] is revealing. In (53) a [Follow-up] after [INF] is impossible because of the following [Re-opener]; in (54), on the other hand, R is followed by F, which indicates the end of the/ embedded checking exchange. In neither case is an F possible after [INF]. Notice that the F test is not intended for eliciting/re-eliciting exchanges, where, in the simple case, F never occurs until after R • . By this discussion I have wanted to show that sequences involving non-initial openings, realized by a Q move, fall into two groups. The first is analysable as two closely related exchanges, informing and eliciting: { INF (F) }{Qi ... The second can be analysed as a superordinate informing exchange which needs a subordinate checking ,or re-opening exchange for its completion: {INF(

~~) R

(F)}

The following types of Q move presuppose an antecedent: Type A:

F ELLIPTED

{INF 0}{Qi R (F)}

Q asking for supplementary information A: [INF] I've got some rich relations B: Qi whereabouts? A: R Canterbury B:

2

F

Q with pronominal anaphoric reference A: [INF] I've bought you a present aren't you going to tell me what it is? B: Qi

A:

R

B:

F

Type B: 3

F DELAYED

asking for repetition ('what did you say?') A: [ INF] you're hopeless sorry? B: Qc you're hopeless R A: Q

B:

F

{ INF Qc R (F) }

138

4

Q asking for alarifiaatio~ ('do you mean?')

A: B: A: B:

[INF] Qc

Type C:

Q

5

6

R

we used a Ladybird book what the ones they have in England? yes

P =

FEEDBACK

{INF Qr R (F)}

Q demanding aonfirmation ('you mean?')

A: B: A:·

[INF] Qr

B:

(F)

R

he's about to go Ken is? yeah

Q refleating reaation

A: B: A: B:

[INF] Qr R

I managed that did you? mhm

(F)

[Follow-ups], indicated by round brackets, are more or less expected; they are least expected after re-opening exchanges. 6.3.2

Initiators

'Initiator' is a superordinate term for a heterogeneous set of linguistic items used in turn-initial position. The following items were observed in the data, classified according to their main pragmatic function. The voiced hesitation [e:m~ has been included for comparison (see Fig 6:2). Syntactically, the items differ even within the subgroups. Group 1, for instance, includes the interjection hey, the verb look, the adverbial now, t~e conjunci a~yway, the disjunct aatually, and the prepositional phrases apart from that, by the way, and in other words. The question may arise why and, but, and so, which occur as connecters in any kind of language production, not necessarily interactive, should be regarded as initiators. The reason is that they proved to be unexpectedly common in Q openings (cf Table 6:2, P 139), I mean, you know, and you see are included as typical features of dialogue which, besides occurring in ~bdial and final position, are frequently found in utterance-initial position.

/

/

139

ITEM

PRAGMATIC FUNCTION

1

actual7,y anyway apart from that by the way hey in other words "look now

attract attention

2

ah and but oh

reflect reaction

GROUP

80

3

4

Fig 6:2

I mean well, you know you see [e :m]

reflect attitude mark hesitation

Initiators classified

If. the items in group 1 occur in a separate TU and precede [Focus] or [Elicit] moves, they are regarded as moves in their own right, ie [Frames]. This is indicated by capital letters in the examples analysed. Otherwise, the distinction move/act will not be commented on in the following discussion, which concentrates on the distribution of initiators in the exchange.

PREVIOUS APPROACHES In previous approaches to the analysis of some of these items in utterance-initial position, they have usually been regarded as syntactic components. Oh and ah are generally treated as interjections in grammars and dictionaries, ie as words expressing various emotions. In most of the instances of initial oh and ah in the data, however, the emotional character appears to be, at most, very weak. Quirk et al (1972:414) point out their use as initiators, although as such retaining much of their original quality, judging by their examples. James (1972) discusses the difference in meaning between the 'interjections' ah and oh in sentence-initial position. In a statement such as Oh/Ah, you're leaving she finds that

140

both oh and ah reflect that A has just realized• that Bis leaving but that oh implies a note of surprise, whereas ah indicates that the information is significant. Anyway and now are generally classified as conjuncts and adverbials. Ilowever, Quirk et al _(1972: 674) state that concessive conjuncts, including anyway, indicate that 'an addition is being made to a process of reasoning' and simi~arly that now in its capacity as a logical connecter marks transition, ie the leading over to a 'new stage in the sequence of thought' (1972:667). This description corresponds to the use of anyway and now in the present data. Quirk et al also mention that items such as by the way are used for marking transition. Svartvik (1980) found that well was basically used as a 'social device', ie to establish and maintain social relationship, an opinion that he shares with Brown(1977). Added to this basic function, Svartvik mentions specific and optional technical discourse functions, such as hesitatbr, floorhold~r and initiator, and comes to the conclusion that well has two main functions which he calls 'qualifier' and 'frame'. As a qualifier, well is typically initial and serves as a link between previous and following discourse. As a frame, it is typically non-initial and embedded in the discourse and may open as well as close~ topic. (Note that my definition of is not identical with Svartvik's but that my are not very different from his qualifiers.)

DEFINITIONS OF INITIATORS I have indentified three variants of initiator, referred to as , , and ,.defined as follows (cf also chapter 4): mark a boundary in the discourse and introduce [Focus] or Q moves. They generally look forward and involve a shift in topic/aspect by introducing something new in relation to· the immediately preceding discourse. They can also be used to resume the thread of discourse after a digression. continue a sequence already in progres? and introduce R or F movis. They always look backward to ~~e immediately preceding discourse but also forward, since they reflect / the speaker's attitude to what. preceded or follows.

/

'~

141

serve to connect two successive speakers' utterances. They can initiate any of the moves in a Q/R exchange. Since they are 'less pronounced' variants of and they share qualities of both. They differ from and by being part of.the first TU of the move they initiate.

, , and are illustrated in (55)-(57): (55)

[Frame] Q

(56)

Q

R

(57)

...

/WELL# "/MAY I ask# /what goes !INTO that paper NOW# 1.1 :8 /what !is he NOW# /WELL# he's /just 'on [ei:] THE!OLOGY de'partment# 5.9:116

but it's /still in the :car at :King's :CROSS# - + Q well /can you can you 'not get CLOSER# [INF]

4.1:81

The [Frame] well in (SS) initiates a new beginning and indicates aspect shift. Well in (56) not only stresses the link between Q and R, but, as an R-prefix, it also signals B's attitude to the following Rand to the preceding Q. Well in (57) is similar to the one in (55) in that it marks an aspect boundary and to the one in (56) in that it links Q to the preceding move .. Note the softening effect of well in (57) which is not present in (SS). In principle, I regard initiators as acts whose function is a result of their position in very much the same way as when a Q· followed by another Q serves as a act (see section 7.1.2). The same items uttered in a non-initial position act differently: (58)

,,, but the /second 'time the 'cycle comes ROUND# [a:m] - . "/WELL# . you're be/ginning to !PETRIFY a bit# 3.6:68

The function of well in (58) is not obvious. Besides being a turnholder, it seems to have a framing effect. [a:m] ANV well COMPAREV' The voiced hesitation [a:m] was found as a Q/R initiator almost as often as well but there was an interesting difference in dis-

142

tribution; [a:m] was somewhat more common with Q than with R, and weii was much more common with R than with Q. This is reflected in Table 6:1. Table 6:1

Distribution of [a:m] and weii with Q and R ITEM

Q

R

TOTAL

[a:m] weii

103 47

96 184

199 231

TOTAL

150

280

430

In addition to the information in Table 6:1, there were two cases ' where [a:m] occurred together with weii in Q moves and 22 cases in R moves. One conclusion that can be drawn is that [a:m] is much more common than weii in exchange openings, while weii is a typical Rprefix. [a:m] in exchange-initial position can be compared to now, looking forward,whereas weii is comparable to oh, looking both forward and backward. '. Voiced hesitations were more common in telephone talk (54%) than in face-to-face conversation(46%). This is probably relat~d to the greater average number of Q/R exchanges in the telephone conversations, approxim~tely 70~ than in the face-to-face conversations, approximately 30 (not including text 11.1 with an extremely high number of Q/R exchanges and texts 5.8, 5.10, and 6.3 with very few). That speakers are more apt to hesitate in situations requiring an 'effort' could be observed in text 11.1, where the defendant in the law suit often hesitated before responding, and in text 3.3, where the university administr~tor used [a:m] to gain time when trying to formulate the Q move in the clearest possible way. Individual differences should of course not be disregarded. Except for the time-filling function, [a:m] and weii as R-initiators are not doing the same job from a pragmatic point of view. Compare eg (59) and (60): (59)

A: B:

Q were /YOU 'then# a/lone WITH 'her# - R [am] I was a/lone with MOTHER# /YES# /after my :wife left [am] 11.1:51

/

/

143 (6 0)

A:

B:

Q you /said you didn't 'know COLEMANS ((1 syll))# un/til you were your - /mother's SO!LICITORS# /THAT right# 'went to the BANK# is 11.1:1161 R /well NO# /mother "!TOLD me#

'''

[em] I was alone with mother yes in (59) is a perfectly acceptable Ron its own. The qualifying after my wife left is optional (for , see sec~ion 8.3) and [em] seems to be a neutral . Well in (60), on the contrary, not only gives the speaker more time, it also indicates that he is not going to provide an entirely satisfactory R (see section 8.1.1) and the , realized by mother told me is obligatory. Similarly, if [am] in (59) were replaced by well the would become obligatory, whereas [am] no would probably do as Ron its own.

VISTRIBUTION OF INITIATORS Table 6:2 gives the distribution of initiators as observed in the corpus. It shows that well is by .far the most common initiator of any move in a Q/R exchange. Its most prominent function is to serve as an R prefix. The next most common item, oh, is also an R-prefix in most cases. Table 6:2 presents only lexical items, so the occurrence of [a:m] with the different moves is not included. Table 6:2

Distribution of initiators

GROUP

ITEM

1

actuaZZy anyway apart from by the way hey in other words Zook now

FOCUS 2 2 1

3 2 10

2

ah and but oh

roTAL

I mean weZZ you know you see

R

F

INITIATORS

TOTAl

1 5 1 1 1 3 2 24

2 5 1 1 2 3 5 12

14

3 7 1 1 2 3 5 26

38

31

17

48

1 81 56 94 23

9 82 58 117 27

255

293

4 180 2 1

5 258 2 2

1

2

1

8

1 2 1

81 56 13 27

63

40

1 2 5 4

4

177

64

48

12

26

184

1 16

1 18

60

BO

3

Q

8 18

2 11 1 2

2 47 1

16

50

184

17

20

60

187

267

30

265

248

65

63

86

459

608

1

144

If the total number of initiators per move type is related to the total number of move types, the figures in Table 6:3 emerge ( followed by are not included). Table 6:3

Number of initiators per move type

TYPE of MOVE

TOTAL Nr of MOVES

FOCUS

INITIATORS per MOVE % Nr

R F

1297 1294 528

30 265 248 65

(39) (20) (19) ( 1 2)

TOTAL

3196

608

( 19)

77

Q

Not unexpectedly, initiators are most typical of [Focus) and least typical of F moves. Note that the figures for initiators with Q and Rare almost equal. The reason is that I have included and, but, and so which only initiate Q (and [Focus)) moves. Initiators in re-opening exchanges are not included; they hardly occur. Returning to Table 6:2, it appears that all items except in otheP woPds, by the way, so, and ah occurred with [Focus) moves, generally as , only occasionally with Q, but in this case much more often as than as . Only three items (not including I mean, which I find somewhat doubtful) oh, ah, and well, served to introduce Rand F moves, but they were, on the other hand, all the more frequent. Well as a was considerably more common than well as a . InitiatoP

+

[Focus]

(61)

[Frame) [Focus)

[a:) /LOOK# [ i ) . /DO. [aw) /I[?) /I[?) - 'felt I :really MUST get 'round ((the)) DEPARTMENTS# 3; 4: 1

(62)

+ [Focus)

/but - [am) . I was !wondering ACTUALLY# /I sup!pose that ,,, 9.1:362

(63)

[Frame) [Focus)

/WELL# \' ... ( (let me)) /try and . 'help you a !LITTLE about : THIS# 11 . 1 : 1 0 54/ /

145

The fact that look in (61) constitutes the very first TU in text 3.4 does not necessarily mean that it is the first thing A says. Part of the conversation may have preceded but has either not been recorded or not transcribed. But look definitely points forward to A's message. By contrast, both but in (62) and well in (63) 16ok back and tie up with what preceded before directing attention forward. The well, however, marks the boundar; between the previous· and the following discourse much more efficiently than the but. Initiator+ Q (64)

[Frame] Q

(6 5)

+Q

C6 6)

[Frame] Q

/NOW# /have you COM! PLAINED about 'this# and

3.3:330

/that was your first !MEETING#

/WELL# /Captain and Mrs KAY# /lived in a on their /OWN#

.

5.9.112 !FLAT# 11.1:678

The [Frame] now in (64) means something like now let's look at this from another angle and introduces a new stage in the discussion. The and in (65) connects a Q initiating a new exchange with the previous Q/R exchange. An alternative would be: wds this your first meeting, ie a non-conducive . Note the similarity between and this was your first meeting. and this was your first meeting, wasn't i t . There were no examples of and as a with Q·(cf but), whereas so occasionally served as i . The [Frame] well in (66) seems to imply let's proceed. From these examples, it appears that items belonging to group 1 introduce a new topic or a new aspect much more efficiently than those classified in groups 2 and 3. Consequently, it is hardly surprising that no items belonging to this group served to initiate Rand F moves, where initiators have very different functions. Initiator+ R (67)

A: Q B: + R

V

/wWere is THAT /going to be# /oh MILES# /right 'out [a:m] the GOLF course#

to:wards 6.2:958

146 (68)

A: Q B: + R

/are these *COPIES#* *well*. /THAT's a COPY#

\

1.4:437

Oh is emphatic and reinforces whatever state~ent is made in R. Not only docs it frequently reinforce polar decisions, oh yes and oh no, but it tends to collocate with expressions involving the speaker's assessment (see further p 147). WeZZ signals indirectness and inadequacy, ie the speaker's awareness that R will not provide exactly the information that the questioner was after. In (68) the question whether there are aopies is never answered; Q is only partly answered. Initiator (69)

A:

+ F

Q

B: R A: F (70)

A:

Q

B:

R

A: F

you'd /never 'think of :drawing a LINE 'down the 'page# be/fore you :START# I '/HAVE# I /do NOW# /AH# 6.2:238 /YES# /YES# [e :m] •'

/who /who /what SCHOOL ·do you WANT#

"/SOIC# - /WELL# I' 11 /TRY#

8.1:685

The function was found to be more typical of F moves than the function. I interpret ah in (69) as you're right I remember, indicating that A suddenly recollects what had just slipped his mind. Oh in the same place would have had a different effect; it would have served as a reinforcement in the same way as it does with R. WeZZ in (70) indicates restriint just ·as it does when an R prefix. If ah and weZZ had been instead of I take it that they would only have been less loaded.

COMMENTS. ON INITIATORS A few interesting points can be made about some individual items. As was pointed out by James (1972) referred to earlier (p 139), oh and ah do not have the same pragmatic meaning; oh implies surprise at realizing a fact while ah indicates that the information received is significant. And I mentioned (above) that they are not doing the same job in the discourse. Compare eg: \,

(71)

oh yes oh I see

ah yes ah I see

/

/

147

Oh serves to reinforce yes and I see. With ah it is the other way round; yes and I see are the reinforcing elements. Possible paraphrases are: oh yes oh I see ah yes ah I see

= =

'you're quite right' 'that's really surprising' I

I do understand'

·- 'that's really interesting'

Oh, more than any other initiator, occurred in collocations with a slight variation in function:

signalling reinforcement a) knowledge of yes and no

b)

expressions reflecting attitude or emotion:

2

emphasizing vagueness

3

signalling reaction or attitude a) followed by complete R:

b)

followed by ellipted R:

yes/no oh { I know/I don't know sure

oh

{

~~r~ry

good he was terrible I'm not surprised

I think oh { I WduZd think I heard it oh { it's/it was we get on quite well

because ... oh { by train six or seven

In addition to the difference in meaning between oh and weZZ mentioned earlier (p 146) I have found that oh is more common with Rs to yes/no-Qs, where it serves to reinforce the polarity decision, and that well occurs more frequently with Rs to WH-Qs, signalling indirectness of some kind. And, but, and so frequently linked a Q opening a new exchange to the preceding discourse, a feature characteristic of the communicative situation in which one speaker's utterance is a reaction to the previous speaker's utterance. Their specific functions differ: and agrees, but objects, and so concludes. Note that all the items listed as initiators can serve as , and that some of them, notably the items classified in group· 3, can serve as . The diagram in Fig 6:3 gives an overall idea of what initiators do:

148

TYPE OF INITIATOR

REALIZED BY

OCCURS WITH

SIGNALS

now anyway

[Focus] [Elicit]

topic/aspectshift

weti oh

[Response] [Follow-up]

reaction assessment

weti oh and but

[Focus] [Elicit] [ Resp'onse] [Follow-up]

same as and but weaker

Fig 6:3

,.

\

Initiators

In column two, I have listed the most common realizations of initiator per move category. Only oh and weii introduce all types of move. Whether they occur in a separate TU as and or in a non-separate TU as depends very much on the move concerned; [Focus] moves prefer , and F moves prefer , while are mo.re common with Q and R moves. Initiators other than oh and weii are bound to particular moves, as indicated in Table 6:2. The difference between and on the one hand and on the other is mainly a matter of emphasis. and are the more _loaded terms, reflecting more of the inherent meaning of the items involved. And the fact that they form a separate TU gives them considerably more weight. However, I regard as less neutral than , since they express the speaker's reaction an'd attitude, which do not. I also regard as more efficient when it comes to marking boundaries in the discourse. This is entirely in the nature of things; introduce new exchanges, _introduce continuing moves. 6.4

COMMENTS ON EXCHANGES

A survey of the exchange types that occurred in the corpus is presented in Appendix IV. It appears that the predominating pattern was {QR}, 55%, followed by {QR F}, 33%. Other patterns were much less frequent, eg {QR Qr R (F)}, 6%, and {Q Qc RR (F)}, 4%. ',,

.,,,,·

/ /

/

/

149

7 The Q move

The Q move can be described in terms of internal structure and interactive function in the discourse. It consists of one or more acts, minimally an obligatory PRIMARY act and an optional SECONDARY act. It initiates SUPERORDINATE eliciting and re-eliciting exchanges and SUBORDINATE checking and re-opening exchanges. Briefly, primary Q acts serve as elicitations and elicit an R; secondary acts serve as complements. The classification of primary Q acts into different types is based on 'what A wants B to do' and consists of four groups (Fig 7:1). Providing an adequate R (group 1) presupposes a literal interpretation of Q. Making a decision (group 2), on the contrary, presupposes an indirect interpretation of Qs which formally correspond to Q acts listed in group 1 but which are responded to in very much the same way as and .

GROUP

A WANTS B TO:

1

Provide an adequate R in terms laid down in Q

TYPE OF Q ACT

prepare the way for and introduce 'focal Qs' which initiate eliciting exchanges. There are two variants'. One, referred to as , occurs in the [Focus] move and was dealt with in section 6.3.1; the other, referred to as , serves either as an introductory Q move or is part of the [Elicit]. There are four subcategories: (a) (b) (c) (d)

PREFATORY PREPARATORY FRAMING PUSHED-DOWN

constituted by far the most common category. Preparatory are characteristic of rule-bouna dialogue, while most of the dialogues in this data involved casual conversation.

A is an act which is sub.ordinate either to the focal Q or to a pushe·d-down . Its primary function is to provide additional information, intended to help B answer Q. A secondary function seems to be to lead R in one direction or the other. The term 'clue' was introduced by Sinclair & Coulthar~, whose definition applies to classroom interaction and could therefore not be taken over directly (cf 1975:41). Examples (31) and (32) illustrate my definition: (31)

B: A:

B:

that /wasn't the !guy I MET# /WAS it# *[a:m]* 4.2:193 */when we* !SAW the BUILDING# -

The additiqnal information given in the , when we sa~ the building, serves as a specification of Q; it tells the answerer in more detail what guy is referred to, which makes it easier to respond. This is the primary function of the .

/

167

(32)

B:

Q

A:

R

well /what are we 'doing 'this WEEKEND# to/mOI'I'OW I've got . :a DANCING 'cZ.ass {in the MORNING#}# - foI' a [am] well I've /nothing DOWN anyway. {at /ALL#}# 4.1:15

The delimits the scope of R, and the answerer is expected to respond on the basis of 1 the information given. He cannot possibly disregard what is said in the . The two basic questions involved in the description of : 'Why did A add a in the first place?l and 'How does the affect R?' will be considered in relation to the following subfunctions: (a) (b) (c) (d) (a) (33)

SPECIFYING Q PROVIDING BACKGROUND INFORMATION EXPRESSING AN OPINION SUGGESTING AN R SPECIFYING Q A:

Q

C:

R

you /don't think 'Lambert'll be "HURT# /DO you# that [a:] that /I. /I. don't WANT +to+ TAKE one# +/NO#+ /[M]# 1.8:558

By adding the A explains and specifies Q which is rather vague on its own and may lead to misunderstanding, since speaker C cannot be expected to follow A's train of thought exactly. Specifying are often introduced by I mean, you know, or equivalent expressions. (b) (34)

PROVIDING BACKGROUND INFORMATION A:

Q

B:

R

[a] /what did it look 'like !THEN# because [n] /now they've 'aZ.Z. moved :OUT# and /moved to a diffeI'ent !PLACE# ((do you /know I'm not [?] I'm)) - - /don't think I can REALLY REMEMBER# /very CLEARLY# 5.9:772

The answerer does not always need the information given in the , on the other hand, do not seem to be ~pecific~lly B-oriented. Non-specifying often have a purely social function; they contribute to makipg the conversation more lively, personal and relaxed. But they may also have a restrictive effect on R by leading it in a certain direction. and acts were more frequent in some conversations than in others. Four out of 16 '[Elicits] in text 6.3, the radio interview, for instance, contained a or a or a combination or both. and in order to make his point clear, but when talking to a group of staff, he avoided being over-explicit by leaving out than II! ! THIS# oh /CERTAINLY# */YES#* */YES#* 3.4:376 /YES# /how many 'people in FACT# /float IN and 'out of the DEPARTMENT# [ a :ml /oh you've :met them !ALL NOW# 1.5:722 ((oh)) I've /met them ALL# /YES#

In (42) the addressee's comment on what the R move revealed is paid attention to; in (43) it passes unheeded. There are probably many reasons wh·y this is so, but one obvious reason is reflected in the prosodic markings which indicate much greater emphasis and surprise in (42) than in (43). The Qr move is similar to the F move in that it marks A's attitude to R. What distinguishes the two is that Qr re~opens the exchange and elicits an R while F closes it; Fis terminal, Qr is not.

I

176 \

'

Qr and Rare partly realized by the same,items, eg really, oh, and tags. Whether such items ·act as Qr or F depends on the way they are said, reflected in prosody; the speech situation and/or the people involved may play a more or less important role. Items realizing acts express: • •

emotional evaluations: that's dreadful, isn't it? queries equivalent to Did you say?/Do you mean? r~ally, n6, tonight surprise: oh did he?, are you indeed? tentative conclusions equivalent to you mean ... ?

• •

In (44) the Qr function is unlikely, since a missing ratification by B would not be felt as absent: (44)

B: C:

R

B:

/comes 'down 'WEEK:ENDS# /cor !BLIMEY /what a {LONG} :WAY# ((it)) "/IS# it's /DIFFICULT# /YES# -

8.2:683

Similar sequences have'.been analysed as FF and not as Qr R. The pragmatic meaning of Qr moves varies along a scale, from query at one end to acknowledgement at the other, with a corresponding decrease in elicitative force, as illustrated in Fig 7:12. The corpus contained 82 Qr exchanges in all. query +R

Fig 7:12 7. 3

acknowledge

-R

Scale of elicitative force in Qr

SUMMING UP THE Q MOVE

Table 7:4 indicates that and were used' roughly to the same extent in both channels, that and occurred more often in face-to-face conversation, which also contained the only instances of , and that was more often found in telephone conversation. The larger number of in face-to-face conversation is probably due to the communicative situation as such, which seems to favour this type of Q, eg to get opinio~s confirmed or agreed upon, or simply elicit an R from the other' party so that ' the conversation can go on. The larger number of in face-to-face talk seems, strangely enough, to point to a/'

////

Table 7:4

Distribution of Q types in all conversations

TEXT CATEGORY FACE-TO-FACE INTERVIEW EXAMINATION TELEPHONE

174 7 15 83

TOTAL

279

Table 7:5 TONE

(24%) (44%) (10 %) (23%)

256 6 62 132

(36%) (37'!,) (43%) (37'!,)

(23%)

456 (3.7%)

205 3 65 73

(28%) (19%) (45%) (20 %)

68

(9 %)

5 ( 1 %)

10 (1%)

1 15

(1%) (4%)

2 (1%) 47 (13%)

8 (2%)

346 (28%)

84

( 7'l,)

54 (4%)

18 (1%)

TOTAL

3 (0%)

721 16 145 358

3 (0%)

··1240

Pitch contour per Q type kQ:permit>

RISE FALL LEVEL

30 245 4

(11 %) (88%)

TOTAL

279

(100%) 456 (100%) 346 (100%)

(1%)

277 (61 %) 172 (38%) 7 (1 %)

94 (27'!,) 250 (72%) 2 (1%)

TOTAL

52 26 6

(62%) (31%) (7%)

33 (61 %) 21 (39%)

10 (56%) 7 (39%) 1 (5 %)

2 (67%) 1 (33%)

498 722 20

84

(100%)

54 (100%)

18 (100%)

3 (100%)

1240

~

--..

178

greater demand for oral feedback in a situation where people can see each other. But the presence or absence of oral feedback in telephone conversation is most certainly subject to the particular type of call; social calls were too few' in the corpus to make a comparison possible. The occurrences of in the telephone calls were switchboard requests, and the three instances of in face-to-face conversation constituted the first pair parts of pre-sequen~es (cf section 7.1.2). The different distribution of Q types in the radio interview and the courtroom examination is intTresting; 44% of the Qs in the interview consisted of but only 10% in the examination, which should be compared with the figures in the third column which indicate that 45% of the Qs in the courtroom consisted of . On the one hand, the figures seem to reveal that the interviewer's Qs were neutral and objective, on the other, that the Qs in the courtroom, the majority of which were conducive, were aimed at putting the defendant in a disadvantageous position. This is not altogether true, however, considering that the courtroom Qs were recapitulations of the previous day's questioning so that mdst of the Rs ·could be anticipated. Table 7:5 shows that a falling tone dominated for and that a rising tone dominated for . This was expected. The 61% rising tone for was less expected, since they are already marked by interrogative word-order (cf Crystal 1969_:4). And, surprisingly enough, as many as 72% of the had a falling tone (cf Table 7.1 which shows that 23% were falling even without lexical markers). Another interesting finding is that 31% of all served as such even with a falling tone (cf Table 8:8):

/ /

/

/

179

8 The R move

8.1

PRIMARY ACTS

Rs can be described in terms of WHAT B responds TO and BY WHAT B responds.· This i~ illud'trated in Fig 8: 1. WHAT B responds TO

BY WHAT B responds

SUBCATEGORY

Q complement

ANSWER

Q performative

REPLY

comply



Fig 8:1

R procedure

Rs directed to the complement constitute ANSWERS; Rs directed towards the performative ar~ REPLIES, as described in section 3.4. Answers are realized by complies (cf Fig 8:3) and , Replies by ~nd , whereas , which -occupy an intermediate position, have mostly been analysed as Answers. As will appear, however, it is sometimes .difficult to draw the line between different R choices, especially between and (indicated by overlapping brackets in Fig 8:1), sometimes also between and . This is due to the analysi~s position as a judge 'after the fact'; there is no reliable way to decide whether or not Bis sincere, although it seems that most speakers cooperate. In what follows I will take it for granted that the speakers are both honest and cooperative unless there are obvious signs t; the contrary. The way B responds depends not only on his ability but also on his willingness to furnish adequate infor~ation. This is illustrate.cl in Fig 8:2, which is a generalization, but as such quite helpfui.

/

180

R ACTS

ABLE

aomp1..y



Fig 8:2

WILLING

INFORMATIVE

+ + +

+ +

+ +

-

!

(!)

!

!

EXPLICIT

+

MANOEUVRING

+

+ +

Factors affecting R

I will return to Fig 8:2 and its applications when discussing the different R procedures. aomp1..ies give adequate information explicitly:

(1)

A: B:

Q /WHEN is it# R /four THIRTY {TO/MORROW#}#

1.4:1118

give adequate information implicitly: (2)

A: B:

Q /do you want 'people ~o come to the :REGISTRY R

'office# /not.MANY#

7.3:785

provide inad~quat~ information (3)

A:

Q /so . so /what 'happened about·:buying the 'house in !HIGHGATE#

B:

R

'''

/we1..1.. "I'm !sti1..1.. 'waiting to 'hear from the :COUNCIL# 7.3:366

are conscious avoidance manoeuvres ~hich also give inadequate information: (4)

A: B:

Q [a:m] well /have you any OTHER {SUGGESTIONS#}# R we1..1.. he /didn't GIVE me any#

8.2:1105

do not even try to answer Q but are overt declarations that B ignores R or does not know how to respond: (5)

B: A: (B: A:

Q

R

+((but /would))+ you +((1 syll))+ !get !that !UP 'any'where# . /I don't KNOW# .

1.4:531

Whichever way B chooses to respond, by a aom~1..y, an , etc, the R act may be expanded by an ELABORATio'N. There .are three

/

/

181

very different types of elaboration which I will call , , and (see further, . section 8.2). Thus , the R move consists of at least one R act, a primary act, which may be accompanied by one or more secondary acts, elaborations. Prim~ry acts respond to Q; elaborations are optional additions, as illustrated in (6): (6)

A:

Q

B:

but - do /you 'think 'that IS# **or /is it !TOO. {/WILD#}#** **/its# [i] it's a /bit** STRiDENT# but [i] it's a /kind of :THING# that /looks /doesn't look 'bad on C(the)) 'bare WALLS# 1.4:793

In the following description of R strategies with the different Q categories (Fig 8.:3) I shall start with aompZies and finish with . Elaborations will, in principle, be dealt with separately. Only aompZies and provide the information required, the former explicitly, the latter implicitly. They do so because Bis both able and willing to give an appropriate answer. and , on the contrary, only pretend to answer Q; in reality, the exact information is not provided. are inadequate in that they either do not 1each 'the point' by giving too little information, or are beside the point by giving information that is not really asked for. In most cases, this does not seem to be due to B's unwillingness to answer but rather to his inability to produce an R. that is exactly to the point. In the case of , Bis either able but unwilling to answer, or'he does not want to confess that he does not know. As to , finally, B may be regarded as willing but not able to answer (cf Fig 8: 2) . The term 'comply' is used as an umbrella term for appropriate (direct and indirect) answers to , , , and (the latter is included although can hardly be discussed in terms of direct/indirectness).What constitutes a aompZy depends on the nature of the preceding Q, as illustrated in Fig 8:3. The subcategories of R listed in Fig 8:1 will be discussed in relation to the Q categories listed in Fig 8:3. Complies will then be dealt with as , , , and , respectively.

/

182

Q CATEGORY

TYPE OF R







Pig 8:3

\

Compliea

Rs to ALT Qs, which respond to and , will be dealt with separately (section 8 .1. 3). Qs with a double function, ie Qs whi.ch can be interpreted literally eg as and indirectly, eg as , are typically realized in the same way as and , and even . Since Rs to such Qs are typically yes and no, they will be dealt with in the account of . Such answers as do not require assumptions other than those entailed by Q itself. (cf Lakoff 1973:456) will be called complies. Rs that answer Q but do so in a vague fashion will be referred to as indirect complies. Direct and indirect complies are exemplified in (7) and (8) in the form of : (7)

A:

Q

B: (8)

/what in 'fact did you :DO# if you /hadn't 'done - [a:m] - 'English since *0- 1 level#* 1 • 5: 99 */maths and* PHYSICS#

,I /mean. [a] "/when would I 'have to sort of COM!MIT my'self {to /THIS#}# [a:m] A: [a: m] - (B: /cos it would DE!{PEND what 'else came} :UP /YOU know# A: /SURE {YEAH#}# [a:m - - a:] /WELL# /I would 'only 'say. [5i] (clears throat) the !sooner the "!BETTER# 3.2:354 B:

Q

The in (7) supplies exact information and is therefore direct, while the vague time indication in (8) places this in the indirect category. Note that the R move in (8) .consists of two acts, the first of which, sure yeah, comments on the part of the eliciting move, cos it would depend ... , and does not answer Q. That the answerer is uncertain. is obvious. Not only does he hesitate but he also begins his utterance with well, ' '

..

/

/

183

which is a marker of indirectness with all types of R. (9)

A:

Q

B: A: B:

((where /ARE you#)) ; [Ml# where /ARE you# . /weZZ I'm :stiZZ at :COLLEGE# I'm /just going to go down to

:SOHO# 7.2:481,

(10)

A:

Q

B:

[w] Jwhat 'time do you :bSUALLY 'go and have. {LbNCH#}# . /WELL#. I /USUALLY 'go a'bout# /quarter 'past !TWELVE# /something Zike !THAT# 7.3:1405

On the surface, the in (9) and (10) appear to give all and only the information asked for. I regard them both as indirect, however, following Lakoff (1973) who found that weZZ is used when B feels that R is insufficient, either because A has to fill in information on his own or ~ecause Bis about to give additional information. In (9), A doe.s not get the answer he expected. The word stiZZ indicates tnat Bis expected to be somewhere else; he probably feels that he is not in a position to provide the R that A had expected. This is anticipated by weZZ, as it were, and to make up for the insufficiency B gives additional information through a act, I'm just going ... Qin (10) is a in disguise. According to the context what A is mainly interested in is not'B's usual lunch time but rather whether B would like to have lunch with him. In other words, what literally is a is really a face-saving device, a pre-Q to an invitation (cf section 7.1.2). WeZZ implies something like I see what you're driving at ... Similarly, ellipted Rs which are felt to be insufficient are initiated by weZZ: (11)

A: B:

/where would you 'be between 'mid Ju'ly and 'and [a:] the 'end of :ADGUST#. /{WELL} .in !HU.DDERS'FIELD# /OR#. in /COVENTRY# 3.2:367

where .A is given a choice.instead of pretise information, and in: (12)

C:

A:

/what !sort of QUESTIONS# [am] - - - weZZ a /Zot of [am] - - - 'aZose questions on history of ART# . 1.3:316

184

where the information is too vague. But generally, ellipted Rs are inherently direct, since they do no more than 'fill the gap' \ by replacing the Q-word. like may be initiated by oh: (13)

B: A:

/how're you !F~ELING# /OH 'HELLO# - /e /open FROM# /ten o'c7,ock untii :si:: o'c7,ock "WEEK# 'du:r>ing the */YUP#* *((and then))* on /SATURDAYS# /f:r>om !ten o'c7,ock 8. 2: 46 'unti7, !!TWO o'cfock#

.

Q it Id be /GREAT# /WOULDN° 1 T it# R I [Ml# -

- 4 .1. 1

Let.us consider the different realizations of , ordered from 'least certain' to 'most certain' and followed by minimal but unambiguous Rs, illustrated in Fig 8:7. Assuming that the paraphrases are correct, Paui went to YORK appears at either end of the·interrogative scale. This implies that' falling and rising tones have a stronger contrastive effect on declarative utterances than on tag-Qs, so that Paui went to YORK shows the highest degree of uncertainty and asks' for affirmation whereas Pau7, went to YORK reflects the lowest degree of uncertainty and only requires acknowledgement. I wish to emphasize that Fig 8:7 is not intended to ~emonstrate that pitch contrast always distinguishes a Q from a non-Q at least not in a larger linguistic context; it only marks the difference in degree of certainty. As to realizations of R, minimal mas a signal of acknowledgement/agreement is sufficient in (f), and possibly in (e); the he did is added as a disambiguator in\~c), and probably in (b) . /

/

215

LIKELY POSITIVE R

IMPLICATION

Q

+Q-ness

YES

a) Paul went to YORK

did he go to York

b) Didn't Paul go to YORK

I thought he went to York

YES (he did)

c) Didn't Paul go to YORK

I'm pretty sure he weiit to York

YES he did

d) Paul went to YORK DIDN't he

I think he went to York

YES

e) Paul went to YORK DIDN't he

I'm sure he went to York

YES/m

f) Paul went to YORK

so he went to York

'

-Q-ness

m

FUNCTION affinn

'

acknowledge

Fig 8:7 Rs to : from least to most certain I am fully aware that Fig 8:7 is a generalization. Compare eg (98) with (c) above: (98)

A:

Q

B:

( . laughs) - d'you /not !THINK *so* *[k)#* /[~]#I/see what you M~AN# 4.1:219

which shows that what type of R is unambiguous is not just a matter of Q form. M would hardly be taken for no, due to lack of a disambiguating I don't, even if the had been missing (for conduciveness, cf section 3.3.5; see also Stubbs 1983:110ff for Rs to negative-Qs). Table 8:6 shows the distribution of yes/no~ . Considering that there is no ca~se for ambiguity in cases where the polariti oi .R echoes the polarity of the preceding Q, it is worth noticing that amplified Rs were nevertheless quite frequent in such cases. This is particularly apparent in the POS + NEG categories, with almost as many instances of yes alone as yes+ .

-

216

Table 8:6

Direct Rs to (yes/no+ expand excluded)

Q~

...:i

u

µ.:i

A

-

POS POS NEG NEG

'yes'/'m'

'no'

YES alone +amplify

NO alone +amplify

DUCL DUCL DUCL DUCL

14 28

TOTAL 19 44 4 10

5

13

1 3



2

1

5

2 1 2

;

Ul

POS POS POS POS

+ + + +

NEG NEG NEG NEG

2 12 3 1

2 11 1

4 23 4 1

POS + POS

7

1

8

O' I

c.:,

relation is rather complex compared to the relationship between positive a,nd negative . are either positive or negative and realized by yes or no depending on which of them is true in the particular case. and are both realized by yes or no depending on 10 which is true or not true, 2) the polarity of Q, and 3) the bias of Q which may or may not correspond to the polarity. For the sake of simplicity, let us consider some invented, decontextualized examples with a falling tone as a marker of certainty. If Q has positive polarity and bias (expressing A's assumption), the (= agreement) is yes and the (= disagreement) is no: (103)

Q

,. µ 1-,

ol

p, I

0 ;?:

E,-< µ.i

u

< ~ I

0

µ

I µ.i

u

< ~

>,.

µ

1-,

ol

p,

.... I

µ

,-