iguistics AN INTRODUCTION

d
ingua<
udy
4ERAL EC ED14 4
IRSE ANALYSIS
CANDLIN

Malcolm Coulthard
For Beth
 An Introduction
to
Discourse Analysis

MALCOLM COULTHARD

ppp
PUP
TennesW^^WA library
Cookeville, Tenn.
";ste72
LONGMAN GROUP LIMITED
London

Associated companies, branches and

representatives throughout the world

(C) Longman Group Ltd 1977

All rights reserved. No part of this

publication may be reproduced, stored
in a retrieval system, or transmitted in
any form or by any means, electronic,
mechanical, photocopying, recording or
otherwise, without the prior permission
of the Copyright owner.

First published 1977

Second impression 1978

ISBN 0 582 55087 4

Printed in Hong Kong by

Wing King Tong Co Ltd
 Contents

page
Preface ix

Chapter 1
Introduction 1

Chapter 2
Speech acts 11

Chapter 3
The ethnography of speaking 30

Chapter 4
Conversational analysis 52

Chapter 5
Classroom interaction 93

Chapter 6
Intonation 116

Chapter 7
Discourse analysis and
language teaching 138

Chapter 8
The acquisition of discourse 154

Chapter 9
The analysis of literary discourse 170

References 183

Index 191
 Acknowledgements

Much of the material in this book was originally prepared for

undergraduate and postgraduate seminars in Discourse Analysis,
Sociolinguistics and Psycholinguistics, and subsequently modified in
the light of students’ comments and criticisms.
During the past four years I have benefitted greatly from discus¬
sions with my colleagues in English Language Research, particularly
David Brazil, Tim Johns and Roger Pearce. I owe to them my
perception of many of the crucial problems of discourse analysis
and my understanding of certain difficult concepts in the writings
of Austin, Halliday, Hymes and Searle. I am also indebted to
Hazel Hanlon and Jackie Glover for transforming an often illegible
manuscript into an almost perfect typescript.
My greatest debt is to John Sinclair, teacher and friend for the
past nine years, through whom I came to work in the field of
Discourse Analysis and on whose suggestion I wrote this book. He
is a constant source of insights and ideas, most of which I have
attempted to acknowledge but he will no doubt discover several in
this book which I thought were my own.
Nancy, May 1976
 Preface

There are a number of important reasons why this is a valuable

contribution to the Applied. Linguistics and Language Study series.
Firstly, there are already a number of books in the series for which
the present volume acts as a necessary foundation. David Crystal’s
and Derek Davy’s Advanced Conversational English as well as
Gillian Brown’s Listening to Spoken English both exemplify some
of the points treated in the chapter on Intonation, while Henry
Widdowson’s Stylistics and the Teaching of Literature has an obvious
connection to the final chapter, though more generally concerned
with the processes of interpretation emerging from the critical
evaluation of ethnomethodology as a central theme in the analy¬
sis of discourse presented here. Furthermore Alan Mountford’s and
Ronald Mackay’s forthcoming edited collection of articles in English
for Special Purposes rests theoretically and descriptively on the ideas
and research introduced and commented on in the first six chapters,
and will need to be assessed against the standpoints of chapter
seven. Discourse Analysis and Language Teaching.
Secondly, and supporting that general concern with relationships
in discourse, the connections between utterances and their inter¬
preters, which the books cited above reflect, we have a current
upsurge of linguistic and applied linguistic interest in supra-sentential
analysis. This is not in itself new, as Malcolm Coulthard’s timely
‘rehabilitation’ of J. R. Firth makes clear, nor is it limited to
linguistics in the narrow sense as sources in anthropology (especially
ethnography), philosophy, sociology and psychology all confirm. It
is significant that in a desire to establish a framework within which
to interrelate linguistic form, semantic interpretation and pragmatic
use for the more complete understanding of communicative com¬
petence, we are driven outside the narrow confines of linguistic
structure. Coping with what Coulthard refers to as the ‘time-bomb’
of meaning in his Introduction has fortunately and appropriately
ix
X PREFACE

compelled researchers into human communication greatly to extend

their frame of reference to encompass Prague, Halliday, Labov,
Argyle, Grice, Austin. Searle, Goffman, Hymes and Wunderlich as
well as the recent research of John Sinclair and Malcolm Coulthard
himself in Birmingham. It follows that the subject demands a wide
focus if we are going to make sense of the ways we employ to
diminish the unpredictability of what people mean when they engage
in interaction.
Thirdly, it is central to applied linguistic concerns that the three
crucial problems for discourse analysis isolated very clearly here,
the number of ‘units’ it will be found necessary and sufficient to
propose for a theory of discourse, their respective ‘size’, and among
them the major question of the relationship between form and
function, between ‘grammaticality’ and ‘meaningfulness’, all derive
from the practical task of describing data. As in the case of the
author’s own work with teacher-pupil language, or our own
Lancaster work into doctor-patient communication, this is often in
an applied linguistic cause, though not necessarily a language teach¬
ing one, as the final chapter on The Analysis of Literary Discourse
underlines.
Finally, as with any exciting contribution to a field where many
workers are now engaged, this book is not the last word; one of its
attractions is the way it sparkles with suggestions for future research,
both in applications and in theory.
In the light of these reasons, and in particular of the last, I
should like to concentrate the remarks of this Preface on the impact
of the book on foreign language teaching curriculum design and
methodology.
The last few years have seen an explosion of curricula and
materials with the laudable aim of introducing learners to ‘real’
acts of communication, and premissed, however indirectly, on the
features of discourse portrayed in this book. The explosion has
been marked, however, by notable confusion, inconsistency and
miscategorisation precisely in respect of the three problem areas
referred to earlier. It is obvious that until the questions of ‘size’,
‘quantity’ and ‘realisation’ are adequately met we shall have to
live with uncertainty, but we can, I believe, already draw two major
consequences for language teaching. The first has to do with
curriculum design and derives from the necessary distinction be¬
tween ‘functional’ and ‘discoursal’ value implicit in the ‘rules of
discourse’ discussed here. It suggests that the item-bank of ‘notions’
 PREFACE XI

or speech-acts linked to an assortment of surface structure types is

as incomplete as a traditional list of grammatical patterns as the
basis of language teaching materials aimed at developing conversa¬
tional competence. Of course learners have to be able to construct
grammatically well-formed sentences and be aware of intra-
sentential semantic equivalence, but they also have to manage the
pragmatic values that utterances take on within a discourse. For
that learners need what we might call encounter syllabuses where
they are progressively sensitised to, say, Hymesian components of
speech events so that they can themselves analyse and compare the
realisations of encounter types, both in the foreign language and
between the foreign language and their mother tongue. In order,
for example, to interpret whatever act (let us say suggestion) fills
the discourse slot INTERROGATE in our materials for doctor-
patient communication, the learner has to understand the coherence
of the particular speech event called a consultation. Insofar as he
does not have this knowledge, as with many native as well as non¬
native patients, the sentence underlying the act may be processed
superficially, even the suggestion recognised, without the INTER¬
ROGATE being perceived, with a resulting discoursal incoherence
in terms of patient response. What is more, this sensitisation cannot
merely take the form of a multi-media display; the rules of discourse
are not items to be learned like some incremental list of words,
structures or even simulated events where one rehearses what one
has seen or heard. The meanings of particular communicative
strategies like the meanings of utterances themselves, have to be
negotiated. They involve ‘work’ as the ethnomethodological dis¬
cussion in chapter four and the reappraisal of first language acquisi¬
tion processes in chapter eight makes clear.
To take two practical examples: in the forthcoming multi-media
programme for English Language Teaching entitled Challenges
(Longman/BBC/Institut fur Film und Bild) 1977 the following
types of exercise appear in the Unit ‘’Something to say':

Chain G Step (iv) Note-taking:

Listen to the tape. You will hear extracts from people
talking. You want to decide WHO is talking to
WHOM, WHERE and WHEN.
Chain H Step (iv) Matching:
People from different parts of the world speak English
in different ways. But the same person speaks English
Xll PREFACE

in different ways too—according to the person he is

talking to, for example.
Listen to the tape.
The same man is asking permission from four dif¬
ferent people to do something.
Who do you think he is talking to in each case?
Chain H Step (v) Listening:
We also talk differently to people according to the
event. Listen to the tape. The same two people are
talking, but the event is different each time.
What do you think the event is in each case?

The above exercises call upon the learner’s sociolinguistic awareness,

and refine it in relation to authentic data presented both on audio-
tape and on colour film. The authors see this as an indispensable
foundation for understanding discourse, and for enabling learners
to make sense of themselves in conversation with others.
A second example comes from recent material in experimental
use in some German comprehensive schools (Gesamtschulen), and
reflects the present concern with conversational strategy by the use
of so-called ‘discourse chains’ as shown on the opposite page.
The second consequence has to do with methodology, and relates
directly to the idea of negotiation of meaning and interpretation
suggested above. There is indeterminancy in discourse, and the
designer of materials and the language teacher cannot hope to
explain discoursal meaning in the traditionally regarded ‘cut-and-
dried’ way of treating grammatical rules. He can only begin by
conventionalising, using examples selected on the grounds of their
probable discoursal interpretation in this or that event. These
‘expression rules’ are different in kind to the traditional rules of
grammar not only in that they are not categorial but variable, but
more importantly in that they imply psycholinguistic processes
whose nature is still unclear. We know enough, however, to realise
methodologically that we must avoid a latter-day ‘structuralism’ of
concepts, and site utterances firmly within connected discourse.
Furthermore ways of teaching should shift from teacher-telling to
learner-interpreting within a syllabus whose prime goal is the
development of strategies for discourse processing, rather than an
assembly of items. We may wish, of course, for pedagogic reasons
to stagger our introduction of indeterminacy, but sooner rather than
later learners need to take a leaf out of the ethnomethodologists’
 PREFACE Xlll

book and begin glossing as a learning procedure. We should look

at grading as a process of gradually increasing the questionability of
meaning, and change our teaching tactics accordingly.

Announce Raise an Counter the Object Play down

an intention, objection objection again the argument Agree
make a
suggestion Where is it ? Oh, it’s not You don’t Pah, Well, I
Oh, in Essex. far. You can know my you’re not suppose I
That’s too be there in parents. My a kid any could try.
far. under an dad would more.
hour. have a
Say Goodbye.
X I’ll go and ask Mary, then.

Agree Express Fix a precise All right. See you then.

enthusiasm. time.
I’m going to Fix a date.
a pop
festival on Where is it ? That’s great. Let’s say
Saturday. Oh, in Look at all on Sunday
Do you
fancy
Essex, that’s
not far. We
these groups. 8 p.m.
When shall
L
coming? can be there we meet ? Make other
in under an suggestions. Agree.
hour.
/ 8 is too Right, see you

\
early, 9 at 9 then.
would be
better.
\

(Taken from: Protokoll der 7 Arbeitstagung Bundesarbeitsgemeinschaft Englisch an

Gesamtschulen: “Baukastein Diskursstruktur und Obungstypologie”
Hessisches Institut fur Lehrerfortbildung. Fuldatal, Kassel, 1975.)

Learners need to become analysts of discourse themselves, and

in confronting a foreign language we should help them by encourag¬
ing a use of existing discoursal awareness in their mother tongue
while providing them with a workable model of analysis for the
organising of the data. In addition to appreciation of the setting, the
discourse topic and the discoverable presuppositions of the partici¬
pants, learners have to see through referential meaning to the dis¬
coursal significance of shifts of stress and key, kinesis and stereo¬
typical strategies of conversation. It is here that chapter six on
Intonation, with its discussion of Brazil’s work, is of greatest
pedagogic significance. It should be part of language learning to
realise the communicative potential of linguistic form by a training
XIV PREFACE

in elicitation procedures for the discovery of possible interpretations,

in appreciation of Gumperz’s comment that ‘by signalling a speech
activity a speaker also signals the social presuppositions in terms
of which the message is to be interpreted.’ We can put discovery
methods to good discoursal use by introducing into our materials
the kind of questioning Gumperz proposes for his interethnic com¬
munication studies:

1. What is A trying to achieve by talking in this way?

2. What is it about the way he says it that makes you . .. ?
3. Could he be trying to ... ?
4. How should he have said it if he wanted to ... ?
5. How did B interpret what A said?
6. How can you tell how B misunderstood?
7. How should B have replied to show that he did understand?

Such questions can be supplemented by others designed to probe

the connections between our selective perception as hearers/readers
and the actual cues present in the utterances. At the very least this
would stimulate an analytic attitude to the language under study.
In our experience with both adult ESP students and secondary
school pupils, without this awareness there is no ready link between
‘doing things with words’ and ‘achieving things with words’, and
incidentally little hope for an understanding by the learner of those
communicative ‘errors’ which cannot be accounted for in terms of
surface form.
In pursuing this second methodological consequence, however,
we have to come to terms with a novel dilemma in language teach¬
ing. With sentence meaning in ‘structuralist’ materials we were
protected as teachers from having much social influence, but now
with the need for discoursal insights into human interaction we
have become intimately involved in manipulating human behaviour.
The social implications of communicative language teaching are
such as to provoke concern about the establishment of norms.
In concluding this Preface, let me underline that the value of the
book lies not only in its measured and informed evaluation of the
components of discourse and its possible applications, but as much
in the way that it does not fudge the issues of description and
theory which are still to be solved. Current interest in discourse,
which this Introduction reflects, is notable in the way it ranges not
only across specialist areas within linguistics, but beyond to other
 PREFACE XV

disciplines concerned with describing and explaining human com¬

munication. The crucial contribution has been to have seen the
connections and to have suggested a synthesis.

Christopher N. Candlin, March 1977

General Editor
 1

Introduction

Although it is now forty years since J. R. Firth urged linguists to

study conversation, for

it is here that we shall find the key to a better understanding of

what language is and how it works,

the serious study of spoken discourse is only just beginning and

currently much of the work is being undertaken not by linguists
but by sociologists, anthropologists and philosophers. The explana¬
tion is not hard to find. While all linguists would agree that human
communication must be described in terms of at least three levels—
meaning, form and substance, or discourse, syntax and phonology—
there are disagreements over the boundaries of linguistics.
Firth (1951) asserted that ‘the main concern of descriptive
linguistics is to make statements of meaning’. Part of the meaning of
an utterance is the result of contrasts in the levels of phonology and
syntax, and Firth accepted that in order to isolate meaningful con¬
trasts in these levels ‘we make regular use of nonsense in phonetics
and grammar’, but, he argued, language is fundamentally ‘a way of
behaving and making others behave’ and therefore ultimately the
linguist must concern himself with the ‘verbal process in the context
of situation’. For Firth language was only meaningful in its context
of situation; he asserted that the descriptive process must begin
with the collection of a set of contextually defined homogeneous
texts and the aim of description is to explain how the sentences or
utterances are meaningful in their contexts.
Firth himself did not in fact explore the relation between form
and meaning and his exhortations to others were ignored because
Bloomfield led linguistics away from any consideration of meaning
to a concentration on form and substance, by observing that
linguists ‘cannot define meanings, but must appeal for this to
students of other sciences or to common knowledge’ (1933). The
2 AN INTRODUCTION TO DISCOURSE ANALYSIS

utterance ‘I’m hungry’ could be used by a starving beggar to request

food or by a petulant child to delay going to bed; Bloomfield argued
that linguistics is only concerned with those phonological, lexical
and syntactic features which the utterances share—he felt it was no
concern of linguistics to explain how identical utterances can have
different functions in different situations, nor how listeners correctly
decode the intended message.
For a generation American linguists concentrated massively and
highly successfully on problems within phonology and morphology
—on the existence of the phoneme and the validity of unique
phonemic descriptions; on discovery procedures for isolating
phonemes and morphemes in languages not previously described;
on the mechanical identification of morpheme boundaries and word
classes. When Chomsky redirected linguistics towards the study of
sentence structure, the concerns were still pre-eminently with the
formal features of language:

the fundamental aim in the linguistic analysis of a language L

is to separate the grammatical sequences which are sentences
of L from the ungrammatical sequences which are not sen¬
tences of L and to study the structure of the grammatical
sequences. (1957)
In arguing the independence of grammaticality from meaningful¬
ness Chomsky produced the most famous example of ‘nonsense’ in
linguistics—‘colourless green ideas sleep furiously.’
Earlier linguists, while concentrating on formal aspects of
language had used collections of speech or writing as a source of
examples. Chomsky suggested that not only was a corpus un¬
necessary, it was actually counterproductive. No corpus, however
large, can be adequate because it will never contain examples of all
possible structures and will actually contain misleading data,
performance errors, caused by

such grammatically irrelevant conditions as memory limita¬

tions, distractions, shifts of attention and interest and errors
(random or characteristic) in applying knowledge of the lan¬
guage in actual performance.
The prime concern of linguistic theory, Chomsky argued, is with
the underlying knowledge, the competence of the ideal speaker-
hearer. The underlying competence is the same for all native
speakers and therefore can be studied in the productions of any
one individual, usually the linguist himself who proceeds by
 INTRODUCTION 3

introspection, checking potential sentences for grammaticality

against his intuitions.
The insights achieved by transformational grammarians were
enormous but as time passed the problems became more serious. It
became evident that there was not in fact a uniform native speaker
competence; it became necessary to talk of degrees of gram¬
maticality or acceptability; crucial examples were attacked as
ungrammatical and defended as ‘acceptable in my idiolect’. Mean¬
while the timebomb meaning was ticking away: in the late 1960s
Ross, McCawley and G. Lakoff began arguing that one cannot in
fact describe grammar in isolation from meaning, that powerful
syntactic generalisations can be achieved by making lexical inser¬
tions at an early stage in the generation of a sentence. By 1972
Robin Lakoff was arguing that
in order to predict correctly the applicability of many rules one
must be able to refer to assumptions about the social context of
an utterance, as well as to other implicit assumptions made by
the participants in a discourse.
Thus the results of empirical investigation have forced many trans¬
formational linguists to recognise the importance of context and to
join a series of disciplines converging on the study of situated
speech.

Early attempts at discourse analysis

Although Firth urged linguists to study the total verbal process in
its context of situation he did not do so himself, choosing rather to
concentrate on phonology. In the intervening period there are only
two isolated attempts to study supra-sentential structure, one based
on a written text, by Harris, the other based on a collection of
spoken texts, by Mitchell.
Harris’s article, although it has the promising title ‘Discourse
Analysis’ is in fact disappointing. Working within the Bloomfieldian
tradition he sets out to produce a formal method ‘for the analysis
of connected speech or writing’ which ‘does not depend on the
analyst’s knowledge of the particular meaning of each morpheme.’
He observes that in grammar it is possible to set up word classes
distributionally and produce a class of adjectives A which occur
before a class of nouns N; such a statement captures a powerful
generalisation, even though it is possible to show that a particular
member of the class A, ‘voluntary’ may never occur before a
particular member of the class N ‘subjugation’.
4 AN INTRODUCTION TO DISCOURSE ANALYSIS

Harris suggests that a distributional analysis can be successfully

applied to a whole text to discover structuring above the rank of
sentence. As an example he creates a text containing the following
four sentences.
The trees turn here about the middle of autumn.
The trees turn here about the end of October.
The first frost comes after the middle of autumn.
We start heating after the end of October.
The aim of the analysis is to isolate units of text which are distri-
butionally equivalent though not necessarily similar in meaning;
that is equivalences which have validity for that text alone. From
the first two sentences above one establishes the equivalence of ‘the
middle of autumn’ and ‘the end of October’, not because they are
similar in meaning but because they share an identical environment,
‘the trees turn here’. The next step is to carry over the equivalences
derived from the first two sentences into the next two and this
allows us to equate ‘the first frost comes’ with ‘we start heating’
and of course both with ‘the trees turn here’ which provided the
original context. Thus, in terms of equivalence classes, all four
sentences have identical structure, class X followed by class Y. The
analyst progresses in this way through the text creating a chain of
equivalences and occasionally, as required, introducing a new class
until the whole of the text has been divided into units assigned to
one or other of the classes.
Harris points out that in evaluating his approach the only rele¬
vant questions are ‘whether the method is usable and whether it
leads to valid and interesting results.’ In the twenty years since the
article was published no one has adapted or developed the method,
apparently because the results were not interesting. It may well be
that any purely formal analysis above the rank of sentence is
impossible. Certainly, as Harris himself observes, it is impossible to
describe the structure of paragraphs in terms of sequences of
sentences of particular types—the constraints above the sentence
are stylistic not grammatical, and organisation and sequence can
only be described in semantic terms. For example, a text which
from a grammatical viewpoint consists simply of a sequence of
clauses with no obvious patterning may from a semantic or
functional viewpoint be seen to consist of a sequence of question-
answer pairs.
In marked contrast to Harris, Mitchell’s ‘Buying and selling in
Cyrenaica’ presents a semantically motivated analysis. Working in
 INTRODUCTION 5

the Firthian tradition he specifies the relevant elements of situation

and relevant participants in detail and divides the buying-selling
process into stages on purely semantic criteria, admitting that ‘stage
is an abstract category and the numbering of stages does not neces¬
sarily imply sequence in time.’ He describes three major categories
of transaction—market auctions; other market transactions; shop
transactions—although the second and third are distinguished
mainly by situation as they share the following five stages:
1. salutation
2. enquiry as to the object of sale
3. investigation of the object of sale
4. bargaining
5. conclusion
This is an ideal structure: sometimes stages 1 and 2 do not occur
and stages 3 and 5 may be realised non-verbally. The following is
an example of a shop transaction:

Personality Translation Stage

buyer: Have you a bed to sell? 2

SELLER: I’ve got one but it’s rather expensive. 2
BUYER: Let me have a look at it then. 2
SELLER: Certainly.
If you want it for yourself I will make you a
reduction. 4
buyer: How much is it? 4
SELLER: £4. 4
BUYER: What’s your last price? 4
SELLER: Believe me if it were anyone but you I’d ask
him five. 4
buyer: I’ll make you a firm offer of £3.50. 4
SELLER: Impossible, let it stay where it is. 4
buyer: Listen. I’ll come this afternoon, pay you £3.70
and take it. 4
(Buyer crosses threshold of shop on his way out.)
seller: It still wants some repairs, 5
While this analysis captures some of the structure of the trans¬
action it is arguable that it is not a linguistic analysis at all—the
stages are defined and recognised by the activity that occurs within
them rather than by characteristic linguistic features and, with the
possible exception of stage 4, which when opened by the buyer
6 AN INTRODUCTION TO DISCOURSE ANALYSIS

begins invariably with the formula ‘How much’, there are no

linguistic markers of the transition between stages. Once the stages
have been isolated non-linguistically, Mitchell then characterises
them linguistically by providing examples of the kinds of phrases
and clauses, often ritual, which occur within them.
In any spoken text there are at least four major levels of organisa¬
tion—phonology, grammar, discourse and non-linguist ic. The
structure in each of these levels can be expressed in terms of small
units combining to form larger units—within phonology and gram¬
mar, the traditional concerns of linguistics, the labels and structure
of the units are well established; within discourse nothing is certain:
there are no agreed labels and few agreed structures. For exemplifi-
catory purposes in the figure below the descriptive units proposed
by Sinclair et al (1972) are used, but one could substitute similar
terms from other systems. (The figure attempts to show roughly the
size relationship between units in different levels.)

Phonology Grammar Discourse Non-Linguistic

(Sinclair et al) (Mitchell)
phoneme

syllable
morpheme
foot
word

group
tone group
clause
act
sentence
move
exchange
stage

transaction
Figure 1

As there are four levels of organisation in any text one can

obviously provide four complementary descriptions of the same
text. What is missing from Mitchell's analysis is any description of
the supra-sentential linguistic structure; the description of overall
organisation of these transactions is in non-linguistic terms—they
consist of a series of stages characterised non-linguistically. How-
 INTRODUCTION 7

ever, if the small amount of linguistic evidence provided is repre¬

sentative it should be possible to describe the discourse structure.
For instance, both examples of stage 2, enquiry, consist of a
sequence of Question-Answer pairs and salutation, when it occurs,
involves a contribution from both speakers. This is an instance of
linguistic, not non-linguistic structure —the occurrence of one
linguistic item, an utterance or part of an utterance, constrains the
choices open to the next speaker.
The structure, or constraints on the next speaker, cannot be
expressed in grammatical terms however; the linguistic form of the
utterance is almost irrelevant; what is structurally important is its
linguistic function and it is evidence of this kind which points to the
existence of another level, discourse, between grammar and non-
linguistic organisation. Sequences, which from a grammatical view¬
point are a random succession of clauses of different types can be
seen from a functional viewpoint to be highly structured.

Language function
Discourse, then, does not consist simply of a string of grammatically
well-formed utterances or sentences. The following examples from
Labov (1970) are grammatically unexceptional yet noticeably odd:
A: What is your name?
B: Well, let’s say you might have thought you had something
from before but you haven’t got it any more.
A: I’m going to call you Dean.
A: I feel hot today.
B: No.
In both examples B’s contribution obviously breaks rules for the
production of coherent discourse. One of the fundamental aims of
discourse analysis is to discover these rules, but an even more
fundamental question is the nature of the units whose structure and
occurrence the sequencing rules will describe.
Labov (1972) emphasises that the first and most important step
is to distinguish ‘ what is said from what is done’; that is, discourse
analysis must be concerned with the functional use of language.
Thus, for all discourse analysts the unit of analysis is not the gram¬
matically defined ‘clause’ or ‘sentence’, although the unit may very -
frequently consist of one clause or sentence. Hymes (1972) labels
the unit ‘speech act’ and insists that it ‘represents a level distinct
from the sentence and not identifiable with any single portion of
8 AN INTRODUCTION TO DISCOURSE ANALYSIS

other levels of grammar, nor with segments of any particular size

defined in terms of other levels of grammar’.
While the relations between the basic units of discourse are
generally agreed to depend on their respective functions there is, as
yet, no consensus on how many functions there are. At one extreme,
Austin (1962) suggests there may be as many as 10,000, at the other
Sinclair et al (1972) postulate only 22, while Searle (1969) contem¬
plates an intermediate number but suggests that some functions
may subsume others.
There are similar problems with the size of the basic unit. Labov
(1970, 1972), Sacks (passim), Schegloff (1968, 1972) and Jefferson
(1972, 1973) regard the utterance as the basic unit of analysis, but
this is almost certainly the result of working with simple and
occasionally constructed data. Sinclair et al observe that they also
began with utterance as the basic unit, but, in dealing with examples
like the following, came to feel the need for a smaller unit, which
they called a move. Moves can be co-extensive with utterances, but
some utterances, like A’s second, contain two moves:

A: Can you tell me why do you eat all that food?

B: To keep you strong.
A: To keep you strong, yes, to keep you strong. Why do you
want to be strong?

Any attempt to apply the analyses suggested by Labov, Sacks,

Schegloff and Jefferson to the above example quickly demonstrates
that in fact their analytic unit is not the utterance but something
equivalent to the move.
Confusion over the size of basic unit does, however, frequently
occur, particularly among those using the findings of discourse
analysis to produce language teaching materials. It is not at all
clear what is the size of unit to which Wilkins’ (1972) functions
relate—some of them appear to be conceived as utterance or move
length, some as a sequence of utterances.
The final problem for discourse analysis is to show how the
functional categories are realised by formal items—what is the
relationship between ‘request’ or ‘question’ and the grammatical
options available to the speaker. Some analysts like Sacks and
Schegloff assume that their categories are intuitively recognisable
from the label, others like Labov feel the need to attempt to write
rules to explain how a given lexico-grammatical structure comes to
realise a given function in a given situation.
 INTRODUCTION 9

Methodological preliminaries
The research reported in the succeeding chapters comes from a wide
range of disciplines with differing ideas on what constitutes rele¬
vant and acceptable data. As we saw above Firth argued for a
text-based description, Chomsky for a total reliance on intuition.
Lyons (1968) suggests that there are three degrees of idealisation
between raw data and the idealised sentences of Chomsky’s compe¬
tence.
The first stage is regularisation in which the analyst ignores such
phenomena as slips of the tongue, hesitations, repetitions, self¬
editing and so on. The second stage is standardisation in which one
ignores variation and treats whatever data one is examining as
homogeneous—thus at the phonemic level, different pronunciations
of the same ‘word’ are treated as if they were the same; at the level
of discourse, variants of a misapprehension sequence are all regarded
as occurrences of the same unit. This is an essential step in any
classificatory system, for in the final analysis all utterances can be
shown to be unique. However there are currently disagreements
among linguists over the degree of standardisation and the amount
of variation which can be successfully described. (Sankoff 1974.)
The third stage of idealisation involves decontextualisation, which
separates sentences from their contexts of use or occurrence and
treats them as self-contained and isolated units.
Most of the work described in the following chapters works with
transcripts which are in an unregularised form, but in fact analyses
the data as if it were both regularised and standardised. The work
by philosophers on speech acts, however, is based entirely on de-
contextualised data, and data which in fact takes no account of any
preceding or succeeding sentences—in other words there is no
concern with interaction even on an idealised level.
In attempting to distinguish descriptions of decontextualised data
from those of contextualised data Widdowson (1973) suggests
several useful pairs of terms—usage/use; sentence/utterance;
locution/illocution; text/discourse; cohesion/coherence. Gram¬
marians are concerned with rules of usage which are exemplified in
sentences', discourse analysts with rules of use which describe how
utterances perform social acts;

a sentence is an instance of usage in so far as it is discoverable

in an utterance, but in so far as that utterance makes a state¬
ment of a particular kind it is an instance of use.
10 AN INTRODUCTION TO DISCOURSE ANALYSIS

Sentences combine to form texts and the relations between sentences

are aspects of grammatical cohesion', utterances combine to form
discourse and the relations between them are aspects of discourse
coherence. Both the following examples are coherent, but only the
first is a cohesive text, with the second sentence linked to the first
by ellipsis:

A: Can you go to Edinburgh tomorrow?

B: Yes, I can.
A: Can you go to Edinburgh tomorrow?
B: B.E.A. pilots are on strike.

Widdowson’s terms and distinctions have been adopted and used

through the book in an attempt to reduce confusion. Many analysts.
Transformational Grammarians, Generative Semanticists and
Ethnomethodologists alike, often use sentence and utterance inter¬
changeably; in what follows these and the other terms will not be in
free or elegant variation.
 2

Speech acts

While linguistics restricted itself for a generation to a concentration

on form, the study of meaning was left to linguistic philosophers,
who concentrated on the sense, reference and implications of
sentences and parts of sentences. In 1962 J. L. Austin observed
that while it had long been the assumption of philosophers that,

the business of a ‘statement’ can only be to describe some state

of affairs or to ‘state some fact’, which it must do either truly
or falsely,

more recently they had come to realise that this was not always the
case. Some sentences, which look like statements, or as Austin
prefers to call them constatives, are not intended to record or impart
information about facts: for example some sentences like ‘the King
of France is bald’ are strictly nonsense, despite unexceptional gram¬
matical form; others, ethical propositions, are ‘perhaps intended,
solely or partly, to evince emotion or to prescribe conduct, or to
influence it in special ways’. Austin focuses on one group of such
sentences, which he labels performatives, in which the saying of the
words constitutes the performing of an action:

‘I name this ship the Queen Elizabeth’—as uttered when

smashing the bottle against the stern.
‘I do’ (re. take this woman to be my lawful wedded wife)—as
uttered in the course of a marriage ceremony.
‘I give and bequeath my watch to my brother’—as occurring
in a will.

In saying ‘I name this ship the Queen Elizabeth’’ the speaker is not
describing what he is doing, nor stating that he is doing it, but
actually performing the action of naming the ship; from that
moment the ship is named. One interesting confirmation that it is
by saying the words that one performs the action is that very
12 AN INTRODUCTION TO DISCOURSE ANALYSIS

frequently one can insert the word ‘hereby’—‘I hereby name this
ship.. The uttering of the words alone is, however, not sufficient
—while the performative utterance is ‘usually a, or even the leading
incident’ in the performing of the acts of naming, marrying or
bequeathing, it is rarely if ever the ‘sole thing necessary if the act
is to be deemed to have been performed’. Austin stresses the con¬
ventional nature of the performative act and the fact that an agreed
procedure must be followed. There are four conditions which must
be satisfied if the performative act is not to misfire.
1. There must exist an accepted conventional procedure, having
a certain conventional effect, that procedure to include the
uttering of certain words by certain persons in certain
circumstances.
By this condition Austin draws attention to the fact that there is a
limited number of performative acts and one cannot arbitrarily
adapt a procedure to perform what appears to be a similar act—
there is a procedure for christening babies but not dogs, for naming
ships but not houses. For some acts procedures differ in different
countries—no one, whatever his religion, can divorce his wife in
England by saying ‘I divorce you; I divorce you; I divorce you’;
some acts are possible in one language community but not in
another—there is no formal procedure in Modern English for
insulting someone, to match that used by German students to
initiate duels in the inter-war years. This is not, of course, to say
that one cannot insult someone in English, but simply that one
cannot insult them by saying ‘I insult you’.
2. The particular persons and circumstances in a given case
must be appropriate for the invocation of the particular
procedure invoked.
This condition emphasises the fact that the uttering of the correct
and appropriate words is insufficient to achieve the successful per¬
formance of the act: the words must be uttered by the appropriate
person—the blacksmith in Gretna Green may read the marriage
service as well as any parson, but the ceremony is still invalid;
while even the appropriate person cannot utter the appropriate
words in inappropriate circumstances—one of the umpires in the
Test match when Leonard Hutton scored his record 364 claimed
later that Hutton was technically out lbw at 332, but, as no one on
the fielding side appealed, the umpire was unable to pronounce him
out.
 SPEECH ACTS 13

3. The procedure must be executed by all participants both

correctly
4. and completely.

These conditions cover misfires which occur despite the existence

of a conventional procedure and the presence of the appropriate
participants in the appropriate circumstances. The problems may
be verbal or non-verbal. The marriage ceremony includes yes/no
questions, ‘Do you take this woman . ..’ but ‘yes’ is not an accept¬
able answer; the ceremony has a fixed point for the ring to be
placed on the finger—failure to produce the ring or placing the
ring on the finger at a different point in the ceremony would again
cause the act to misfire.
So far we have seen that the uttering of certain words by appro¬
priate people in appropriate circumstances can constitute the
performing of certain conventional acts, but how does one recog¬
nise and distinguish such performative utterances? All Austin’s
initial examples have the verb in the simple present active form
with a first person singular subject, ‘I name this ship’, and this is
apparently significant since neither ‘I am naming this ship’, nor ‘he
names/named this ship’ nor ‘this ship is named by me’ can act as
performatives. However, it soon becomes apparent that there are
performative utterances which have no subject or verb at all, ‘guilty’
pronounced by the foreman of a jury or ‘out’ by an umpire, and
reluctantly Austin is forced to conclude that there are in fact no
formal criteria which distinguish performative from non-performa¬
tive utterances.
The achievement so far has been to isolate

a class of utterances, linguistically quite heterogeneous, which

have in common that, in virtue of non-linguistic conventions, to
issue them (happily) counts as doing this or that.
(Warnock, 1973)

In one important sense these performative utterances are idioms—

the meanings of the individual words are not of great importance
and synonyms cannot be substituted—it is the uttering of pre¬
determined words in a fixed sequence in a few highly conventional¬
ised, at times ritual, situations, which constitutes the performing of
the action. If performative utterances were restricted to such
situations their existence would be an interesting but not particularly
significant fact about language use. However, Austin noticed that
the concept of the performative utterance, of doing something by
14 AN INTRODUCTION TO DISCOURSE ANALYSIS

saying something, had a more general application, for in saying

‘I promise’, ‘I apologise’, ‘I warn you’ one actually performs the
acts of promising, apologising and warning. Thus these utterances
also are performative, but are crucially distinct from the first group
in that there are no rule-governed conventions restricting their use—
anyone can make a promise to anyone in any place at any time.
Interestingly, because these utterances are not part of a conven¬
tional procedure their linguistic form is not predetermined; there
are cases in which

to utter the words ‘the ice over there is thin’ to a skater is to

issue a warning... without its being the case that there is a
statable convention at all, such that the speaker’s act can be
said to be an act conforming to that convention.
(Strawson 1964)

The form of the performative utterance may be explicit, like ‘I warn

you that X’ but it is much more likely to be primary, simply ‘X’.
This, of course, raises the question of how one recognises a given
utterance as performative. Even the explicit performatives create
problems, because although, again, one might hope that the gram¬
matical form of first person singular subject and present simple
active verb would be distinctive, in fact not all utterances of this
form are performative,

‘ I promise only when I intend to keep my word.’

‘On page 49 I protest against the verdict.’

nor do all explicit performatives have this form.

‘Passengers are warned to cross the track by the bridge only.’

‘Notice is hereby given that trespassers will be prosecuted.’

With primary performatives the problems are even greater; it may,

at times, be difficult even to decide which performative is being
realised: ‘I’ll be there first’, could well be a prediction, promise or
threat. There may be lexical clues, but these are highly unreliable:
both ‘there’s a bull in that field’ and ‘there’s a dangerous bull in that
field’ may be primary performative warnings or simple statements
of fact.
Here Austin comes up against the central problem in discourse
analysis, the interface between form and function. The difficulty is
to explain how a relatively small number of grammatical options
can realise a relatively large number of discourse functions, and
 SPEECH ACTS 15

how both listener and analyst can successfully recognise which

function or performative is being conveyed. Austin suggests that
the problem is not, in fact, too difficult because,

any utterance which is in fact performative should be reducible

or expandible or analysable into a form with a verb in the first
person singular present indicative active... Thus ‘out’ is
equivalent to ‘I declare, pronounce or call you out’, ‘guilty’ is
equivalent to ‘I find, pronounce, deem you to be guilty’.

The discussion has now come full circle. We first established that
there was a set of utterances of the form ‘ I + present simple active
verb’ which were performatives; then it became apparent that not
only were there constative utterances with the same grammatical
form, but also performatives with other grammatical forms which
often did not even include the performative verb. We then sug¬
gested, however, that those utterances which were actually per¬
formative, but did not have the form ‘I + present simple active
verb’ were ‘reducible, expandible or analysable into that form.
This revives the question of how one recognises whether an utter¬
ance of the form ‘I + present simple active’ is performative or
constative. In the following lists the first column contains explicit
performatives, the third column constatives while the status of
those in the middle column is doubtful.

EXPLICIT PERFORMATIVES CONSTATIVES

I bid you welcome I welcome you

I apologise I am sorry I repent
I criticise you I blame you I am shocked by you
I approve X I approve of X I feel approval of X

Austin suggests four tests for deciding which way utterances in the
middle column are being used.

(i) does the saying of the words constitute the performing of an

act? This can be tested by asking ‘did he really of a particular
utterance. It is not possible to question whether a person
actually bid another welcome, the uttering of the words consti¬
tutes the action, but one can ask following the utterance ‘I wel¬
come you’, ‘Did he really welcome him?’, and the answer and
therefore the classification of the utterance is likely to depend
on the circumstances.
(ii) could the action be performed without uttering the words?
16 AN INTRODUCTION TO DISCOURSE ANALYSIS

One can be sorry just as one can repent without saying any¬
thing; one cannot apologise silently.
(iii) is the action something that can be done deliberately and
voluntarily? One can be ‘willing to apologise’ but not ‘willing
to be sorry’—one is either sorry or not, though one can be
willing to say that one is sorry.
(iv) can the utterance be literally false? Austin sees this as a
crucial distinction between constatives which can be true or
false and performatives which can only be happy or unhappy.
Despite saying ‘I am sorry’ it need not be true that one is
sorry; if one says ‘I apologise’, however, it cannot be false that
one has apologised—the apology may be insincere and the
speaker may have abused the procedure but that is another
matter.

These criteria enable one to assign successfully utterances of the

form ‘I + present simple active verb’ to the class of explicit perfor¬
matives or constatives, the one being subject to a test of happiness,
the other to a test of truth. However, yet again in this argument,
having taken two steps forward we must take one back. There is a
class of utterances which Austin labels expositional performatives,
or expositives in which ‘the main body of the utterance has generally
or often the straightforward form of a statement’, which is subject
to a test of truth, but prefacing the statement is a verb phrase, like
I argue/conclude/testify/admit/predict, which satisfies all the
criteria for performatives:

I argue that there is no backside to the moon.

It doesn’t take long to realise that even ‘I state’ satisfies the perfor¬
mative test. This is initially very disconcerting because the whole
drift of the argument so far has been concerned with distinguishing
performatives and constatives. However, Austin observes, the
situation

is not so bad: we could distinguish the performative opening

part (I state that) which makes clear how the utterance is to be
taken, that it is a statement (so distinct from a prediction etc),
from the bit in the that-clause which is required to be true or
false.
It is now only a small step to realise that all utterances previously
labelled constative, even those with the grammatical form ‘I +
present simple active verb’ are in fact primary performatives which
 SPEECH ACTS 17

are ‘expandible or analysable into a form “I state that I +

present simple active verb’”. Thus Austin has argued us into the
position that others have reached by different means—that in saying
anything one is performing some kind of act.
When Austin argued that any primary performative can be
expanded into an explicit form he suggested that ‘“out” is equiva¬
lent to “I declare, pronounce or call you out”’. The very fact that
he felt able or constrained to offer three explicit performatives as
alternative versions of one primary performative suggests that there
are complications here. How many different functions or perfor¬
matives does one need to recognise? Austin suggests that the number
of functions is equivalent to the number of performative verbs,
which he estimates at between one and ten thousand.
There are two questions to be asked at this stage: the first, posed
by Searle (1965) is whether there are ‘some basic illocutionary acts
to which all or most others are reducible’—this is a very pertinent
question in the light of the virtual synonymy of ‘declare’ ‘pro¬
nounce’ and ‘call’ in the example above. The second question is
whether it is wise to rely on natural language as a means of categori¬
sation—should ‘I order you to’ ‘I request you to’ ‘I beg you to’
‘I entreat you to’ be necessarily regarded as different speech acts?
Certainly all could be expansions of the same primary performative
‘put down that gun’ when uttered by speakers with differing status
relative to their addressee. The converse of accepting all and only
the performative verbs as labels for the various functions is that
there may be certain actions achieved through saying something,
like ‘insulting’, which cannot be labelled simply because there is a
gap in the lexical system, no performative verb ‘I insult you that’.
As Searle (1969) observes
we might have had an illocutionary verb ‘rubrify’ meaning to
call something red. Thus, ‘I hereby rubrify it’ would simply
mean ‘It’s red’. Analogously we happen to have an obsolete
verb ‘macarize’, meaning to call someone happy.
Having demonstrated that in fact all utterances are performative
Austin reconsiders the senses in which ‘to say something may be to
do something’ and concludes that in ‘issuing an utterance’ a speaker
can perform three acts simultaneously: a locutionary act which is
the act of saying something in the full sense of ‘say’; an illocution¬
ary act which is an act performed in saying something, the act
identified by the explicit performative; and a perlocutionary act, the
act performed by or as a result of saying. Thus:
18 AN INTRODUCTION TO DISCOURSE ANALYSIS

Act A or Locution
He said to me ‘Shoot her’ meaning by ‘shoot’ shoot and
referring by ‘her’ to her.
Act B or Illocution
He urged (or advised, ordered, etc.) me to shoot her.
Act C or Perlocution
He persuaded me to shoot her.

It is not Austin’s intention to suggest that in speaking one has the

option of performing one or another of these types of speech act; in
fact one normally performs all three simultaneously, but it is useful
for analytic purposes to isolate them.
Austin first distinguishes locutionary and illocutionary acts.
While
to perform a locutionary act is in general, we may say, also and
eo ipso to perform an illocutionary act,
the interpretation of the locutionary act is concerned with meaning,
the interpretation of the illocutionary act with force. Austin glosses
‘meaning’ unhelpfully as the use of language with ‘a certain more
or less definite “sense” and a more or less definite “reference” ’, but
Strawson (1973) clarifies the matter by asking what a listener would
need to know, so that he could be said to know ‘the meaning of
precisely what was said’ on a given occasion. He points out that a
complete mastery of the linguistic system, syntax and semantics, is
almost always insufficient; any stranger listening to a tape-recording
of the utterance ‘John will get here in two hours from now’, would
not know the person referred to by ‘John’ or the time and place by
‘here’ and ‘now’. Thus meaning must be seen as an amalgam of
grammatical, lexical and extra-textual information, and it is the
function of the locutionary act to transfer this meaning from speaker
to listener. There is still a sense in which the listener may, however,
not have understood ‘how what was said was meant’, that is whether
the illocutionary force of the locution concerning John was assertion,
prediction or warning.
Many utterances are the simultaneous performance of locutionary,
illocutionary and perlocutionary acts, and Austin observes that ‘it is
the distinction between illocutionary and perlocutionary which
seems likeliest to give trouble’. Basically an illocutionary act is a
linguistic act performed in uttering certain words in a given context,
a perlocutionary act is a non-linguistic act performed as a conse¬
quence of the locutionary and illocutionary acts. The illocutionary
 SPEECH ACTS 19

act, being achieved through the uttering of certain words, is

potentially under the control of the speaker; provided he uses the
correct explicit performative in the appropriate circumstances he
can be certain that the act will be happy—no one can prevent some¬
one from warning or advising them, except by refusing to listen.
The perloeutionary act, however, is the causing of a change in the
mind of the listener, so that he becomes ‘alarmed’, ‘convinced’,
‘deterred’. The act is the effect of the utterance on the listener but
this is not an effect governed by convention—there is no conven¬
tional or reliable way of ‘convincing’ or ‘deterring’ someone. I may
warn you in order to deter you but may in fact simply succeed in
encouraging or even inciting you.
For this reason Austin feels it necessary to distinguish between
perloeutionary object, basically the intended result of the illocution¬
ary act and perloeutionary sequel, the unintended or secondary
result. Thus

the perloeutionary object of one illocution may be a sequel to

another: for example, the perloeutionary object of warning, to
alert someone, may be a sequel of a perloeutionary act which
alarms someone.
It is in this way that Austin solves the problem, raised earlier, of
accounting for those performative actions like ‘insult’ for which
there is no performative formula. He argues that some acts have
only sequels not objects,

thus I may surprise you or upset you or humiliate you by a

locution, though there is no illocutionary formula ‘I surprise
you by ... ‘ I upset you by .. . ’, ‘ I humiliate you by ...
Unfortunately Austin does not pursue the investigation of perlocu-
tionary objects and sequels but obviously such a study could reveal
persuasive and oratorical techniques and form the substance of a
companion volume How to achieve things through words.
From the discussion so far it is evident that Austin attaches con¬
siderable importance to speaker’s intention. Some discourse analysts.
Sacks and his colleagues in particular, would want to argue that the
illocutionary force of an utterance is what it is taken to be by the
listener rather than what it is intended to be by the speaker on the
grounds that neither listener nor analyst can ever be sure of the
speaker’s intention because it is never available for examination,
but the listener’s interpretation is evident in his response and it is
this which determines the progress of the interaction. For Austin,
20 AN INTRODUCTION TO DISCOURSE ANALYSIS

however, working not with extracts from conversations but with

isolated invented sentences, illocutionary force is inextricably
bound up with speaker’s intention—he argues that if a listener
misinterprets an utterance the speaker should be regarded not as
having (accidentally) produced a different illocutionary act but as
having produced none at all—‘the performance of an illocutionary
act involves the securing of uptake’, that is intended uptake. The
unstated assumption is that each locution has one illocution¬
ary force, but as we have seen above a primary performative is
potentially ambiguous and in all areas of language speakers can
and regularly do exploit ambiguity. Searle (1969) offers an
example

suppose at a party a wife says ‘It’s really quite late’. That

utterance may be at one level a statement of fact; to her inter-
locuter, who has just remarked on how early it was, it may be
(and be intended as) an objection; to her husband it may be (and
be intended as) a suggestion or even a request (‘Let’s go home.’)
as well as a warning (‘You’ll feel rotten in the morning if we
don’t.’).

Unfortunately Searle does not pursue the problems for any intention-
based description which this example raises.
Strawson (1964) focuses on the problem of intention in some
detail. He observes that a speaker has much more control over the
illocutionary force than the perlocutionary effect of his utterance for

in general a man can speak of his intention in performing an

action with a kind of authority he cannot command in pre¬
dicting its outcome.

He accepts Austin’s position that the illocutionary force of an

utterance is that intended by the speaker, but stresses that in order
to achieve uptake of his illocutionary act the speaker has to make
his intention clear to the listener—one cannot warn someone unless
they realise you are intending to warn them. For this to be possible
clearly ‘there must exist, or he must find, a means of making the
intention clear’. The explicit performative is the clearest formula¬
tion; if a speaker uses a primary performative it is because he is
confident that clues in the context or preceding text will make his
intention clear and if he is in doubt he can add a ‘force-elucidating
comment’, like ‘this is only a suggestion’, or ‘I’m warning you’.
Intention is, however, of little importance in the discussion of
perlocutionary acts; not only is it irrelevant to reveal one’s per-
 SPEECH ACTS 21

locutionary intent, it may even be counterproductive—as Strawson

points out if one’s intention is to show off or impress ‘recognition
of this intention might militate against the effect and promote an
opposite effect’.
Austin’s work has aroused a great deal of interest and criticism
among philosophers. He expected major difficulties with the
illocutionary/perlocutionary distinction but in fact there have been
more with the locutionary/illocutionary one. Problems arise with
utterances like ‘I warn you there’s a bull in that field’ in which,
one can argue, to know the meaning of the locutionary act is already
to know the illocutionary force. Cohen (1969) asks

in what way does the illocutionary force of such an utterance

differ from that part of its meaning which belongs to it in virtue
of its performative prefix?

and argues that illocutionary forces do not in fact exist. Strawson

(1964) accepts that in explicit performative utterances the meaning
may exhaust the force, but in primary performatives the meaning
‘though it limits does not exhaust the force’. Searle (1969), in similar
vein to Cohen, argues that

where a certain force is part of the meaning, where the meaning

uniquely determines a particular force, these are not two differ¬
ent acts, but two different labels for the same act,

but reaches the conclusion that there are only //locutionary acts.
These criticisms are unhelpful and appear to pun on the meaning of
‘meaning’. As Forguson (1973) observes,

even if there are cases in which meaning completely determines

force it isn’t the same thing as force.

The locutionary/illocutionary/perlocutionary distinctions, despite

all criticisms and difficulties with definition remain useful and
suggestive concepts in any discussion of speech acts.

Searle
Since the death of Austin the most significant work in speech act
philosophy has been that of John Searle, who has attempted to
detail some of the rules which govern the effective production of
certain speech acts. While Austin suggested four conditions govern¬
ing the ‘happy’ production of ritual or archetypal performatives,
he suggested no conditions or rules for other performatives. Searle
22 AN INTRODUCTION TO DISCOURSE ANALYSIS

(1965) attempts, by a detailed discussion of one non-conventional

illocutionary act, ‘promise’, to

explicate the notion of illocutionary act by stating a set of

necessary and sufficient conditions for the performance of a
particular kind of illocutionary act and extracting from it a set
of semantical rules for the use of the expression (or syntactic
device) which marks the utterance as an illocutionary act of
that kind.

Searle, while accepting the notion of illocutionary act, does not

accept the locutionary act, but instead proposes the existence of the
propositional act which carries the content of an utterance. To
many this will seem little more than a quibble but it enables Searle
to be more explicit about the way the content is expressed. He
suggests that a propositional act includes the subsidiary acts of
referring to someone or something and of predicating some property
or act of that to which one has referred. The propositional act, he
emphasises, is not represented by the entire utterance, but only by
those portions of the sentence which do not include ‘the indication
of illocutionary force’. Thus, he suggests, an utterance consists of
two (not necessarily separate) parts, a proposition and a function
indicating device which marks the ‘ illocutionary force ’. He observes
that function-indicating devices in English

include word order, stress, intonation contour, punctuation, the

mood of the verb and finally the set of so-called performative
verbs.
In each of the following utterances, Searle suggests, the speaker
expresses the proposition that John will leave the room, that is he
predicates the action of leaving the room of John, though only in
the second does he perform the illocutionary act of ‘asserting’.

Will John leave the room?

John will leave the room.
John, leave the room!
If John will leave the room I will leave also.

Before he attempts to clarify the nature of the rules which govern

the linguistic realisation of illocutionary acts Searle discusses and
distinguishes the two major types of rule, regulative and constitutive.
Regulative rules, as the name implies, are concerned with conditions
on the occurrence of certain forms of behaviour—‘Children are
 SPEECH ACTS 23

forbidden to play football on the grass’; constitutive rules define

the behaviour itself—‘A player is offside if ... If the children
ignore the notice they will still be playing football though inciden¬
tally breaking the law; if they ignore the offside rule they are no
longer technically playing football, for football has no existence
apart from its constitutive rules.
In the study of language use both sets of rules are important. All
interaction has regulative rules, usually not explicitly stated, which
govern greetings, choice of topic, interruption and so on and as
Hymes (1972) points out the rules vary from community to com¬
munity. Constitutive rules in speech are those which control the
ways in which a given utterance of a given form is heard as realising
a given illocutionary act. Searle’s aim in this article is to describe
the constitutive rules for the illocutionary act of promising.
Although Searle acknowledges the existence of a variety of
‘function indicating devices’ he does not investigate how they can
be used to mark a given utterance as a primary performative with
the illocutionary force of a promise. Instead he simplifies matters
considerably by his decision to confine his discussion

to full-blown explicit promises and [to] ignore promises made

by elliptical turns of phrase, hints, metaphors, etc.

In other words his concern is not with deciding whether a poten¬

tially ambiguous utterance is intended as a promise but with how
an utterance of the form ‘I promise X’ can ‘happily’ secure uptake
as a promise. He suggests that five rules govern the making of a
promise:
propositional content rule—in a promise an act. must be predi¬
cated of the speaker and it must be a future act—a speaker
cannot promise to have done something nor promise that
someone else will do something.
preparatory rules—(a) a promise is defective if the promiser
doesn’t believe the promisee wants it done or even if the thing
promised is something the promisee doesn’t want done other¬
wise whatever the speaker’s intention it will act as a warning.
(b) a speaker cannot promise to do something he would be
expected to do anyway—as Searle observes any husband who
promises his wife not to be unfaithful in the next week is likely
to provide more anxiety than comfort.
sincerity rule—the speaker must intend to perform the action;
24 AN INTRODUCTION TO DISCOURSE ANALYSIS

although as Searle admits it is in fact possible for someone to

promise with no intention at all of honouring it, but then he is
abusing the procedure.
essential rule—the uttering of the words counts as the under¬
taking of an obligation to perform the action.

A major difference between Austin and Searle lies in the deriva¬

tion of the illocutionary force of an utterance—Austin argues it is
the successful realisation of the speaker’s intention, Searle that it is
the product of the listener’s interpretation of the utterance. Prepara¬
tory rule (a) makes this clear. This provides for a speaker saying
‘I promise I’ll be there by three o’clock’, feeling certain in his own
mind that he has committed himself and yet unwittingly having
performed the illocutionary act of ‘warning’ or ‘threatening’,
because the hearer doesn’t in fact want him to be there by three
o’clock.
Searle observes that if the analysis he proposes for ‘promise’ is
of any general interest ‘then it would seem that these distinctions
should carry over into other types of speech act’. Searle (1969)
attempts such an extension. For instance for the speech act of
‘ordering’

the preparatory conditions include that the speaker should be

in a position of authority over the hearer, the sincerity condi¬
tion is that the speaker wants the ordered act done and the
.essential condition has to do with the fact that the speaker
intends the utterance as an attempt to get the hearer to do the
act.

Using the same framework Searle offers an analysis of ‘request’,

‘assert’, ‘question’, ‘thank’, ‘advise’, and ‘warn’. He notes that
certain conditions overlap and this leads him to question whether
there are ‘some basic illocutionary acts to which all or most of the
others are reducible’. While not committing himself to a global
answer he certainly feels that there are some illocutionary acts
which can be usefully regarded as special cases of others—‘asking
questions’, he suggests, is a special case of ‘requesting’; a fact which
explains

our intuitions that an utterance of the request form ‘tell me the

name of the first President of the United States?’ is equivalent
in force to an utterance of the question form ‘What is the name
of the first President of the United States?’
 SPEECH ACTS 25

The question of basic illocutionary acts or functions is of great

relevance to anyone hoping to produce a comprehensive descriptive
system. If there are, as Austin suggested, up to 10,000 different
illocutionary acts, a descriptive system which gave all equal weight
would be unmanageable. Austin himself grouped them into five
major classes: verdictives, typified by the giving of a verdict by a
jury, arbitrator or umpire—acquit, grade, estimate, diagnose;
exercitives, which are the exercising of powers, rights or influence—
appoint, order, advise, warn; commissives, which commit the
speaker to doing something, but also include declarations or
announcements of intention—promise, guarantee, bet, oppose;
behabitives, a miscellaneous group concerned with attitudes and
social behaviour—apologise, criticise, bless, challenge; expositives,
which clarify how utterances fit into ongoing discourse, or how they
are being used—argue, postulate, affirm, concede.
Searle does not discuss Austin’s classification but the evidence
suggests that he would have considerably more than five basic
illocutionary acts, while the only descriptive system which claims
to be able to describe a spoken text exhaustively is that of Sinclair
et al (1972) which contains 22 acts at primary delicacy.
While his early work is mainly concerned with isolating the
conditions governing the happiness of explicit performative utter¬
ances, Searle (1975) tackles the much knottier problem of how
listeners correctly interpret primary performatives, or, as he prefers
to call them, indirect speech acts. For illustrative purposes he focuses
on indirect directives or requests, and suggests that the possible
realisations can be usefully grouped into six categories:

1. Sentences concerning
hearer’s ability; Can you pass the salt?
2. Sentences concerning
hearer’s future Will you
pass the salt?
action; Are you going to
3. Sentences concerning
speaker’s wish or I would like (you to pass) the salt.
want;
4. Sentences concerning
hearer’s desire or Would you mind passing the salt?
willingness;
5. Sentences concerning It might help if you passed the salt.
reasons for action; I don’t think you salted the potatoes.
26 AN INTRODUCTION TO DISCOURSE ANALYSIS

6. Sentences embedding
either one of the
above or an explicit
performative; Can I ask you to pass the salt?
(therefore not really
a separate class).
Searle observes that the first three types refer to the three felicity
conditions on directive illocutionary acts which he proposed in 1969
—respectively preparatory, concerned with the listener’s ability;
propositional content concerned with the futurity of the action; and
sincerity concerned with the speakers wanting the listener to per¬
form the action. He assimilates group 4 to group 5 arguing, not
entirely convincingly, that ‘both concern reasons for doing A’ ‘since
wanting to do something is a reason par excellence for doing it’.
From the data he is able to draw the following generalisations:

S can make an indirect directive by:

1. either asking whether or stating that a preparatory condition
concerning his ability to do A obtains;
2. either asking whether or stating that the propositional con¬
tent condition obtains;
3. stating that the sincerity condition obtains, but not by asking
whether it obtains;
4. either stating that or asking whether there are good or over¬
riding reasons for doing A, except where the reason is
that H wants or wishes etc., to do A, in which case he can
only ask whether H wants, wishes, etc. to do A.

These generalisations represent a description of the data—they

categorise and circumscribe the available ways of producing an
indirect directive—but there is no attempt to explain why these are
the options and the only options, nor how a hearer, faced with an
utterance like ‘Can you pass the salt?’ sets about deciding whether
the speaker intends it as a question or a request. Indeed, Searle
admits that the hearer

needs some way of finding out when the utterance is just a

question about his abilities and when it is a request,
but observes unhelpfully that

it is at this point that the general principles of conversation

(together with factual background information) come into play.
 SPEECH ACTS 27

For those interested in the analysis of spoken discourse it is a great

disappointment that both Searle and Austin have confined their
attention to invented examples, particularly in view of Austin’s
insistence that

the total speech act in the total speech situation is the only
actual phenomenon which, in the last resort, we are engaged in
elucidating.

However, the influence of Austin in particular is evident in all

aspects of research into spoken discourse. Workers in a whole
variety of disciplines—anthropology, sociology, linguistics, psycho¬
linguistics, applied linguistics—have adapted and modified the
concept of a speech act in their search for adequate analytic cate¬
gories. As the remaining chapters will make clear virtually everyone
acknowledges some debt to speech act philosophy.

Performatives and grammar

Recently there has been a growing interest within generative
semantics in the relationship between the grammatical rules which
generate sentences and the pragmatic rules which govern their use
as performative utterances. The first attempt to incorporate ‘speech
acts’ into linguistic theory was by Boyd and Thorne (1969) in a
novel analysis of modal sentences. Their general approach was to
‘treat the modal verbs as indicators of the illocutionary potential of
the sentences in which they occur’. They suggested that traditional
grammarians would have distinguished the following pair of sen¬
tences in terms of tense.

He lives in Edinburgh.
He will live in Edinburgh.

whereas they preferred to see the difference as a distinction between

statement and prediction. The other modals are analysed on similar
principles, though the analyses are more complex and less accept¬
able. While their analysis was flawed and quickly rejected, the idea
of incorporating performatives into the grammatical deep structure
was adopted and developed.
Ross (1970) argues that all sentences must be derived from deep
structures containing one and only one ‘explicitly represented per¬
formative as its highest clause’. He observes that transformational
grammarians have long accepted that sentences like ‘Go’, where no
subject need appear in the surface structure but where a second
28 AN INTRODUCTION TO DISCOURSE ANALYSIS

person subject is understood, should be derived from structures

which actually contain a noun phrase ‘you’ as subject in an em¬
bedded clause. Ross presents a series of highly intricate but con¬
vincing grammatical arguments to support the claim that declarative
sentences like

(a) Prices slumped.

(b) I like you when you giggle.
(c) Even Rodney’s best friends won't tell him.

must also be analysed as implicit performatives and derived from

deep structures containing an explicit performative verb, which is
later deleted. He therefore proposes that the underlying structure
for sentence (a) is not, as had been supposed, that represented in
tree (i) but rather that in tree (ii)

S C

NP VP NP

prices V

slumped VP
+ performative
+ declarative prices V

slumped
(i) (ii)
Ross admits that there is an alternative pragmatic analysis, which
would support derivation (i) by arguing that ‘certain elements
are present in the context of a speech act and that syntactic pro¬
cesses can refer to such elements’. His own preference for the
performative analysis at this stage is simply because it does work,
whereas the pragmatic one, while theoretically promising, has not
been attempted. However, later, Ross (1975) in a discussion of the
rules governing the interpretation of indirect speech acts argues that
one cannot ‘relegate syntactic and pragmatic processes to different
components of a grammar’. He concludes that

rather than it being possible for the ‘work’ of linking surface

structures to the sets of contexts in which these structures can
be appropriately used to be dichotomised into a set of
 SPEECH ACTS 29

pragmatic rules and a set of semantactic rules, it seems to be

necessary to postulate that this work is to be accomplished by
one unified component in which rules concerned with such
pragmatic matters as illocutionary force, speaker location ...
and the rules concerned with ... semantic matters ... and
[those] with syntactic matters ... are interspersed in various
ways.
Although Ross argues strongly for this mixed component
description, there are still major disagreements within generative
semantics, as Sadock (1975) reports, over whether one should have
separate rules of conversation to govern the interpretation of utter¬
ances as Gordon and Lakoff (1971) proposed, or whether the
grammar generates an utterance tagged with its illocutionary poten¬
tial as Ross proposes.
The evidence is that pure linguists and discourse analysts, both
drawing on speech act philosophy for their own purposes, are now
approaching common ground and that the results of their research
will soon begin to cross-fertilise.
 3

The ethnography of speaking

Chomsky set the goal of linguistic theory as the description of the

ideal speaker-hearer’s competence, his knowledge of grammaticality,
or whether or not putative sentences are part of his language. In
emphasising a concern with the grammatical rather than the appro¬
priate, Chomsky was willing to demonstrate simply the grammatical
relationships between ‘he hit me’, ‘it was me that he hit’, ‘it was
him that hit me’, without attempting to explain why one and not
another might be appropriate to a particular situation. Hymes
(1971) argues that Chomsky’s definition of competence is too
narrow—linguistics ought to concern itself with communicative
competence, the speaker’s ability to produce appropriate utterances
not grammatical sentences.
He suggests that ‘an adequate approach must distinguish and
investigate four aspects of competence: systematic potential; appro¬
priateness', occurrence', feasibility’. By systematic potential he
refers to ‘whether and to what extent something is not yet realised’
and suggests that ‘it is to this that Chomsky in effect reduces com¬
petence’. Appropriateness includes ‘whether and to what extent
something is in some context suitable, effective or the like’. Thus,
two features can vary independently: schizophrenic’s talk is often
marked by grammatical but inappropriate utterances, as in this
example already quoted from Labov (1970),
A: What is your name?
B: Well, let’s say you might have thought you had something
from before, but you haven’t got it any more.
A: I’m going to call you Dean.
while Albert (1972) reports that among the Burundi appropriate
but ungrammatical utterances occur frequently in certain situations
—differences in rank require a peasant-farmer to make ‘a rhetorical
fool of himself’ when his adversary is a prince or herder although
at other times he ‘may show himself an able speaker’.
 THE ETHNOGRAPHY OF SPEAKING 31

A speaker’s competence also includes knowledge about occurrence,

‘whether and to what extent something is done’. This theoretical
dimension provides for the fact that members of a speech com¬
munity ‘are aware of the commonness, rarity, previous occurrence
or novelty of many features of speech, and that this knowledge
enters into their definitions and evaluations of ways of speaking’.
The final dimension feasibility is concerned with ‘whether and to
what extent something is possible’. Hymes (1972) refers to the
experience of J. R. Fox working among the Cochiti of New
Mexico, who was unable to elicit the first person singular possessive
form of ‘wings’ on the grounds that the speaker, not being a bird,
could not say ‘my wings’—‘only to become the first person in
Cochiti able to say it, on the grounds that “your name is
Robin” ’.
Any utterance, or extended piece of discourse, can be described
in terms of these four dimensions. Thus Hymes suggests that
Leontes’ speech in Act II of The Winters Tale is ungrammatical,
appropriate, individual and difficult, while the bumbling speech of
the Burundi peasants is ungrammatical, appropriate, common and
awkward. Even if the scope of linguistics were expanded to cover
these four aspects of competence Hymes feels it would still be too
narrow and in his dissatisfaction feels the need to propose ‘a second
descriptive science of language’, the ethnography of speaking,
concerned not simply with language structure, but with language
use, with ‘rules of speaking ... the ways in which speakers associate
particular modes of speaking, topics, or message forms, with particu¬
lar settings and activities’ (1972). Any description of ‘ways of speak¬
ing’ will need to provide data along four interrelated dimensions:

(i) the linguistic resources available to a speaker—how many

different styles he can choose from;
(ii) supra-sentential structuring—how many differently struc¬
tured linguistic events, like trials, religious ceremonies,
debates, songs, are recognised;
(iii) the rules of interpretation by which a given set of linguistic
items comes to have a given communicative value;
(iv) the norms which govern different types of interaction.

Obviously, any attempt to produce a description in these terms

would be an enormous and perhaps impossible undertaking, and
thus all the work so far attempted within this framework is neces¬
sarily partial.
32 AN INTRODUCTION TO DISCOURSE ANALYSIS

The speech community

The initial task is to delimit the group of speakers for whom one is
going to produce ‘rules of speaking’. Hymes (1972) stresses that it
is not adequate to define a group as all those who have access to a
particular language or dialect; he argues that it is possible for
speakers to share formal linguistic features, phonology, grammar,
lexis, but still to be unable to interpret accurately the other’s
messages. For example, Labov (1972), in a discussion of aspects of
language use among adolescent New York negroes, presents utter¬
ances like ‘your momma’s a peanut man’, or ‘your mother’s a
duck’, which are superficially intelligible but whose real significance
as ritual insults is not available to most English speakers.
Speakers who apparently share the same language may also have
different ‘norms as to greetings, acceptable topics, what is said next
in a conversation’, how speaking turns are distributed and so on.
For example. Sacks, Schegloff and Jefferson (1974) suggest that for
American English there is a conversational rule that only one
speaker speaks at a time, whereas Reisman (1974) observes that in
Antigua ‘the start of a new voice is not in itself a signal for the
voice speaking to stop or to institute a process which will decide
who is to have the floor’. Any group which shares both linguistic
resources and rules for interaction and interpretation is defined as a
speech community and it is on such groups that ethnographers of
speaking concentrate.
Although the defining instance of a speech community is the
sharing of one linguistic variety, most communities have several
between which they switch. Blom and Gumperz (1972) report an
investigation in the Norwegian village of Hemnesberget where all
the residents speak both the standard language. Bokmal, and the
local dialect Ranamal. Bokmal is the language of formal education,
official transactions, religion and the mass media, but the local
dialect still enjoys great prestige, and by ‘identifying himself as a
dialect speaker both at home and abroad, a member symbolises his
pride in his community’. In any interaction speakers have a choice
of two varieties; Blom and Gumperz were interested in the factors
which influenced the use of one rather than the other. After close
observation and analysis of tape recordings of free speech they
concluded that locals would typically use local dialect except in
situations ‘defined with respect to the superimposed national
Norwegian system’. Even then, in the community administration
 THE ETHNOGRAPHY OF SPEAKING 33

office, where the standard language prevailed, clerks were observed

to switch depending on topic and,

when residents step up to a clerk’s desk, greetings and inquiries

about family affairs tend to be exchanged in the dialect, while
the business part of the transaction is carried on in the standard.

In other words, it appeared that topic could only cause speakers to

switch from standard to dialect—whereas in a standard language
situation talk about family affairs might be conducted in dialect, in
a gathering of friends and kin speakers would never switch to the
standard language, even if the topic were national or official matters.
To test their hypothesis Blom and Gumperz arranged to tape-
record the conversation of two groups of local residents, both self-
recruited and consisting of close friends and relatives. On both
occasions the investigators first stimulated discussion among the
group and then as the conversation progressed interjected questions
and comments feeling that the greater the range of topics the greater
the chance of a switch to the standard language. In fact, as predicted,

in several hours of conversation . . . marked by many changes

in topic, [they] found a number of lexical borrowings but not a
clear instance of phonological or grammatical switching.

At this point it would seem possible to write rules for the speech
community of Hemnesberget to predict the choice of one or other
speech variety for one type of speech event, conversation. However,
Blom and Gumperz recorded conversations among two more
groups, one composed of members of a formerly active peer group
who had spent the past few years away at university, returning only
in the summer, the other comprising three speakers from families
who tended to dissociate themselves from the local community. The
students claimed to be pure dialect speakers, but for them topic
was a significant variable: non-local topics evoked ‘a tendency to
switch towards standard phonology while preserving some morpho-
phonemic and lexical’ features of the dialect. For the other group
the local dialect was only used for local anecdotes, humour, and
attempts to provide local colour and the standard language was the
normal speech style. Thus, what appeared to be one speech com¬
munity was in fact at least three, each sharing the same linguistic
varieties but having different rules for their use.
These results are disturbing because they threaten the concept of
a speech community—just as some grammarians have retreated to
34 AN INTRODUCTION TO DISCOURSE ANALYSIS

a study of their own idiolect, the ethnographers of communication

may well be reduced to writing rules for two-member speech com¬
munities.

Speech styles
Any ethnography of speaking must describe the linguistic options
open to the speech community. As we have seen the residents of
Hemnesberget had a choice of two major varieties, the local dialect
and the standard language. Ferguson (1959) suggests that speakers
of Swiss German, Arabic, Greek and Tamil are faced with a similar
choice, this time between two standard languages—a high form
typically used in sermons, speeches, lectures, news broadcasts, and
a low variety used in conversations, political and academic discus¬
sion, ‘folk’ literature. Americans, according to Joos (1962), have a
choice not between major varieties but between five different degrees
of formality within the one standard language; Labov (1968) pro¬
vides supporting evidence, from the differential occurrence of post¬
vocalic /r/» of four degrees of formality.
Geertz (1960) reports that Javanese has three major styles which,
unlike those suggested by Joos and Labov for English, are recog¬
nised and named by speakers of the language—‘krama’, ‘madya’,
and ‘ngoke’, high, mid and low. These styles share some linguistic
features with other levels, but also have unique lexical items and
grammatical constructions. In addition there is a set of ‘honorifics’,
mostly referring to ‘people, their parts, possessions and actions’,
which occur independently of the first style-defining set of choices
and raise the style ‘half a notch’. These honorifics can only occur
with the low and high styles, at least in the usage of the educated
townsman, who thus has five recognisably distinct varieties to
choose between. Thus the question ‘are you going to eat rice and
cassava now?’ could have any of the five realisations on page 35 de¬
pending on the context and relative status and familiarity of the in¬
terlocutor—a townsman would use low with a friend, mid with a
non-intimate and high to a high official from whom he would in
turn receive low.
Halliday, McIntosh and Strevens (1964) suggest that it is possible
to make much finer distinctions and argue that a speaker’s linguistic
repertoire consists of a large number of registers, or varieties distin¬
guished according to use. This is an intuitively attractive concept,
for speech communities within the same language area could then
be distinguished by the range of registers they had available, but
 THE ETHNOGRAPHY OF SPEAKING 35
R Q
*§. £/)
-S
a ^ Co
R
w o ^ ^ -§.
O e a ~e. <=)- R
z a -s*: 3 =3
§ ^ R
w «© -s* S?
£=-> 3
C K s « & 3
§
c
-S.
;&* .Q Q 3 <n
w 3
</] ^ Q S'S' St
u ^ -St; <3 J ^ 3
H s: 3 "<5J
.« cv.
W Q C\ . V5P’s ^3
-I ^ a ^ 3 3
p* t* c g a
<3 a
S §/$£ § ^ I |
o 3r ^ 3 CO *“^>
§ § | C <3
CX> 33. <fc
Oj
~ <3 3
§ *3^ t-
^ <SJ Q §-| jjj-
^ ■§. § < K Oj <; oo X -St; ^1
«

£ 5
o 3
c 3 3
3 >>
Q
*n

co
> ‘CU
CO
t/5
C/3 Q
CO
O
-St;

K
-a a
c J
cO Q
-it:
<L> Q
O ^0
co <o

3
Sn. 3
C3 <3
<L> -R ■§.
CU §?
3
•§. C
5

ox)
c ^3 S?
3
'o
OX) -C> ■S' a
3

c
a
c
t?
3 cu &
3 £
O cu O
■S'
>3
Q
g
a
co

Q
<D
3 Q
u §■
ctf § R 5P

J
— cd c3
> rn <N
-
J
36 AN INTRODUCTION TO DISCOURSE ANALYSIS

registers prove difficult to isolate and define. Halliday et al observe

that a sports commentary, church service and school lesson are
‘obviously linguistically distinct’, and suggest that there is a register
appropriate to each. This raises the question of how different
registers can be recognised and isolated and they suggest that while
there will be grammatical differences the major distinctions between
registers will be lexical. However, the claim that
some lexical items suffice almost by themselves to identify a
certain register: ‘cleanse’ puts us in the language of advertising,
‘probe’ of newspapers, especially headlines, ‘tablespoonful’ of
recipes or prescriptions, ‘neckline’ of fashion reporting or
dressmaking instructions,
is worryingly naive. In fact there are no restrictions on the concept:
a register can apparently vary in size and importance from that
of dressmaking to that of scientific English (Huddleston et al, 1968);
and registers are circularly defined: the language used in dress¬
making patterns is the register of dressmaking and the register of
dressmaking is that used in dressmaking patterns.
Hymes (1974) suggests that it is more useful to see a speech
community as comprising a set of styles, where style is used in the
neutral sense of ‘a way or mode of doing something’. Whereas style
has often been used as a concept to account for variation according
to author, setting or topic it has never been used as the general basis
of description. This, Hymes suggests, is now possible if one exploits
the long recognised fact that a linguistic item can be described along
two axes, vertical and horizontal, paradigmatic and syntag/natic, or
choice and chain. For example, for the noun group ‘the dog’, one
of the things a description must do is account for the other items
which could occur as paradigmatic choices instead of ‘the’; the
other thing a description must do is to characterise the set of items
which can occur following ‘the’.
Drawing on Ervin-Tripp (1972) Hymes suggests that the concept
of syntagmatic relations can be generalised to handle the co¬
occurrence of items over larger stretches and this will allow one to
‘characterise whatever features go together to identify a style of
speech in terms of the rules of co-occurrence among them’. The
concept of paradigmatic choice can be similarly generalised to one
of alternation to cope with the choice between styles.
The concept of style may seem very close to that of register but
there is a crucial difference: registers are mainly defined and recog¬
nised by topic and context-specific lexis—the register of sermons is
 THE ETHNOGRAPHY OF SPEAKING 37

the language used in giving sermons; styles, however, as the rules of

alternation emphasise, are not mechanically connected to particular
situations—speakers may choose among styles and their choices
have social meaning. One of the most reliable ways of making
people laugh is to adopt a style inappropriate to a particular con¬
text or message.
It is of course one thing to define a style as a set of co-occurrent
choices; it is another to isolate different styles. As Hymes observes
the relevant speech styles of a community cannot be arrived at
mechanically for one could note an infinite number of differ¬
ences and putative co-occurrences.
The aim, therefore, is to isolate significant speech styles, that is
ones that speakers can distinguish and use.
Hymes accepts that some stylistic features may be present in a
piece of discourse without defining a significant style; their presence
may simply convey a ‘tinge or character’, but not be an organising
principle and this is close to the everyday definition of style. How¬
ever, Hymes recognises two kinds of groupings of stylistic features
which do constitute organised use—those which colour or accom¬
pany the rest of what is done, stylistic modes, and those which can
be said to define recurrent forms, stylistic structures. A principal
aspect of stylistic modes is a set of modifications entailed in con¬
sistent use of the voice in a certain way, as in singing, intoning,
chanting, declaiming. As an example of the importance of mode
Hymes refers to the basic distinction among the Wolof of Senegal
between ‘restrained’ and ‘unrestrained’ speech, distinguished
principally by paralinguistic features; restrained speech being
characterised as low pitched, breathy, slow, soft with final pitch
nucleus, unrestrained as high pitched, clear, fast, loud with initial
pitch nucleus.
Stylistic structures, as the name implies, are verbal forms organised
in terms of defining principles of development or recurrence. One
kind of structure is the organisation of sentences and utterances into
larger units such as greetings, farewells, prayers; the other is the
systematic exploitation of arbitrary linguistic features which
Sinclair (1971) calls latent patterning—at the rank of word
poets frequently use such features as initial consonant, final syllable,
positioning of stress to add an extra layer of patterning which we
recognise as alliteration, rhyme and metre respectively. Repetitions
at regular intervals of these patterns create structures we call verse
forms.
38 AN INTRODUCTION TO DISCOURSE ANALYSIS

Hymes calls these structures elementary or minimal genres, and

observes that both kinds of groupings of features, modes and struc¬
tures, enter into more complex groupings called complex genres.
A church service would be an example of a complex genre, con¬
taining the elementary genres of hymn, psalm, prayer and sermon,
and evidencing the stylistic modes of singing, chanting and perhaps
declaiming.
The work of Bricker (1974) on Mayan provides a useful exemplifi¬
cation of Hymes’ concept of style. She notes an initial division into
formal and informal genres, the formal comprising myth’, ‘prayer’,
‘song’, ‘contemplation’, ‘planning’, ‘war’, ‘argument’, and ‘frivo¬
lous talk’, the informal three other types of ‘frivolous talk’, ‘gossip’
and ‘discussion’. All the formal genres ‘are structurally alike: they
are expressed as semantic couplets’; the informal genres have no
common structure, but specifically avoid couplets. The following
prayer illustrates the couplet structure.

Well grandfather.
Lord:
How long have you been waiting here for my earth?
How long have you been waiting here for my mud?
I am gathering together here;
I am meeting here.
I see the house of poverty;
I see the house of wealth
Of His Labourer,
Of His tribute-payer.
Holy Esquipulas, thou art my father;
Thou art my mother.

Obviously the organising principle is the highly marked semantic

and syntactic parallelism. These couplets happen to be from a
prayer but apparently could just as easily come from a song, for
songs closely resemble prayers in their context, content and function
—the crucial distinction is in the stylistic mode, ‘prayers are simply
recited, while songs are sung to a musical accompaniment’. Natur¬
ally some genres will have a more rigid and overt structure than
others—indeed until recently many considered that conversation
had no identifiable structure at all. Hymes suggests that for conver¬
sation the distinctive modes and structures are simply more difficult
 THE ETHNOGRAPHY OF SPEAKING 39

to identify, and the work of Sacks, Schegloff and* Jefferson, described

in detail in Chapter 4, provides growing evidence of a high degree
of structuring in conversation.

Speech events
Hymes stresses that it is essential to distinguish a genre, which is a
unique combination of stylistic structure and mode, from the ‘doing’
of a genre. In order to emphasise the distinction between genre and
performance, a distinction frequently obscured by users of a
language, who often employ the same label for both, Hymes sug¬
gests the categories of speech event and speech act to parallel com¬
plex and elementary genres. All genres have contexts or situations
to which they are fitted and in which they are typically found. Some
genres, like ‘conversation’, can occur appropriately in a wide range
of situations, some, like ‘prayer’, are highly restricted; however, it
is a defining criterion of a genre that it is a recognisable style and
therefore can be used in inappropriate situations. The cultural
implications of an inappropriate use of a particular genre like
prayer may of course differ; in one culture the result may be
laughter, in another death.
Speech events occur in a non-verbal context, the speech situation,
which may or may not affect the choice of genre and ‘it is for speech
events and speech acts that one writes formal rules for their occur¬
rence and characteristics’. Speech events are the largest units for
which one can discover linguistic structure and are thus not coter¬
minous with the situation; several speech events can occur suc¬
cessively or even simultaneously in the same situation, as for in¬
stance with distinct conversations at a party.
The relationship between speech events and speech acts is hier¬
archical, ‘an event may consist of a single speech act, but will often
comprise several’. Speech acts may often consist at the grammatical
level of single sentences but they are not equivalent to them. Rather
they are functional units, similar to Austin’s speech acts and they
derive their meaning or value not from the grammatical form but
from the speech community’s rules of interpretation. Hymes notes
that for English

a sentence, interrogative in form, may be now a request, now a

command, now a statement; a request may be manifested by a
sentence that is now interrogative, now declarative, now im¬
perative in form.
40 AN INTRODUCTION TO DISCOURSE ANALYSIS

One ultimate aim of the ethnography of speaking is an exhaustive

list of the speech acts and speech events of a particular speech
community, though the descriptive framework is currently ‘heuristic’
and ‘quite preliminary’. Already work by Sacks (passim) and
Sinclair et al (1972) suggests that there is a need for more than two
functional units—Hymes offers as examples of speech acts ‘request’,
‘command’, ‘greeting’, ‘joke’, but Sacks has shown that greetings
and some jokes consist of more than one speech act and yet are only
parts of a single event. There have been several detailed descriptions
of speech events; one of the clearest is Labov’s (1972) discussion of
‘ritual insults’.
There is a speech event which occurs in many American negro
speech communities and goes by a series of different names, ‘the
dozens’, ‘sounding’, and ‘signifying’ being the most common. The
event is a contest, with definite winners and losers, in which one
speaker pits his linguistic skills against another in the production of
ritual insults (speech acts). The form of the insults varies from
place to place and over time; Labov reports that for an earlier
generation sounds consisted of ‘rhymed couplets, always sexual,
and aimed at the sexual degradation of the object’:

iron is iron and steel don’t rust

but your momma got a pussy like a greyhound bus.

These couplets could follow each other in any order and there was
no question of a speaker improvising during a contest—the winner
was the speaker with the best memory or the best delivery. Currently
in New York, however, sounding is an adolescent habit and the
rules have changed: couplets are rare and the sounds are often not
sexual; instead successive sounds are related semantically and the
skilful sounder is the one who can build on and develop the previous
sound. In other words the skill now resides in quick thinking rather
than a good memory. Each sound is evaluated by the audience,
which is an essential participant in the speech event. The opening
utterance ‘almost always contains a reference to B’s mother’; the
second speaker’s reply is based on the first utterance, and ‘to the
extent that it is an original or well-delivered transformation of A, B
may be said to have won’. The first speaker can continue with an
entirely new sound, but if he can build on B’s contribution ‘it is
usually evaluated more highly’. The series continues until one
speaker drops out or is judged by the audience to have lost.
The following sequence begins with an exemplificatory sound
 THE ETHNOGRAPHY OF SPEAKING 41

produced for the investigators but responded to, as if it were a

real sound, by one of the other boys.

A: Your father got brick teeth.

B: Your father got teeth growing out his behind.
{laughter from all the other boys)

This is a winning effort; B takes A’s sound ‘and adds to it elements

of absurdity and obscenity which are positively evaluated by the
audience’. A tries to top this but fails and is crushed by B’s further
transformation of the original:

A,: Yeah, your father, y-got, your father grow, uh, uh, grow
hair from, from between his, y’know. {one boy laughs)
B2: Your father got calluses growin’ up through his ass, and
cornin’ through his mouth, {all laugh)

B, seizing his advantage, begins another round,

B3: Your father look like a grown pig. {all laugh)

but A, unable to think of a suitable response switches to a personal

rather than a ritual insult, and makes fun of the B’s father’s stam¬
mer:

A3: Least my—at least my father don’t be up there talking

uh-uh-uh-uh-uh-uh.

In themselves sounds are superficially harmless for utterances can

only function as sounds if they are patently untrue. However, such
is the importance placed on winning that the speaker who is losing
is often tempted to change the nature of the speech event into the
more overtly aggressive one of real insults.

Components of speech events

So far the discussion of speech acts and speech events has concen¬
trated on stylistic mode and structure and for many acts and events
these are the defining criteria—a song is a song whoever sings it; at
least in our culture. However, some genres are performed for
specific purposes in specified places with particular participants. An
Anglican baptism traditionally takes place beside the font with
seven essential participants—the parson, the unbaptised baby, the
parents and three godparents, and the definition and description of
42 AN INTRODUCTION TO DISCOURSE ANALYSIS

the speech event requires participants and situation, as well as

style, to be specified.
For every speech event Hymes recommends that the ethnographer
initially provides data on structure, topic, participants, setting,
purposes and channel (spoken, written, whistled, drummed), so that
knowing the possible parameters one can check whether an appar¬
ently irrevelant one is in fact irrelevant. Hymes reports that Arewa
and Dundes (1964) investigating the uses of language among the
Yoruba, observed that proverbs were only used by adults and were
always spoken, but pressing the point discovered they could also be
drummed, in a slightly altered form, and used by children as part
of a formulaic apology. In other words by being aware of the
possible parameters the ethnographer can more easily and success¬
fully discover the constraints on the performance of genres, and the
defining criteria of particular speech events.

Setting
All speech events occur of necessity in time and space—sometimes
it is one of the defining criteria of an event that it occurs at a specific
time or in a specific place. Foster (1974) describes a series of fifteen
agricultural festivals which the Iroquois celebrate at appropriate
points during the year. At two of the festivals two speech events
concerned with asking the Creator for successful crops occur, the
Tobacco Invocation and the Skin Dance, while at the other thirteen,
the major speech event is the Thanksgiving Address. Salmond
(1974) on the other hand, reports a speech event among the Maori,
the Encounter Ritual, which can occur at any time but only in a
‘marae’, a complex consisting of carved meeting house and court¬
yard for orators. Closer to home we also have speech events tied to
a particular time—special church services for Easter, or the Queen’s
Christmas message; or to a particular place—there is a very
restricted number of places where marriages can be solemnised or
litigation occur. Even when a speech event is not restricted to a
particular setting, the setting may affect either the stylistic mode—
people tend to speak in hushed tones in church; or the stylistic
structure—Geertz reports that the Javanese ‘would be likely to use
a higher level to the same individual at a wedding than in the street’.
Hymes stresses that the ethnographer must also take note of the
‘psychological setting’ of an event—the cultural definition of an
occasion as formal or informal, serious or festive. Frake (1972)
compares litigation among the Subanun and the Yakan. For both
 THE ETHNOGRAPHY OF SPEAKING 43

litigation is ‘an integral speech event concerned with settling dis¬

putes by means of a ruling formulated by neutral judges’; the major
difference is not in the event itself but in its place in the overall
structure of the culture. The Subanun divide activities sharply into
festive and non-festive; litigation is festive behaviour and often
accompanied by eating, drinking and merrymaking. The festive
nature of the occasion conditions the choice of style—both litigants
and judges employ esoteric legal language, often arranged into verse
form and sung to the tunes of drinking songs. Yakan litigation on
the other hand occurs in a very informal atmosphere and the process
is initially indistinguishable from ‘a group of people talking to¬
gether’. The underlying structure of both speech events is very
similar, but the psychological setting and resulting style very
different.

Participants
Traditionally speech has been described in terms of two participants,
a speaker who transmits a message and a listener who receives it.
Hymes argues that there are at least four participant roles addressor,
speaker, addressee and hearer or audience, and that while conversa¬
tion may require only an addressor and an addressee, other speech
acts require different configurations. As we saw, ritual insults
require three participant roles, one being an audience whose func¬
tion is to evaluate each contribution. When one considers the neces¬
sary participants for a whole speech event the situation is often
more complex. Sherzer (1974) describes a speech event among the
Cuna called ‘chanting’ in which two chiefs perform a ritualised
interaction in front of an audience—one chief chants, and at the
end of each verse the other responds ‘thus, it is so’. The responder
is essential; a chief cannot chant on his own. Then, when the chant¬
ing is over a third participant, the chief’s spokesman addresses the
audience directly and interprets for them.
There are some speech events which have only one human
participant—for instance in our culture some forms of prayer.
Sherzer (ibid) describes disease-curing events among the Cuna
where the participants are the curer and a group of wooden dolls,
‘stick babies’, which are considered to carry out the actual business
of curing once they have been told, as addressees, what to do.
Hymes points out that non-humans can also be taken as addressors,
citing an occasion when following a clap of thunder an old Indian
asked his wife if she had heard what the thunder had said.
44 AN INTRODUCTION TO DISCOURSE ANALYSIS

Any description of a speech community must include data on who

and what can fill the participant roles, and in what speech events
and speech acts. Some speech events simply require that certain
participant roles be filled—anyone can act as audience to a play or
ritual insult; other events require participants of a particular age,
sex, kinship relation, status, role or profession—only Cuna chiefs
can chant; initiation or puberty rites are almost invariably sex
specific; while in most Maori tribes only male elders can deliver
speeches on the ‘marae’. In other events turns to speak are regu¬
lated by relations between particular participants—Albert (1972)
reports that among the Burundi turns to contribute to a debate
are strictly controlled by relative status, with the most important
speaking first, the least important last; while the Wolof have a
rule that in greetings the lower status speaker begins first (Irvine
1974).
Certain participant features and particularly certain kinds of
relationships between participants directly condition the choice of
linguistic items in speech. As we mentioned above the choice of a
particular styleme in Javanese reflects the degree of familiarity of
the speakers and their relative status; the residents of Hemnesberget
use the standard language with strangers; and many European
languages use the choice between singular or plural second person
pronoun, ‘tu’ or ‘vous’, ‘du’ or ‘sie’, to a single addressor to mark
familiarity or distance.

Purpose
All speech events and speech acts have a purpose, even if occasion¬
ally it is only phatic. Sometimes several events share the same style
and are distinguished only by purpose and participants or setting.
Hymes notes that among the Wai Wai of Venezuela the same genre,
the ‘oho-chant’, is used for a series of speech events which are
distinguished according to their function in marriage contracts,
trade, communal work tasks and invitations to feasts.
Frake (1972) reports four speech events among the Yakan, dis¬
tinguished by purpose—mitin, ‘discussion’, qisum ‘conference’,
mawpakkat ‘registration’, and hukum ‘litigation’. Initially, to the
outsider there is no difference—no special setting, clothes or para¬
phernalia, no activity other than talk. Mitin is the most general and
apparently refers to unfocussed, purposeless conversation in which
all participants have equal speaking rights. Qisum is a discussion
with a purpose, some issue such as when to plant rice has to be
 THE ETHNOGRAPHY OF SPEAKING 45

decided and again all participants have equal speaking rights, but
this time the event has a recognisable end when a decision is
reached. Mawpakkat is a negotiation over a disagreement, its pur¬
pose is to reach a settlement, and now the participants are divided
into two protagonistic sides. Finally, hukum is concerned with a
disagreement arising over an offence, the purpose is to reach a legal
ruling based on precedent and this requires additionally a court
comprising a set of neutral judges.
Hymes observes that ‘the purpose of an event from a community
standpoint may not be identical to the purposes of those engaged in
it’. At every level of language individuals can exploit the system for
personal or social reasons or artistic effects. Irvine (1974) describes
a speech event among the Wolof, the ‘greeting’, which ‘is a neces¬
sary opening to every encounter, and can in fact be used as a defi¬
nition of when an encounter occurs’. Relative rank determines who
greets whom—it is customary for the lower ranking party to greet
the higher and there is a proverb ‘when two persons greet each
other, one has shame, the other has glory’. However, individuals do
not always wish to take the higher status position because along
with prestige goes the obligation to contribute to the support of low
status persons. For this reason a higher status person may indulge
in ‘self-lowering’ by adopting the opening role, and Irvine observes
that no one ever asked her for a gift if they had not first managed
to take the lower status role in the interaction.

Key
Within key Hymes handles the ‘tqne, manner or spirit’ in which an
act or event is performed. He suggests that acts otherwise identical
in setting, participants, message, form, etc. may differ in key as
between mock and serious, perfunctory and painstaking. Sacks has
observed that the first question one must ask of any utterance is
whether it is intended seriously and Hymes emphasises the signifi¬
cance of key by observing that when it is in conflict with the overt
content of an act, it often over-rides it. Thus ‘how marvellous’
uttered with a ‘sarcastic’ tone is taken to mean the exact opposite.
The signalling of key may be non-verbal, by wink, smile, gesture
or posture, but may equally well be achieved by conventional units
of speech like the aspiration and vowel length used to signal empha¬
sis in English. The Wolof greeting discussed above is normally
begun by the lower status speaker and responded to by the higher
status; there are also paralinguistic features associated with each of
46 AN INTRODUCTION TO DISCOURSE ANALYSIS

these roles, classifiable on the dimensions of ‘stress’ and ‘tempo/

quantity’:
STRESS TEMPO/QUANTITY

Noble s (— high, — loud) t (— rapid, — verbose)

Griot S (+ high, -f loud) T (+ rapid, + verbose)
Thus the opening greeting normally has the associated paralinguistic
features ST, the response st. However, if a speaker wishes to indicate
that the status assigned by his role is at variance with his true status
he does this by using an inappropriate stress pattern—a speech
style sT will sometimes be used by a noble who has taken the role
of initiator but wants to indicate that (he knows) he is being polite.
He is showing deference (initiator role and T) even though he doesn’t
have to (s).

Channels
Under channel the description concerns itself with the ‘choice of
oral, written, telegraphic, semaphore, or other mediums of trans¬
mission of speech’. Most genres are associated with only one channel
and an attempt to use a different channel, as in the case of the
drumming of Yoruba proverbs, necessitates some changes. The
development of radio and television has created a situation in which
some speech events have enormous unseen and unheard audiences,
which subtly affect the character of the event. What is superficially
a round-table discussion or a cosy fireside chat is in fact an oppor¬
tunity to attempt indirectly, to sway a nation’s opinions. The chan¬
nel itself has even allowed the creation of new speech events, the
sports commentary and the quiz show, with their own highly
distinctive stylistic mode and structure, prescribed participants,
typical setting and key.

Message content
Hymes suggests that ‘content enters analysis first of all perhaps as
a question of topic, and change of topic’. For many events and acts
topic is fully predetermined and invariable, though for others,
particularly conversation, topic is relatively unconstrained. In some
communities topic may have little effect on style, in others it may
be strongly marked. Geertz (1960) observed that the Javanese used
lower stylemes ‘when speaking of commercial matters, higher ones
if speaking of religious or aesthetic matters’; while as mentioned
above topic was a condition of code-switching in the village of
 THE ETHNOGRAPHY OF SPEAKING 47

Hemnesberget. Although Hymes stresses the importance of message

content, it is an aspect of the speech event virtually ignored by
ethnographers of speaking.
The main components of speech events have been artificially
separated for the purpose of exposition; any detailed description
of a speech event must include information on all components and
on the inter-relations between them, though Hymes suggests that
the relative importance of particular components will vary from
community to community: ‘for one group rules of speaking will be
heavily bound to setting; for another primarily to participants; for
a third perhaps to topic.’

Rule breaking
A successful ethnography of speaking will describe the normative
structure of all the speech acts and events of a given speech com¬
munity by detailing for each act and event the necessary configura¬
tion of components and style. Norms, of course, are not always
adhered to and each community has its own rules for interpreting
rule-breaking. When Blom and Gumperz replayed the tape of the
Hemnesberget students discussion to other residents they at first
refused to believe that the speakers were members of the same
speech community and then, when they recognised the voices,
expressed disapproval of their ‘artificial speech’. On this occasion
the rules were apparently not broken deliberately—some of the
participants were also surprised at their own code-switching—but
very frequently rule-breaking is deliberate and for specific effect.
Salmond (1974) suggests that the main justification for writing a
detailed description is that

only when the rules are laid down as economically as possible

and all the options are made clear can an outsider appreciate
the manipulation that people practise.

She reports that at the Maori ‘rituals of encounter’ deliberate rule¬

breaking results in a great loss or gain of prestige—the unsuccessful
contender leaves the ‘marae’ in utter humiliation, the successful is
greatly honoured having proved himself above the constraints that
bind ordinary people.
On one famous occasion one Maori group from a part of the
country where women were allowed to speak was visiting another
where women were not. The hosts opened the oratory but when it
came to the guests turn that was a problem, for the most senior in
48 AN INTRODUCTION TO DISCOURSE ANALYSIS

rank was an old chieftainess. After a moment’s hesitation she began

to speak. Immediately there was a protest from the hosts but the
chieftainess calmly ignored them, continued her speech to the end
and then said
‘You Arawa men, you tell me to sit down because I am a
woman, yet none of you would be in the world if it wasn’t for
your mothers. This is where your learning and your grey hairs
come from!’
then turning her back on them she bent over and flipped up her
skirts ‘in the supreme gesture of contempt’. Most rule-breaking is
less flamboyant and less risky than this.
In the 16th century, English, like many modern European
languages, distinguished two second person singular pronouns, ‘you’
and ‘thou’. It was customary for nobles to use ‘you’ reciprocally, to
receive ‘you’ from their inferiors but to address them as ‘thou’. If
a speaker broke the ‘rules’, the rule-breaking was meaningful and
thus Sir Edward Coke was able to insult Sir Walter Raleigh at his
trial by addressing him as ‘thou’:
All that he did at thy instigation, thou viper; for I thou thee,
thou traitor.
Ervin-Tripp (1972) presents a similar insult:
policeman: What’s your name, boy?
doctor: Doctor Poussaint. I’m a physician ...
policeman: What’s your first name, boy?
doctor: Alvin.
Ervin-Tripp observes that the policeman insulted the doctor three
times. Firstly, he employed a social selector for race, in addressing
him as ‘boy’; secondly, he treated the reply as a failure to answer,
a non-name; thirdly he repeated the term ‘boy’ emphasising the
irrelevance of the name Dr Poussaint. So Ervin-Tripp points out
communication had been perfect in this interchange. Both were
familiar with an address system which contained a selector for
race available to both blacks and whites for insult, condescen¬
sion or deference, as needed. Only because they shared these
norms could the policeman’s act have its unequivocal impact.

Norms of interaction
All communities have an underlying set of non-linguistic rules which
governs when, how and how often speech occurs. Thus the Anang
 THE ETHNOGRAPHY OF SPEAKING 49

value speech highly and the young are trained in the arts of speech,
while for the Wolof speech, especially in quantity, is dangerous and
demeaning. French children are encouraged to be silent when
visitors are present at dinner; Russian children are encouraged to
talk. Among the Arucanian there are different expectations of men
and women, men being encouraged to talk on all occasions, women
to be silent—a new wife is not permitted to speak for several
months.
Even within North-Western Europe there are surprising differ¬
ences. One ethnographer reports how, when he was researching in
Iceland, neighbouring Eskimos would visit once a day for an hour
to check that all was well. During the hour there would be no more
than half a dozen exchanges, and all the rest of the time was spent
in silence. Another ethnographer describes staying with in-laws in
Denmark and being joined by an American friend who, despite
warnings, insisted on talking with American intensity until ‘at
9 o’clock my in-laws retired to bed; they just couldn’t stand it any
more’.
Other norms govern the physical distance at which speech events,
particularly conversations, take place. Watson and Graves (1966)
report that compared with Americans, Arab students confront each
other more directly when conversing, sit closer, are more likely
to touch each other and speak more loudly, behaviour which is
often interpreted as aggressive or over-friendly by Americans.
Again, there are cultural differences in the ways in which speakers
deal with encoding difficulties, occasions when they are forced to
stop midway through a grammatical structure. Hymes (1972) sug¬
gests that for white middle-class Americans the normal hesitation
behaviour is to pause, and often fill the pause with ‘urn’, ‘er’, and
then to continue, while for many blacks, the normal pattern is to
recycle to the beginning of the utterance, (sometimes more than
once). Recycling is a feature sometimes evident in children’s speech
but may be interpreted by whites as a defect in adults.
The clashes of norms described so far may produce some personal
discomfort, tension and even unjustified censure, but normally the
norm-breaking is accepted as the performance of someone who
doesn’t share the same norms. Major problems can arise, however,
when participants assume that they do share the same norms. Polgar
(1960) discovered that Mesquaki Fox children interpreted the nor¬
mal loudness of voice and directness of American English teachers
as ‘meanness’ and ‘getting mad’, and even more serious has been
50 AN INTRODUCTION TO DISCOURSE ANALYSIS

the misinterpretation of the behaviour of thousands of Negro chil¬

dren in New York schools. These children were observed to be fail¬
ing at school—on all standard tests of reading, verbal and non-verbal
abilities they were grossly retarded; in individual interviews with
educational psychologists they said very little and many of them
appeared to be ‘linguistically deprived’. Bereiter and Engelmann
(1966) working with four-year-olds in Urbana claimed that they
communicated by gestures, ‘single words’ and ‘a series of badly-
connected words or phrases’ such as ‘they mine’ or ‘me got juice’.
They observed that the Negro children could not ask questions and
‘without exaggerating . . . these four-year-olds could make no state¬
ments of any kind’. From observations like these Bereiter and
Engelmann concluded that one must treat these children as if they
had no language at all, and therefore they devised a pre-school
program using formal language drills to teach the children English.
Even when the teaching was underway, they observed that some
children, having been taught

two plus one equals three . . . would continually lapse into

amalgamations, ‘two pluh wunic’k three’. Having done this
they were no longer able to substitute other numbers for the
‘one’, it having become fused with the beginning sound of
‘equals’.
Labov (1969) argues vehemently that the concept of ‘verbal
deprivation’ has no basis in social reality—it is a nonsense created
by educational psychologists who ‘know very little about language
and even less about Negro children’. Labov argues that to say the
children have no language or are even linguistically deprived is a
complete misunderstanding—they come from a culture where
linguistic ability is highly valued, as is evident from the importance
of ‘sounding’ and the fact that the verbally most skilful boy is
usually the peer group or gang leader. The truth is that the children
do not choose to display their abilities at school because they are
not ones the school values and because the school is a hostile
situation.
Negro children faced with a large, though friendly coloured
interviewer, let alone a white interviewer as used in Bereiter and
Engelmann’s tests, choose to produce monosyllabic, non-committal
answers, whereas white middle-class children are willing to chatter
away in the ‘same’ situation. It is because the situation or rather
the speech event is assumed to be the same, with the same norms of
 THE ETHNOGRAPHY OF SPEAKING 51

interaction, that the psychologist and sociologist alike feel able to

compare the performance of different class and ethnic groups—
only when one realises that the norms are not the same can one
perceive the uselessness of the exercise.
In a striking demonstration that the psychologists were wrongly
inferring from what the children said to what they were able to say,
Labov took a rabbit into a classroom where Negro children were
dutifully ‘learning English’:

T. This is a book. What is it?

P. It is a book.
T. What colour is the book?
P. It is a red book.

He explained to the children that he had a rabbit that was very shy
but if a few of them would take the rabbit into the next room and
talk to it, it would be quite happy. In the next room everything
these children said to the rabbit and each other was tape recorded
and they rapidly displayed grammatical sophistication far in excess
of the structures being drilled to their fellows next door.

Rules of interpretation
All speech communities have unique rules for interpreting the
messages conveyed both verbally and non-verbally, and thus the
same formal realisations may have different values in different
communities. One of the aims of the ethnography of speaking is to
describe and systematise the interpretive rules used by members of
the speech community. In fact those working within the ethno¬
graphy of speaking have not, so far, made a great deal of progress
in this direction, but scholars in other disciplines, working for
different purposes on English conversation, have made considerable
advances, which are described in detail in the next chapter.
 4

Conversational analysis

Hymes (1972) stresses that it is heuristically important for the

ethnography of speaking ‘to proceed as though all speech has
formal characteristics of some sort as manifestation of genres’,
though he admits that the very notion of casual unmarked speech
points to the fact that there is a ‘great range among genres in the
number and explicitness of formal markers.’ Perhaps for this reason
those ethnographers of speaking who have provided detailed
structural descriptions have focused on well defined and often
ritualised events—greetings (Irvine 1974), ritual encounters (Salmond
1974). chanting (Sherzer 1974), ritual insults (Labov 1972), while
those who have looked at conversation have examined not its
structure, but factors affecting the choice of code or style (Blom
and Gumperz 1972, Gumperz 1964, Geertz 1960).
Until very recently most of the advances in conversational analysis
had been made by a small group of sociologists. Sacks, Schegloff
and Jefferson, who, paradoxically, stress that they work with
conversational materials,

not because of a special interest in language, or any theoretical

primacy [they] afford to conversation,
(Schegloff and Sacks 1973)

but because they see conversational analysis as a first step towards

achieving a ‘naturalistic observational discipline’ to deal with
details of social interaction in ‘a rigorous, empirical and formal
way’. Nevertheless, their findings are useful to and usable by any¬
one interested in the structure of conversation.

Turn-taking
One of the basic facts of conversation is that the roles of speaker
and listener change, and this occurs with remarkably little over¬
lapping speech and remarkably few silences. Sacks (ms) suggests
 CONVERSATIONAL ANALYSIS 53

that there is an underlying rule in American English conversation—

‘at least and not more than one party talks at a time.’ This is not an
empirical fact, because there are obviously many instances of short
pauses and short overlaps, but rather a normative or ‘observably
oriented to’ feature of conversation, in other words it is a rule used
by conversationalists themselves. If more than or less than one party
is talking it is ‘noticeable’ and participants set out to ‘remedy’ the
situation and return to a state of one and only one speaker. If the
problem is more than one speaker one of the participants usually
quickly yields the floor,

lori: But that wz—Then you wentuh Fre:ds

ellen: Wer-we left—we left—
ben: Lno. That’s the time we left Fre:ds

while if the problem is silence other speakers begin speaking, or

indicate their intention to speak by noises like ‘er’ or ‘mm’. In
other words turns to speak typically occur successively without
overlaps or gaps between them. Overlapping is dealt with by one
speaker ending his turn quickly, gaps between turns by another
speaker beginning his turn or simply indicating that his turn has
begun and incorporating the silence into it.
A second feature of conversation is that speaker change recurs,
and this presents problems for the participants—how can they
achieve change of speaker while maintaining a situation in which
at least, but not more than, one speaker speaks at a time?
Sacks suggests that a current speaker can exercise three degrees
of control over the next turn. Firstly, he can select which parti¬
cipant will speak next, either by naming him or by alluding to him
with a descriptive phrase, ‘the Right Honorable Member for Bexley
South’. If he selects the next speaker, the current speaker usually
also selects the type of next utterance by producing the first part of
an adjacency pair (see below p. 70), for example a question or a
greeting which constrains the selected speaker to produce an appro¬
priate answer or return greeting.

doctor: Hello Mrs Jones

patient: Hello Doctor
doctor: Hello Catherine
child: Hello
The current speaker’s second option is simply to constrain the next
utterance, but not select the next speaker, while the third option is
54 AN INTRODUCTION TO DISCOURSE ANALYSIS

to select neither and leave it to one of the other participants to

continue the conversation by selecting himself.
Sacks emphasises that these options are in an ordered relationship
—the first over-rides the second and the second over-rides the third.
If the current speaker selects the next speaker he alone should talk
next; Sacks notes that when an unselected speaker takes a turn
already assigned to a selected one, the right of the selected speaker
to speak next is usually preserved:

A (to C): Tell us about yourself so we can find something bad

about you.
B: Yeah hurry up.

Importantly, these selection techniques operate only utterance by

utterance: there is no mechanism in conversation by which the
current speaker can select the next-but-one speaker—choice of the
next speaker is always the prerogative of the current speaker if he
chooses to exercise it. In more formal speech situations—class¬
rooms, courtrooms, formal discussions—it is, of course, quite pos¬
sible for one speaker, whose role assigns him extra authority, to
select the speakers for several successive utterances.
While these speaker options explain how the next speaker comes
to be selected, or to select himself, they do not explain how the next
speaker knows when the current speaker has finished, and therefore
when he can begin, although this is obviously essential if he is to
avoid overlap or silence. Sacks suggests that in fact next speakers
are not concerned and can never be concerned with actually com¬
pleted utterances, because one can never be sure that an utterance
is complete—it is always possible to add more to an apparently
complete utterance, and speakers frequently do so. Therefore, next
speakers are concerned with points of possible completion. In
developing this idea. Sacks observes that turns consist of one or
more sentences, with a sentence being defined as a unit which has
its completion recognised on its completion, and that it is not
complete recognisable by participants; also it can be monitored,
from its beginning, to see from its beginning what it will take
for its completion to be produced, in such a way that, on its
completion, its completion may be recognised. (Sacks ms)
Speaker change takes place at the end of sentences', if the next
speaker or next action has been selected, the next speaker will take
over at the end of the sentence during which the selecting was done;
if the current speaker has not selected, any participant may self-
 CONVERSATIONAL ANALYSIS 55

select at the end of any sentence. Thus, a speaker is vulnerable at

every sentence completion whether he selects next speaker or action
or not, and even if he gets past one sentence completion he is
equally vulnerable at the end of the next sentence. Turns to speak
are valued and sought and thus the majority of turns in any conver¬
sation consist only of a single sentence, unless permission has been
sought for a longer turn, perhaps to tell a story or a joke.
The argument so far is that conversation is made up of units
which are recognisable as either incomplete or possibly complete
and that next speakers can begin as soon as a current speaker has
reached a possible completion. This fact, it is suggested, explains
the low incidence of overlap and silence. However, the ability to
come in as soon as a speaker has reached a possible completion
requires a high degree of skill on the part of participants—they need
to be able both to analyse and understand an ongoing sentence in
order to recognise when it is possibly complete, and also to produce
immediately a relevant next utterance. Do speakers have this
ability?
Jefferson (1973) argues that the recipient of an ongoing utterance

has the technical capacity to select a precise spot to start his

own talk ‘no later’ than the exact appropriate moment.
She gives three kinds of example. First she shows that speakers
can, without a pause, produce a completion to a prior speaker’s
otherwise complete utterance:

ben: An’ there—there wz at least ten mi:les of traffic

bumper tuh bumper.
ethel: —because a’that.

Much more impressive are instances of recipients coming in at just

the right moment with their own proposed completion of an as yet
uncompleted sentence:

louise: No a Soshe is someone who-pis a carbon copy of

their friend.
Roger: Ldrinks Pepsi.
A variant of this is when the recipient is able to predict the ending
of the sentence and attempts to say the same thing at the same time:

dan: The guy who doesn’t run the race doesn’t win it,
but ’e doesn t lose it.
ROGER: _B't lose it._
56 AN INTRODUCTION TO DISCOURSE ANALYSIS

Thus speakers demonstrably do have the ability to place their

entries with great precision. Nevertheless, unintentional overlaps
still occur, frequently caused by self-selection. If the current speaker
has not selected a next speaker, a self-selecting speaker, beginning
at a possible completion, may well overlap with the current speaker
who has decided to continue, or with a second self-selecting speaker.
The problem is usually ‘remedied’ quickly by one of the speakers
yielding the floor; when the overlap is the result of two self-selecting
speakers there appears to be a rule that the ‘first starter’ has the
right to continue.
On occasions when both speakers cease speaking simultaneously,
as in the example above, there is the question of who has next turn;
if one speaker was producing a completion of the other’s utterance,
as Roger is, the speaker whose turn was completed takes the next
turn. If, in other circumstances, an overlap is ended by one speaker
stopping first, if only by a syllable, he takes the next turn:

agnes irThats about a:ll,

guy: '-What else.
guy; The hat,
(0.5)
agnes: En the hat,

Silence between turns similarly creates a problem and participants

feel that a silence is attributable, usually to some intended next
speaker. Schegloff and Sacks (1973) quote a report of a silence.

He hadtuh come out tuh San Francisco. So he called hhh from

their place, out here to the professors, en set up, the, time, and
hh asked him to hh—if they’d make a reservation for him
which they did cuz they paid for hiz room en etcetera en he
asked them tuh:; make a reservation for iz parents. En there
was a deep silence, she said, at the other end ’e sez Oh well
they'll pay for their own uh-hh- room and accommodations.

They observe that the silence was noted by the speaker as not his,
but he then transformed it into a pause by continuing.
There is a very low tolerance of silence between turns and if the
intended next speaker does not begin almost at once the previous
speaker is likely to produce a post completer, which is either a
question, noticing the silence such as ‘Didn’t you hear me’, or a
.narked repeat of his utterance. To avoid this, speakers who have
 CONVERSATIONAL ANALYSIS 57

not yet formulated what they want to say tend to indicate their
intention to speak by ‘erm\ ‘um\ ‘mm’, or an audible intake of
breath and thereby incorporate the silence into their turn.
We have seen that the basis of conversation is that at least and
not more than one party speaks at a time, and that the system for
allotting turns works only one turn at a time. Sacks et al (1974)
distinguish different speech exchange systems according to the
organisation of turn-taking. They observe that whereas in conversa¬
tion turns are allocated singly, in some systems, debates or law suits
for example, there is a high degree of pre-allocation of turns. The
most extreme form is the system governing public discussion among
the Burundi, where ‘the order in which individuals speak in a group
is strictly determined by seniority of rank.’ (Albert 1972). Turns
come in a fixed order and only when everyone has had a turn and
the cycle has recommenced can the first speaker speak again.
Sacks et al (ibid) suggest that different turn-taking systems pro¬
duce differently structured turns. In a pre-allocated system there
are no interruption pressures, turns tend to be longer and for these
reasons consist of a series of linked sentences. In a turn-by-turn
allocation system there are strong pressures from other partici¬
pants wanting to speak, and the turn is typically only one sentence
long.
There are several techniques open to the speaker who wishes to
continue speaking past a particular ‘possible completion’. The
simplest technique is to employ what Sacks calls an utterance in-
completor—these are items such as ‘but’, ‘and’, ‘however’ and
other clause connectors, whose importance in conversation is that
they turn a potentially complete sentence into an incomplete one.
This, of course, is not a particularly sophisticated technique,
because a self-selecting speaker, waiting for the first ‘possible
completion’ may have already planned or even begun his turn—
Ferguson (1975), in an examination of eleven hours of conversation,
noted that 28% of interruptions occurred after conjunctions.
The successful floorholder produces utterances which
while they could perfectly well be composed of talk that, in
terms of ‘possible sentence completion’, would have one, or
more than one, such occurrence within them., are built in such
a way as to not have possible completions within them.
(Sacks ms)
One technique is to begin with an incompletion marker, ‘if’, ‘since’,
or any other subordinator, which informs the other participants that
58 AN INTRODUCTION TO DISCOURSE ANALYSIS

there will be at least two clauses before the first possible completion.
A speaker can also prestructure a fairly large unit of speech by such
devices as ‘I’d like to make two points’ or simply ‘firstly’:

Now the second big area of course is the question of how you
handle incomes and I myself very strongly believe that we have
to establish in Britain two fundamental principles. First of
all.. .
(Denis Healey; Analysis', BBC Radio 4)

None of these devices can guarantee that the speaker keeps the
floor but they do force the other speaker into a position where he
must interrupt and be seen to be interrupting. Speakers reject inter¬
ruptions, if they choose not to yield the floor, by speaking more
loudly, more quickly and in a higher pitch; often the surface gram¬
mar or phonology breaks down, and frequently there is a reference
to the interruption.

atha : they have at their disposal enormous assetsrand their policy

pitman: Llook can I just

come in on thatrlast year

atha: Lyes in a moment if you may and when

i’ve FiNiSHEDpthen you’ll know what the point is

pitman : L- yes I M SO SORRY

(Money at Work BBC 2)

A non-speaker who wishes to speak, but is unable to find a suit¬

able entry spot has the option of simply breaking in, though this is
frequently heard as rudeness.

pitman: but there aren’t enough people who need that service at
two thirty at nightpdown to my particular sta

widlake: LyOU talk like the chairman of British

Rail

or of indicating by repeated short, single-tonic, utterances his desire

 CONVERSATIONAL ANALYSIS 59

for the floor. This is a technique which children master early as the
following example of a two-year-old shows.

mother: .-not sure what she’s been saying

father: not at all today
tom: Lah ah ah ah

mother: ohrwell at least well I’ll need to know when

father: darpling
tom : Uh -ah ah ah ah

mother: she comes pin for my con sorry darling.

father:
tom: Lsto stop talking

At the other extreme a non-speaker who is offered the floor but

doesn’t want it may simply remain silent until the speaker continues
(Kendon 1967, de Long 1974) or produce a minimal response to
confirm, agree or express interest, or use the whole of his turn to
produce a possible pre-closing ‘alright’, ‘okay’, ‘so’, ‘well’,
(Schegloff and Sacks 1973), and thereby indicate that he has nothing
further to add and is willing to close the topic.
The evidence adduced so far to describe and explain speaker
change has been almost entirely grammatical and semantic. We
have seen how grammatical markers and considerations of meaning
turn certain points in utterances into ‘possible completions’ and it
has been suggested that at each ‘possible completion’ a speaker is
vulnerable. There is, however, evidence to suggest that speakers
signal paralinguistically and kinesically to the other participants at
which possible completions they are actually willing to relinquish
the floor.
De Long (1974) reports a detailed analysis of a series of conver¬
sations between four- and five-year-old pre-school children, which
shows a marked correlation between certain body movements and
change of speaker. The transcription noted eight basic movements,
including ‘up’, ‘down’, ‘left’, ‘right’, ‘forward’ and ‘backward’, for
eight parts of the body, the head, the trunk and the left and right
arms, hands, and fingers. Analysis showed that two movements co¬
occurred, either simultaneously or in rapid succession to signal a
termination. The first was a leftward movement of the head, the
second a downward movement by the head, arms or hands indi¬
vidually or in any combination.
60 AN INTRODUCTION TO DISCOURSE ANALYSIS

De Long stresses that

to say that the intention to terminate verbalisation, willingness
to yield the floor, is signaled by downward and leftward move¬
ments does not mean that every time a left in the head is accom¬
panied by a down in the head or in other parts of the body the
speaker intends to terminate.
In fact it is only when such signalling occurs at possible completions
that termination is signalled. To support his analysis de Long
describes two apparent exceptions: a long gap between two utter¬
ances, and the only recorded interruption.
The gap occurred when one child ended ‘I’ll show you what’s
brown’, tilting his head to the left but not making any downward
signals. During the next few seconds he made intermittent down¬
ward and leftward movements while the other child remained
silent; then ‘after no less than an eight second delay we find left
and downward successive signalling in the head’. Immediately, the
other child began speaking.
As confirmation of the significance of this signal, the only inter¬
ruption is found to occur immediately following a series of down¬
ward movements by several parts of the body and a succession of
downs and lefts in the head which the interrupter apparently took
as speaker change signals.
Kendon (1967) suggests that one important factor enabling the
smooth change-over of speaker is gaze. He notes that, while listen¬
ing, A typically looks at B with fairly long gazes broken by very
brief away gazes, but while speaking A looks at and away from B
for more equal periods. Focusing on the ends of utterances of more
than five seconds, Kendon notes that
usually the person who is bringing a long utterance to an end
does so by assuming a characteristic head posture and by look¬
ing steadily at the auditor before he actually finishes speaking.
The auditor, who spends most of his time looking at the speaker, is
able to pick up these signals and tends to respond by looking away
just before or just as he begins to talk. By this time of course the
initial speaker is looking at the auditor and can pick up the signal
that he has accepted the offer of the floor.

Such changes in behaviour which precede the utterance itself

clearly make it possible for each participant to anticipate how
the other is going to deal with the actual point of change of
speaker role, perhaps facilitating the achievement of smoother
... change overs.
 CONVERSATIONAL ANALYSIS 61

Kendon notes that fewer than a third of the utterances which ended
with an extended gaze were followed by silence or delayed response,
as compared with almost three-quarters of those that ended without
the speaker looking up. In larger groups the situation is more
complicated, but Weisbrod (1965) studying a seven person discussion
group, found that the person whom the current speaker last looked
at before ending was most likely to speak next.
As we observed above turns to speak are valued and speakers
compete for them. One of the points at which a speaker is vulner¬
able is when he pauses within a phrase—Ferguson (1975) discovered
that almost a third of interruptions occur following ‘fillers’ such as
‘urn’, ‘er’, and ‘y’know’. Kendon observed that speakers tend to
avoid gazing during hesitant speech—they only spend 20% of the
time gazing during hesitant speech as compared with 50% of the
time during fluent speech—probably as a defence against interrup¬
tion. In Kendon’s data interruptions were rare but in those cases
where a small battle for the turn occurred both speakers stared
fully until the conflict had been resolved.
Duncan (1973, 1974) suggests that the cues for speaker change
can be grammatical, paralinguistic or kinesic or any combination of
all three. A listener may claim the speaking turn when the current
speaker gives a turn signal, defined as the display, at the end of a
phonemic clause, of at least one of a set of six cues. The cues are

1. Intonation: the use of any pitch level/terminal juncture

combination, other than 22
2. Paralanguage: drawl on the final syllable or on the stressed
syllable of the phonemic clause.
3. Body motion: the termination of any hand gesticulation or
the relaxation of a tensed hand position.
4. Sociocentric sequences: the appearance of one of several
stereotyped expressions, such as ‘but uh’, ‘or something’,
‘you know’, labelled sociocentric by Bernstein (1962).
5. Paralanguage: a drop in paralinguistic pitch and/or loud¬
ness, in conjunction with a sociocentric sequence.
6. Syntax: the completion of a grammatical clause involving a
subject-predicate combination.

Duncan (1974) observes that the occurrence of a speaker turn signal

does not necessarily condition a change of speaker, the listener
always has the right to decline but the more cues displayed simul¬
taneously the greater the likelihood that the listener will take over
62 AN INTRODUCTION TO DISCOURSE ANALYSIS

the floor. Duncan emphasises the importance of the speaker turn

signal by showing that in his data every smooth exchange of the
speaker role followed a speaker turn signal, while every attempt by
an auditor to claim the turn while no cues were being displayed
resulted in simultaneous turns.
Duncan (1973) suggests that the speaker turn signal can be over¬
ridden by a claim suppressing signal, which consists of hand gesticu¬
lations, and is apparently almost totally effective—‘of 88 auditor
turn claims in response to turn signals, only two such claims were
made when the suppression signal was additionally being displayed.’
Once the auditor takes over the speaking role he typically dis¬
plays a speaker-state signal, which consists of at least one of four
cues: the averting of gaze noted by Kendon; the initiation of gesticu¬
lation; sharp audible intake of breath; and paralinguistic overloud¬
ness.
In any description of turn taking there is the problem of what
constitutes a turn, and while most analysts accept that nods of
agreement and murmurs of assent do not count, there are some
important differences of opinion. Duncan uses the term back channel
behaviour to cover contributions which do not constitute a turn but
which provide the speaker with useful information as his turn pro¬
gresses. Under this heading Duncan includes ‘sentence completions’,
‘requests for clarification’ and ‘brief restatements’, all of which for
Sacks would be complete turns—indeed Duncan and Niederehe
(1974) express some disquiet that
for some of the longer back channels, particularly the brief
restatements, the boundary between back channels and speak¬
ing turns became uncertain. On an intuitive basis, some of these
longer back channels appeared to take on the quality of a turn.
Duncan’s main concern is to describe speaker change, and for
this purpose it may be sufficient to categorise all utterances as either
‘back channels’ or ‘turns’; for those interested in describing the
structural relations between utterances, however, this is not suffi¬
cient—the options open following a ‘request for clarification’ are
very different from those following a ‘brief restatement’.

Discourse units and discourse function

Discourse does not consist simply of a succession of turns—a string
of grammatically well-formed utterances. The following examples
from Labov (1970) quoted in Chapter 1 are grammatically unexcep¬
tional yet noticeably odd:
 CONVERSATIONAL ANALYSIS 63

1. A: What is your name?

B: Well, let’s say you might have thought you had some¬
thing from before, but you haven’t got it anymore.
A: I’m going to call you Dean.
2. A: I feel hot today.
B: No.

In both examples B’s contribution obviously breaks rules for the

production of coherent discourse, and one of the major aims of
conversational analysis is to discover these rules and to describe the
conversational structures they generate. An initial and fundamental
question is the nature of the units whose structure and occurrence
the sequencing rules will describe.
Harris (1960) observed that traditionally grammatical description
has taken the sentence as its upper limit and it is instructive to
discuss the reasons for this. A grammatical description provides the
structure(s) of a given unit in terms of allowable combinations of
units at the rank below. An essential feature of any grammatical
description is the specification not only of what structures can occur
but also of those structures which cannot occur. Thus a grammar of
English would allow the following sentences,

I bought these chairs yesterday.

Yesterday I bought these chairs.
These chairs I bought yesterday.

but not.

Yesterday these chairs bought I.

These chairs bought yesterday I.
I these chairs yesterday bought.

The speaker’s choice of one of the grammatical options depends on

cohesive and stylistic considerations. Once one comes to look at
choices above the sentence, however, all the choices are stylistic
ones. There is no way of describing paragraph structure in terms
of allowable combinations of the units next below, sentences; any
collection of sentence types in any sequence can constitute a para¬
graph and ‘rules’ about paragraph writing take the form of advice
about ‘topic sentences’ and the alternation of long and short sen¬
tences.
However, while it appears that structure describable in terms of
formal units ends at the sentence, patterning of functional units
64 AN INTRODUCTION TO DISCOURSE ANALYSIS

occurs above the sentence and across utterance boundaries. In the

following example it is impossible to describe or even contemplate
constraints on B's utterance in grammatical terms but in functional
terms his options are highly restricted.

A: Where’s the typewriter?

B: In the cupboard.
Is it in the cupboard?
Look in the cupboard.
I think it’s in the cupboard.

In other words the argument is that whereas it is impossible to

provide a structural description of a conversation in terms of
'declarative followed by moodless clause’, or ‘interrogative followed
by declarative’ it is possible to provide a meaningful structure in
terms of Question and Answer, Challenge and Response. Thus
Labov (1972) argues that the first and most important step is to
distinguish ‘What is said from what is done\ and stresses that the
unit of analysis is not the grammatically defined clause or sentence
but a functional unit, which may of course be realised by a single
clause or sentence.
Any attempt to characterise discourse structure in terms of
functional units must confront the problem of grammatical realisa¬
tion—how do the four major clause types, ‘declarative’, ‘interro¬
gative’, ‘imperative’ and ‘moodless’ realise a multiplicity of
different functions, and how can a hearer correctly interpret which
function is intended.
Labov (1970, 1972) focuses on answers and sketches out a series
of interpretive rules to explain how a second utterance comes to be
heard as an answer to a preceding one. The simplest relationship is
between a question and an elliptical answer,

A: Are you going to work tomorrow?

B: Yes.

Here a simple rule can account for the relation

If A utters a question of the form Q-S, and B responds with

an existential E (including yes, no, probably, maybe etc.) then
B is heard as answering A with a statement E-Sr

A more complex relationsnip holds between the following pair of

utterances:
 CONVERSATIONAL ANALYSIS 65

A: Are you going to work tomorrow?

B: I’m on jury duty.

Grice (1975) argues that there is an underlying constraint on all

conversationalists to ‘be relevant’ and for this reason A will assume,
at least initially, that there is a proposition known to both which
connects B’s response to his question; i.e. ‘if someone is on jury
duty he cannot go to work.’ To account for this type of relationship
Labov proposes the following rule;

If A makes a request for information Q-S15 and B makes a

statement S2 in response that cannot be expanded by rules of
ellipsis to the form X S: Y, then S2 is heard as an assertion
that there exists a proposition P known to both A and B:
Y S2 then (E) S,
where (E) is an existential operator, and from this proposition
there is inferred an answer to A’s request: (E) Sv

This rule makes clear the crucial importance of shared know¬

ledge in conversation; not simply shared rules for the interpretation
of linguistic items, but shared knowledge of the world, to which a
speaker can allude or appeal. Labov notes that this rule is invariant:
A must inspect B’s utterance to see if he can detect an underlying
linking proposition and ‘failure to locate such a proposition may
reflect a real incompetence’:

linus: Do you want to play with me, Violet?

violet: You’re younger than me. (Shuts the door)
linus: (puzzled) She didn’t answer my question.

Linus does not share the knowledge to which Violet appeals and
therefore is unable to hear her utterance as an answer to his ques¬
tion.
These first rules are concerned with explaining how utterances
following questions come to be heard and interpreted as answers.
Labov also discusses how some utterances, declarative in form,
come to be heard as questions. He presents the following extract
from a therapy session,

therapist: Oh, so she told you.

patient: Yes.
therapist: She didn’t say for you ...
patient: No.
66 AN INTRODUCTION TO DISCOURSE ANALYSIS

therapist: And it never occurred to her to prepare dinner.

patient: No.

and observes that it consists of a series of pairs where ‘the first

utterance is a statement and the second is “yes” or “no”, and it
seems that a statement is functioning as equivalent to a yes-no
question.’ It is, of course, certainly not the case that any statement
can be followed by ‘yes’ or ‘no’.

A: I don’t like the way you said that.

B: *Yes.

A: I feel hot today.

B: *No.

Labov suggests that the statements in the therapy extract are acting
as requests for confirmation and have the same compelling force as
requests made in question form; but how ‘is it that we regularly
and reliably recognise these as requests and not as assertions?’ The
interpretive rule depends again on shared knowledge. Whenever
there are two participants in a conversation, A and B, one can
distinguish ‘A-events’, things that A alone knows about, ‘B-events’,
things that B alone knows about, and ‘AB-events’, things that are
known to both. Using this terminology one can state a simple
interpretive rule:

If A makes a statement about a B-event it is heard as a request

for confirmation.

The interpretation of such utterances as requests for clarification

depends crucially of course on speaker A’s assumptions about
speaker B’s knowledge being correct. In the following example the
assumptions were wrong and A’s intended request for confirmation
was heard as a statement of new information.

A: There’s no playgroup next week then.

B: Oh isn’t there.

A third major group of speech acts comprises commands or

requests for action. Labov argues that in analysing these one must
take account of sociological concepts: ‘notions of role, rights, duties
and obligations associated with social rules.’ Labov and Fanshel
(ms) formalise the pre-requisites for an utterance imperative in
form to be heard as a valid request for action:
 CONVERSATIONAL ANALYSIS 67

If A addresses to B an imperative specifying an action X at a

time Tx and B believes that A believes that
1. (a) X should be done for a purpose Y {need for the action)
(b) B would not do X in the absence of the request (need
for the request)
2. B has the ability to do X
3. B has the obligation to do X or is willing to do it
4. A has the right to tell B to do X,
then A is heard as making a valid request for action.
Imperative utterances which fail to satisfy one or more of these
pre-conditions are, in Austin’s terms, infelicitous, and may be
variously interpreted as cheeky, insulting, joking or simply irrele¬
vant.
The rule so far only covers those utterances in which there is a
close fit between intended function and formal realisation, that is
imperative commands, but as Labov and Fanshel admit these are
the minority of cases. They therefore offer a rule for indirect
requests, which can instructively be compared to Searle’s suggestions
in Chapter II and which demonstrate how a valid request can be
made without resorting to the imperative mood.
If A makes to B a request for information or an assertion to
B about
(a) the existential status of an action X,
(b) the time Tx that an action X might be performed,
(c) any of the preconditions for a valid request for X as given
in the Rule for Requests
and all other pre-conditions are in effect, then A is heard as
making a valid request of B for the action X.
They cite as an example an utterance from a therapy session,
A: wellyouknow, w'dy’mind takin’ thedustrag an’ just dust-
around?
and explain that this is interpreted as an indirect request through
being a request for information about the third precondition for
valid request, B's willingness.
Labov (1972) recognises that ‘there may be such a thing as
premature formalisation’ which the ethnomethodologists are anxious
to avoid, and that a linguist may be led to set up ‘paradigms of
discrete features where only open sets are found in reality.’ How-
68 AN INTRODUCTION TO DISCOURSE ANALYSIS

ever, we are now in a position to appreciate the advantages of a

formalisation which isolates the rules underlying a given speech act.
Labov and Fanshel observe that although they ‘are accountable to
the text as it actually occurs, [their] discourse rules represent a
general grammar of possible speech actions and ways of executing
them.’ Thus while A’s indirect request to ‘dustaround’ actually
occurred in the form quoted above, there are many alternative ways
in which it could have been realised by questioning or asserting
other pre-conditions:

(a) Existential Status Have you dusted yet?

You don’t seem to have dusted
this room yet.
(b) Time Reference When do you plan to dust?
I imagine you will be dusting
this evening.
(c) Pre-conditions #
la. Need for action: Don’t you think the dust is
pretty thick?
This place really is dusty.
lb. Need for the request: Are you planning to dust this
room?
I don’t have to remind you to
dust this room.
2. Ability: Can you grab a dust rag and
just dust around?
You have time enough to dust
before you go.
3a. Willingness: Would you mind dusting
around?
I’m sure you wouldn’t mind
picking up a dust rag and just
dusting around.
3b. Obligation: Isn’t it your turn to dust?
You ought to do your part in
keeping this place clean.
4. Rights: Didn’t you ask me to remind
you to dust this place?
I’m supposed to look after this
place, but not do all the work.
 CONVERSATIONAL ANALYSIS 69

Obviously the examples above are just a few of the large number
of possible indirect formulations of this particular request, but the
important fact to remember is that all formulations work by refer¬
ring either to the existential status or time of the action, or to the
pre-conditions of the request.
If one is interested in the process by which speakers generate
rather than hearer’s interpret text, the problems are much greater.
Labov and Fanshel admit that there is
an unlimited number of ways in which we can refer to the pre¬
conditions, and this poses a serious problem if we want to make
firm connections between these discourse rules and actual sen¬
tence production.
In fact the difficulties are not insurmountable; there is only an un¬
limited number of ways of making a given indirect request if one is
considering it in isolation; in reality the constraints of the preceding
discourse, the current topic, the facts of the situation and the current
speaker’s intentions for the progress of the succeeding discourse
will all reduce the choice enormously.
These rules for indirect requests are highly significant. In them¬
selves they successfully account for the interpretation of individual
utterances and delimit the range of possible alternative realisations;
more importantly they point the way for others to characterise other
speech acts and take further steps towards a generative model of
discourse.

Conversational structure
Before attempting a description of discourse structure one must
first decide on the size of the basic unit. Labov (1970, 1972), Sacks
(passim), Schegloff (1968, 1972) and Jefferson (1972, 1973) use
utterance or turn. Sinclair et al (1972) observe that they also began
with utterance as the basic unit but, in dealing with examples like
the following, came to feel the need for a smaller unit, which they
called move. Moves can be co-extensive with utterances, but some
utterances, like A’s second, contain two moves.

A: Can you tell me why do you eat all that food;

B: To keep you strong.
A: To keep you strong, yes, to keep you strong.
Why do you want to be strong?

Any attempt to apply the analyses suggested by Labov, Sacks,

Schegloff and Jefferson to the above example quickly demonstrates
70 AN INTRODUCTION TO DISCOURSE ANALYSIS

that, in fact, their analytic unit is not the utterance but something
equivalent to the move.
Sacks observes that a conversation is a string of at least two
turns. Some turns are more closely related than others and he
isolates a class of sequences of turns called adjacency pairs which
have the following features: they are two utterances long; the utter¬
ances are produced successively by different speakers; the utterances
are ordered—the first must belong to the class of first pair parts,
the second to the class of second pair parts-, the utterances are
related, not any second pair can follow any first pair part, but only
an appropriate one; the first pair part often selects next speaker
and always selects next action—it thus sets up a transition relevance
and expectation which the next speaker fulfils, in other words the
first part of a pair predicts the occurrence of the second, ‘Given a
question, regularly enough an answer will follow’ (Sacks 1967).
There is a class of first pair parts which includes Questions, Greet¬
ings, Challenges, Threats, Warnings, Offers, Requests, Complaints,
Invitations, Announcements. For some first pair parts the second
pair part is reciprocal. Greeting-Greeting, for some there is only
one appropriate second, Question-Answer, for some more than one.
Complaint-Apology / Justification.
Adjacency pairs are the basic structural units in conversation.
They are used for opening and closing conversations,
[Hi there
\ Hello
[ Bye then
) Bye
and are very important during conversations both for operating the
turn-taking system by enabling a speaker to select next action and
next speaker, and also for enabling the next speaker to avoid both
gap and overlap.
It is, however, no difficult matter to discover a question not
followed by an answer and this raises a question about the status of
the pair. Sacks argues that, whereas the absence of a particular
item in conversation has initially no importance because there are
any number of things that are similarly absent, in the case of an
adjacency pair the first part provides specifically for the second and
therefore the absence of the second is noticeable and noticed. He
observes that people regularly complain ‘You didn’t answer my
question’ or ‘I said hello, and she just walked past’.
 CONVERSATIONAL ANALYSIS 71

Sacks further suggests that it is a feature of conversation that

a person who has asked a question has ... a reserved right to

talk again, after the one to whom he has addressed the ques¬
tion speaks. And in using this reserved right he can ask a
question. (1972a)

This chaining rule allows for an indefinitely long sequence of the

form Q.A.Q.A.Q.A. Such sequences in fact rarely occur in conver¬
sation although they are typical of parts of doctor/patient interviews
and courtroom cross-examinations.
Sequences consisting of different types of pair are, however, quite
common. Sacks suggests that some of the structural complexity in
conversation arises from speakers avoiding potentially embarrassing
or annoying situations. Whenever a speaker makes a Request, an
Invitation, an Offer, he is exposing himself to a possible rejection,
so speakers typically avoid the possible rejection by producing what
Sacks calls a pre-sequence, another pair which determines whether
his Invitation is likely to be accepted.

jack: Say what ya doin?

pre-invitation
judy: Well, we’re going out. Why?
jack: Oh, I was just gonna say come out and
come over here and talk to the people.

Participants can recognise pre-sequences fairly easily and indicate

in their replies whether the Invitation is likely to be accepted.

A: Say what you doin’ tonight?

B: Oh, I’m just...

The very recognisability of the first part of a pre-sequence some¬

times results in ‘collapsing’, in which the next speaker produces
what is really an answer to the first part of a notional next sequence.
In the following example utterance 1 was produced by a five-year-
old and the actual reply was utterance 4; utterances 2 and 3 repre¬
sent one form an uncollapsed version might have taken:

(1. Can you fix this needle?

pre-request
j 2. Sure.
j 3. Will you?
request
14. I’m busy.
72 AN INTRODUCTION TO DISCOURSE ANALYSIS

The ellipsis in fact allowed the five-year-old to protest that an

intended question had been misanalysed as a pre-request and
respond,
5. I just wanted to know if you can fix it.

As an alternative analysis one can suggest that utterance 1 is one of

Labov and Fanshel’s indirect requests which refers to the second
pre-condition on request, B’s ability; B’s reply ‘I’m busy’ is heard
as a refusal because it denies the third pre-condition, his willingness.
The structures described so far have been linear, one pair
followed by another; there are also cases of embedding, of one
pair occurring inside another. Schegloff (1972) calls these embedded
pairs insertion sequences. Sometimes, either because he doesn’t
understand, or because he doesn’t want to commit himself until he
knows more or because he’s simply stalling, a next speaker pro¬
duces not a second pair part but another first pair part. The sugges¬
tion is ‘if you answer this one, I will answer yours’.

A. I don’t know where the -wh - this address / / is. Q

/B. Well where do -
Insertion Qi
sequence Which part of town do you live?
A. I live four ten East Lowden. Ai
B. Well you don’t live very far from me. A

One question which immediately arises is in what sense is the

pair QiAi inserted into the pair QA; surely this is treating conversa¬
tion as an accomplished product rather than a developing process,
because A may never occur. Schegloff argues that

The 0 utterance makes an A utterance conditionally relevant.

The action the Q does (here, direction asking) makes some
other action sequentially relevant (here, giving directions by
answering the 0). Which is to say, after the Q, the next speaker
has that action specifically chosen for him to do and can show
attention to, and grasp of, the preceding utterance by doing the
chosen action then and there. If he does not, that will be a
notable omission.

In other words, during the inserted sequence the original question

retains its transition relevance, and if the second speaker does not
then produce an answer it is noticeably absent in exactly the same
way as it would be if there were no intervening sequence, and the
questioner can complain about the lack of answer in exactly the
 CONVERSATIONAL ANALYSIS 73

same way. Adjacency pairs are normative structures, the second

part ought to occur, and thus the other sequences are inserted
between the first pair part that has occurred and the second pair
part that is anticipated.
Jefferson (1972) proposes another type of embedded sequence,
the side sequence. She observes that the general drift of a conversa¬
tion is sometimes halted at an unpredictable point by a request for
clarification and then the conversation picks up again where it left
off. The following example is children preparing for a game of
‘tag’.

steven: one, two, three, (pause) four, five, six, (pause) eleven,
eight, nine, ten.
susan: Eleven?—eight, nine, ten.
steven: Eleven, eight, nine, ten.
nancy: Eleven?
steven: Seven, eight, nine, ten.
susan: That’s better.

Whereupon the game resumes.

The side sequence begins with a questioning repeat—an interro¬
gative item indicating that there is a problem in what has just been
said, ‘and its work is to generate further talk directed to remedying
the problem.’ Questioning repeats occur typically after the ques¬
tioned utterance has been completed, because only then can one be
sure that the speaker is not going to correct himself or explain the
unclear item. An interrupting questioning repeat is liable to attract
a complaint not a clarification, ‘if you’d just let me finish’.
Jefferson suggests initially that the misapprehension sequence has
a three-part structure, consisting of ‘a statement of sorts, a mis¬
apprehension of sorts, and a clarification of sorts.’ The example
above is in fact more complex consisting of a statement followed
by two misapprehension and clarification pairs. So far the side
sequence looks rather like Schegloff’s insertion sequence. There are,
however, two major differences: firstly, the first item, the statement,
is not a first pair part, the other items are in no sense inserted and
there is no expectation of who should speak at the end of the
sequence or of what type of utterance should follow; secondly,
while the sequence misapprehension-clarification looks like a pair,
there is actually a compulsory third element in the sequence, an
indication by the misapprehender that he now understands and
74 AN INTRODUCTION TO DISCOURSE ANALYSIS

that the sequence is now terminated—‘That’s better’ in the example

above, or ‘yeah’ in the example below.

Statement: If Percy goes with - Nixon I’d sure like that.

Misapprehension: Who?
Clarification: Percy. That young fella that uh—his
daughter was murdered. (1.0)
Terminator: Oh yea:h. Yeah.

Because the first item, the statement, is not a first pair part, the
conversation cannot resume with the second pair part as after an
insertion sequence, and there remains the problem of a return.
Jefferson observes that

it is not merely that there occurs a return to the on-going

sequence, but that to return to the on-going sequence... is a
task performed by participants.

She suggests that the return can be effected either as a resumption

or as a continuation—a resumption is achieved by attention getters
such as ‘listen’ or ‘hey you know’, which mark that there is a
problem in accomplishing a return, while continuations, attempted
by ‘so’ or ‘and’ are directed to ‘covering-up’ the problem, to pro¬
posing that there is no trouble, sometimes where that may not be
the case. Thus the full system is

statement: A: And a goodlooking girl

comes to you and asks
you, y’know,
' misapprehension: B: Gi (hh) rl asks you to—

1 clarification C: Wella its happened a lotta

times.
termination B: Okay okay go ahead.
continuation: B: So he says ‘no’ . . .

Chaining, insertion sequences and side sequences are examples

of obvious local structuring in conversation. Sacks (1967) suggests
that there is a much more pervasive form of structuring which he
calls tying. The analytic system described so far would enable one
to divide a stretch of conversation into say a series of Q-A pairs,
but would not allow any statements to be made about relationships
 CONVERSATIONAL ANALYSIS 75

between the pairs and the evidence that they did occur and only
could have occurred in that order. Information of this kind comes
from features which show how one utterance is tied to previous
ones, and tying is realised typically by pronouns and pro-verbs,
although adverbs like ‘too’, ‘as well’, and contrastive stress are also
used.

A: Where was everybody last night?

B: Well, John and Lisa went to the movies.
A: DID THEY?

Cohesive devices of this kind are described and discussed in detail

in Halliday and Hasan (1976).
Importantly, tying doesn’t just refer to a relationship between
successive utterances—the pronoun ‘they’ for instance refers back
to ‘John and Lisa’ for as long as it is used, tying utterances together
over long stretches. The fact that utterances and pairs of utterances
are not isolated but tied to preceding utterances means that
speakers must understand what has gone before in order to produce
a correctly tied utterance. Sacks argues that a speaker can’t not tie
and that this is very important because in tying a speaker is forced
to show whether he did or did not understand what went before.
Thus tying simultaneously fulfils two functions, it is cohesive and it
displays the speaker’s understanding of previous utterances.

Topic
Sacks observes that initially he did not attempt to describe topic
because it seemed to be

the sort of thing in which direct content considerations were

necessary and ... [he] couldn’t proceed in [his] usual fashion ...
to extract relatively formal procedures. (1968)

Now, he considers structure to be massively present.

An initial question is what sort of things can and do form topics
in conversation? Some topics are not relevant to particular conver¬
sations because

it is a general rule about conversation that it is your business

not to tell people what you can suppose they know. (1971)
and the suitability of other topics depends on the person one is
talking to.
We experience, see, hear about events all the time; some are
76 AN INTRODUCTION TO DISCOURSE ANALYSIS

‘tellable’, some aren’t, and of those that are tellable some are
tellable to everyone, some have a restricted audience, some must be
told immediately, some can wait and still retain their interest. For
instance, if one’s sister becomes engaged, some relatives must be
told immediately, others on a first meeting after the event, whereas
some of one’s friends might not know the sister or even that one has
a sister, and for them the event has no importance or even interest.
This concept of tellability or newsworthiness is difficult to apply
to a particular item in a particular conversation, but it is used by
conversationalists all the time. They constantly analyse what is said
for its newsworthiness, ‘why that now and to me’, and someone
who consistently produces talk which is not newsworthy is regarded
as a bore.
The pressures on people to transmit relevant news are increased
by the existence of the telephone—one no longer needs to wait
until one meets one’s friends or relatives nor does one need to make
special or difficult journeys to pass on information. Sacks has a
good example of this in a tape of a series of telephone conversations.
A and B are friends; B works at a local department store; A was
driving past the store in the morning when she noticed an incident
outside involving police cars (part of transcript is reproduced on
p. 83); knowing it was B’s day off she rings up to tell her about the
news, thus fulfilling her obligations of monitoring the world for her
friend and reporting relevant events. However, B has a second
friend, who also works at the store, who didn’t ring up to tell her
about it and whom she then rings to discover what all the com¬
motion was about. This puts C in a difficult position; she has been
caught out failing to keep her friend informed and she takes the
only possible way out, that of denying that what happened was
newsworthy.

C: It was nothing, uh—in fact I dn’t even say anything to

Willy about it.

In other words, if she didn’t consider it worth telling her husband

to whom virtually everything is newsworthy, it must have been an
insignificant event.
Thus we see there are certain things which one must say to parti¬
cular people and certain things which are tellable if one happens to
meet them. This leads on to the idea of reason for a call or visit—
it is a basic assumption of all except chance encounters that the
person who initiates the encounter has some reason for so doing.
 CONVERSATIONAL ANALYSIS 77

and if there is no such reason people regularly feel the need to state
this, ‘I was just passing’, ‘I just felt like giving you a call’. Conver¬
sations tend to begin with the topic which is the reason for the
encounter and then move on to other topics; though, of course, the
association of ‘reason for call’ with ‘first topic’ can be exploited
by producing a false reason for the call and introducing the real
reason as just another topic later in the conversation.

Topic change
Sacks (1971) observes that in a conversation which is progressing
well talk drifts imperceptibly from one topic to another, and
suggests that the relative frequency of marked topic introduction is
some measure of the quality of a conversation. Turns must display
‘why that now’, and ‘speakers specifically place almost all their
utterances’, and the most usual way of doing this is by tying gram¬
matically and topically to what has gone before.
However, as Sacks (1968) stresses, talking topically and talking
about some topic chosen by another speaker is not the same thing
at all. One can perfectly well have a sequence in which successive
speakers talk in a way topically coherent with the last utterance,
but in which each speaker talks on a different topic.

That is to say, ‘talking topically’ doesn’t consist of blocks of

talk about ‘a topic’. And when one presents a topic, except
under rather special circumstances, one may be assured that
others will try to talk topically with what you’ve talked about,
but you can’t be assured that the topic you intended was the
topic they will talk to.

Speakers are aware of this as a problem and have ways of formu¬

lating a topic to make it more likely that other speakers will talk to
it. Sacks exemplifies with a hypothetical speaker who wants to talk
about surfing. He could introduce the topic by saying ‘I went surf¬
ing yesterday’, but that allows topically coherent utterances of the
form ‘I did X yesterday’; that is, ‘surfing’ has been presented as
one of a class of activities and a topically coherent next utterance
can focus on another of that class. A better opening he suggests
would be ‘I went to Malibu yesterday’. Malibu is also one of a
class, this time a class of places, but it is a known feature of the
place that people surf there, and predisposes the next speaker to at
least begin with a reference to surfing, if only to say that they don't
like it. Thus a possible sequence would be
78 AN INTRODUCTION TO DISCOURSE ANALYSIS

I was at Malibu yesterday

Yeah? I was at County Line
How was it?
Too low tide. (Sacks ms)

Topic conflict
The phenomenon of topic drift can be frustrating at times for
conversationalists. Everyone has had the experience of failing to
get in at the right time with a good story or experience, and then
seeing it wasted because the opportunity never recurs. At times,
within conversations, there is competitive talk when two speakers
want to develop the topic in different ways; both fight because they
know there will be no further opportunity to say what they want to
say. In the following example Roger wants ‘New Pike’ to be the
topic, Jim, ‘P.O.P.’

ROGER; Isn’t the New Pike depressing?

ken; Hh. The Pike?
ROGER; Yeah! Oh the place is disgustingrAny day of the week
jim: '-I think that
P.O.P. isrdepressing its just—-
ROGER: LjJut you go—you go—take—
jim: Those guys are losing money.
ROGER: But you go down—dow . down to the New Pike there’s
a buncha people oh ; : and they’re old and they’re
pretending they’re having fun . but they’re really not.
ken: How c’n you tell? Mm?
ROGER: They’re—they’re trying make a living, but the place is
on the decline, ’s like a dergenerate place
jim : Uo’s P.O.P.
ROGER: Y’know?
jim: P.O.P. is just—
ROGER: Yeah its one of those pier joints y’know?
jim: Its a flop! hehh. (Sacks 1968)
As we saw earlier utterances normally relate back to the previous
utterance —here Roger and Jim compete by skip-connecting, relating
back to the last-but-one utterance, their own. Each time one of them
gets a turn he declines to talk about the previous speaker’s topic
and reasserts his own. Skip-connecting is not an uncommon pheno¬
menon, but apparently speakers only skip-connect over one utter¬
ance and thus, Ken’s entry with what is a challenging question ‘How
 CONVERSATIONAL ANALYSIS 79

c’n you tell’ in fact preserves Roger’s topic. Jim in his next turn is
forced to produce a normally connected utterance, but still is able
to use it to assert P.O.P. as a possible topic, ‘So’s P.O.P.’

it appears that Jim, seeing he loses out, making a gesture of

acquiescence while holding onto P.O.P. gets from Roger an
acquiescence in including P.O.P. in the talk, that is, in Roger’s
next ‘Yeah its one of those pier joints’ he connects to Jim’s last.
(Sacks 1971)
Once this competition has been resolved the conversation moves
forward again.

Stories
For Sacks a story is any report of an event—it may be only one
sentence long, but is usually longer, and therefore presents a prob¬
lem for any intending teller because, as we noted earlier, other
speakers are likely to self-select on the first possible completion.
The intending story-teller therefore needs to seek a suspension of
the usual turn-taking machinery and he achieves this by a floor-
seeker or story-preface.

Wanna hear a joke?

or
I was at the police station this morning.

The function of these utterances is to select not next speaker as is

usual in the turn-taking system, but next speaker but one, and to
guarantee this speaker a stretch of uninterrupted talk in which to
accomplish the story. Usually anything but a direct refusal is suffi¬
cient to allow the speaker to begin.

A: I was at the police station this morning.

B: Big deal.
A: ‘Big deal’, yeah, somebody stole all my radio equipment
outta my car.
(Sacks 1967)

Any suspension of the turn-taking system creates another problem,

how do the other conversationalists know when the system begins
operating again, in other words how do they know when the story¬
teller has finished? Jokes have recognisable endings, stories don't,
and thus a second function of the preface is to provide some infor¬
mation about what it will take for the story to be over, so that the
listeners will know when to react.
80 AN INTRODUCTION TO DISCOURSE ANALYSIS

Once someone has told a joke or a story others are likely to

follow. Any joke can follow any joke, but successive stories are
topically related and highly pre-specified. Thus if the first speaker
talks about his holidays, the second is also expected to tell a first
person story about his holidays; if someone tells a ‘relative’ story,
‘My mother said X this morning’, a next story will be about some¬
one standing in a similar relationship to the speaker. A story which
does not have this degree of parallelism is less appropriate,
although sometimes there is no alternative. In mixed company, if
the men are producing first person rugby stories the only option
open to the women, if they want to remain in the conversation is to
produce stories about close relatives’ rugby exploits.
However, if across a range of topics one participant shows him¬
self unable to produce appropriate stories he also shows himself to
not be a member of the group with which he is conversing.

Topical coherence
The referential and descriptive items within any story are related in
highly complex ways and each occurrence serves to reinforce and
re-emphasise the topic. Sacks (1972a) presents some techniques for
analysing topical coherence by focusing on the story

The baby cried. The mommy picked it up.

Initially he suggests that most people will ‘hear’ the story in the
same way and will agree with the following ‘facts’. Firstly, although
there is no genitive in the story, the mommy who picks up the baby
is the baby’s mommy; secondly that the two events occur sequen¬
tially; and thirdly that the second event occurs because of the first
event. He sets out to produce a descriptive apparatus that will
account for these facts, and, because it is ‘overbuilt’, for similar
facts in other stories.
Sacks introduces the concept of membership categorisation device
to handle a descriptive category, for example ‘sex’, which com¬
prises one or more subordinate concepts or categories, for example
‘male’ and ‘female’ and a set of rules which enables one to pair
population members with a category. There are a large number of
membership categorisation devices and, because some words have
more than one meaning, some categories will be members of more
than one device. Thus, while ‘baby’ along with ‘mommy’ and
‘daddy’ is a member of the device ‘family’, it is also a member of
the device ‘stage of life’ along with ‘adult’ and ‘adolescent’.
 CONVERSATIONAL ANALYSIS 81

Sacks then introduces two rules, the economy rule and the con¬
sistency rule. The economy rule states that ‘if a member uses a
single category from any device then he can be recognised to be
doing adequate reference to a person’; the consistency rule states
that once one member of a device has been used ‘other categories
of the same collection may be used to categorise further members
of the population’. From the consistency rule he derives a corollary,
which he calls a hearer's maxim.

If two or more categories are used to categorise two or more

members of some population and those categories can be heard
as categories from the same collection then: Hear them that
way.

Thus ‘mommy’ and ‘baby’ are heard as being co-members of the

device ‘family’ but how do we hear that this particular mommy
and baby are related? Sacks further notes that many devices are
duplicatively organised, that is the population is seen to consist of
a series of such devices and is analysed as such—the population
consists of a large number of families, not a large number of
unrelated mothers, fathers, children, babies, and each device has a
number of each. It is for this reason that the mommy is heard as the
mommy of the baby, and for exactly the same reason the story

The first baseman looked round, the third baseman scratched

himself.

will be heard as implying that the basemen are in the same team.
Sacks next introduces the concept of category-bound activities.
He suggests that some activities are appropriate to, or done by,
members of certain devices—thus ‘crying’ is an activity bound to
the category ‘baby’ when it is a member of the ‘stage of life’ device.
Some devices he notes, incidentally, are organised in a positional
way, that is there is an ordered relationship like baby, child,
adolescent, adult, and in this case category-bound activities can be
instanced to praise or degrade. For babies and young children,
crying, in certain circumstances may be the norm and an absence
of crying is a cause for praise, ‘what a big boy you are’, while
alternatively an older child who cries may be told not to ‘be a
baby’. Because, in this story, the baby is crying and this is an
activity bound to the category ‘baby’ in the ‘stage of life’ device,
both meanings of baby are simultaneously present.
The idea of membership outlined in this article is developed in
82 AN INTRODUCTION TO DISCOURSE ANALYSIS

Schegloff (1972). Schegloff points out that any speaker wishing to

refer to a place or location has a relatively large number of possible
formulations—as he writes he could describe the location of his
notes as ‘right in front of me, next to the telephone, on the desk, in
my office, in the office ... in Manhattan, in New York City . . .’
While all these ‘correctly’ describe the location of the notes, on any
occasion of actual use not all of them are ‘correct’. The problem
Schegloff poses is ‘how is it that on a particular occasion of use
some term from the set is selected and other terms rejected?’ The
answer depends partly on whom one is talking to and partly on the
topic.
Whatever the topic of the conversation the speaker must member¬
ship his listener, put him into one of two or more mutually exclusive
boxes. Each time a topic changes the listener must be re-member-
shipped, and during a conversation the same person may be
membershipped as a doctor, a rugby player, a liberal, a gardener,
a bridge player. Speakers usually membership their friends cor¬
rectly but may make mistakes with strangers, and shoppers member-
shipped as shop-assistants can become very annoyed. Sacks (1968)
reports an exchange on an aeroplane,

passenger: Do you have a cigarette?

stewardess: No we don’t. They don’t provide that service any
more.

The stewardess assumes quite naturally that she has been member-
shipped as a stewardess and that the question is addressed to her in
that role. She indicates this in her use of ‘we’ in the reply, and
replies, on behalf of the organisation, that cigarettes aren’t available
any more. Had she taken it that the passenger was membershipping
her as a stranger not a stewardess. Sacks argues, she might have
offered him one of her own cigarettes.
Thus membershipping is not simply a feature recoverable from
what is said, it is also a vitally necessary determiner of what one
says. As Schegloff points out before a speaker can produce even a
location term he must membership his hearer, and if he gets it
wrong one gets sequences like

A: I just came back from Irzuapa.

B: Where’s that?

When in doubt about how to membership someone speakers play

safe and use a presequence
 CONVERSATIONAL ANALYSIS 83

A: D’you know where the Triboro Bridge is?

B: Yeah.
A: Well you make a right there.
The fact that there is a diversity of possible formulations for the
same person, place or event allows for much greater topical co¬
herence because the speaker can choose categories from the same
device. Sacks (1968) and Schegloff (1972) quote the following piece
of a telephone conversation:
ESTELLE: Well, I just thought I’d - re - better report to you
what’s happen’ at Bullock’s today. Well I-v-got outta
my car at five-thirty. I drove aroun’ an’ at first I
had t’go by the front door a’ the store,
JEANETTE: Eyeah.
ESTELLE: An’ there was two p’leece cars across the street,
andeh - colored lady wan’tuh go in the main
entrance there where the silver is an all the gifts an’’
things.
JEANETTE: Yeah,
ESTELLE: And they, wouldn’ let ’er go in and he hadda gun
he was holding a gun in ’is hand a great big long
gun
JEANETTE: Yeh.
ESTELLE: An’nen over on the other side, I mean to the right
of there, where the employees come out there was a
whole oh musta been tenuh eight’r ten employees
stanning there.
It is immediately obvious that once Estelle has given the name of
the store where the incident happened, all other places are described
in relation to it. The police cars were ‘across the street’; the
coloured lady wanted to go ‘in the main entrance’, ‘where the silver
is’; the eight or ten employees were standing ‘over on the other
side’, ‘to the right of there’. Bullocks is the topic and all the loca¬
tions are formulated to emphasise this. Had the teller, for instance,
been coming out of the store across the road, the police cars would
have been parked in front of the store and the incident would have
been ‘across the street’. The choice of location terms follows the
consistency rule. Bullocks is the topic and the way in which the
places are formulated emphasises its centrality. As Sacks observes:
the phenomenon of being ‘parked across the street from’ is
obviously one sort of characterisation which turns on not only
84 AN INTRODUCTION TO DISCOURSE ANALYSIS

where you are but what it is that is being talked about, and
where that is.

The analysis of conversation

The techniques described above are designed to handle parts of
conversations, short sequences, topics, stories. Sacks (ms) asks
w'hether one can use conversation as an analytic unit. The basic
question is whether there are some universal features which all
conversations share, or whether conversations consist of a random
collection of smaller units in no fixed sequence. He suggests that
greetings are close to being universal in conversation and although
they sometimes don't occur, on some of these occasions their
absence is noticeable, which suggests that conversationalists feel
they are an almost invariant feature.
There are two important features about greetings: firstly, they
occur at the very beginning of a conversation, and cannot be done
anywhere else in the conversation; secondly they allow all the
speakers a turn, right at the beginning of the conversation.

Hello there you two

Hi
Hi there

There are two major types of occasion on which a conversation does

not open with a greeting. Firstly conversations between people who
do not consider themselves co-conversationalists, for example
strangers. They are not on greeting terms and therefore do not
exchange a greeting. The speaker who opens must demonstrate in
his first utterance why he is beginning the conversation:

Excuse me. Can you tell me the way to ...

or
Hey. You’ve dropped your book.

The other conversations which typically don’t open with a greeting

are telephone conversations. Schegloff (1968) argues that although
the person who answers the telephone may say ‘hello’ this is not a
greeting, it is an Answer to the Summons from the caller embodied
in the ringing of the telephone. Following this indication that the
channel is open there is often a greetings sequence to begin the
conversation proper, although sometimes, if the answerer simply
answers with ‘hello’ there is first a checking sequence to make sure
the caller is talking to the right person.
 CONVERSATIONAL ANALYSIS 85

Summons: Telephone rings.

Answer: A: Hello.
Greeting ('B: Goodmorning.
sequence |A: Oh hi.

Following the opening sequence the conversation consists of a

series of one or more topics, though occasionally, as Schegloff and
Sacks (1973) observe, the conversation may be closed before
speakers reach the first topic:
Am I taking you away from your dinner?
Are you busy?
It is a general rule that the caller or visitor introduces the first
topic, and as we noted earlier if there is no specific reason for the
call or visit this is often explicitly stated—‘I was just dropping by’.
There are of course exceptions, most notably when the caller has
been asked to call and wants to be told why, and also when the
called (wants to give the impression that he) has been trying to
contact the caller, and uses his second turn not simply to reply to
the greeting but to initiate the first topic.
Where you been all day. I’ve been trying to get hold of you?
Even if the called doesn’t initiate, the first topic may still not be
the ‘reason for the call’. We mentioned above that sometimes a
caller may not want the real reason to occur in the distinctive first
topic slot and may therefore substitute another. At other times con¬
versationalists may not feel they have anything sufficiently important
to be preserved as the ‘reason for the conversation’ and there are
ways of talking past the first topic slot.

A: Hello there.
B: Hello.
A: What’s new with you?
B: Not much, and you?
A: Nothing. (Sacks ms)
The endings of conversations are also things that have to be
achieved—speakers don’t just stop speaking. Conversations virtually
always end with a closing pair, composed of ‘goodbye’, ‘good¬
night’, ‘see you’, and so on. However, the closing sequence can only
occur when a topic has been ended and other speakers have agreed
not to introduce any new topics. Arriving at a point where a closing
sequence can begin requires a certain amount of work.
86 AN INTRODUCTION TO DISCOURSE ANALYSIS

As we noted earlier topics frequently merge one into another.

There are, however, ways of bounding topics to produce a clean
ending. One way involves one party producing a proverbial or
aphoristic summary or comment on the topic which the other party
can agree with.
dorinne: Uh—you know, its just like bringin the—blood up
THERESA: Yeah well. THINGS UH ALWAYS WORK OUT
FOR THErBEST
dorinne: '-Oh certainly.
Another technique is for the speaker to indicate that he has noth¬
ing further to add to the topic by using his turn to produce simply
‘alright’, ‘okay’, ‘so’, ‘well’, often lengthened and with a falling
intonation contour. In doing this the speaker ‘passes’. This allows
the next speaker the choice of either introducing an entirely new
topic, because the constraints of topical coherence have been lifted,
or of also passing and turning the first speaker’s offered possible
pre-closing into a pre-closing sequence. Then, as neither speaker
has raised a new topic they can move into a closing sequence and
end the conversation:
Topic Theresa: Yeah well. Things uh always work
bounding out for the pbest.
sequence dorinne: LiOh certainly.
Pre-closing fi dorinne: Alright Tess.
sequence (^Theresa: uh huh. Okay.

Closing fidorinne: G’bye.

sequence 1 Theresa: Goodnight.
(Schegloff and Sacks 1973)
In that example both participants agreed that the conversation
had gone on long enough; however, sometimes one speaker wants
to end but for some reason is unable to achieve a topic bounding
sequence and is then forced into a different type of pre-closing:
either a statement which presents a reason for stopping,
I gotta go, baby’s crying,
or an offer to allow the other speaker to stop.
Well, I'll letchu go. I don’t wanna tie up your phone.
This is costing you a lot of money.
Okay I’ll letcha go back to watch your Daktari.
 CONVERSATIONAL ANALYSIS 87

Again, these are only possible pre-closings and especially the latter
kind may not be accepted, the other speaker may deny that he
wants to get away, though if he does accept they can then move
straight into the closing sequence:

B has called to invite C, but has been told C is going out to

dinner

"B: Yeah. Well get on your clothes and get

Pre-closing out and collect some of that free food and
sequence | we’ll make it some other time Judy then.
^C: Okay Jack.
Closing f B: Bye bye.
sequence 1 C: Bye bye.

These examples contain only the essentials of a closing, an

achieved pre-closing sequence and closing pair, but closings may
include making arrangements, re-emphasising arrangements made
earlier, restating the reason for the call, as well as many repetitions,
and continue for many utterances—Schegloff and Sacks quote a
‘modest’ example containing twelve utterances.
The slot after the ‘possible pre-closing’ is the one provided for
introducing any topic which has not yet received mention, and new
topics can be introduced after a pre-closing or even a closing,
provided they are marked as being misplaced.
caller: Okay, thank you.
CRANDALL: Okay dear.
caller: OH BY THE WAY I’d just like to ...

Any items inserted during the closing after earlier opportunities

have been passed up have the status of after thoughts and this posi¬
tion can be exploited in order to take away the importance of a
piece of news—one doesn’t normally forget really important items.
Schegloff and Sacks (ibid) quote the following example which
occurs, at the end of a fairly long telephone conversation, following
the pre-closing, when both speakers have apparently indicated they
have no more topics.
A: I - uh ;;; I did wanna tell you an I didn’ wanna tell you
uh;; last night. Uh because you had entert - uh, company.
I-I-I had something - terrible t’tell you. Soruh
B: Lhow
terrible is it?
88 AN INTRODUCTION TO DISCOURSE ANALYSIS

A: Uh, tuh . as worse it could be.

(0.8)
B: W-y’ mean Ada?
A: Uh yah.
B: What’she do, die?
A: Mm;:hm.

Stylistic features of conversation

In a fascinating series of lectures given during 1971 Sacks suggests
that conversation has much of the additional phonological, gram¬
matical and thematic patterning which is usually thought to typify
works of literature.
In a discussion of an extended version of the ‘skip-connecting’
passage (p. 78) he points out marked phonological patterning. There
is a series of words, ‘depressing’, ‘disgusting’, ‘degenerate’, ‘de¬
cline’, ‘decrepit’, in which the first two phonemes are identical,
displaying ‘reverse rhyme’. There are also words and phrases which
echo each other because they share phonemes—‘degenerate’, and
‘pier joint’; ‘walkin aroun drunk’, and ‘all kindsa fun’; ‘alcoholics’
and ‘all kindsa things’.
Sacks argues that features such as these cannot be rejected as
chance because they occur too frequently and any rejection on the
grounds of implausibility ignores the fact that speakers are all the
time achieving effects of similar sophistication and complexity.
These phonological echoes are evidence of how closely attentive
speakers are to each other—a speaker’s choice of one formulation
rather than another is partly determined by the phonological
patterning of the previous text and the alternative formulations.
Texts also display marked lexical patterning. This same frag¬
ment has a large number of marked contrast terms, for example,
‘these’ and ‘those’; ‘go to’ and ‘come from’; ‘in’ and ‘out’; ‘you’
and ‘they’; ‘men’ and ‘ladies’; ‘new’ and ‘old’; ‘ever’ and ‘never’;
‘pretending’ and ‘really’; ‘depressing’ and ‘fun’. The use of such
contrasting terms is particularly appropriate at this point in
the conversation because of the topic conflict which we discussed
above.
In the following example from a group therapy session Roger is
complaining about the way in which he has to describe his preferred
occupation—he wants to describe it as life-pervasive.
ROGER: When I say I wanna be something, its not just that I
wanna be this, its just I-I-I just - that’s the only thing
 CONVERSATIONAL ANALYSIS 89

I tell people that I wanna be an artist. Its really a

whole way of life ...

You visualise yourself uh living a certain way

I see it as a whole picture

I don’t see it that way at a*

I - How am I gonna live, what am I gonna do for a

living, and the whole scene.

And uh since most people don’t think along these

lines . ..
One must remember that this is unplanned conversation and that
while there is a vast number of ways to describe his problem he
consistently uses specifically visual, artist-appropriate, terms. ‘You
visualise yourself’; ‘I see it as a whole picture’; ‘I don’t see’; ‘the
whole scheme’; ‘along these lines’.
This topic-appropriate choice of terms applies to many meta¬
phorical uses of language. Sacks presents another extract from a
group therapy session where the patients have been talking round
the subject of ‘sex’ without actually mentioning it. The therapist
observes.
Well so far, all of you skirted around the subject. That
see(hh)ms to b(h)e predominantly uh on your minds at any rate.
The problem facing the therapist in producing this utterance is to
indicate that he knows the topic they have been talking around,
without actually introducing it himself. He does this by choosing an
expression, ‘skirt around’, which both means ‘to allude to’ and
also itself alludes, by a pun on ‘skirt’, to the hidden topic.
Stories in conversation tend to be created anew for each retelling;
jokes usually have a fixed form and can be ruined by slight altera¬
tions. Sacks focuses on the structural complexities of the following
‘dirty joke’.
ken: You wanna hear muh—eh my sister told me a story last
night . . . There—There was these three girls. And they
were all sisters. And they’d just got married to three
brothers . . . So, first of all, that night, theya’re—on their
honeymoon the—uh mother-in-law, says—(to ’em) well
90 AN INTRODUCTION TO DISCOURSE ANALYSIS

why don’tcha all spend the night here an’ then you cn
go on yer honeymoon in the morning. First night, th’
mother walks up t’ the first door an' she hears this
uuuuuuuuuuhh! Second door is HHOOOHH! Third door
there is nothin. She stands there fer about twunny five
minutes waitin fer sumpna happen—nuthin. Next morn¬
ing she talks t’ the first daughter an she sz—wh how
come ya—how come y’went YEEEEEEAAHAGGHH
last night, ’n daughter sez well, it tickled mommy, ’n
second girl how come ya screamed. O mommy it hurts.
hh third girl, walks up t’ her—why didn’ ya say any¬
thing last night. Wyou told me it was always impolite t’
talk with my mouth full.

The joke has a simple temporal ordering—the events are told in

the order in which they are said to have happened. There is also a
more complex sequential organisation by which each of the events
depends for its significance on its position with relation to other
events—-‘next morning’ has its relevance because there was a ‘last
night’; in order to understand what is going by reference to a
‘second door’ one uses that there was a ‘first door’. The joke has
two major units, the ‘first night’ and the ‘next morning’; the ‘first
night’ poses a problem which is solved the ‘next morning’, and the
solution is nicely placed as the punch line which closes the second
sequence and the joke itself.
The mother in the joke acts as a guide—the listener sees the
events through her eyes. The fact that she moves from first door to
second and second door to third as soon as she hears a noise shows
that it is noise which is important and enables the listener to see
that lack of noise is the important event at the third door. Her
questions next morning elicit explanations of why the noises were
different and the meaning of the silence.
The economy and organisation of the events in each sequence is
admirable. The joke depends on the fact that silence in the third
room is unexpected, but the joke doesn't actually say the mother
was surprised. In order to show that something is not normal one
needs at least but not more than three instances, two normal and
one non-normal. The events could be presented in any order, but
then it would be necessary to comment on the unexpectedness of
one of them. The joke presents two ‘normal’ events first and
reinforces this by the fact that the mother moves from one door
 CONVERSATIONAL ANALYSIS 91

to the next; the third event is seen as different from the first two
and the mother waits. The second sequence, ‘next morning’ pre¬
serves the sequence and allows the explanation of the unexpected
event to occur last, after two ‘normal’ explanations can predis¬
pose the listener to hear the punning response in its intended
meaning.
The punch-line to this joke has an extra dimension to it. The joke
was told to Ken by his 12-year-old sister, and it embodies adolescent
girls’ objections to parental rules and interference. The cleverness
of the punch-line is that it uses one rule to explain the breach of
another and incidentally complains that sometimes it is impossible
to satisfy one’s parents.

Concluding remarks
The descriptions of ‘conversational mechanisms’ offered by Sacks,
Schegloff and Jefferson provide detailed evidence of the high degree
of structuring in everyday conversation.
^ There are, however, major difficulties facing anyone who attempts
to use their analyses on his own data, because the descriptive cate¬
gories are, in the main, not defined. The reader must generalise
from a few examples to his own definition of ‘threat’, ‘warning’,
‘challenge’, ‘clarification’, ‘terminator’, and hope that he has cor¬
rectly isolated the criterial features. This difficulty arises because
there is no attempt to relate form to function, to explain how a
given configuration of linguistic symbols comes to convey a given
conversational meaning.
- A second problem is that Sacks, Schegloff and Jefferson do not,
and do not claim to, offer an analysis which will cope with the
whole of any text. Sacks’ method is to focus on interesting and
potentially rewarding stretches of text and to produce an analysis
by a process of comparison with apparently similar stretches in
other texts, and an appeal to intuition.
So far the major analytic category is the pair, of which there are
a fairly large number—many of which are known only by name,
while the definitions of others are vague. However, a large propor¬
tion of utterances are not pairs and the intending analyst is unsure
how to handle them.
A related problem is that it is difficult to know what constitutes a
variant of a particular structure. Question and Answer are reciprocal
parts of a pair and it is ^feature of pairs that the second part follows
immediately on the first. Schegloff presents exceptions in the form
92 AN INTRODUCTION TO DISCOURSE ANALYSIS

of insertion sequences when one pair is embedded within another

with the assumption ‘you answer this and I'll answer yours’. Sacks
provides the example quoted earlier

B: Yeah. Hurry up.

where B’s utterance comes betw-een the first and second parts of a
pair, but we have no idea what this utterance is an example of,
what similar utterances would look like, or whether in fact many
pairs have one or more non-paired utterances inserted.
This raises the whole question of what constitutes a structure.
Jefferson presents the side sequence as consisting of three utterances
—a statement, a misapprehension and a clarification, but an early
example shows that there can be at least two misapprehensions and
clarifications, and later examples show many other unanalysed and
disregarded utterances occurring during the sequence. Schegloff and
Sacks (1973) present a clear analysis of simple closing sequences
and then mention that some are more complex and contain extra
items—for which they provide no analysis.
A fourth problem is that Sacks and Schegloff appear to use
‘utterance’ and ‘turn’ interchangeably but whereas many utterances
consist of a single sentence, some don't. Speakers can achieve at
least two things during an utterance—for instance end one pair and
begin a next,

A: How are you feeling?

B: Fine thanks.
B: And you?

but it is not at all clear whether here B’s utterance constitutes a

single turn as it would for Duncan (1974) or whether it is two
turns, each members of a different pair.
The final problem is what importance to attach to intonational
and paralinguistic features. Although Sacks and Schegloff sometimes
mark pitch movements and stress, it is difficult to know how they
use them, and in fact their whole work is characterised by a lack of
explicit definitions—it is assumed that speakers and analysts can
recognise items for what they are.
Thus, in conclusion, one must say that although the work of
Sacks and his colleagues presents a great many insights into the
workings of conversation and some useful analytic tools, it suffers
from a lack of explicitness and formalisation, which hinders others
anxious to make use of it.
 5

Classroom interaction

One type of verbal interaction, that between teachers and pupils,

has received an enormous amount of attention during the past thirty
years from researchers in a wide range of disciplines. There is a vast
literature and comprehensive reviews can be found in Medley and
Mitzel (1963), Kliebard (1966), Biddle (1967), Meux (1967), Weick
(1968). Naturally, such studies have educational priorities: none is
concerned with simply observing the interaction and describing its
linguistic structure, turn-taking system, or productive, interpretive
and sequencing rules. Many studies, such as those of Gallagher and
Aschner (1963) and Taba et al (1964) concerned with levels of
thinking, produce an analysis in terms of abstract categories several
stages removed from the linguistic data.
There are, however, three descriptive systems which do attend
closely to the linguistic data—Barnes (1969), Flanders (1970) and
Bellack et al (1966). Barnes has two major educational tenets: that
pupils should be encouraged to participate and draw on their own
knowledge and experience as much as possible; and that teachers’
questioning should be more concerned with stimulating thinking
than eliciting factual information. The descriptive system he pro¬
poses concentrates on two aspects of the interaction: (a) pupils'
participation where he considers the amount and type of pupil
participation and the way the teacher handles the turn-taking system
and guides the development of the topic; (b) teachers' questioning
where he divides all the teachers’ questions into four categories:
factual, reasoning (open and closed), other open questions (not
requiring reasoning), social.
The results of such an analysis are necessarily statistical and the
‘meaning’ of the analysis is the relationship between the ideal state
of affairs provided by the educational theory and the actual data.
Thus Barnes notes that the pupils’ participation is too low; they ask
too few questions and when they are willing to contribute as in the
following extract their contributions are not built upon:
94 AN INTRODUCTION TO DISCOURSE ANALYSIS

T: You get the white . . . what we call casein . . . that’s . . .

er . . .protein . . . which is good for you . . . it’ll help to
build bones . . . and the white is mainly the casein and so
it’s not actually a solution . . . it’s a suspension of very
fine particles together with water and various other things
which are dissolved in water .. .
P,: Sir, at my old school I shook my bottle of milk up and
when I looked at it again all the side was covered with . . .
er . . . like particles and . . . er . . . could they be the white
particles in the milk?
P2: Yes, and gradually they would sediment out, wouldn’t they,
to the bottom?
P3: When milk goes very sour though it smells like cheese
doesn’t it?
P4: Well, it is cheese isn’t it, if you leave it long enough?
T: Anyway, can we get on . . . We’ll leave a few questions for
later.

Barnes found the results of the analysis of questions similarly

depressing: ‘what most impressed us in these figures was the pre¬
dominance of factual over reasoning questions in the three arts
lessons,’ and the fact that entirely open questions hardly ever
occurred. Again the interpretation is one step away from the
linguistic description and coloured by the educational theory—in
some cultures, which place a high value on rote learning, factual
questions may be more highly regarded than reasoning ones and
the following example might not be regarded as an illustration of
how some teachers are obsessed with terminology:

T: Where does it go before it reaches your lungs?

P: Your windpipe. Miss.
T: Down your windpipe . . . Now can anyone remember the
other word for windpipe?
P: The trachaea.
T: The trachaea . . . good . . . After it has gone through the
trachaea where does it go to then? . . . There are a lot of
little pipes going into the lungs . . . what are those called?
Ian?
P: The bronchii.
T: The bronchii . . . that’s the plural . . . what’s the singular?
What is one of these tubes called? Ann?
P: Bronchus.
 CLASSROOM INTERACTION 95

T: Bronchus . . . with ‘us’ at the end . . . What does ‘inspira¬

tion’ mean ... ?
The descriptive system proposed by Barnes is, intentionally,
partial, and only handles selected items in the data; one which
categorises all the data is Flanders (1970). The basic system com¬
prises ten categories, seven for teacher talk, two for pupil talk, and
one for ‘silence or confusion’.

1. Accepts feeling
Response 2. Praises or encourages
3. Accepts or uses ideas of pupils

Teacher
4. Asks questions
Talk
5. Lecturing
Initiation 6. Giving directions
7. Criticising or justifying authority

Response 8. Pupil response

Pupil talk
Initiation 9. Pupil initiation

Silence 10. Silence or confusion

This system was designed to be used in ‘real time’ for coding

lessons as they happen, not retrospectively from recordings or
transcripts, and the analysis consists of a series of symbols, recorded
at three-second intervals, from which one can discover what was
happening at any moment during the lesson.
To anyone involved in the analysis of language function, the
concept of a temporal rather than a linguistic unit is counter¬
intuitive, and of course the analysis will not distinguish one six-
second question from two three-second ones. However, Flanders is
not attempting to capture features of the interactive use of language
in his analysis.

The major feature of this category system lies in the analysis

of initiative and response which is a characteristic of interaction
between two or more individuals. To initiate, in this context,
means to make the first move, to lead, to begin, to introduce an
idea or concept for the first time, to express one’s own will. To
respond means to take action after an initiation, to counter, to
96 AN INTRODUCTION TO DISCOURSE ANALYSIS

amplify or react to ideas which have already been expressed,

to conform or even comply to the will expressed by others.
This at first seems similar to Sacks’ notion of the first and second
parts of pairs, but in fact the relationship is looser. Flanders’ major
interest is in topic and who controls it, how far and how often pupil
contributions are utilised by the teacher. Thus any teacher utterance,
of whatever grammatical form or illocutionary force, which is based
on something a pupil has said is regarded as a response.
Initiative P: The rain of the desert would make many plants
grow.
Response T: Mary thinks plants would grow because of the
rain. Would you agree or disagree Jerry?
We can see here a similar concern to that of Barnes and in Flanders’
terms many of Barnes’ examples show the teacher failing to pro¬
duce responding behaviour at crucial points.
As a characterisation of the discourse Flanders’ system obviously
ignores many features—while concentrating on the control and
development of topic it neglects what may be crucial aspects of the
turn-taking system. Sacks, as we noted earlier, claims that a speaker
asking a question of another, named participant is exerting the
greatest possible control over the discourse. In the example above
Flanders ignores this significance of the question. There are other
examples of this difference in emphasis. In one illustration Flanders
evokes a situation where the pupils are contributing eagerly and
‘the teacher’s only (sic) role is to decide who talks next’, by point¬
ing, nodding or calling out names. He suggests that here the
teacher’s role is trivial and ‘one of the better ground rules we have
used is to ignore the teacher'. In fact, in terms of turn-taking, this is
a highly significant activity.
Possibly because of the crude division of all utterances into
instances of initiative or response, possibly because of the temporal
rather than linguistic unit of analysis, Flanders discovers no larger
structures, except statistical tendencies which form patterns charac¬
terised by a predominance of teacher or pupil initiatives.
The system proposed by Bellack et al (1966) is the most sugges¬
tive for those interested in techniques of discourse analysis. His
system has several points in its favour: firstly, the analysis is in
terms of linguistic not temporal units; secondly he has intuitively
more acceptable ideas about initiating and responding behaviour,
seeing it as structurally not topically reciprocal; thirdly he intro-
 CLASSROOM INTERACTION 97

duces an extra category, reacting, to cope with teacher utterances

which are related to, but not called for by, pupil utterances;
fourthly, his categorisation of utterances is in terms of discourse
function rather than pedagogical function. He suggests that all the
interaction can be described in terms of four moves:

Structuring Structuring moves serve the pedagogical function

of setting the context for subsequent behaviour by either
launching or halting/excluding interaction between students
and teachers. For example, teachers frequently launch a class
period with a structuring move in which they focus attention
on the topic or problem to be discussed during that session.

Soliciting Moves in this category are intended to elicit (a) an

active verbal response on the part of the persons addressed;
(b) a cognitive response, e.g. encouraging persons addressed to
attend to something; (c) a physical response. All questions are
solicitations, as are commands, and requests.

Responding These moves bear a reciprocal relationship to

soliciting moves and occur only in relation to them. Their
pedagogical function is to fulfil the expectation of soliciting
moves; thus students’ answers to teachers’ questions are
classified as responding moves.

Reacting These moves are occasioned by a structuring,

soliciting, responding, or prior reacting move, but are not
directly elicited by them. Pedagogically, these moves serve to
modify (by clarifying, synthesising, or expanding) and/or to
rate (positively or negatively) what has been said previously.
Reacting moves differ from responding moves; while a respond¬
ing move is always directly elicited by a solicitation, preceding
moves serve only as the occasion for reactions.

The following is an example of the analysis applied to a piece of

data:

T: STR: Let’s turn to American investments abroad.

SOL: You suppose we do invest much money outside of
the U.S.?
P: RES: Yes.

The concept of move is a powerful one. Bellack notes that utter¬

ances typically consist of one or two moves and can never consist of
 98 AN INTRODUCTION TO DISCOURSE ANALYSIS

more than three and in this case the sequence must consist of a
responding or reacting move, followed by structuring and soliciting
moves. He notes that:

moves occur in classroom discourse in certain cyclical patterns

or combinations, which we designated teaching cycles. A
[typical] teaching cycle begins either with a structuring or a
soliciting move . .. continues with a responding move by the
student addressed and ends with an evaluative reaction by the
teacher.
He isolates in all, 21 different structural cycles and suggests that
‘styles of pedagogical discourse can be described in terms of cycle
activity, percentage of teacher-initiated cycles and distribution of
cycle types.’
There are, naturally, some problems in using Bellack’s analytic
system, and this is a useful point at which to discuss the adequacy
of all the analyses discussed in this and the previous chapter, in the
light of four criteria intended for all linguistic descriptions suggested
in Sinclair (1973).

1. The descriptive apparatus should be finite, or else one is not

saying anything at all, and may be merely creating the
illusion of classification.
2. The whole of the data should be describable; the descriptive
system should be comprehensive. This is not a difficult
criterion to meet, because it is always possible to have a
‘ragbag’ category into which go all items not positively
classified by other criteria. [Of course] if we find that 95%
of the text goes into the ragbag we would reject the descrip¬
tion as invalid.

While making apparently innocuous demands—that the system

should have a demonstrably finite number of items and be able to
handle the whole of a specified corpus—these two criteria, if
accepted, disqualify all systems except those of Flanders and
Bellack. As Labov (1972) notes. Sacks and his colleagues believe it
is inappropriate to attempt to describe ‘all the data’ at this stage,
and Labov himself only attempts to handle fragments. Nevertheless,
Sinclair’s point is a valid one, Austin (1962) suggested that there
might be as many as 10,000 performative verbs, but as Searle (1969)
points out there are still likely to be performative actions for
which there is no performative verb, suggesting that the system is,
at least potentially, infinite; while on the other hand some of
 CLASSROOM INTERACTION 99

Austin’s performatives—‘beg’, ‘entreat’, ‘request’—may be the

same illocutionary act seen from different status and social view¬
points.

3. There must be at least one impossible combination of

symbols.

This is the basic notion of linguistic structure discussed in the

previous chapter. If one cannot place constraints on particular
sequences of items occurring, the only statements one can make are
stylistic and significantly this rules out Flanders’ system. Only
Bellack makes true structural statements about the whole of his
field of discourse indicating which structures can and which cannot
occur.

4. The symbols in the descriptive apparatus should be pre¬

cisely relatable to their exponents in the data ... if we call
some phenomenon a ‘noun’, or a ‘repair strategy’ or a
‘retreat’ we must establish exactly what constitutes the class
with that label.

Bellack’s analytic scheme fails on this criterion but he is in good

company; Sacks for instance makes no attempt to circumscribe the
formal realisations of ‘question’, ‘threat’, ‘denial’, but simply pro¬
vides examples, assuming that others have the competence to
recognise future instances as members of particular categories.
Labov alone, in the work described in the last chapter, accepts the
dangers of ‘premature formalisation’ and attempts to write pro¬
ductive and interpretive rules for a small number of speech acts.
The descriptive system outlined in Sinclair et al (1972) and based
on tape recordings of interactions in English classrooms is, perhaps
not surprisingly, the only one which does not fail on Sinclair’s four
criteria; it has of course other shortcomings which will be discussed
later.

The English used by Teachers and Pupils

Sinclair et al see their ultimate goal as a descriptive system which
will cope with all forms of discourse, but argue that there are advan¬
tages in beginning with a formal situation like the classroom be¬
cause, whereas ‘in day-to-day conversation complex rules pertain
. . . within the classroom the mechanisms are much more apparent.’
To describe the interactions inside the classroom Sinclair et al
devised a rank scale model based on the principles outlined for
 100 AN INTRODUCTION TO DISCOURSE ANALYSIS

grammatical models by Halliday (1961). The basic assumption of a

rank scale is that a unit at a given rank, for example, word, is made
up of one or more units of the rank below, morphemes, and com¬
bines with other units at the same rank to make one unit at the
rank above, group. The unit at the lowest rank has no structure;
for example in grammar, morpheme is the smallest meaningful
unit, and cannot be subdivided into smaller meaningful units. How¬
ever, if one moves from the level of grammar to the level of phono¬
logy, morphemes can be shown to be realised by a series of
phonemes. Similarly, the smallest unit at the level of discourse will
have no structure, although it is realised by clauses, groups or
words at the level of grammar.
Each rank above the lowest has a structure which can be
expressed in terms of the units next below. Thus, the structure of a
clause can be expressed in terms of nominal, verbal, adverbial and
prepositional groups. The unit at the highest rank is one which has
a structure that can be expressed in terms of lower units, but does
not itself form part of the structure of any higher unit. The very fact
that one can talk of the structure of a unit implies that there are
certain combinations of smaller units which cannot occur (Sinclair’s
third criterion) and thus a unit about which one can state no con¬
straints in terms of the unit next below has no structure. It is for this
reason that sentence is regarded as the highest unit of grammar—
above is the domain of style.
The major advantages of describing new data with a rank scale
are two-fold: firstly, no rank has any more importance than any
other and secondly, if one discovers new patterning it is a fairly
simple process to create a new rank to handle it. This is in fact
what happened. Initially there were only two ranks, utterance and
exchange; utterance was defined as everything said by one speaker
before another began to speak, exchange as two or more utterances.
However, there are problems with such an analysis; the following
example has three utterances, but how many exchanges?

T: Can you tell me why do you eat all that food?

Yes.
P: To keep you strong.
T: To keep you strong. Yes. To keep you strong.
Why do you want to be strong?

An obvious boundary occurs in the middle of the teacher’s second

 CLASSROOM INTERACTION 101

utterance, which suggests a unit smaller than utterance, which was

labelled move.
At this stage the rank scale had as its highest unit exchange,
consisting of utterances consisting of moves, but when it became
evident that a large number of exchanges had boundaries inside
utterances, utterance as a unit of discourse was abandoned. Further
work suggested that many moves, like the teacher’s first contribution
in the example above, consisted of smaller discrete units, acts, and
later still it was suggested that exchanges combined to form trans¬
actions. The eventual system offered a description of lessons in
terms of five ranks, with each related to the one above by a ‘consists
of’ relationship.
Verbal interaction inside the classroom differs markedly from
desultory conversation in that its main purpose is to instruct and
inform and this difference is reflected in the structure of the dis¬
course. In conversation, topic changes are unpredictable and un¬
controllable, for as Sacks has shown, a speaker can even talk ‘on
topic’ without talking on the topic intended by the previous
speaker. Inside the classroom it is one of the functions of the teacher
to choose the topic, decide how it will be subdivided into smaller
units and cope with digressions and misunderstandings.
Obviously there are many different kinds of structure and organi¬
sation in any text and one could approach a lesson from a peda¬
gogical viewpoint and produce a pedagogical structure showing the
stages through which the lesson passed. Sinclair et al accept that
there will be both linguistic and pedagogical structure to a lesson;
what is in doubt is how large are the units for which there is
linguistic structure, how small are the units over which pedagogical
considerations exert a control, and what degree of overlap there is.
Figure 2 overleaf shows how they suggest that the largest units of
discourse will overlap with the smallest of non-linguistic organisa¬
tion, and the smallest of discourse with the largest of grammatical
organisation.
Sinclair et al emphasise that they have been ‘constantly aware of
the danger of creating a rank for which there is only pedagogical
evidence.’ In fact for the largest unit, lesson, they are unable to
provide a structure, and it thus has the same status as paragraph
in grammar. Transactions do have a structure, expressed in terms
of exchanges, and Sinclair et al note that the boundaries of trans¬
actions are marked almost always by frames, realised by four words,
‘well’, ‘right’, ‘now’, ‘good’. When used to indicate boundaries these
102 AN INTRODUCTION TO DISCOURSE ANALYSIS

words are strongly stressed, uttered with a falling intonation and

followed by a short pause. Their normal meaning is suppressed—
‘now’ has no time reference, ‘right’ or ‘good’ no evaluative func¬
tion—though at other places in the lesson these same items are used
normally.

Non-Linguistic
DISCOURSE Grammar
Organisation

course
period LESSON
topic TRANSACTION
EXCHANGE sentence
MOVE clause
ACT group
word
morpheme

Figure 2 Levels and Ranks

Teachers very frequently follow a frame, which indicates to the

pupils that one transaction has ended and another is beginning,
with a focus, a metastatement which tells them what the transaction
is going to be about:

frame: well
focus: today I thought we’d do three quizzes.

Often a transaction ends with a second metastatement, a conclusion,

which summarises what the transaction was about or has achieved
before the class moves on:

conclusion: what we’ve just done, what we’ve just done is

given some energy to this pen.
frame: now

It is instructive to compare these findings for the classroom with

those of Sacks and Schegloff for conversation. The teacher, because
his role involves the choice of topic is able to indicate in advance
what a chunk of discourse will be about, while the following
example would be anomalous in conversation just because conver¬
sationalists do not have this degree of control:
 CLASSROOM INTERACTION 103

well
today I thought we’d talk about my holidays in France.

However, as we saw in the previous chapter, conversationalists can

‘close down’ a topic by using items identical to those Sinclair et al
isolated as frames in classroom discourse, and while they cannot
produce metastatements about the future content of the discourse
they can and certainly do produce retrospective metastatements.
Here, then, we see a structural difference reflecting status differences
between the participants.
Transactions have a structure expressed in terms of exchanges—
they begin with a boundary exchange, which consists of a frame
and/or focus and this is followed by a succession of informing,
directing or eliciting exchanges. Informing, directing and eliciting
exchanges are concerned with what is more commonly known as
‘stating’, ‘commanding’ and ‘questioning’ behaviour though these
terms are not used for reasons explained below.
The structure of exchanges is expressed in terms of moves and
for illustrative purposes we will concentrate on eliciting exchanges.
Remembering Sacks’ work on Question-Answer sequences one
might have expected the eliciting exchange to consist typically of a
teacher question followed by a pupil answer and a series of eliciting
exchanges to produce the pattern T - P, T - P, T - P, but this in
fact is not the case; the structure is rather T - P - T, T - P - T,
T - P - T. In other words the teacher almost always has the last
word and two turns to speak for every pupil turn. This goes some
way to explaining the consistent finding that teachers talk on
average for two-thirds of the talking time. The teacher asks a
question, the pupil answers it and the teacher provides evaluative
feedback before asking another question.

T: Those letters have special names. Do you know what it is?

What is one name that we give to these letters?
P: Vowels.
T: They’re vowels, aren’t they?
T: Do you think you could say that sentence without having
the vowels in it?

It is suggested that the three move eliciting structure is the nor¬

mal form inside the classroom, for two reasons: firstly, answers
directed to the teacher are difficult for others to hear and thus the
repetition, when it occurs, may be the first chance some children
 104 AN INTRODUCTION TO DISCOURSE ANALYSIS

have to hear what their colleague said; secondly and more impor¬
tantly, a distinguishing feature of classroom discourse is that many
of the questions asked are ones to which the questioner already
knows the answer and the intention is to discover whether the
pupils also know. Often answers which are ‘correct’ in terms of the
question are not the ones the teacher is seeking and therefore it is
essential for the teacher to provide feedback to indicate whether a
particular answer was the one he was looking for. This three part
structure is so powerful that if feedback does not occur it is in
Sacks’ terms ‘noticeably absent’, and this is often a covert clue that
the answer is wrong.

T: (elicit) Can you think why I changed ‘mat’ to ‘rug’?

P; (reply) Mat’s got two vowels in it.
T: (feedback) 0

T: (elicit) Which are they? What are they?

P: (reply) ‘a’ and ‘t\
T: (feedback) 0

T: (elicit) Is ‘t’ a vowel?

P: (reply) No.
T: (feedback) No.

Moves combine to form exchanges; moves themselves consist of

one or more acts. The status and relationship of moves and acts in
discourse is very similar to that of words and morphemes in gram¬
mar. Just as the word is the minimal free form so the move is the
minimal contribution a speaker can make to an exchange; just as
some words consist of a single morpheme so some moves consist of
a single act.
The category act is very different in kind from Austin’s illocu¬
tionary acts and Searle’s speech acts. Acts are defined principally
by their function in the discourse, by the way they serve to initiate
succeeding discourse activity or respond to earlier discourse activity.
The definition of the acts is very general; elicitation for instance
has its function ‘to request a linguistic response’, directive ‘to
request a non-linguistic response’. The analysis thus does not
attempt to distinguish for example between ‘request’, ‘ask’,
‘entreat’, ‘beg’, ‘enquire’. However, as a descriptive system within
the Hallidayan framework it allows the concept of delicacy—
initially crude or general classifications can at a secondary stage be
 CLASSROOM INTERACTION 105

more finely distinguished. Thus, if it is possible to show that moves

containing an act ‘request’ are followed by a different class of
moves from those containing an act ‘entreat’, this distinction can
be built in at secondary delicacy—Straker-Cook (1975) demonstrates
how this can be done in an adaptation of the system to handle
lectures.
Precisely because the definitions of the acts are very general the
system requires only twenty-two, which, it is claimed, are sufficient
to describe the whole of the corpus, though not necessarily all class¬
room discourse let alone all discourse. Adaptations for other dis¬
course types, broadcast interviews (Pearce 1973) and doctor-patient
interviews (Coulthard and Ashby 1977), shows that while a
few new acts are needed and some of those already isolated are
possibly classroom specific, the system is in fact both flexible and
generalisable.
As a result of the asymmetrical status relationships and the fact
that one participant controls the topic and the right of the others to
contribute, there are several acts in classroom discourse solely
concerned with turn-taking. Obviously there has to be some
linguistic etiquette inside the classroom; if thirty pupils shouted out
in answer to every elicitation there would be chaos. There are
several ways in which teachers select next speaker. Sometimes they
nominate a child:
Joan, do you know who these people were,
sometimes children are required to bid by raising their hand or
shouting ‘sir’ or ‘miss’ and the teacher then nominates one of those
who have bid:
T: (elicit) Anyone think they know what it says?
P: (bid) HANDS RAISED
T: (nomination) Let’s see what you think, Martin.
P: (reply) Heeroglyphs.
T: (evaluation) Yes, you’re pronouncing it almost right.
At times with a new or difficult class a teacher may find he needs
to insist on the speaker-selection process. The following example
comes from a lesson given by an experienced teacher to a class he
hasn’t taught before, when he felt it necessary to provide cues to
the children to raise their hands and bid.
T: (cue) Hands up.
(elicit) What’s that?
106 AN INTRODUCTION TO DISCOURSE ANALYSIS

P: (bid) RAISED HAND

T: (nomination) Janet.
P: (reply) A nail.
T: (feedback) A nail, well done, a nail.

Interestingly the second, responding move in an exchange,

typically the pupil’s, does not have a complex structure in terms of
acts, while the third, follow-up move, typically the teacher’s and
concerned with fitting the reply into the ongoing discourse, can
consist of up to three acts—an accept, which takes the information
offered into the discourse, and evaluation which assesses its worth
and relevance, and a comment which contributes new related infor¬
mation. A typical exchange could look like this:

Eliciting exchange

initiating move Can anyone have a shot, (elicit)

a guess at that one ?

responding move Cleopatra. (reply)

follow-up move Cleopatra. (accept)

Good girl. (evaluation)

She was the most

famous queen, wasn’t (comment)
she?

Discourse structure
The system devised by Sinclair et al has five ranks and it is instruc¬
tive to compare these with the units proposed by Sacks—‘turn’,
‘pair’, ‘sequence’, ‘topic’ and ‘conversation’, which are remarkably
similar. One can usually equate ‘turn’ with ‘move’, ‘pair’ and
‘sequence’ with ‘exchange’, and ‘topic’ with ‘transaction’.
Scheflen (1964) suggests that there is evidence that speakers
signal kinesically the boundaries of a hierarchy of discourse units —
the sentence, the point, the position and the presentation—which
appear to approximate to the exchange, transaction and interaction
respectively. Kendon (1972) reports a more detailed study using
kinesic and paralinguistic evidence in which he also isolated four
 CLASSROOM INTERACTION 107

units—the discourse, the locution group cluster, the locution group

and the locution. Although the units have been isolated for different
purposes and on different criteria there is a significant similarity in
size, which the presentation below attempts to capture.

Sinclair Sacks Scheflen Kendon

Interaction Conversation Presentation
Transaction Topic Position Discourse
Locution Group
Exchange Point
Cluster
Move Turn — Locution Group
Act — Sentence Locution

Form and function

The major problem for a system which sets out to analyse discourse
is to show how the functional categories are realised by formal
items. Sinclair et al observe that while their categories of elicitation,
directive and informative are frequently realised by interrogative,
imperative and declarative structures respectively there are occasions
when this is not so. The opportunity for variety arises from the
relationship between grammar (in the broad sense) and discourse.
The unmarked form of a directive may be imperative, ‘Shut the
door’, but there are many marked versions, using interrogative,
declarative and moodless structures.

Can you shut the door?

Would you mind shutting the door?
I wonder if you could shut the door.
The door is still open.
The door.

To handle this lack of fit between grammar and discourse Sinclair

et al suggest two intermediate areas where distinctive choices can
be postulated situation and tactics. Situation refers to all relevant
factors in the environment, social conventions and the shared
experience of the participants; tactics handles the syntagmatic pat¬
terns of discourse, the way in which items precede, follow and are
related to each other. ‘It is place in the structure of the discourse
which determines ultimately which act a particular grammatical
item is realising, though classification can only be made of items
already tagged with features from grammar and situation.’
108 AN INTRODUCTION TO DISCOURSE ANALYSIS

Situation
In situation knowledge about schools, classrooms, one particular
moment in a lesson is used to reclassify items already labelled by
the grammar. Usually the grammatical types declarative, interro¬
gative, imperative and moodless, realise the situational categories
statement, question, command and response, but this is not always
so. Of the sixteen possible combinations—declarative command,
and so on—there is only one, imperative statement, which they say
they cannot instance.
The interrogative, ‘what are you laughing at’, is interpretable
either as a question, or as a command to stop laughing. Inside the
classroom it is usually the latter, and in the following example,
where the teacher has just played the children a tape of a man with
a ‘funny’ accent in order to discuss reactions to accents with them,
he still has to work quite hard to convince the pupil that he is ask¬
ing a question and not issuing a command requiring the cessation of
the activity and a conventionalised acknowledgement/apology.

teacher: What kind of a person do you think he is?

Do you—what are you laughing at?
pupil: Nothing.
teacher: Pardon?
pupil: Nothing.
teacher: You’re laughing at nothing, nothing at all?
pupil: No.
It’s funny really ’cos they don't think as though they
were there they might not like it and it sounds rather
a pompous attitude.

The girl’s mistake lies in misinterpreting the situation and the

example demonstrates the crucial role of situational information in
the analysis and interpretation of discourse. Labov (1972), it will be
remembered, handles the problem in terms of A and B events; here
he would say that the teacher assumed it was an A-B event that
laughter was permissible at this point but in fact it was only an A
event.
Sinclair et al suggest that there are four questions one needs to
ask about the grammatical form of a clause in order to be able to
analyse it as the realisation of a particular function.

1. If the clause is interrogative is the addressee also the sub¬

ject?
 CLASSROOM INTERACTION 109
2. What actions or activities are physically possible at the time
of utterance?
3. What actions or activities are proscribed at the time of
utterance?
4. What actions or activities have been prescribed up to the
time of utterance?
Using the answers to these questions they formulate three rules to
predict when a declarative or interrogative will be realising some¬
thing other than a statement or question. Figure 3 shows dia-
grammatically the classification of interrogatives by means of these
rules; the classification of other grammatical structures works in
the same way.

action command
proscribed

. action_ _ command
feasible
-subject =. polar _
addressee question
modals: action not_question
^ can, could, _ feasible
I Will, would,
N going to
T _wh- . question
E action not question
R " proscribed
RH action which should
o . have been but has_ . command
G not been performed
A all other
T verbs
I _ all others__question
V
E action_command
proscribed
.subject =#_
addressee
. action not___ question
proscribed

Figure 3 The classification of an interrogative by situation

Rule 1
An interrogative clause is to be interpreted as a command to
do if it fulfils all the following conditions:
110 AN INTRODUCTION TO DISCOURSE ANALYSIS

(i) it contains one of the modals can, could, will, would, (and
sometimes going to)',
(ii) if the subject of the clause is also the addressee;
(iii) the predicate describes an action which is physically pos¬
sible at the time of the utterance.

1. Can you play the piano, John. command

2. Can John play the piano? question
3. Can you swim a length, John? question

The first example is a command because it fulfils the three condi¬

tions—assuming there is a piano in the room. The second is a
question because the subject and addressee are not the same person.
The third is also heard as a question if the children are in the class¬
room, and the activity is not therefore possible at the time of the
utterance. However, Sinclair et al predict that if the class were at
the swimming baths example three would be interpreted as a com¬
mand and followed by a splash.

Rule 2
Any declarative or interrogative is to be interpreted as a
command to stop if it refers to an action or activity which is
proscribed at the time of the utterance.

1. I can hear someone laughing. command

2. Is someone laughing. command
3. What are you laughing at. command
4. What are you laughing at? question

The declarative command, as in the first example, is very popular

with some teachers. It is superficially an observation, but it’s only
relevance at the time of utterance is that it draws the attention of
‘someone’ to their laughter, so that they will stop laughing.
Examples two and three, though interrogative in form, work in
exactly the same way. Example four is only interpreted as a ques¬
tion when laughter is not regarded as a forbidden activity.

Rule 3
Any declarative or interrogative is to be interpreted as a com¬
mand to do if it refers to an action or activity which teacher
and pupil(s) know ought to have been performed or completed
and hasn’t been.
 CLASSROOM INTERACTION 111

1. The door is still open. command

2. Did you shut the door. command
3. Did you shut the door? question

Example one states a fact which all relevant participants already

know; example two is apparently a question to which all participants
know the answer. Both serve to draw attention to what hasn't been
done in order to cause someone to do it. Example three is a question
only when the teacher does not know whether the action has been
performed.
Obviously these rules are very similar to those proposed in
Labov (1970, 1972) and Labov and Fanshel (ms). Many linguists
object that one cannot appeal to participants’ knowledge of the
world or even of the situation because it is not available to the
analyst. Bever and Ross (1966), discussing the example ‘Everyone
should read the Bible. Deuteronomy is one of the great books of the
world’, argued that in order to handle this the descriptive apparatus
would need to include such information as the fact that Deuteronomy
is one of the books of the Bible. In other words it would be
necessary to build speaker’s knowledge into the grammar, which is
obviously impossible. However, rules such as those proposed by
Labov and Sinclair et al manage to explain interpretation of utter¬
ances by appealing to speakers’ knowledge without characterising it.
In this way they reflect what speakers do all the time; they con¬
tinuously ‘membership’ their co-conversationalists and if they
make wrong assumptions their utterances are misinterpreted.
Successful discourse rules must predict and cope with such ambi¬
guities; those of Labov and Sinclair et al certainly handle some of
them.

Tactics
In grammar items are classified according to their structure: the
relative position of subject and verb determines which clause is
declarative, interrogative or imperative. In situation information
about the non-linguistic environment is used to reclassify gram¬
matical items as statements, questions or commands. However, the
discourse value of an item depends also on what linguistic items
have preceded it, what are expected to follow and what do follow.
Such sequence relationships are handled in tactics.
The definitions of the discourse acts, informative, elicitation and
directive, make them sound remarkably similar to statement, question
112 AN INTRODUCTION TO DISCOURSE ANALYSIS

and command but there is a major difference, which can perhaps

best be described as potential v actual. While elicitations are always
realised by questions, directives by commands and informatives by
statements, the relationship is not reciprocal: questions can realise
many other acts and the expression ‘rhetorical question’ is a recog¬
nition of this fact. Statements, questions and commands only realise
informatives, elicitations and directives when they are in an initiat¬
ing position; an elicitation is an initiating question whose function
is to gain a verbal response from another speaker—questions occur
at many other places in the discourse but then their function is
different, and this must be stressed. A question which is not intended
to get a reply is realising a different act from one which is; the
speaker is using the question for a different purpose and this must
be recognised in the description.
Spoken discourse is produced in real time and the system of
functional categories attempts to deal with the ‘now-coding’ aspect
of speech. Speakers inevitably make mistakes, or realise that they
could have expressed what they intended much better. A teacher
may produce a question which he fully intends as an elicitation and
then change his mind. Obviously he can’t erase what he has said,
and he doesn’t tell the children to ignore it, but he does signal that
the children are not expected to respond as if it were an elicitation.
In the ‘what are you laughing at?’ example discussed above, the
teacher abruptly changes course in the middle of a question. This is
rare and signals to the class that what has gone before should be
regarded as if it had never been said, deleted completely.
More frequently, as in the example below, the teacher follows
one potential informative, directive or elicitation with another,
usually a more explicit one, signalling paralinguistically, by intona¬
tion, absence of pausing, speeding up his speech rate, that he now
considers what he has just said a starter. Starters are acts whose
function is to provide information about or direct attention or
thought towards, an area in order to make a correct response to the
initiation more likely. But this function is only post-factum, when
the teacher perhaps realises that the intended elicitation was not
specific enough.

teacher: What about this one? This I think is a super one.

Isobel, can you think what it means?
pupil: Does it mean there’s been an accident further along
the road?
 CLASSROOM INTERACTION 113

The teacher begins with a question which appears to have been

intended as an elicitation. She changes her mind and relegates it to
a starter. The following statement is in turn relegated by a second
question which then does stand as an elicitation. In any succession
of statements, questions and commands the pupil knows that he
usually only has to respond to the final one; only the final one has
the status of an initiation. This ‘rule’ can lead to an incorrect
response if the pupil doesn’t fully understand what the teacher is
saying. In the following example a quoted question is heard as an
elicitation.
pupil: Well, he should take some look at what the man’s
point of view is.
teacher: Yes, yes.
But he wasn’t asked that question don’t forget. He
was merely asked the question ‘Why, why are they
reacting like thisV
pupil: Well, maybe it’s the way they’ve been brought up.

Observations
One of Sinclair’s criteria is that the descriptive system should
attempt to cope with all the data for only then can one fully appre¬
ciate the inadequacies. Any attempt to use this descriptive system
reveals one major inadequacy—it can only nominally handle
monologue. Sacks, here, has no problem because ‘turns’ can be of
any length, but an analysis which attempts to divide turns or moves
into smaller units has great difficulty with lengthy utterances. At
the simplest level there is the question of whether a two-minute lec¬
ture by the teacher is one inform or a series: a more difficult prob¬
lem comes at third part in the exchange when the teacher produces a
comment on the pupil’s utterance which appears to drift into an
inform beginning a new exchange. In the example below it is diffi¬
cult to know where or indeed whether to draw the line:
T. elicit: See if you can see anything that you think is
rather peculiar.
P. reply: There’s two aitches.
T. accept: Yes.
T. evaluation: Some are are duplicated aren’t they?
T. comment: There are two symbols for one sound.
? I haven’t been able to find out why this is.
? I’ve got some books that you can look at just
now to see if you can find out just why it is.
114 AN INTRODUCTION TO DISCOURSE ANALYSIS

Pearce (1973) argues that perhaps the best solution is to see

larger contributions as a different type of discourse, not interactive
in the same way and therefore not suitable for this type of analysis
which would then simply cope with the points of speaker change at
the beginnings and ends. A skilful conversationalist can produce
highly structured monologues, composed in real time, which are
more suited to an analysis devised for written texts. The following
example is a larger extract from the interview with Denis Healey
quoted in Chapter IV.

. . . now I think one can see several major areas concerning the
unions which require to be considered there’s first the question
with which the industrial relations act is mainly concerned and
that is how industrial disputes are handled . . . now I believe
myself the TUC and the CBI and it’s interesting that they agree
on this are right in saying that the government by trying to be
judge and jury in its own case has really so tainted the idea of
government interference in industrial negotiation that it’s
desirable to have a conciliation service set up as in the United
States independent of government altogether ... as you know
the Labour Party and the TUC have pretty well reached agree¬
ment in details of this and er I believe the CBI and the TUC
are very likely to do the same unless Mr. Heath upsets it by
insisting on interfering and this is one very important side er of
the act . . . now the second big area of course is the question of
how you handle incomes and I myself very strongly believe
that we have to establish in Britain two fundamental principles
... first of all that differentials that’s to say the gap between
what one man earns and a slightly more skilled man earns have
got to be rather smaller than they have been in the past if the
people at the bottom of the ladder are to have a chance of a
decent life. . . secondly, that increased earnings earnings
above the average have got to depend on increased produc¬
tion and the real trouble is that over the last two years every
single thing the present government has done has been deliber¬
ately calculated to widen differentials rather than reduce them
and to reward people who don’t work for their living rather
than those who do. ...

A major criticism of the descriptive system is by Mountford

(1975b) who observes that
 CLASSROOM INTERACTION 115

the model as constructed sets out to analyse the products of...

discourse activity, to account for the resultant data in terms of
a descriptive apparatus that is applied to the discourse ex post
facto.

He argues that the analysis concentrates on the product rather than

the process and maintains that ‘ a model of discourse analysis should
be able to account for how participants understand discourse as a
communicative activity.’ Similar criticisms have been voiced infor¬
mally by others working within an ethnomethodological framework.
This criticism appears to be based on a misunderstanding of the
generality of the descriptive system. Possibly because Sinclair et al
talk in terms of ‘analysis’ and ‘categorisation’, base their descrip¬
tion on tape-recorded data and provide extensive analysed examples,
Mountford and others assume that what has been produced is
simply an analytic grid to slot items into. Had Sinclair et al stressed
the tactics component and their version of interpretive rules the
description may have seemed closer to Mountford’s ideal which
‘should reflect the interpretive and compositional procedures of the
participants as participants'.
In fact some of the interpretive procedures discussed above, such
as the way in which a missing third part in an eliciting exchange is
interpreted as a negative evaluation, and the downgrading of poten¬
tial elicitations by succeeding ones, seem to be just the kind of
devices to handle discourse as a developing process.
Widdowson (1975) goes so far as to suggest that ‘descriptions of
use in terms of precise rules may give an inaccurate picture of how
people use language.’ The work of Brazil (1975), reported in detail
in the next chapter, suggests that, on the contrary, at the level of
exchange one can describe exhaustively the options open to
speakers, and that, at least at this rank, precise rules are both pos¬
sible and appropriate.
 6

Intonation

Many of the descriptive systems outlined above have relied, at times

quite heavily, on intonation in deciding which categories individual
utterances or parts of utterances belong to. Austin refers on several
occasions to the importance of ‘tone of voice, cadence, emphasis’
and ‘intonation’; Sinclair et al use high falling intonation as the
distinguishing feature of frames', while Sacks and his colleagues are
obviously using intonational clues when they transcribe utterances
declarative in form as questions. However, no one currently in¬
volved in discourse analysis marks intonation continuously in their
transcriptions; appeal to intonation is apparently spasmodic, if not,
haphazard, and usually occurs when differences are perceived for
which there can be no other explanation.
Most descriptions of intonation are, however, of little use to those
interested in coherence and cohesion or productive and interpretive
rules. Crystal (1969) is a major attempt to state exhaustively the
intonational options in English, but, although he accepts that

statements of the meaning of the prosodic contrasts [he has]

described must indeed be the dominant aim of the linguist,
he feels that in the current state of knowledge it is impossible to
make any valid generalisations about meaning. O’Connor and
Arnold (1959) are more daring. They suggest that

the contribution that intonation makes is to express in addition

to and beyond the bare words and grammatical constructions
used, the speaker s attitude to the situation in which he is
placed.
However, their attempts to provide valid generalisations about the
attitudinal meanings of the 10 major tones they isolate only serve
to demonstrate the difficulties and dangers of which Crystal warns.
For instance, they suggest that a speaker uses a low falling intona¬
tion with a statement to indicate that it is definite and complete in
 INTONATION 117

the sense that it is a ‘separate item of interest’; in addition the

intonation conveys a 'cool, calm, phlegmatic, detached, reserved,
dispassionate, dull, possibly grim or surly attitude on the part of the
speaker’. Their own examples demonstrate the difficulties admir¬
ably:

You’ve got Jipstick on your collar again.

It’s getting Jate.
It is hard to hear the first utterance as ‘cool’, ‘calm’ or ‘detached’,
and the second as ‘grim’ or ‘surly’. In fact even though O’Connor
and Arnold offer a hundred different ‘meanings’ including such
fine distinctions as ‘mild surprise but acceptance of the listener’s
premises’, ‘critical surprise’ and ‘affronted surprise’, these meanings
depend, as Brazil (1975) observes, far too much on ‘the co¬
occurrence of particular lexical items’, and this casts doubt on the
validity of the whole system.
Halliday (1967) while ‘not suggesting that the view of intonation
as carrying emotive meanings is to be rejected ’ asserts that in many
cases intonation patterns ‘can be systematised into a formal gram¬
matical statement’. His argument is that all meaning differences in
language are referable to either lexis or grammar and, since English
intonation is never lexical, the contrasts must be absorbed into a
grammar suitably elaborated to receive them. He distinguishes five
major tunes or tones identified according to pitch movement:

Tone Visual symbol Tonic Movement Terminal pitch

tendency

1 \ falling low

rising high
('
r falling-rising high

3 —/ rising mid

4 (rising)—falling—rising mid

5 (falling)—rising—falling low

At a very general level Halliday suggests that tone choices are

concerned with the simple opposition between certain and un¬
certain polarity:
118 AN INTRODUCTION TO DISCOURSE ANALYSIS

if polarity is certain, the pitch of the tonic falls; if uncertain, it

rises. Thus tone 1 is an assertion or a query not involving
polarity; and tone 4, which falls and then rises is an assertion
which involves or entails some query. Tone 2 is a query . . . and
tone 5 which rises and then falls is a dismissed query, one
countered by an assertion. Tone 3 avoids a decision; as an
assertion, it is at best confirmatory, contingent or immaterial.
We can exemplify the way these general meanings are given
specific, even attitudinal significance in declarative utterances. For
declaratives, with certain polarity, tone 1 is the neutral or unmarked
choice; all other tones have meaning as marked choices. The double
lines / / indicate the boundaries of the tone group, the number at
the beginning indicates the tone, while the syllable in italics is the
tonic.
//II saw him yesterday / / (neutral)
/ / 2 I saw him yesterday / / (contradictory) (‘challenging’
‘oppressive’ ‘defensive’ ‘indignant’
etc.)
/ /3 1 saw him yesterday / / (non-committal) (‘disengaged’
‘unconcerned’ ‘discouraging’)
//4 I saw him yesterday / / (reservation) (‘there’s a ‘but’ about
it’)
II51 saw him yesterday / / (committed) (‘involved’ ‘assertive’
‘superior’, ‘encouraging’)
Thus, if he so wishes, the speaker has four major options to
indicate his degree of involvement with the information and his
interlocutor.
Although the examples have been and usually are presented
as free-standing isolated utterances this is not the way they occur.
Very many clauses, both utterance-initial and utterance-medial,
are connected topically to a previous one—in the example above
‘him’ is referring back to someone identified earlier. We could
imagine the following extract from a dialogue;
A: //1 When did you see John! I /
B: //1 I saw him yesterday / /
Halliday observes that all items in clauses can be divided into
given and new, with new information typically occurring at the end
of the clause and carrying the tonic as in the example above.
Decisions about the placing of the tonic Halliday discusses as
tonicity choices. There are occasions when the items at the end of
 INTONATION 119

the clause are given and the new information occurs earlier—this
Halliday labels marked tonicity:

A: 111 Did he leave a message for you/ /

B: II 1 No I saw him yesterday / /

Halliday similarly regards as unmarked cases those clauses which

consist of a single tone group. However, it is possible to have more
than one tone group in a clause when the amount of information to
be transmitted is dense—typically news bulletins have a high ratio
of tone groups to clauses, but the examples of marked tonality are
not rare in conversation:

A: 1/2 Did he write to you/ /

B: II1 No I saw him//1 yesterday/ /

No one has yet attempted to take account of the way speakers

manipulate the tonality and tonicity systems to structure their
messages, but obviously these two systems, intimately concerned
with the packaging of information and decisions about what a
speaker takes to be ‘given’ and what he regards as ‘new’, are likely
to be of great importance; just as one speaker must membership
another to decide on an appropriate formulation of place, time or
whatever, so he must equally membership to divide his message
into bits of given and new information.

Intonation in a model of discourse

As we mentioned above Halliday regards a speaker’s option to
break up a stretch of language into a large or small number of tone
groups as having linguistic significance in itself. Thus in the follow¬
ing example the occurrence of a tone group boundary at the end of
the first clause in (b) would be seen as realising a term in a system
which opposed the two versions as a semantically relevant choice.

(a) //when I’ve finished Middlemarch I shall read Adam

Bede//
(b) //when I’ve finished Middlemarch//I shall read Adam
Bede//

Brazil (1975) argues that this is a consequence of taking a gram¬

matical view of intonation and that the full significance of intona¬
tion can in fact be derived from choices associated with each
separate tone group. That is, in any sequence of tone groups the
fact that a speaker ends one group at a particular point can be
120 AN INTRODUCTION TO DISCOURSE ANALYSIS

attributed to the fact of his choosing afresh from a set of para-

digmatically available options; the resultant division of the utterance
into fewer or more tone groups is a simple reflex of successive
choices of this kind. It is pertinent here to point to evidence in a
related field and to invoke Laver’s view that the tone group is ‘the
most likely candidate for the unit of neurolinguistic pre-assembly’:

There is good evidence for believing that... the tone group is

handled in the central nervous system as a unitary behavioural
act and the neural correlates of the separate elements are
assembled ... before the performance of the utterance begins.
(1970)
In any phonological analysis there is the problem of what categories
to use and what phonetic data count as instances of the same
category. Brazil proposes no new categories—he accepts Halliday’s
five tones and a distinction of key, ‘high’, ‘mid’ and ‘low’, first
proposed by Sweet in 1906—what Brazil does do is to offer a new
and exciting interpretation of the significance of these phonological
options.

Tone
Brazil suggests that for any tone group a speaker has a fundamental
choice between a falling tone and a falling-rising tone, Halliday’s
tones 1 and 4.
\ / x
/ / when I’ve finished Middlemarch/ / I shall read Adam Bede / /
X V' >
\ \ f
/ / when I’ve finished Middlemarch/ / I shall read Adam Bede/ /
\ v
Whatever additional meanings are realised by these alternative tone
choices, the first utterance is certainly to someone who is expected
to know already that the speaker is reading Middlemarch but for
whom the speaker’s future intentions are an item of news; in the
second utterance the question of the speaker’s reading Adam Bede
has somehow already arisen and he is offering, as news, information
about when he will read it. Significantly, this interpretation is main¬
tained when the order of the clauses is reversed:
x \ /
/ /1 shall read Adam Bede / / when I’ve finished Middlemarch/ /
\ / \
/ / I shall read Adam Bede / / when I've finished Middlemarch//
v \
 INTONATION 121

Halliday (1970) suggests that a co-occurrence of tones 1 and 4

expresses an unequal relationship between the two parts of the

utterance; one part (tone 1) is presented as the main informa¬
tion and the other (tone 4) as some kind of accompanying
circumstance.

While this is a relevant observation it does not help us to a satisfy¬

ing explanation of how these two utterances come to have their
different meanings.
Many researchers have commented on the fact that a speaker can
never say exactly or explicitly all that he wants to communicate—
he must assume shared knowledge, opinions and meanings. As we
have seen Labov (1970, 1972) attempts to distinguish, for analytic
purposes shared from individual knowledge with his concept of A,
AB and B events, but there remains the problem of how one decides
and indeed who decides whether a particular utterance is an event
of a particular kind. In the following example A assumed his utter¬
ance was a B-event, B assumed it was an A-event.

A: So there’s no playgroup next week then.

B; Oh isn’t there?
A: No, I was asking.
B; Oh I don’t know.

Brazil argues persuasively that a speaker can in fact indicate, by

the choice of the falling-rising tone, those tone groups whose content
he regards as part of the shared, already negotiated, common
ground, as A-B events; whereas choice of falling tone indicates that
the content is shared. Indeed Brazil speculates that

it may be no accident that so many of the so-called ‘pause

fillers’ that abound in spontaneous English speech are cut to
one of two patterns: the ‘you know’ pattern, and the ‘I mean
to say’ pattern.
To the fall-rise tone, the ‘you know’ type, he gives the label
referring (symbol r); to the falling tone, the ‘I mean to say’ type,
proclaiming, (symbol p). In the examples which follow the symbol
is placed at the beginning of the tone group and the tonic syllable
is italicised:

//r when I’ve finished Middlemarch / fp I shall read Adam

Bedell

The referring tone on ‘Middlemarch’ indicates this is part of the

122 AN INTRODUCTION TO DISCOURSE ANALYSIS

shared knowledge; the proclaiming tone on ‘Bede’ that this is new.

For the purposes of presentation Brazil uses invented examples,
but the binary choice between referring and proclaiming and its
relevance in the structure and transmission of messages can be seen
in the following extract from a doctor/patient interview:

//r I’ve come to see you / / p with the rash I / r I’ve got on my
chin lip and underneath / / r which has developed / / p in
the past three days / / r well it’s irritating / / r and at work / /
r with the dust / / / / r us being a clothing factory / / r well I
find it’s irritating lip makes me want to scratch it 11

The decision as to which parts of the message can be regarded as

shared, and marked as such by use of referring tone, is the speaker’s
and is made in the light of his moment-by-moment assessment of the
state of play. In the above example the fact that the patient has
come to see the doctor, that his rash—once it has been identified
as the reason for the visit—is on the chin and that it has ‘developed’,
are all visibly evident; that it is underneath his chin, and has
developed ‘in the last three days’ are not at all evident and are
therefore proclaimed. The reason for the use of referring tone for
‘//with the dust// us being a clothing factory//’ is less clear
because there is not sufficient information available to the analyst
about the relationship between this doctor and patient—it may be
that this doctor knows the patient well enough for the patient to
assume that he knows where he works, and that this is therefore
shared knowledge; but this may be an example of the common use
of referring tone for information which the speaker wishes to slip
into the situation as if it had been already negotiated. A less inno¬
cent example would be the prefacing of a statement with

/ / r as we all know II...

which can usefully be compared for effect with

lip as we all know II...

Once he has established the referring/proclaiming distinction,

realised by Halliday’s tones 4 and 1, Brazil suggests that for each
there is a marked or intensified form, which the listener hears as
indicating some measure of extra involvement on the part of the
speaker. For the falling, proclaiming, tone, the intensified form is
the rise-fall, Halliday’s tone 5, symbol p + :
 INTONATION 123

/ Ip when I’ve finished Middlemarch// (the plain proclaiming

form)
//p+ when I’ve finished Middlemarch// (. . . and not a
moment before!) / N*

It may be less obvious that the rising tone, Halliday’s tone 2,

stands in the same kind of relationship to the fall-rise. The rising
tone has been most frequently exemplified in single tone-group
utterances, and so has come to be associated in a common-sense
way with ‘questions’; although, in fact, many questions have falling
intonation and when rising pitch is associated with interrogative
function the fact can be related to the more general notion of
reference. In each of the examples introduced so far, the rise can
commute with the fall-rise to produce a more committed alter¬
native, while in the following examples the alternation accounts for
the distinction between mild and insistent questions,
\ t
II r had he read it / /

// r + had he read it / /

and ‘mild’ and ‘insistent’ conditions.

\ ? \
II r had he read it / / p he would have understood //
V
II r + had he read it / /p he would have understood //
The low-rising tone, Halliday’s tone 3, realises a neutral option
whereby the speaker can avoid making the tone group either pro¬
claiming or referring. Use of this tone has been variously character¬
ised in teaching manuals as ‘careful ’, ‘ uninvolved ’ and ‘ patronising ’.
The common factor in all such labels seems to be a kind of with¬
holding or withdrawal from the interactive situation—a disengage¬
ment which can be thought of as occupying the opposite pole to the
intensified options discussed above. Such a view is quite consistent
with another function often attributed to the low rise, namely that
of marking the incompleteness of the utterance to focus upon the
continuity of the discourse rather than upon the process of working
towards a convergence of views is indeed to disengage oneself.
Extensive use of this tone is a feature of certain distinctive speaking
styles, such as that adopted when telling stories to young children
or delivering broadcast news headlines.
124 AN INTRODUCTION TO DISCOURSE ANALYSIS

The five tones can now be related to each other as follows:

,-unmarked : fall-rise

r-refer

-intensified : rise

r-unmarked : fall

-proclaim

-intensified : rise-fall

-neutral low rise

Such a representation shows clearly the interdependence of two

different dimensions of meaning—the refer/proclaim and the
involved I uninvolved. Attention to the first leads us to a considera¬
tion of how the parties to a conversation adjust the presentation of
information to take account of the others’ assumed knowledge and
viewpoint, while the second provides an indication of the engage¬
ment of the individual speaker.

Key
The key system offers three choices, high, mid, low, and thus for
each successive tone group a speaker must choose a pitch above, at,
or below the level which for him is the norm in the circumstances.
Factors like excitement or anger can easily affect both the level of
the mid term and the range from high to low, and also the pitch
may vary considerably during the course of an extended utterance.
However, leaving aside such complications for the moment, and
assuming—as seems not improbable—that there is a progressive
stepping down in a lengthy utterance from one tone group to the
next, every tone group functions as an exponent of one of the three
phonological categories:
high--.-

mid

low
 INTONATION 125

Brazil proposes a phonological unit above the tone group which

consists of an indefinite number of high key choices, a mid key
choice and an indefinite number of low key choices:

(High 1... n) Mid (Low 1... n)

This structure suggests that any high key tone group(s) can be
thought of as being bound to a succeeding mid key tone group and
any low key tone group(s) as bound to a preceding one. High key
carries the implication ‘there is more to follow’, low key ‘this is
said in a situation created by something that went immediately
before’. In discourse terms, one sets up expectations, the other has
prerequisites.
While choices in the tone system were essentially interpersonal,
concerned with shared and new information, choices in the key
system are textual, concerned with relations among linguistic items.
The options open to a speaker when he embarks upon a new tone
group are:
-contrastive —— high key
-marked -—-
tquauve IOW Key

L- neutral- — mid key

A speaker’s choice of high key is heard as contrastive—this is

most evident when the contrasts are within the linguistic system.

high* mid low

p he didn't say a word
r but we know he speaks English

Here the implication ‘its not that he cant speak English’ is achieved
by associating the contrastivity derived from the key system with
one of the terms in a grammatical system, namely polarity. The
key system can be similarly used to achieve lexical contrasts, in the
following example between ‘read’ and ‘speak’ or ‘say’, but here the
contrasts are created by the text for this specific purpose and do not
necessarily exist outside it.
p he didn’t say a word
r but we know he reads English
* Brazil uses a three-column presentation to indicate key, each new tone
group begins on a new line, and the position of the first word indicates
whether the key is high, mid or low.
126 AN INTRODUCTION TO DISCOURSE ANALYSIS

These examples show that the listener must interpret the content
of high key tone groups as being from a closed set of possibilities—
while lexical classes are open and one item does not necessarily
contrast with any other, the speaker indicates that the listener
should hear two particular items or sets of items as contrastive in
this particular context. High key thus allows the speaker to contrast
any item with any other:
high mid low
p we’re going to Margate this year
r not Bognor
Such an utterance fits a situation where it is already recognised
that Bognor is the only alternative that need be taken into account.
It may be that the other participant has asked about Bognor, or
perhaps, as long-standing acquaintances, they both know that the
possibilities are limited to two. In either case, the speaker takes for
granted the existence of a closed set of options in the common
ground already negotiated.
The case can be compared with the choice of proclaiming tone
with high key:
p we’re going to Margate this year
p not 5ognor
Here, the speaker includes as part of his contribution to the shared
knowledge the information that Bognor was the alternative. To put
the matter a slightly different way, use of high key with referring
tone assumes that a potential contrast has already been established,
either within the referential framework that the participants
uniquely share or in the wider social context:
r I’ve used most of it
r but you can have what’s left if you like
With proclaiming tone, on the other hand, the speaker can exploit
a contrast not yet assented to by the other party:

high mid low

r as soon as he’d finished eating
p he changed into tennis gear
The likelihood of the speaker continuing with something like ‘... of
all things!’ supports the interpretation that he is here making a
polar distinction between ‘changing into tennis gear’ (quite in¬
appropriate. inconsiderate, or otherwise unacceptable in the circum-
 INTONATION 127

stances), and ‘getting on with the washing-up’ or ‘doing anything

else’. His message includes, of course, the assumption that such a
proper/improper classification will be acceptable to the hearer.
The meaning of low key derives, in various ways, from a condi¬
tion of equivalence which, in a given conversational context, a
speaker may assume exists between two successive tone groups in
the discourse
high mid low
p we gave it to our neighbours
p the Robinsons
The selection of low key for ‘the Robinsons’ signals the speaker’s
expectation that the hearer will already know who the neighbours
are, that this tone group is no more than a restatement of infor¬
mation already proclaimed. The choice of high key for ‘the
Robinsons’ would have contrastive implications signifying perhaps,
‘the Robinsons—not the other neighbours, the Smiths’. There is, of
course, a third option; the speaker could select mid key, the neutral
option, so-called because it adds no meaning other than that
brought by the choice of proclaiming or referring tone. Thus in
p we gave it to our neighbours
p the Robinsons
the fact that ‘ the Robinsons ’ are ‘ our neighbours ’ is proclaimed as
new information.
At this point it may be helpful to bring together in a single dia¬
gram the two sets of options from which a speaker must select in
producing every tone group.
- unmarked fall-rise
r refer
- intensified rise

[- unmarked fall
L- proclaim
rtone - intensified rise-fall

•-neutral low-rise

contrastive high key

r marked restrictive low key
Lkey
L neutral mid key
128 AN INTRODUCTION TO DISCOURSE ANALYSIS

Key choices at utterance boundaries

We have so far characterised the meaning of the tone and key
options available to a single speaker during a single utterance. How¬
ever, discourse consists of successive utterances by different
speakers and we must now consider whether the relationships
between successive tone groups produced by different speakers are
the same as those between successive tone groups from the same
speaker.
We can certainly account for the function of low key at the
beginning of utterances by a simple extension of what was said
about low-key choices in medial position.
high mid low
A: p what did you think of the acting
B: r O/zvier
p was very good
Here the function of low key, in narrowing the amount of common
ground from ‘acting’ to ‘Olivier’, is very similar to its function in
the following single speaker sequence:
r I can’t say I thought much of the acting
r O/zvier
p was very good
A similar example, but this time with an exchange boundary in the
middle, is:
A: p where do you live
B: p in the A venue
A: p which end of the Avenue
Here, the second question is heard as requesting a ‘narrowing’ of
the information supplied in response to the first. There is a conver¬
sationally recognisable sense in which the second is ‘subordinated’
to the first, in a way that would not be true of, say, a high-key
question, ‘'which Avenue?’.
The choice of high key at the beginning of an utterance can be
similarly explained as an extension of what was said earlier about
contrastivity. In the following example the choice of high key can
be seen as contrasting ‘Elizabeth’ with all others and insisting on
her uniqueness in a way that a mid-key choice would not do.

p this is Elizabeth
p Peter’s wife
 INTONATION 129

However, there is an alternative interpretation: high key may be

selected to mark off the ensuing part of the discourse as in some
way distinctive from what has preceded it. To put it another way,
the speaker may be underlining either the separateness of Elizabeth,
in a context where there is the likelihood of confusion about her
identity, or the separateness of what he is now doing—turning,
perhaps, from one group of guests at a reception and starting on a
new batch of introductions. Both acts fall within the definition of
contrastivity, but this extension involves recognising that the closed
set within which a contrast is defined may have sections of the
discourse for its members.
The frames which Sinclair isolated, and no doubt the possible
pre-closings of Schegloff and Sacks, are uttered in what we now
recognise as high key
high mid low
P r‘ght
p so that the end of the first quiz
In fact high key occurring with or without a framing item can be
an indication of a higher level boundary. A doctor, having ex¬
hausted one line of enquiry will sometimes begin a new sequence of
questions in high key, and elsewhere in the interview use it as a
mark of transition from diagnosis to prescription. Similarly, when
a lecturer, having indicated that he has several things to say about
a particular topic continues
r firstly
he is marking a contrast between what he said in his preamble and
the matter of the first stage of his exposition. In this case, the infor¬
mation carried by key may be said to be redundant, since ‘firstly’
itself serves as an indicator of the structure of the discourse, but
later, he may say
r when she’d finished Adam Bede
and co-selection of high key would then be the only signal to his
audience the sentence was to be heard as in some sense initiating a
new stage in the proceedings.
The cross-utterance meanings associated with key choices so far
have all been anaphoric, however, there are significant cataphoric
meanings for utterance final tone groups. As we noted above Brazil
postulates a unit above the tone group consisting of one mid key
plus an indefinite number of high and low keys:
130 AN INTRODUCTION TO DISCOURSE ANALYSIS

(High 1... n) Mid (Low 1... n)

The significance of this structure lies in the fact that any utterance
ending in high key is part of an incomplete structure and thus con¬
strains the hearer to respond in some way, whereas an utterance
ending in low key is part of a potentially complete structure and
there is an implication of finality which has to be countered if the
hearer is to continue. Mid key sets up no particular expectations,
though there may, of course, be constraints referable to other
causes, for instance co-occurrence of interrogative mood.
These very broad generalisations can be illustrated by reference
to:

high mid low

A: p where is he now
B: p in bed

Here, the high-key response articulates the presumption that A will

respond to what B is presenting as a surprising/scandalising/thought
provoking or otherwise comment-worthy piece of news. If, however,
speaker B continues with something like:

B: p in bed
r if you can credit it

he still exploits the meaning-potential of high key, but then removes

the compulsion to respond. It is the fact of ending the utterance in
high key that adds a meaning which, in this example, could be
glossed as ‘what do you think of that?’. The comparable low-key
response:

A: p where is he now
B: p in bed

discourages further reaction, because of its suggestion that there is

nothing more to be said. If high key marks the response as ‘likely
to be surprising’, low key marks it as ‘only to be expected’. By
choice of the equative option, B implies that A really knows—or
could guess—the answer to his question. Seen in the context of the
speakers’ mutual understanding, and from the point of view of B,
no new matter is presented. Any reaction on A’s part would amount
to a rejection of B's assessment of the situation. Such reactions as
are possible therefore sound challenging or otherwise aggressive.
 INTONATION 131

high mid low

A: p where is he now
B: p in bed
A: p is he really
But as with high key, the constraint is associated with the final
choice; a continuation in mid key retains the internal implication—
'as you may have guessed’—but removes the prohibition to respond.
B: p in bed
p he was too sick to do anything else
A: p oh
If A’s reaction to the information he seeks is itself in low key, this
similarly has the effect of inhibiting further response from B:
A p where is he now
B p in bed
A p I see

Tone selection
The prospective significance of tone selection can be considered
first in the light of the commonly reported observation, in teaching
manuals and elsewhere, that substitution of rising tones for falling
tones frequently changes statements into questions. There are two
reasons why such an assertion needs careful scrutiny. Firstly, it
takes the categories ‘statement’ and ‘question’ for granted. It
ignores the complex relationship existing between discourse cate¬
gories like elicitation and informative on the one hand, and the
lexico-grammatical entities that realise them on the other hand.
Secondly, the observation is usually made with respect to utterances
which are co-terminous with a single tone group. I shall try to
relate it to what has been said about referring and proclaiming
tones, assuming that it is the referring (‘rising’) tones that need to be
explained, since the imputation of an interrogative function implies
a constraint upon the next speaker to respond.
(a) A: pi think we can manage it
r provided Tom will help
(b) A: pi think we can manage it
B: r provided Tom will help
The major difference between the two examples above is that in the
first the condition ‘provided Tom will help’ is simply referred to as
something all participants are fully aware of, whereas in the second.
132 AN INTRODUCTION TO DISCOURSE ANALYSIS

B’s utterance is heard as a request for confirmation that it is being

taken for granted. B’s contribution expects a response.
The general point seems to be that if an utterance has only
referring tones it serves to present matter as presumed common
ground and the next speaker is constrained to confirm that it does,
in fact, have that status. So

high mid low

r you’re taking raincoats

is equivalent to ‘am I right in thinking you’re taking raincoats’.

With the fall-rise this is no more than a neutral request for confir¬
mation. With the rise,

r+ you’re taking raincoats

there is probably a measure of incredulity, arising from the greater

involvement: ‘I really am right in thinking ...’.
There is an exception to the generalisation in the case of low key
tone groups. In

A: pi think we can man age it

B: r provided Tom will help

the condition is referred to as agreed, rather than as presumed

common ground, and so expects no response. Indeed, it has the
usual effect of low key in implying finality and so inhibiting re¬
sponse.

Intonation and exchange structure

We are now in a position to take note of a third factor which contri¬
butes to the significance of a particular tone and key choice—the
place occupied by the utterance in exchange structure. Brazil sug¬
gests that

(a) prospectively, significant choices contribute to the definition

of relationships between elements of exchange structure;
(b) retrospectively, significant choices help to define relation¬
ships between exchanges or between higher units.
In the following three-part exchange we shall focus on the exponent
of each move in turn:

A: Where does Tom live B: In the Avenue A: Yes

 INTONATION 133

Where does Tom live

In its retrospective aspect, high key would be structurally contras¬
tive though the actual extent of what is contrasted must be deduced
from other evidence. Following a lengthy discussion of affairs as
they concern ‘Harry’, the utterance may mark a major boundary
between large units of a discourse; minimally, it may separate one
exchange from a preceding, and incomplete one. In other words key
can indicate that the speaker is aware he is beginning an ‘insertion
sequence’.
high mid low
A: r can you think of anyone who could help
B: p where does Tom live
An elicitation in low key requests a ‘narrowing’ of information
previously supplied:
A: p I think everyone’s moved
r since you were here
B; p where does Tom live

In both cases, the effect of the key choice is to indicate a special

relationship between the current exchange and what has gone before
(though the effect may have to be explained with reference to the
function of the exchange as a component of a higher unit). The
neutral, mid-key choice signifies no special relationship: the elicita¬
tion is then freestanding with respect to those anaphoric functions
that key realises.
When the whole move consists of a single tone group, that key
choice which is initial is also final and will therefore have prospec¬
tive significance within the exchange. If high key is chosen, then
the observance of the sequence (Hi ... n) M (Lj ... n) will prevent
a low key response, but this will be the only constraint attributable
to key, since the utterance is marked by its grammar as expecting a
response thus the tendency of high key to coerce and of low key to
inhibit the hearer from responding will not operate.

In the avenue
The only effect of a high key choice here would be to determine the
nature of the constraints upon the next speaker: a high key response
is presented as comment-worthy news which expects an explicit
reaction within the exchange, and will also, by the operation of the
sequencing rule, preclude the possibility of a low key reaction. The
134 AN INTRODUCTION TO DISCOURSE ANALYSIS

finality associated with low key would tend to terminate the ex¬
change, so that the third-part 'yes’ we have included in our example
might not, in fact, occur.

Yes
The exponent of the third element in our specimen exchange
represents the first speaker’s reaction to the information he has
elicited. The key choice associated with it has significance as an
indicator of his stance relative to that information. We may perhaps
take it as ‘normal’ that he asks a question because he does not
know the answer. Having received the information, he is therefore
in no position to judge its accuracy. A predictable reaction in these
circumstances would be a low key ‘yes’, nothing more than a
minimal acknowledgement that the requirements of the elicitation
have been met.
The first speaker may, however, have views of his own about the
information he elicits. In this case, a mid-key ‘yes’, meaning ‘yes,
that’s what I thought too’ signifies his willingness to go along with
B’s response, while a third possibility is that, in his capacity as
perhaps school teacher or quiz master, he already possesses the
information; the purpose of the elicitation has been to find out
whether the second speaker has it as well. In this case, the ‘yes’
will be informative, and will be in high key. The possibilities can be
illustrated by the following imaginary classroom exchanges, the
classroom being the place where the full range is most likely to
occur:

high mid low

teacher: p how old are you
pupil: p eleven
teacher: p yes (‘That’s what
I wanted to know’)
teacher: p what do you make it
pupil: p eleven
teacher: p yes (‘That’s what I make it, too’)

teacher: p how many are there in a cricket team

pupil: p eleven
teacher: p yes (‘That’s the correct answer’)

We have not yet exhausted the interest of the third move in the
type of exchange structure we have been examining. This is, in fact.
 INTONATION 135

the structural position at which attention to key promises to be most

revealing.
Two other major options at third place are to produce a repetition
or a reformulation of the previous speaker’s response and again the
speaker has the full range of key options. A low key repetition is
simply a verbal acknowledgement that the addressee has heard the
response, while by selecting low key for a reformulation he indicates
that, within the frame of reference the parties share at the moment
of utterance, his reaction proposes no new matter.

high mid low

A: p where does Tom live
B: pin the A venue
A: jp in the A venue
j p on the other side of town

What can count as a reformulation depends upon socially-

constructed equivalence: in the exchange above, and in appropriate
circumstances, it might for instance include ‘too far away to be of
any use’, ‘he’s made it up with his family’ and ‘so he really did go
bankrupt’. The same range of repetitions and reformulations can
co-occur with mid-key. In such a case, however, the restrictive
implication is absent. The reaction is heard as a simple restatement
of the response, either in its own formal terms or in different terms.
If it is the former, then its function appears to be to do no more
than ‘register’ the new matter:

A: p where does Tom live

B: pin the Avenue
A: p in the Avenue

But as a mid-key contribution, it leaves the hearer with the option

of continuing the exchange, so that, for instance, he may either
react as recipient of the speaker’s reiteration of the matter with
mid-key ‘yes’; or—as informant—confirm that it corresponds with
his previous utterance, with a high key ‘yes’. In neither event is
very much achieved: it seems appropriate to think of such sequences
as a kind of conversational marking time, during which existing
convergence is reinforced—or perhaps time is bought—before new
matter is presented for negotiation. In recordings of doctor/
patient interviews, an audible mid key ‘registration’ of a response
seems often to accompany the making of a written record.
Mid-key reformulations are more likely to elicit a fourth-part
136 AN INTRODUCTION TO DISCOURSE ANALYSIS

reaction, and for very good reason: by contrast with the low-key
reformulation, the hearer has the opportunity to indicate whether
the presumed equivalence is valid or not, and by contrast with the
repetition new and potentially contentious matter is introduced.
High key repetitions and reformulations can occur after any
element in exchange structure, not simply third place items, and all
are effectively new elicitations. These observations allow us to
suggest a linguistic rather than a contextual explanation for Labov’s
‘requests for confirmation’. In the first example below the patient
doesn’t respond to the doctor’s repetition, in the second he does:

patient: well I had’m er a week last Wednesday

doctor: a week last Wednesday
how many attacks have you had

patient: I felt a tight pain in the middle of my chest

doctor: tight pain
patient: you know like a dull ache .. .

The crucial difference is in the intonation. Neither is marked by a

rising pitch contour but the first is in low key and marked as an
acknowledgement, the second is high key and marked as an elicita¬
tion.
All the discussion of exchange structure so far has varied key but
held tone constant, using only proclaiming examples. For the sake
of completeness I will briefly discuss the effect of substituting
referring tone.

7. Where does Toni live. Referring tone here would usually mark
the elicitation as an ‘echo question’: ‘Am I right in thinking you
asked “Where does Tom live?” ’. It thus has retrospective signifi¬
cance, in that it relates the exchange it initiates with a previous one.
Co-occurring high or low key may have the usual contrastive or
restrictive implications: with unmarked tonicity, high key means
‘Are you enquiring where Tom lives?’; with low key the meaning
is something like ‘Do you only want to know where he lives?’.

2. In the Avenue. With referring tone would probably be heard as

a suggestion, carrying the implication that the first speaker was able
to assess its accuracy in spite of his having asked for the information.
High key would indicate expectation of a reaction, mid key would
permit one. Probably with a positive response, low key would be
impossible; with negative response, low key implies that the first
 INTONATION 137

speaker is being told something he already knew: / /r not in the

A venue / /. Such a response is exchange terminating.

3. Yes. Recipient ‘yes’ implies that the speaker knew that much
already: the matter of the response was already common ground.
It is likely to be followed by a new elicitation which makes clear
that it was not really the information he wanted: for instance ‘What
number in the Avenue?’. Informant ‘yes’ occurring in the class¬
room or in the course of a guessing game similarly indicates that
the proffered information is certainly correct, but it is not quite
what was wanted. If the speaker produces repetitions or reformu¬
lations with referring tone these are heard as new elicitations,
regardless of the key choice.
It has not been possible to do full justice to Brazil’s description of
intonation in this chapter; but there is enough to convince readers
that they ignore intonation at their peril.
 7

Discourse analysis and

language teaching

Like other branches of linguistics language teaching has, until

recently, concentrated on grammatical rather than communicative
competence. Wilkins (1972b) observes that although there have been
major changes in the methodology of language teaching over the
years the underlying principle has remained the same:

it has been assumed that units of learning should be defined in

grammatical terms, although the precise sequence in which they
occurred would be influenced by pedagogic considerations.

This is something of an exaggeration: on the one hand many courses

have had lexis rather than or as well as syntax as an organising
principle; on the other early work by Palmer (1934) and Abbs et al
(1969) shows an intuitive appreciation of the importance of speech
acts and communicative function.
However, as a generalisation Wilkins’ observation has a great
deal of truth; as he points out even courses which encourage dia¬
logue and improvised drama are still structured grammatically and
the 'situations that are created are pedagogic, bearing little resem¬
blance to natural language use’. One can see the conflict of
grammatical and communicative criteria clearly in question-answer
sequences. In order to practice particular grammatical structures
students may be drilled to produce responses to questions which are
grammatically correct but unusual or even deviant in terms of
discourse rules:

Q: What is this?
A: This) . , .
| is a book.

Q: Where is the typewriter?

A: The typewriter is in the cupboard.
 DISCOURSE ANALYSIS AND LANGUAGE TEACHING 139

Rarely, if ever, are students taught the much more frequent mood¬
less responses: ‘a book’, ‘in the cupboard’.
We saw in Chapter 3 how norms of interaction differ from com¬
munity to community and obviously these must be acquired as
well as the language. It has been suggested, only semi-humorously,
that one result of introducing foreign students to the present con¬
tinuous in English with appropriate actions, ‘I am opening the
window’, ‘I am cleaning the board’, is to create the impression that
one of the quirks of the English is to keep commenting on what
they are doing.
The alternative to a grammatically structured syllabus is one
which is structured communicatively, where the students learn to
produce communicative acts in a relevant sequence and acquire at
any one time only those aspects of grammar necessary for the
realisation of a particular act. Johns (ms) points out that this raises
the question of structuralisation: how can the student be assisted to
relate a particular structure ‘to the overall framework of the
language’. In other words instead of being presented with a coherent
grammar of the language and having to construct for himself
realisation rules for particular functions, the student may be given
little more than a series of guidebook phrases for greeting, apologis¬
ing or complaining and have to construct his own grammar of the
language.
A major problem for anyone wanting to write a language course
based on communicative or functional principles is that whereas
there are many detailed grammatical descriptions of English there
is currently, as we have seen in earlier chapters, no adequate
functional or discourse description, and there is unlikely to be one
for several years to come.
Wilkins (1972a) recognises this and proposes an interim ‘ad hoc’
framework which he suggests can be refined and modified in the
light of future research. One added advantage of a communicative
or notional syllabus is that it need not be language specific, but can
be designed for a series of culturally related speech communities,
that is communities with a large degree of overlap in their norms of
interaction. Wilkins, assuming that Western Europe is a fairly
homogeneous speech area, asks

what are the notions that the European learner will expect to
be able to express through the target language,

and sets out to describe them in some detail.

140 AN INTRODUCTION TO DISCOURSE ANALYSIS

One set of notions, the semantico-grammatical ones, are un¬

remarkable and are probably covered in some form by most existing
courses—notions of ‘time’, ‘space’, ‘quantity’, ‘matter’, ‘deixis’
and, following Fillmore (1968), ‘case’. His categories of communi¬
cative function are, however, a new departure.
He observes that ‘language learning has concentrated ... on the
use of language to report and describe’ but claims that these two
functions ‘are by no means the only ones that are important for the
learner of a foreign language’. He therefore sets out to classify the
functions utterances can perform and also to suggest, for English,
the realisations which should be taught first. The kinds of function
he has in mind are ‘judgement’, ‘approval’, ‘disapproval’, ‘suasion’,
‘prediction’, ‘greeting’, ‘sympathy’,‘gratitude’, ‘flattery’,‘hostility’,
‘information asserted’, ‘information sought’, and so on. Each of
these functions is glossed in broad terms and followed on the left-
hand side of the page by ‘a list of vocabulary items falling within
or closely related to its semantic field’; while on the right-hand side
of the page he offers suggested grammatical realisations for students
at an early stage of learning. Thus,

Information sought: ‘Question’;

—request, question, ask. Information seeking is likely to be
an important aspect of a learner’s
language use.
(a) Interrogatives
(b) Declaratives + question
intonation
(c) Question-word questions
When
Where Iar
.... How + J much
What
Who
What (time)

‘Request’:
Would you shut the window, please
(Would you mind shutting...)
(1972a)

There are many criticisms one can make of this framework as

Wilkins readily admits: some of the categories overlap; some of the
realisations sound stilted or odd; it may not be an accident that
 DISCOURSE ANALYSIS AND LANGUAGE TEACHING 141

there are no suggested realisations for the notions ‘sympathy’ and

‘flattery’; and there is no indication of the size of the communicative
acts, whether they are clauses, utterances, exchanges or transactions.
Widdowson (1975) asserts that the only test a pedagogic grammar
must pass is whether it works and thus theoretical criticism is
irrelevant, but Wilkins has not yet produced a design for a com¬
municatively structured syllabus, let alone evidence that it works or
is even workable.
Jakobovitz & Gordon (1974) working on similar lines do offer at
least a sketch of the first ten lessons of an intermediate course—
interestingly half the course is concerned with ‘describing’ and
‘reporting’, one of the major faults of non-communicative courses
according to Wilkins.

Lesson 1 Greeting and Leave-taking

Lesson 2 Making Requests: Part 1
Lesson 3 Making Requests: Part 2
Lesson 4 Extending Invitations
Lesson 5 Making Apologies
Lesson 6 Describing Events: Part 1
Lesson 7 Describing Events: Part 2
Lesson 8 Reporting Events: Part 1
Lesson 9 Reporting Events: Part 2
Lesson 10 Reporting Events: Part 3

The major questions one must ask of any such syllabus of course
is how the students learn the formal aspects of the language, which
of the possible grammatical realisations of a given function are
taught and why these and not others. For Lesson 2 Jakobovitz &
Gordon suggest three major categories of request:

A. Asking Informational Questions

A1—that take yes/no answers;
A2—Other
B. Requesting Agreement
B1—for personal opinion or feeling;
B2—for proposed action
C. Asking for permission

and for one of the categories, ‘requesting agreement’, they offer the
following grammatical structures, two of which sound odd to the
British ear:
142 AN INTRODUCTION TO DISCOURSE ANALYSIS

Bl. For personal opinion or feeling:

(i) S + don’t you think so?
(It’s a beautiful day, don’t you think so?)
(ii) S + isn’t it?
(It’s a beautiful day, isn’t it?)
(iii) S + wouldn’t you agree?
(We’re much better off here, wouldn’t you agree?)

These are of course ‘merely suggestions’ and thus difficult to

criticise, but again one wonders how the student is assisted to build
up his grammar of the language from the instances he is offered.
Jakobovitz & Gordon do not suggest that there should be explicit
supportive grammar work but are very much aware of the need to
re-use structures presented in earlier lessons later in the course.
Abbs et al (1975), one of the few published courses to be func¬
tionally rather than grammatically structured, claim that

the learner is taught strategies for handling particular language

functions such as identifying people and places, expressing
personal tastes, emotions, moods and opinions, giving informa¬
tion, making suggestions, giving advice and so on. The structural
contents have been selected as being appropriate to the particu¬
lar function, rather than as an unrelated series of structures
arranged in order of supposed linguistic difficulty.

It is necessary at this point to consider the phrase ‘functionally

structured’—the implication is that functions should be taught in a
certain order either because some are more ‘important’ than others
for social or professional reasons, or because certain functions can
be usefully grouped together. Neither of these criteria appear to
apply to the Abbs’ materials and one suspects that in fact the
organisation is covertly structural. The first few functions are intro¬
duced in the following order:

Identification
Invitations
Likes and dislikes (1)
Description: People
Description: Places
Impatience
Not knowing
The Past (1)
Surprise and disbelief
 DISCOURSE ANALYSIS AND LANGUAGE TEACHING 143

There are two major differences between the notional and

functional categories suggested by Wilkins and Abbs and those
used by the discourse analysts whose work has been discussed in
the earlier chapters: firstly they are truly ‘notional’ in that there is
no indication of where they are derived from nor any suggestion
that there are reliable realisation rules; secondly the authors are
concerned with single units, usually utterances and have no overt
concept of discourse structure.
Jupp and Hodlin (1975), in their book. Industrial English, exploit
an intuitive concept of exchange structure but the only materials to
explicitly use the insights of discourse analysis are Johns (ms) and
Candlin et al (ms).
Johns (ms) presents selected materials from two courses, one for
foreign teachers and the other for intending seminar participants.
The first set of materials, based on the analysis of classroom
language proposed by Sinclair et al (1972) and outlined in Chap¬
ter V, is intended for ‘non-native teachers, and trainee teachers
required to teach English by Direct Method, or another subject
through the medium of English’. The course comprises eight units
with each unit consisting of:

(a) an explicit presentation of a simplified version of the

descriptive model;

(b) intensive listening to extracts from recordings of authentic

lessons, with as wide a range as possible of teachers/
teaching styles/subjects/ages of pupils, for analysis;

(c) language laboratory simulation of the analysed features of

classroom interaction, through imitation and drill-like
exercises.

By the time the student-teachers reach unit 4 they have covered the
three-part exchange structure, and are practising different ways of
responding to pupil utterances and ways of controlling turn-
taking:

Naming
(The following drill sequence comes after the student has listened
to, and analysed, two authentic recordings of teachers getting pupils
to name objects’.)
144 AN INTRODUCTION TO DISCOURSE ANALYSIS

Drill one
I want you to get your pupils to identify the objects shown in the
pictures above

1. // 1 what’s / this 11 (a hacksaw) //1 + yes it’s a // 1

hacksaw //
2. /I 1 what’s / this // (a fountain pen) // 1 + yes it’s a // 1
fountain pen // etc.
Drill two
This time I want you to get your pupils to identify the same objects,
but this time you cue them to put their hands up: every time it is
Arthur who makes the first bid: and so he is the one you nominate.

1. II 3 hands up what’s // 1 this //

II 2 Trthur // (a hacksaw) // 1 -f yes it’s a // 1

hacksaw // etc.
Drill three
Here are some new pictures: this time Arthur does not get the
answer right, so you have to ask the next pupil who bids, whose
name is Brenda: if Brenda gets it wrong, the next is Charlie, and
after Charlie, Diana.
 DISCOURSE ANALYSIS AND LANGUAGE TEACHING 145

1. // 3 hands up what’s // 1 this //

// 2 Arthur (a saucepan) II 3 no I/
I/ 2 Brenda (a motorcar) // 1 + yes it’s a // 1
motorcar //

2. // 3 hands up what’s //I this 11

/1 2 Arthur (a spoon) // 3 no //
// 2 Brenda (a hammer) // 3 no //
// 2 Charlie // (a screwdriver) // 1 + jcj it’s a // 1
■screwdriver // etc.

The second set materials, for seminar discussion strategies, are

based on Johns’ own analysis and are aimed particularly at over¬
seas social science postgraduate students, who frequently come from
undergraduate courses taught entirely by lecture to postgraduate
courses structured round participatory seminars. Johns suggests
that in seminars ‘control is up for grabs’ and that, as a consequence,
students are faced with five types of linguistic problem: turn¬
taking; handling discontinuity; linking; mitigation; and repair
work.
There are ten units in this course; the following extracts come
from one on ‘prefaces and suppletion’. By this stage the students
have been introduced to six ways of responding to previous infor-
matives, amplification, contradiction, counter, restriction, explana¬
tion, consequence. Now the student is first given practice in
identifying prefaces in authentic texts and in imitating them on
gapped recordings; next he practises producing suppleted moves
after a fixed preface of the form ‘I’d like to come in here if I
may’; finally he makes his own amplification on hearing a
signal.

Drill one
Preface + Amplification
You will hear an amplification made in a short form. I want you to
make the same amplification in a long form using the ‘not only ..
but’ construction.

1. ‘Mary’s very pretty’ ‘I’D LIKE TO COME IN HERE IF

‘Intelligent too.’ I MAY. NOT ONLY IS MARY
VERY PRETTY, BUT SHE’S IN¬
TELLIGENT, TOO.’
 146 AN INTRODUCTION TO DISCOURSE ANALYSIS

2. ‘The theory needs ‘I’D LIKE TO COME IN HERE IF

to be discussed.’ I MAY. NOT ONLY DOES THE
‘And tested.’ THEORY NEED TO BE DIS¬
CUSSED, BUT IT NEEDS TO BE
TESTED, TOO’. (8 examples)

Drill four
Preface + Amplification
This time you will hear six extracts from the discussion: this time
you must make your own amplification of the speaker’s remark
when you hear the signal.

1. ‘...there are, however, many many more undergraduates

today.’
2. ‘...what it was always possible to do was to determine
what class your children should belong to.’ (6 examples)

Johns observes that in these materials there is a triple progression

from: (i) universal to restricted preface; (ii) idealised to authentic
prompt; (iii) controlled to freer production. The suggested prefaces
are formal if not stilted, but Johns suggests that it is the act of inter¬
rupting which causes the foreign students their problems—once
they have a formula they are prepared to use it ‘in a real discussion
taking place in real time’. As Widdowson observes success is the
only ultimate criterion and these materials are rated enjoyable and
successful by the students for whom they are designed.
Like Johns, Candlin et al (ms) focus on a particular Special
Purpose, this time the language needs of foreign doctors working in
casualty departments. Again like Johns, their materials are based
on a previous detailed functional description of native speaker
interaction, and attempt not simply to teach single functions but to
show the doctors how to open and close interviews, how to parti¬
cipate in other types of exchange, how to build exchanges into
longer sequences, how to manipulate the turn-taking system. The
doctors first learn greeting, then eliciting, then interrogating; at this
point they are able to participate in the following exercise generating
the first few exchanges of a simulated interview:

one learner as patient, other as doctor. Doctor uses a GREET

and an ELICIT, after which the patient describes the nature,
cause, etc. of his complaint. At an appropriate moment, the
doctor inserts an INTERROGATE.
 DISCOURSE ANALYSIS AND LANGUAGE TEACHING 147

Written English
By no means all students learning English need to be able to
speak it—many simply want to read material unavailable in trans¬
lation. However, there is only one major attempt to apply a theory
of discourse to reading materials—English in Focus. This series has
as its target audience the rapidly growing number of students
world-wide with elementary knowledge of English, who need to
improve their ability to read and write in English in order to pursue
their studies in science or technology at tertiary level. The editors,
Allen and Widdowson, observe that there is now an abundant
supply of basic language courses and a growing number of specialist
study skills courses like Candlin et al (1973), but virtually nothing
to assist the learner to make the transition. The books set out to
teach quite a lot of grammar through exercises ‘designed to focus
on points which are particularly important in scientific writing’,
but the innovation is the attempt to teach directly certain important
communicative acts.
The materials are based on a description of written discourse out¬
lined in Widdowson (1973) and developed in Mountford (1975a).
Widdowson observes that in written, as in spoken texts it is impos¬
sible to establish the rhetorical nature of an utterance by reference
to the occurrence of certain linguistic elements; each of the illocu-
tions must be defined in terms of the communicative act it performs.
As an illustration he focuses on ‘the most common or elemental’
type of explanation, in which one event or state of affairs is repre¬
sented as accounting for another, and in which what is to be
accounted for is in some sense known:

John stopped because his brakes had jammed.

Under no circumstances, however, could this count as a scientific

explanation; for the scientist the concern is

not so much with accounting for an event or state of affairs by

reference to another, as with accounting for an event or state of
affairs as a particular instance of a general rule.

Thus scientific explanations consist of two distinct sentences or

locutions, one being an observation or prediction, the other a
generalisation. For the scientist there are two explanatory pro¬
cedures, inductive, when he proceeds from observation to generalisa¬
tion, and deductive when the generalisation leads to a prediction.
In discourse terms the only difference is sequence.
148 AN INTRODUCTION TO DISCOURSE ANALYSIS

If a car’s brakes fail when

(observation) approaching a sharp bend at
speed, the car continues straight
inductive + ahead and crashes.
explanation Every body remains in a state
of uniform motion in a straight
(generalisation) line unless acted upon by forces
from outside.

The labels ‘observation’ and ‘generalisation’ above are bracketed

to emphasise that although in other contexts and isolated from each
other these particular locutions could realise the illocutions
‘observation’ and ‘generalisation’, here they combine to realise the
illocution ‘explanation’. Both Widdowson and Mountford, follow¬
ing the ethnomethodologists, stress this shifting nature of discourse
—readers must continually work at interpreting a text and must
frequently revise their initial interpretation in the light of the occur¬
rence and value of succeeding locutions.
Building on the work by Widdowson, Mountford (1975a) char¬
acterises, more formally, another twelve scientifically relevant ill¬
ocutions — ‘assert’, ‘generalise’, ‘infer’, ‘explicate’, ‘interpret’,
‘define’, ‘exemplify’, ‘illustrate’, ‘describe’, ‘report’, ‘observe’
and ‘predict’ — by what Searle (1965) calls their essential condit¬
ion. Thus

generalise: p reduces all instances of R to a class,

infer: p is a logical consequence or conclusion drawn
from facts and indications.
define: (i) p specifies the characteristics of R by placing R
in a class and distinguishing R from other
members of the class.
(ii) p states the meaning of expression q.

English in Focus attempts to teach students the realisations of the

major illocutions isolated by Widdowson and Mountford. Allen and
Widdowson (1974a) argue that:

the difficulties which students encounter arise not so much from

a defective knowledge of the system of English but from an
unfamiliarity with English use and consequently their needs
[must] be met by a course which . . . develops a knowledge of
how sentences are used in the performance of different com¬
municative acts.
The four major communicative acts which the books set out to
 DISCOURSE ANALYSIS AND LANGUAGE TEACHING 149

teach are classification, definition, generalisation and explanation.

Glendenning (1973) introduces classification first and then in the
next chapter presents the following three-part diagram to demon¬
strate the relationship between the realisations of classification
(a + b), definition (a + b + c) and generalisation (a + c).

Mountford (1975b) introduces students to inductive explanations

with the following diagram which also indicates the relationship of
observations to instructions and results.

The students are first given practice in the conversion of instruc¬

tions and results to observations and then in different ways of link¬
ing observations and generalisations. Finally they learn the
150 AN INTRODUCTION TO DISCOURSE ANALYSIS

relationship, between inductions and deductions, and the fact that

when the sequence is reversed the linking is achieved by ‘thus’,
‘therefore’, ‘hence’ and ‘consequently’.

- Deduction
Generalisation Observation
Wrought iron is If you heat a bar to
malleable Induction bright red and hammer it
to shape, it will spread
without cracking

Like the Johns and Candlin materials the English in Focus series
in based on an explicit theory of discourse. However, although all
the books are in the same series and cater for audiences with
identical linguistic problems, though differing scientific subjects,
there are surprising differences. Some are perhaps trivial—what
Allen and Widdowson call ‘directions’, Glendenning and Mountford
classify as ‘instructions’, and each book introduces the illocutions
in a different sequence. What is perhaps more disturbing is the fact
that there is no agreement on the necessary number of basic illo¬
cutions—what is one to make of the fact that students of workshop
practice and physical science need to be able to understand and
produce ‘inductions’ and ‘deductions’ while students of mechanical
engineering apparently need only ‘inductions’—do they never have
hypotheses which they need to test? Why do physical scientists not
need to be able to make ‘predictions’ when the others do? From
this point of view Maclean (1975) appears to belong to another
series entirely—the only communicative act her medical students
are introduced to is ‘description’; not for them the problems of
‘classification’ ‘this disease is a skin disease’, let alone definition
‘this disease is a skin disease which we can treat with aureo-
mycin’.
Widdowson stresses repeatedly the importance of position in
sequence in determining the illocutionary force of any particular
locution. The relationship between locutions, and therefore the
illocutionary force, may be marked by a sentence connector, but
sometimes this relationship has to be inferred as in the following
example from Allen and Widdowson (1974b), where the third
sentence is in a ‘therefore’ or ‘conclusion’ relationship to the pre¬
ceding two:
 DISCOURSE ANALYSIS AND LANGUAGE TEACHING 151

‘Inorganic acids consist only of hydrogen and an acid radical.

Hydrochloric acid consists of hydrogen and the chloride
radical, and sulphuric acid consists of hydrogen and the
sulphate radical. They are inorganic acids.’

Allen and Widdowson (1974a) argue that whereas a native speaker

can almost always correctly identify illocutionary acts which are
not formally marked, a non-native speaker needs practice:
one way of doing this is to ask him to insert expressions into
the sentences of [a] passage which make explicit what their
illocutionary function is.
In the application of this theoretical principle students are
presented with a passage followed by an exercise requiring them to
insert particular connectives between particular statements. The
following exercise comes from the first chapter of Allen and
Widdowson (1974b):
The Properties of Air
(1) The earth is surrounded by a layer of air. (2) This is between
150 and 200 km thick and is called the atmosphere.
(3) Air is invisible and therefore it cannot be seen. (4) But it
occupies space and has weight in the same way visible sub¬
stances do. (5) This fact is illustrated in Problems A and B.
(6) Air, then takes up space and has weight. (7) The atmo¬
sphere, therefore, weighs down on the surface of the earth.
(8) However, this weight cannot be felt pressing on us because
air not only exerts a downward pressure, but it also exerts
pressure upwards and sideways, and this pressure is balanced
by the equal pressure which our blood exerts in all directions.
(9) In short, air exerts pressure in every direction.
EXERCISE C Relationships between statements
Place the following expressions in the sentences indicated.
Replace and re-order the words in the sentences where neces¬
sary.
EXAMPLE
since (3)
Air is invisible and therefore it cannot be seen. (3)
= Since air is invisible, it cannot be seen.
(a) consequently (3) (d) it follows that (7)
(b) however (4) (e) nevertheless (8)
(c) in short (6) (f) then (9)
152 AN INTRODUCTION TO DISCOURSE ANALYSIS

The justification of this type of exercise is superficially convincing

but there are several worrying features. Firstly, although Allen and
Widdowson argue that the ‘expressions’ make explicit the ‘function
a particular sentence is meant to fulfil’ and also assert that it is
‘crucial for the student... to understand which statements are
meant to be illustrations, qualifications, conclusions and so on’,
there is no attempt to link the two. Students are not introduced to
the illocutions ‘illustration’, ‘qualification’, ‘conclusion’, let alone
told of the relationship between these illocutions and their appro¬
priate expressions—to the student it is presented as, and some
would say never becomes more than, an exercise in text manipu¬
lation.
Secondly, one must question the validity of some of the equiva¬
lences which the students are being taught. In the exercise above
students are led, by items (c) and (0 to assume that ‘in short’ is
equivalent to and replaceable by ‘then’—‘In short, air takes up
space and has weight’. Although one’s intuitions about discourse
are much less reliable than those about grammar, it is difficult to
accept that this could occur as a paragraph-initial illocution.
Thirdly, in later exercises students are required not simply to
replace one sentence linker with another, but to insert an item
where there was nothing before, in order to ‘clarify’ the illocution¬
ary force; at times this produces very odd sounding discourse. In
the following example, quoted earlier, the students have been
instructed to insert ‘for example’ and ‘therefore’;

Inorganic acids consist only of hydrogen and an acid radical.

For example hydrochloric acid consists of hydrogen and the
chloride radical and the sulphuric acid consists of hydrogen
and the sulphate radica. Therefore they are inorganic acids.

One of the oddities of this passage is that ‘for example’ implies that
the two acids are accepted as ‘inorganic’ but then ‘therefore’ implies
that this is a conclusion derived from the two previous state¬
ments.
Fourthly, it is instructive to compare the four text books to dis¬
cover whicn sentence connectors the authors consider important.
Although Winter (1976) offers an exhaustive list and Mountford
suggests a classification into ten sub-groups, there is apparently no
consensus between the five authors of the Focus books. Indeed their
choices appear almost random—out of fifty-three words and phrases
offered to the students in the four books, only four—‘consequently’.
 DISCOURSE ANALYSIS AND LANGUAGE TEACHING 153

‘however’, ‘for example’ and ‘therefore’ occur in every book—and

thirty-one are seen as important by only one of the authors.
One final theoretical point—a major question in language teach¬
ing is how far one should use authentic text on which to base
materials. The Focus books do not. Allen and Widdowson (1974a)
defend this decision on two grounds: firstly, in this way they are
able to ‘avoid syntactic complexity and idiosyncratic features of
style which would be likely to confuse students’; and secondly, by
composing passages it is possible to ‘ foreground features of language
which have particular communicative value’.
There are of course arguments on both sides and obviously the
search for suitable authentic texts can be arduous, but the dis¬
advantages of created text are equally great. Subject specialists have
criticised the scientific content; language teachers have observed
that many passages, like the one above on ‘properties of air’ are
not acceptable English; but an even more serious indictment is that
the passages are surely, in Widdowson’s terms, text, that is exempli¬
fications of the rule system, rather than discourse. In fact the books
are, in some respects, a more subtle version of the earlier, ridiculed
mistake:

Where is the typewriter?

The typewriter is in the cupboard.
 8

The acquisition of discourse

The focus and success of research into child language acquisition is

both conditioned and constrained by the linguistic framework
within which the researcher works. As the major focus of linguistics
has moved from the native speaker-hearer’s grammatical compe¬
tence (Chomsky 1957, 1965) to semantic deep structure (Fillmore,
Lakoff, Ross, McCawley) and now to pragmatics (Gordon & Lakoff
1971), contextualised utterances (Fillmore 1973, Lakoff R, 1972)
and the structure of conversation (Labov 1972), so research into
child language acquisition has changed its focus from surface gram¬
matical structures to underlying semantic roles and now to a con¬
cern with pragmatics and interaction.
The major research in the early 60’s concentrated on the structure
of young children’s utterances at the two-word stage, and attempted
to write rules to account for the regularities—in Widdowson’s (1973)
terms sample utterances were regarded as sentences exemplifying
the child’s grammar. Three independent investigations produced
very similar descriptions in terms of two major word classes, called
pivot and open by Braine (1963), modifier and noun by Brown and
Bellugi (1964), and operators and non-operators by Miller and
Ervin (1964). Braine’s pivot words were few in number and typically
occurred first in the utterance, with a structure P + O; much less
frequent were the structures O + P and O + O, while utterances
consisting of a single P or two P’s do not occur.
P + 0 0 + P 0 + 0
Gregory see boy push it Mommy sleep
my milk move it milk cup
more melon oh-my see
Andrew all broke boot off papa away
I sit mama come pants change
Steven want baby bunny do find bear
get ball want do two checker
 THE ACQUISITION OF DISCOURSE 155

Brown (1973) observes that at the time,

this convergence of three independent and geographically well-
separated studies, on a pivot grammar description for the first
sentences was widely accepted as a major datum of develop¬
mental psycholinguistics.
However, there were problems. For a start the pivot-open class
distinction didn’t appear to make sense linguistically, although it
could be counter-argued that there is no reason why a child’s first
grammar should be structured in ways acceptable or even compre¬
hensible to adults. More disturbingly, if the division into classes was
correct the grammar should be generative, in other words it should
be possible to produce utterances which, had the occasion arisen,
could have been produced by the child. However, the P + O gram¬
mar would generate ‘my hot’ (P + O) or ‘broke come’ (O + P),
utterances of a type which. Brown notes, ‘have been mighty slow in
turning up’.
Thus the first major attempts to produce a text-based grammar
of children’s utterances failed. Chomsky (1957) argued that corpus-
based grammars must fail because it is impossible to derive deep
structure from surface structure and for the same reason suggested
that children must be innately pre-programmed to learn language.
McNeill (1970) took the innateness hypothesis to its extreme and
argued that it is the basic universal syntactic relations which are
innate and that the child’s experience with language simply provides
information about language-specific surface structures. Thus, in
essence, McNeill proposes deriving a young child’s utterances from
an adult deep structure with extra deletions, arguing that although
children are limited to uttering single words at the beginning of
language acquisition, they are capable of conceiving of some¬
thing like full sentences.
Mid-way between McNeill and Braine are Bloom (1970) and
Schlesinger (1971). Bloom instances the use of the same utterance,
‘mommy sock’ in two entirely different contexts, and argues that in
one the underlying structure is possessor/possessed, in the other
subject /object; in other contexts in addition, this same noun/noun
structure can also be attributive, ‘party hat’ or locative, ‘sweater
chair’. In fact. Bloom suggests that at the two-word stage children
are actually able to handle 13 grammatical relations and the under¬
lying structures must account for these, even though on the surface
many of the differences are neutralised. The reason for suggesting
that this particular set of relations exists and is not simply being
156 AN INTRODUCTION TO DISCOURSE ANALYSIS

imposed on the data by the listener is that they begin to emerge in

surface grammatical forms as the child’s utterances increase in
length and surface complexity.
Thus we can see that until very recently research into language
acquisition just like linguistic theory has been obsessed with the
description of grammatical rather than communicative competence.

Acquisition of language function

Dore (1973, 1975) argues that it is pointless to describe the utter¬
ances of young children in purely grammatical terms; the most
important fact about such utterances is that the children are using
them to do things. Dore therefore proposes a description in terms
of speech acts, using the model outlined by Searle (1969). Searle’s
speech acts consist of a propositional act—referring and predicating
—and an illocutionary act—‘stating’, ‘warning’, ‘questioning’, etc.
Dore suggests that the young child’s one word utterances can be
usefully characterised as primitive speech acts, which have no
predication but consist simply of a rudimentary referring expression,
for example ‘doggie’ or ‘bye bye’ and a primitive force, typically
indicated by intonation. The same item or rudimentary referring
expression can occur with different forces and thus realise different
primitive speech acts:
‘mama’ with a falling terminal intonation contour was used
in circumstances where the child merely labelled his mother
or some doll as the mother; ‘mama’ with a rising terminal
contour was used to ask if an object belonged to his mother
. . . and it was used with an abrupt rising-falling contour to call
his mother.
Dore suggests that all the one-word utterances of the two children
he studied were realisations of one of nine primitive speech acts:
Primitive speech act Description of example

Labelling M touches a doll’s eyes, utters /aiz/, then

touches its nose, utters /nouz/; she does
not address her mother and her mother
does not respond.
Repeating M, while playing with a puzzle, overhears
her mother’s utterance of doctor (in a con¬
versation with the teacher) and M utters
/data/; mother responds Yes, that's right
honey, doctor, then continues her conver¬
sation; M resumes her play with the
puzzle.
 THE ACQUISITION OF DISCOURSE 157

Primitive speech act_Description of example

Answering Mother points to picture of a dog and asks
J What's this?', J responds /bauwau/.
Requesting J tries to push a peg through a hole and
(action) when he cannot succeed he looks up at his
mother, keeping his finger on the peg, and
utters /a?a?a?/ (with constant contours
and minimal pause between syllables); his
mother then helps him push the peg, say¬
ing Okay.
Requesting M picks up a book, looks at her mother and
(answer) utters /bukf/ (where arrow indicates a
rising terminal contour); mother responds
Right, it's a book.
Calling J, whose mother is across the room, shouts
/mama/ loudly (where ~ indicates an
abrupt rising-falling contour); his mother
turns to him and says I'm getting a cup of
coffee. I'll be right there.
Greeting J utters /hai/ when teacher enters room;
teacher responds Hello.
Protesting J, when his mother attempts to put on his
shoe, utters an extended scream of varying
contours, while resisting her; M, in the
same circumstance, utters No.
Practising M utters daddy when he is not present;
mother often does not respond.
The most noticeable absence from this list is any type of state¬
ment or assertion. Halliday (1975) also notes the absence of such
speech acts and suggests that in fact this is not surprising, because
the idea that one can use language to convey information not
known to one’s audience is a sophisticated one. In fact stating is the
‘only use of language in a function that is definable solely by
reference to language’.
The first questions one asks of any descriptive system are ‘why
these categories and no others?’ and ‘how can one recognise items
as exponents of particular categories?’. The most contentious cate¬
gory is that of repeating. Keenan (1974) argues that researchers are
far too ready to classify a child’s utterance simply as an imitation
or a repetition, when the same phenomenon in an adult conversa¬
tion, where it occurs frequently, would be felt to have some extra
significance. One would not be happy with a description which
simply characterised the doctor’s second utterance, in the following
example, as a repetition:
doctor: and when did you get these
patient: a week last Wednesday
doctor: a week last Wednesday
158 AN INTRODUCTION TO DISCOURSE ANALYSIS

As we saw in Chapter VI, Brazil argues that depending on the key

choice, the doctor will be heard as either requesting more infor¬
mation or indicating that he understands and acknowledges the
information given. Keenan suggests that one should similarly see a
child’s repetition as having some function in discourse, and thus, in
the example Dore offers of repeating ‘doctor’, the child’s action
may more usefully be described as performing the primitive speech
act of practising or requesting answer.
Four of the acts which Dore proposes are initiating—requesting
action, requesting answer, calling and greeting-, two are responding
—answering and protesting-, and two are non-interactive—labelling
and practising. Although Dore claims that primitive force is
typically indicated by intonation, in fact this is not true—both the
responding acts are marked by position in sequence, practising acts
by the absence of relevant referents in the situation and greeting
acts are recognisable because there is only a closed set of exponents.
Only three of the acts are in fact distinguished solely by intonation,
labelling, requesting answer and calling, significantly the three acts
exemplified in the ‘mama’ example above.
The basic assumption behind the label ‘primitive speech act’ is
that these uses of language eventually develop into adult-type
speech acts. Dore suggests three developmental stages. (See facing
page.)
Unfortunately, though he claims that his observations of chil¬
dren’s early two-word utterances suggest that the characterisation
of stage II is ‘an adequate representation of the child’s ability’.
Dore provides no examples and no indication of how the realisation
of elementary illocutionary force differs from that of primitive
force. In stage III he places the occurrence of grammaticalisation,
the process by which the illocutionary force operates to select
features of the modality component; thus he suggests that the
illocutionary forces request and demand select respectively the
moods interrogative and imperative, but as we have seen in earlier
chapters this is an oversimplification. However, despite some in¬
adequacies, this approach to the description of language acquisition
is a notable and promising departure.
Halliday (1974, 1975) reports a detailed study of the language
development of one child, Nigel, from the age of 9 months, before
which he had no recognisable communication system, to the age of
18 months when he began to abandon the idiosyncratic system he
had created for himself and learn the adult language. Halliday
 THE ACQUISITION OF DISCOURSE 159

Stage I Stage II

Primitive Speech Act Speech Act

/
primitive
\
rudimentary
/
elementary
\
rudimentary
force referring illocutionary proposition
expression force

predicating referring
expression expressions

Stage III

Speech Act
/
illocutionary sentence
force

modality proposition

verb cases
provides a description of the developing system at six weekly inter¬
vals and, because at this age the corpus is very small, with only
12 distinct utterances or meanings at 10*4 months and 52 at 1614
months, every meaning can be shown as an exponent of one or other
of the descriptive categories. Halliday observes that at this early
stage a child’s communication system is, like all other animal
systems, bi-stratal—‘it has a semantics and a phonology, but noth¬
ing in between’. He suggests that the only way to interpret such a
system is in functional categories. Halliday (1974) brings to the
data
some notion of the developmental^ significant functions that,
on general socio-cultural grounds (as well as from our know¬
ledge of the nature of the adult language), we should expect to
determine the content structure of the child’s proto-language.
He proposes six functions:
(i) instrumental—this is the ‘I want’ function of language;
and it is likely to include a general expression of desire,
some element meaning simply ‘I want that object there
(present in the context)’, as well as perhaps other expres¬
sions relating to specific desires.
160 AN INTRODUCTION TO DISCOURSE ANALYSIS

(ii) regulatory—the regulatory is the ‘do as I tell you’ function

... in the instrumental the focus is on the goods or services
required and it does not matter who provides them,
whereas regulatory utterances are directed towards a
particular individual.
(iii) interactional—the ‘me and you’ function . . . this is
language used by the child to interact with those around
him . . . and it includes meanings such as generalised
greetings, ‘Hello’, ‘pleased to see you’ and also responses
to calls, ‘Yes’.
(iv) personal—this is language used to express the child’s own
uniqueness . . . this includes, therefore, expressions of
personal feelings, of participation and withdrawal, of
interest, pleasure, disgust.
(v) heuristic—the ‘tell me why’ function, that which later on
develops into the whole range of questioning forms that
the young child uses. At this very early stage, in its most
elementary form, the heuristic use of language is the
demand for a name.
(vi) imaginative—the function of language whereby the child
creates an environment of his own ... a world initially of
pure sound, but which gradually turns into one of story
and make-believe.
(Halliday, 1975)

It is instructive to compare the functions which Halliday brings to

the data with the primitive speech acts which Dore derives from it:

Halliday Dore

Instrumental
<j^ Requesting (action)
Regulatory
Interactional Calling. Greeting, Answering
Personal Protesting
Heuristic Requesting (answer); ? Labelling
Imaginative

The only one of Dore’s primitive speech acts not to find a parallel
function is practising, and this is because Halliday (1974) ex¬
cluded it ‘on the grounds that the learning of a system is not a
function of the system’. Similarly there is only one of Halliday’s
functions for which Dore does not offer a parallel speech act—■
 THE ACQUISITION OF DISCOURSE 161

imaginative and, as the realisations are such things as ‘peep-o’,

‘ cockodoodledoo ’ and lion noises, Dore may have simply ignored
them.
At about 17 months there was a sudden change in Nigel’s
language, which Halliday sees as the beginning of the transition to
the adult system—he began to use intonation systematically to
make a binary distinction between two classes of utterances prag¬
matic and mathetic—‘from this point on, for six months or more,
he spoke all pragmatic utterances on a rising tone and all others on
a falling tone’.
Nigel’s pragmatic class subsumed the instrumental and regulatory
functions and now corresponds directly with Dore’s primitive speech
act requesting {action), while the mathetic arises ‘primarily from a
combination ... of the personal and the heuristic’, and is used for
making representation of what the child observes. Thus:

mathetic Daddy got scrambled egg . . .

pragmatic Mummy get foryou scrambled egg

where the message is ‘Daddy’s got some; now get some for me’.
Halliday suggests that it is now a short step to complete utterances
in which one part is presented as a condition on the other,

when newworld finish song about bus

Halliday’s description of adult language comprises three major

functional components, the ideational, interpersonal and textual.
The ideational options relate to the content of what is said and
through them the speaker expresses his experience, thoughts and
knowledge; the interpersonal options indicate the speaker’s role
vis-a-vis the other participants and allow him to express judgments
and attitudes; the textual options ensure a cohesive and coherent
text. Just as Dore suggests how the primitive speech acts develop
into full speech acts, so Halliday suggests how the child’s functions
develop into the more abstract ones of adult language. He observes
that

it seems likely that the ideational component of meaning arises,

in general, from the use of language to learn, while the inter¬
personal arises from the use of language to act.
Thus the mathetic/pragmatic distinction which Nigel chose to
indicate by intonational contrasts is seen as the bridge between the
child and adult systems. One must not, however, equate the
162 AN INTRODUCTION TO DISCOURSE ANALYSIS

mathetic/pragmatic distinction with the ideational/interpersonal,

because at this stage Nigel’s utterances are still unifunctional, either
mathetic or pragmatic, while adult utterances are, crucially, multi¬
functional—every utterance has ideational and interpersonal and
textual features.

Acquisition of productive and interpretive rules

A major difficulty for children is the lack of fit between form and
function; the fact that any given speech act can be realised by
several grammatical structures and that any given grammatical
structure can realise several speech acts. Ervin-Tripp (in press)
suggests that there are five major types of directive:

(i) interrogatives ‘Gotta match’

(li) imperatives ‘Gimme a match’
(iii) imbedded imperatives ‘Could you gimme a
match’
(iv) statements of need or desire ‘I need a match’, ‘I’m
cold’
(v) statements of external condition ‘There aren’t any
matches here’

Drawing on a wide range of tape-recorded data she predicts that the

comprehension of directives
progresses from imperatives ... to imbedded imperatives ...
then to need statements where the fulfilling act must be sup¬
plied. Very late would be interrogative requests and general
statement-requests because of the knowledge required of impli¬
cations.
In other words she suggests that young children will decode the
grammatically and lexically more explicit directives more success¬
fully.
Shatz (1974), however, provides interesting evidence that at an
early age children are equally happy with explicit and implicit
forms of directives. Her data consists of videotapes of three two-
year-olds interacting with their mothers in their own homes and
playing with a Targe pre-school toy’ provided by the experimenter.
Shatz focused on utterances by the mothers which were directive
in intent and interrogative or imperative in form—‘suggestions
implicit or explicit that the child perform an act different from the
one he was performing at the time of the utterance’. The results are
highly significant. Although Ervin-Tripp predicts that the explicit
 THE ACQUISITION OF DISCOURSE 163

imperative directives would be the most successful, in fact the chil¬

dren only responded appropriately to them in 40% of cases while the
less explicit interrogative directives were successful in 52% of cases.
One might assume that the interrogative directives were so success¬
ful because they were what Ervin-Tripp calls imbedded imperatives,
utterances of the form ‘can you shut the door’, but in fact 40% were
pure interrogatives like ‘are there any more suitcases’, where the
requested action, ‘find another suitcase’ is not stated in the utter¬
ance.
We can thus see that at this early age children are not only per¬
fectly aware that a given functional intention can be transmitted by
more than one grammatical structure, but that they can, in ambigu¬
ous cases, choose the correct alternative with a high degree of
accuracy—of the interrogative directives to which the children
responded, only 15% were mistakenly treated as questions, the
other 85% were followed by the appropriate action.
Obviously the children were using something other than formal
grammatical clues to decode their mothers’ intention. Shatz
wondered whether the mothers were providing non-verbal indica¬
tions of the directive intent of their interrogatives, but in fact all the
mothers tended to use more gestures with imperatives than with
interrogatives. There must, therefore, be another explanation. Shatz
proposes a basic ground rule for this type of interaction—‘Mama
says; child does’. The child’s strategy is

find some element, either action or object, in Mama’s speech

which you can act out or act upon. Then perform some action
which communicates to Mama that you have heard, are
responding, are taking your turn in the interaction.

The obvious prediction from such a theory, and one on which she
is now working, is that a child will respond with an action ‘even to
ill-formed directives like “may you shut the door”, and misinterpret
questions like “can you jump” as directives even when they are
intended informationally’.
For the critical observer there are three deficiencies in the data
presented by Shatz: firstly, there is no indication whether interro¬
gatives intended and heard as questions were intonationally (or
gesturally) different from either those intended as directives but
heard as questions or those intended and heard as directives;
secondly, the method followed by Shatz and others deduces a
child’s interpretation of a given utterance from his response to it—
164 AN INTRODUCTION TO DISCOURSE ANALYSIS

yet Reeder (1975) reports that some children while understanding

certain interrogatives as questions rather than directives actually
acted on the proposition as if they were directives; thirdly, the
ground rule ‘Mama says; child does’ is potentially too powerful
because it completely ignores context—for instance it is hard to
imagine that utterances of the form ‘are there any more suitcases’
would invariably be interpreted as directive.
Reeder (ibid) reports an experiment with slightly older children,
aged 38-46 months which avoids these criticisms. He sets out to
provide the children with a series of utterances, all of the form
‘would you like to do P’ and all with rising intonation, which could
only be correctly interpreted as question or directive by reference to
the context and cotext. The utterances were contextualised for the
children in an ongoing interaction in a toy playschool between a
model teacher and model pupils. Thus an utterance intended as a
question would be contextualised by

Peter, in you come and help me feed the animals . . . you

choose which one you want to feed.

Peter then approaches and leaves four pets in quick succession,

turns to the teacher who then asks

"rabbit
would you like to tadpoles
feed the tortoise
mice

For the directive, the context would be ‘Simon, leave the swing,
Peter’s on it’ followed by

M1UC

seesaw
would you like to play on the
roundabout
climbing frame

All the ‘teacher’s’ utterances were pre-recorded and at the

appropriate time the tape recorder would be switched on, so that all
the children heard exactly the same test item. The novelty of this
experiment is that the children are not direct participants but are
being asked to interpret utterances not addressed to them at all.
The children were not asked to remember, repeat or paraphrase
what the teacher said, but to choose between two alternative para¬
phrases also pre-recorded on tape loops and to which they could
 THE ACQUISITION OF DISCOURSE 165

listen as often as they wanted. Thus the children heard utterances

in the form (a) and were asked to decide whether the teacher said
(b) or (c); (on one occasion a child said that the teacher didn’t say
either and then reproduced form (a).)

(a) would you like to do A tone 2 (rising intonation)

(b) do you want to do A tone 1 + (high fall)
(c) I want you to do A tone 1 (mid fall)

Each child heard eight utterances intended as directives and eight

intended as questions and in every case the sole disambiguating
features was contextual information—that is, in the light of what
had just been said and done, the children were asked to decide
whether the teacher was likely to be asking or telling the pupils. The
children were successful 70% of the time and one child interpreted
the utterances ‘correctly’ every time. These results reinforce Shatz
in demonstrating that children do not have great difficulty with
indirect formal realisations of particular functions and at the same
time show that children probably use more general, contextual,
information to decide in ambiguous cases. This suggests that chil¬
dren’s problems may lie in interpreting the context and this still
presents problems for older children as we saw in this example from
Chapter 5:

T: what are you laughing at

P: nothing

While Shatz and Reeder discuss children’s understanding of

directives used by others, Garvey (1975) uses videotapes of children
aged 3%-5% playing together in pairs, to discover how children use
language to direct others. Interestingly children apparently over¬
whelmingly choose the explicit form—94% of the younger children’s
directives and 87% of the older children’s were imperative—
although as we have seen they have no difficulty in understanding
less explicit forms.

Conversational structure
Most of the work in the acquisition of discourse is currently con¬
centrating on single utterances and even parts of utterances, examin¬
ing how the child produces and interprets individual speech acts; so
far there has been little concern with conversational structure and
how the child learns to converse. One major exception is the work
166 AN INTRODUCTION TO DISCOURSE ANALYSIS

of Keenan (1974, 1975) analysing monthly videotapes of her twin

sons, aged 2.9 at the beginning of the research and 3.8 at the end.
One major feature of adult conversation is that successive utter¬
ances are closely related; as we have seen Sacks emphasises that
speakers work to achieve this coherence, while the inability to relate
utterances is taken to be one of the indications of mental illness. It
is therefore of interest to discover when children become able to
hold a coherent conversation. Halliday (1975) notes that before the
age of 18 months Nigel had no concept of exchanging information
and therefore could not indulge in dialogue, while Piaget (1926)
observing children in a kindergarten, suggests that even at the age
of five or six children tend not to address their speech to a co¬
present listener. This would suggest that the habit, if not the skills
of conversation, are acquired late. However, the evidence from
Keenan’s children suggests just the opposite—in the earliest record¬
ings there is strong evidence of the great importance the children
attach to relevance and turn-taking. In the following example the
children maintain a coherent, though meaningless, interaction by
focussing on the phonological shape of the previous utterance and
then modifying it:

A; [Ja:] [JabatJ]
B: [Joibabat]
A: [Joibabat] [JobabatJ] (laughs)
B: [JoibababatJ]
A: [JoibatJ] (laugh)
B: [JoibatJ]
A: [baptj]
B: [JoibatJ]
A: [batji] [bitji] [badi] [bidi] [babi]
B: [badi] (laughing)
(Keenan & Klein 1974)

As Keenan & Klein observe such examples show that ‘even when
the child is unable to maintain a referential talk-exchange he is still
willing to interact verbally’. Indeed their evidence suggests that
this is a crucial development stage, for at 2 years 9 months a third
of the children’s exchanges are sound play, while three months
later it has completely disappeared.
In adult conversation questions and commands are speech acts
which require a response from the listener; assertions, formally at
least, do not, although in most cases they do elicit some form of
 THE ACQUISITION OF DISCOURSE 167

acknowledgement or reaction. The twins, however, ‘observe a

conversational norm which obliges the addressee routinely to
acknowledge the speaker’s utterance’. Keenan and Klein isolate
five types of acknowledgement or relevant response:

(a) basic —direct repetition A: mommy’s silly

acknowledgment B: mommy’s silly
(b) affirmation —explicit agree¬ A: big one
ment B: yes
(c) denial —negation of A: Jack and Jill
opposition B: no Jack and Jill
(d) matching —claim to be A: I find feather
performing a B: yes I find I get one
similar action
(e) extension —new predication A: flower broken
to previous B: many flowers broken
speaker’s topic
The listener is expected to produce one of these acknowledge¬
ments; if he doesn’t the speaker may repeat his assertion until it is
acknowledged:

A: goosey goosey gander . . .

B: [i:] moth [i:] moth
A: goosey goosey gander where shall I wander . . .
B: [i:] moth (4 times)
A: upstairs downstairs in the lady’s chamber . . .
B: [i:] moth (3 times)
A: [i:] moth
B: gone moth allgone
Keenan and Klein comment

in this context ‘goosey goosey gander’ is not a relevant re¬

sponse. It is not accepted by the utterer of ‘[i:] moth’ who, in
turn, perseverates with his utterances until the other stops
singing and takes note.
Conversation can of course, only progress when there is a topic
and as we have seen adults have problems in introducing, maintain¬
ing and ending topics—the problem for children is even greater
because they cannot always be sure that the addressee is even
attending. For this reason the pragmatic operation of introducing a
168 AN INTRODUCTION TO DISCOURSE ANALYSIS

new topic into the discourse (A1 below) is often carried out
separately from the semantic operation of predicating something
about it, (A2):

Al: oh oh oh bell
Bl: bell
A2: bell it’s mommy’s
B2: (unclear)
A3: was mommy’s alarm clock
was mommy’s alarm clock
B3 alarm clock
A4: yeah goes ding dong ding dong
(Keenan 1974)

In such cases the repetition by the second child is a crucial indicator

that he accepts the topic and that the first speaker can go on to
develop it.
In these examples we see children struggling to create coherent
and cohesive discourse utterance by utterance; Garvey (1975) in the
work on children’s requests discussed above suggests that slightly
older children can handle multi-utterance structures. While a
request sequence consists minimally of a request and an acknow¬
ledgement (which includes both verbal and non-verbal response),
there are other optional components. Often there is what Sacks
would call a pre-sequence and she labels preparation of propo¬
sitional content

A: You see that hammer there

B: Yeah
A: Hand it to me

In essence this is a more grammatical or explicit version of the

topic-comment structure ‘bell—its mommy’s’ noted by Keenan.
Almost a fifth of the requests in Garvey’s data included in the
same turn another clause, which she calls an adjunct, and which
serves to provide additional relevant information in a similar way to
the starter identified by Sinclair et al (1972):

A: adjunct: that’s where the iron belongs

request: put it there

Following the request but before the acknowledgement there some¬

times occurs an insertion sequence which she calls a clarification
and which consists of a request-answer pair.
 THE ACQUISITION OF DISCOURSE 169

A: request: hand me the truck

B: 'request: which one
clarification -<
A: answer: the dump truck

A final, optional, element in this structure is the acknowledgement

of the acknowledgement or more accurately of the performing of the
action:

A: request: Tie that on for me

B: acknowledgement: (Fastens tool belt on S)
A: acknowledgement: Thanks

Thus the full request structure which these children are already
able to handle is:
(Presequence) (Adjunct) Request (Clarification) Acknowledge¬
ment (Acknowledgement)
This evidence suggests that by the age of 5% the children have
mastered most of the complexities of conversational structure for
although requests are Garvey’s main concern, she notes that, as one
would predict from adult conversation, presequences and insertion
sequences occur with assertions and questions as well as requests
and she concludes that

it seems reasonable to suggest that learning to produce dis¬

course can be understood as learning to perform the component
behaviours which contribute to the successful execution of
speech acts, learning the relative order of these behaviours, and
learning the appropriate distribution of roles which the alter¬
nating turns require.
Three years ago there was no published work on the acquisition of
discourse; now it is one of the fastest growing and most exciting
areas within linguistics. At the moment it is possible to encompass
all the important research in a chapter; in three years’ time it will
require a book to itself.
 9

The analysis of literary

discourse

Stylistics
Literature is the art form realised entirely through language and
although evaluation and interpretation is the province of the literary
critic it is reasonable to suggest that a detailed analysis of authorial
technique and stylistic features can be more successfully achieved
within a rigorous linguistic framework. In one of the earliest contri¬
butions to stylistics Abercrombie (1965) uses a description of speech
rhythm to elucidate some problems in metrics. Leech (1969) draws
on an analysis of syllable structure in English to describe exhaus¬
tively the possible formal patterning in verse, and shows that in
addition to the common patterns of alliteration, rhyme and asson¬
ance, there exist the less frequently exploited ones of pararhyme,
consonance and reverse rhyme. At the level of grammar Halliday
(1973) demonstrates how William Golding in The Inheritors
characterises his Neanderthal men as primitive creatures, rarely
understanding cause and effect, by particular choices in the transi¬
tivity network. Those working on shorter texts can investigate in
greater detail: Sinclair (1966) in ‘Taking a Poem to Pieces’ offers a
complete grammatical analysis in scale and category terms, and
Thorne (1965) proposes a unique transformational grammar of an
e.e. cummings poem.
Like all other branches of applied linguistics, stylistics depends
on the tools provided by theoretical linguistics; as the techniques of
discourse analysis become more sophisticated and more widely
recognised we can expect their growing exploitation in stylistics. In
what follows I hope to show how a study of question-answer
sequences throws a great deal of light on the way in which certain
crucial dramatic effects are achieved in Othello.
We must of course exercise due care when we apply techniques
developed for the analysis of conversation to invented sequences
created and shaped for an artistic purpose, because some of the
 THE ANALYSIS OF LITERARY DISCOURSE 171

rules and conventions are different. One of Grice’s (1975) conversa¬

tional maxims is ‘be relevant’; this is true in an exaggerated way
of most literary conversations—it is assumed that utterances
are not simply relevant to the current topic, but also to the develop¬
ment of theme or characterisation. Thus many of the audience
complained about Harold Pinter’s early plays because what the
characters said often appeared to have no relevance to any¬
thing.
A second of Grice’s maxims concerns quantity and he para¬
phrases it as ‘make your contribution as informative but not more
informative than is required’. This maxim is frequently broken, at
least from the point of view of a character’s co-conversationalists—
frequently one character is told something he almost certainly
knows in order to indirectly inform the audience. However, despite
these differences we obviously for the most part interpret literary
conversation according to the same rules as ordinary conversa¬
tion.

Questions
Every time a speaker asks a question there is a series of under¬
lying assumptions, all of which must be correct if he is to receive
the answer he seeks. At times of course some of these assumptions
may be incorrect and the response may consist of a challenge to one
or more of the assumptions. Below is a list of eight assumptions
and examples from Othello of responses denying the validity of the
assumptions.

ASSUMPTION RESPONSE DENYING ASSUMPTION

1. Addressee is listening. 0
2. Speaker questions at OTHELLO: Not now sweet Desde-
appropriate time. mona, some other time.
(Ill iii 56)

3. Addressee hears the iago: Go to, farewell.. .

question. do you hear Roderigo?
RODER1GO: What say you?
(I iii 376-7)

4. Addressee understands the OTHELLO: Is he not honest?

question. IAGO: Honest my lord?
(Ill iii 104-5)
172 AN INTRODUCTION TO DISCOURSE ANALYSIS

ASSUMPTION RESPONSE DENYING ASSUMPTION

5. Addressee accepts speaker iago: Though I am bound to

as a person allowed / every act of duty I am
empowered to ask the not bound to that all
question. slaves are free to; utter
my thoughts
(III iii 139-41)
6. Addressee thinks the roderigo: Wilt thou be fast to my
speaker does not know hopes?
the answer. iago: Thou art sure of me...
I have told thee often
and I tell thee again
and again. (I iii 363-5)
7. Addressee is willing to iago: Are all doors locked?
answer. brabantio: Why, wherefore ask
you this? (I i 85)
8. Addressee knows the iago: What had he done to
answer. you?
cassio: I know not (II iii 277-8)

There are, of course, certain situations in which the assumptions are

different, or apply differentially. Thus in a classroom only pupils
are likely to question at inappropriate times; it is common-place
for teachers to ask questions to which they do know the answer;
and sometimes they also ask questions, for particular purposes,
which they know the addressees are unable to answer.
Sacks suggests that the first question we must always ask about
any utterance is whether it is intended seriously; perhaps the second
is whether the speaker is lying. A speaker always has the option of
creating a false impression, or, in the restricted terms of this discus¬
sion of falsely denying an assumption. As always a particular
linguistic option can be exploited. Parents know from long experi¬
ence that children frequently ‘fail’ to hear utterances like
A: Simon, it’s bath time.
and therefore resort to presequences which establish the first
assumption—that the addressee is listening:
A: Simon.
B: Yes.
A: It’s bath time.
 THE ANALYSIS OF LITERARY DISCOURSE 173

Similarly most people have at some time pretended not to hear or

not to understand difficult or embarrassing questions in order to
give themselves time to think. However, the success of such
manoeuvres depends on the skill with which they are carried out—
if the responder is suspected of ignoring another of Grice’s maxims
‘try to make your contribution one that is true’, all is lost.

Othello
One of the major cruces in any literary discussion of Othello is how
he comes to be so easily and quickly convinced in one central scene,
Act III iii, of his wife’s adultery—Muir (1968) observes that at the
beginning of the scene
Othello is perfectly happy in his marriage; at the end he has
decided to murder Desdemona and Cassio. Unless we realise
the extent to which Shakespeare has telescoped the action, we
shall be bound to think that Othello was absurdly prone to
jealousy, instead of ‘not easily jealous’ as he claims at the end
of the play.
I want to suggest that Iago rouses Othello’s suspicion by a sequence
of unanswered questions, not simply because the questions are
unanswered but because they are avoided apparently, but in fact
deliberately, clumsily, which suggests to Othello that Iago is con¬
cealing something.
In order to establish that for most of the play question-answer
pairs occur quite naturally let us glance at this unremarkable extract
from Act II iii:*
iago: What was he, that you followed with your sword?
What had he done to you?
cassio: I know not.
iago: Is’t possible?
cassio: I remember a mass of things, but nothing 280
distinctly; a quarrel, but nothing wherefore.
O God, that men should put an enemy in their
mouths, to steal away their brains; that we should
with joy, revel, pleasure, and applause, transform
ourselves into beasts!
iago: Why, but you are now well enough: how came
you thus recovered?
cassio: It hath pleas’d the devil drunkenness to give place
to the devil wrath; one unperfectness shows me
another, to make me frankly despise myself.
* Line references are to the Arden edition edited by M. R. Ridley.
174 AN INTRODUCTION TO DISCOURSE ANALYSIS

Cassio’s answer to lago’s first question denies our eighth assumption,

that the addressee knows the answer; Iago expresses surprise if not
disbelief but Cassio explains how he comes to ‘know not’. Iago
asks another question and Cassio answers perfectly satisfactorily.
Here, as in most other places in the play, the conventions governing
questions in conversation are carried over into the simulated conver¬
sation.
However, when we come to examine the crucial Act III iii there
are marked differences. At the beginning of the scene Cassio, dis¬
graced and relieved of his command after a drunken brawl, comes
to Desdemona, on lago’s advice, to ask her to intercede on his
behalf. Meanwhile Iago has brought Othello to see the two together;
Cassio seeing Othello approach takes his leave and as he does so
Iago observes

iago: Ha, I like not that. 35

othello: What dost thou say?
iago: Nothing, my lord, or if—I know not what.

Here we see Othello ask for a repetition, or more likely a clarifica¬

tion of what Iago has said—why does he not like it? —and Iago
refuses, then hesitates, then refuses again. This is the first time Iago
has declined to answer a question in the play, let alone a question
from Othello, but in isolation it might not be significant.
Othello asks another question, or perhaps uses an interrogative
to express surprise, for there is no doubt in his mind as his third
utterance shows:

othello: Was not that Cassio parted from my wife?

iago: Cassio, my lord? ... no, sure, I cannot think it.
That he would sneak away so guilty-like, 40
Seeing you coming.
othello: I do believe ’twas he.

Othello presents Cassio’s departure as an AB event and it is there¬

fore surprising that Iago does not confirm but questions. Having
questioned such an obvious fact Iago must offer a reason, and so he
does with a sting in the tail—having apparently dissociated the
departing figure from the still ‘honest’ though disgraced Cassio,
Iago is able to suggest that he departed guiltily.
The oddness of these two unanswered questions is emphasised as
the play returns to normal question-answer conventions with
 THE ANALYSIS OF LITERARY DISCOURSE 175

Desdemona replying to Othello and incidentally confirming the fact

which Iago had questioned:

desdemona: How now, my lord?

I have been talking with a suitor here,
A man that languishes in your displeasure.
othello: Who is’t you mean? 45
desdemona: Why, your lieutenant, Cassio, good my lord.
If I have any grace or power to move you.
His present reconciliation take:
For if he be not one that truly loves you.
That errs in ignorance, and not in cunning, 50
I have no judgement in an honest face,
I prithee call him back.
othello: Went he hence now?
desdemona: Yes, faith, so humbled.
That he has left part of his griefs with me,
I suffer with him; good love, call him back. 55

Forty lines further on Desdemona and Emilia depart leaving

Othello and Iago together. Iago immediately returns to the topic of
Cassio but he still must tread warily—the status difference is strongly
marked by the asymmetrical use of the personal pronouns, ‘you’
and ‘thou’ and Iago ‘bids’ to speak in the same way that children
do to manoeuvre his discourse superior into asking him to ask him
a question:

iago: My noble lord, 94

othello: What dost thou say, Iago?
iago: Did Michael Cassio, when you woo’d my lady,
Know of your love?

The question he asks is amazing when one considers that twenty

lines earlier Desdemona had referred to Cassio ‘that came a-wooing
with you’, but in fact at this point the topic of the discourse is not
particularly important. The crucial feature is the way in which Iago
represents himself as highly reluctant to answer any questions and
trades on his reputation.
Othello sees ‘honest Iago’, the blunt, unsubtle, outspoken soldier
apparently attempting clumsily and with increasing desperation to
avoid committing himself to any opinion about ‘honest Cassio’. In
quick succession Iago avoids six questions:
176 AN INTRODUCTION TO DISCOURSE ANALYSIS

iago: Did Michael Cassio, when you woo’d my lady. 95

Know of your love?
OTHELLO: He did, from first to last... why dost thou ask?
iago: But for a satisfaction of my thought.
No further harm.
OTHELLO: Why of thy thought, Iago?
iago: I did not think he had been acquainted with her. 100
OTHELLO: O yes, and went between us very often.
iago: Indeed?
OTHELLO: Indeed? Indeed: discern’st thou aught in that?
Is he not honest?
iago: Honest, my lord? 105
OTHELLO: Honest? ay, honest.
iago: My lord, for aught I know.
OTHELLO: What does thou think?
iago: Think, my lord?

Iago’s first answer ‘but for satisfaction of my thought’ is the

politest of possible refusals, denying assumption 7, and also allow¬
ing Iago to drop another stone into the pond—‘no further harm’.
His second answer, to the insistent ‘why of thy thought’, is patently
untrue—had he assumed a negative answer the question wouldn’t
have been asked in that form. Although we have much less reliable
intuitions about discourse, an exchange of the form

iago: Did Michael Cassio, when you woo’d my lady

know of your love?
othello: He did not.
iago: As I thought.

would have been very odd, particularly if Iago had, as he claims, no

particular reason for asking the question. As Sacks reminds us a
speaker must make his utterances obviously relevant.
Iago creates more confusion or suspicion by his interested or even
surprised ‘indeed’ and subsequent lack of response to ‘discem’st
thou ought in that?’, and to the AB event question ‘is he not
honest’ he can apparently only lamely pretend he hasn’t under¬
stood and return ‘honest, my lord’, before producing the non¬
committal ‘my lord for aught I know’. Iago here succeeds in
presenting the image of a man wriggling and twisting not to lie but
also not to tell a damaging truth. By the time we reach the last pair
 THE ANALYSIS OF LITERARY DISCOURSE 177

othello: What dost thou think?

iago: Think, my lord?

Iago has apparently run out of invention and is seen to be desper¬

ately pretending that he does not understand the question.
Othello thinks aloud showing Iago the inferences he is making
from this peculiar performance and then pleads for clarification:

othello: Think, my lord? By heaven, he echoes me, 110

As if there were some monster in his thought.
Too hideous to be shown: thou didst mean something;
I heard thee say but now, thou lik’st not that.
When Cassio left my wife: what didst not like?
And when I told thee he was of my counsel, 115
In my whole course of wooing, thou criedst ‘Indeed?’
And didst contract and purse thy brow together.
As if thou then hadst shut up in thy brain
Some horrible conceit: if thou dost love me.
Show me thy thought.
iago: My lord, you know I love you. 120
othello: I think thou dost.

Iago neatly sidesteps Othello’s question and focuses on the sub¬

ordinate clause which he re-affirms—‘you know I love you’.
The cumulative effect of this series of unanswered questions is to
convince Othello that Iago is not being truthful and to search for a
reason. Iago is able to build on the foundations of his suspicious
question-avoiding and gradually become more specific in his
accusations until he can warn

Look to your wife, observe her well with Cassio. 201

The structure of the discourse has returned to normality and the

dramatic development is once again being carried by the content
rather than the form of the discourse.

Later scenes
Many people, on reflection, consider parts of the plot of Othello
unconvincing. Why, they ask, does Desdemona not confess to having
lost her handkerchief and why at a later stage does she not force
Othello to detail his accusations and ‘ocular proof’. On stage, how¬
ever, in the moment by moment development of the play the events
are utterly convincing. Sacks frequently observes that topic is a
178 AN INTRODUCTION TO DISCOURSE ANALYSIS

joint production and at those points in the play when Desdemona

could have cleared herself there is marked topic conflict and as a
result normal interaction breaks down.
The following extract comes from Act III iv and is the first meet¬
ing between Othello and Desdemona after Iago has aroused ‘the
green-eyed monster’, jealousy. Desdemona is anxious to continue
the topic of their last meeting, the reinstatement of Cassio, Othello
wants to check that she has indeed lost the handkerchief. However,
Othello doesn’t simply ask whether she has lost the handkerchief
but prefaces it by a twenty line disquisition on its magical properties
and the dangers of losing it. Small wonder she should wish she ‘had
never seen it’, but in the context this is for Othello the first hint of
confirmation and he demands an explanation. For Desdemona the
remarkable thing is his tone—she of course does not know what he
has suffered in the short time since she last saw him—so she
responds with surprise ‘Why do you speak so startingly and rash’.
To Othello this seems like a deliberate avoidance of the question
and his repeated question is similarly parried with an expression of
concerned surprise. Othello has been through this before—the last
time someone avoided his questions it later transpired that he was
covering up untold horrors; this time he will not be put off:

desdemona: Then would to God that I had never seen it! 75

othello: Ha, wherefore?
desdemona: Why do you speak so startingly and rash?
othello: Is’t lost? is’t gone? speak, is it out o’ the way?
desdemona: Heaven bless us!
othello: Say you? 80
desdemona: It is not lost, but what an if it were?
othello: Ha!
desdemona: I say it is not lost.
othello: Fetch’t, let me see it.

Desdemona is unable to understand the overriding importance

which Othello attaches to the handkerchief and suspects that he is
manipulating the conversation to avoid talking about Cassio. She
fears Cassio will not arise naturally as a topic and therefore deter¬
minedly reintroduces him. However, she is only partly successful
and from this point on they both stick stubbornly to their own
topic, ‘skip-connecting’ in Sacks’ terms, until Othello explodes and
leaves the stage:
 THE ANALYSIS OF LITERARY DISCOURSE 179

desdemona: Why, so I can sir, but I will not now.

This is a trick, to put me from my suit, 85
I pray let Cassio be receiv’d again.
OTHELLO: Fetch me that handkerchief, my mind misgives.
desdemona: Come, come,
You’ll never meet a more sufficient man.
OTHELLO: The handkerchief!
desdemona: I pray, talk me of Cassio. 90
OTHELLO: The handkerchief!
desdemona: A man that all his time
Hath founded his good fortunes on your love.
Shar’d dangers with you,—
OTHELLO: The handkerchief!
desdemona: I’ faith, you are to blame.
OTHELLO: Zounds! (Exit) 95

Desdemona and Othello meet only once more in private before

the scene in which he stifles her and this time it is Othello who is
unwilling to answer questions—there is no pretence, no attempt to
deny presuppositions, he simply acts as if he hasn’t heard her ques¬
tions. He won’t tell her what his pleasure is; what horrible fancy
grips him; what his words mean; with whom she is false. This is
obviously dramatically necessary—any accusation at this stage
would offer Desdemona the chance to reply and demonstrate her
innocence.

desdemona: My lord, what is your will?

othello: Pray, chuck, come hither.
desdemona: What is your pleasure?
othello: Let me see your eyes, ... 25
othello: Look in my face.
desdemona: What horrible fancy’s this?
othello: (to Emilia) Some of your function, mistress.
Leave procreants alone, and shut the door.
Cough, or cry hem, if anybody come;
Your mystery, your mystery: nay dispatch. 30
(Exit Emilia)
desdemona: Upon my knees, what does your speech import?
I understand a fury in your words.
But not the words.
othello: Why, what art thou?
desdemona: Your wife, my lord, your true and loyal wife. 35
180 AN INTRODUCTION TO DISCOURSE ANALYSIS

othello: Come, swear it, damn thyself,

Lest, being like one of heaven, the devils them¬
selves
Should fear to seize thee, therefore be double-
damn’d,
Swear thou art honest.
desdemona: Heaven doth truly know it.
othello: Heaven truly knows, that thou art false as hell. 40
desdemona: To whom, my lord? with whom? how am I false?
othello: O Desdemona away! away! away! (Act IV ii)

Othello is half-crazed with anger and grief and this is both

reflected in and transmitted by the jerky semi-coherent discourse.
Desdemona finds it difficult to contribute relevantly let alone
defend herself. She makes one last attempt

Alas, what ignorant sin have I committed?

but this merely evokes a rambling series of insults; finally he accuses

her of being a ‘strumpet’ and ‘whore’, ignores her denials, and
storms out.
Only in Act V ii, when it is too late and he has already resolved
to murder her does Othello answer Desdemona’s questions and
reveal his evidence:

desdemona: What’s the matter?

othello: That handkerchief which I so lov’d, and gave thee
Thou gavest to Cassio.
desdemona: No by my life and soul.
Send for the man and ask him. 50
othello: Sweet soul, take heed of perjury,
Thou art on thy death bed.

Thirty-five lines later he stifles her.

Concluding remarks
Most of the work described in earlier chapters has been concerned
with spoken interaction and this particular chapter has applied
insights derived from conversational analysis to illuminate technique
in simulated spoken interaction. Those working on written discourse
have tended to analyse it as monologue and to ignore the fact that
as he reads it the reader interacts with the text and thus an inter¬
active model might also be appropriate for a written discourse.
 THE ANALYSIS OF LITERARY DISCOURSE 181

As you close this book you might like to speculate on the func¬
tion of full stops. Are they perhaps interaction points, places where
the writer thinks the reader needs to stop and ask questions about
the previous sentence, questions whose range I initially restrict
by the structuring of my argument and which I subsequently answer
in the next or later sentences? But now, for me,

the rest is silence.

.
 References

Abbs, B., Ayton, A. and Freebairn, I. Strategies, London, Longman, 1975.

Abbs, B., Cook, V. and Underwood, M. Realistic English, London, O.U.P.,
1969.
Abercrombie, D. Studies in Phonetics and Linguistics, London, O.U.P.,
1965.
Albert, E. M. ‘Culture patterning of speech behaviour in Burundi’, in
Gumperz, J. J. and Hymes, D. (eds.), Directions in Sociolinguistics, New
York, Holt, Reinhart and Winston, 1972, 72-105.
Allen, J. P. B. and Widdowson, H. G ‘Teaching the Communicative
Use of English’, in International Review of Applied Linguistics in Lan¬
guage Teaching, 1974, 1-21.
English in Physical Science, London, O.U.P., 1974.
Arewa, E. O. and Dundes, A. ‘Proverbs and the ethnography of speaking
folklore’, in Gumperz, J. J. and Hymes, D. (eds.), The Ethnography of
Communication. American Anthropologist 66/6, 1964, 70-85.
Austin, J. L. How to do things with Words, Oxford, Clarendon Press, 1962.
Barnes, D. ‘Language in the secondary classroom’, in Barnes, D., Britton, J.
and Rosen, H. Language the Learner and the School, Harmondsworth,
Penguin, 1969.
Bates, E. ‘Acquisition of pragmatic competence’, in Journal of Child Lan¬
guage 1, 277-81, 1974.
Bellack, A. A., Kliebard, H. M., Hyman, R. T. and Smith, F. L. The language
of the classroom. New York, Teachers College Press, 1966.
Bereiter, C. and Engelmann, S. Teaching Disadvantaged Children in the
Pre-School, New York, Prentice Hall, 1966.
Bernstein, B. B. ‘Linguistic codes, hesitation phenomena and intelligence’, in
Language and Speech 5, 1962, 31—46.
Bever, T. G. and Ross, J. R. ‘Underlying structures in discourse’,
unpublished ms. quoted in Labov, W. ‘Rules for ritual insults’.
Biddle, B. J. ‘Methods and Concepts in Classroom Research’, in Review of
Educational Research 37, 1967.
Blom, J.-P. and Gumperz, J. J. ‘Social meaning in linguistic structures:
code-switching in Norway’, in Gumperz, ,T. J. and Hymes D. (eds.), Direc¬
tions in Sociolinguistics, New York, Holt, Reinhart and Winston, 1972,
407-434.
Bloom, L. Language Development: Form and Function in Emerging
Grammars, Cambridge, Mass.; MIT Press, 1970.
Bloomfield, L. Language, New York, Holt, Reinhart, 1933.
Boyd, J. and Thorne, J. P. ‘The deep grammar of modal verbs’, in Journal of
Linguistics 5/1, 1969, 57-74.
Braine, M. ‘The ontogeny of English phrase structure: the first phase’, in
Language 39, 1963, 1—13.
184 REFERENCES

Brazil, D. C. Discourse Intonation, Discourse Analysis Monographs 1,

Birmingham, English Language Research, 1975.
Bricker, V. R. ‘The ethnographic context of some traditional Mayan speech
genres’, in Bauman, R. and Sherzer, J. (eds.), Explorations in the Ethno¬
graphy of Speaking, London, C.U.P., 1974, 368-388.
Brown, R. A First Language, London, Allen and Unwin, 1973.
Brown, R. and Bellugi, U. ‘Three processes in the child’s acquisition of
syntax’, in Harvard Educational Review 34, 1964, 133-151.
Bruner, J. ‘The ontogenesis of speech acts’, in Journal of Child Language 2,
1975, 1-19.
Candlin, C. N., Kirkwood, J. M. and Moore, H. M. Study Skills in
English—Part 1, mimeo University of Lancaster, 1973.
‘Developing Study Skills in English’, in English for Academic Study,
London, British Council ETIC, 1975.
Candlin, C. N, Leather, J. H. and Bruton, C. J. (ms) Doctor Patient Com¬
munication Skills. Draft materials; Applied Linguistics Research Unit,
University of Lancaster.
Chomsky, N. Syntactic Structures, The Hague, Mouton, 1957.
Cohen, L. J. ‘Do illocutionary forces exist?’, in Fann, K. T. (ed.), Symposium
on J. L. Austin, London, Routledge & Kegan Paul, 1969, 420-444.
Coulthard, R. M. and Ashby, M. C. ‘A linguistic description of doctor-
patient interviews’, in Wadsworth, M. and Robinson, D. eds. Studies in
Everyday Medical Life, London, Martin Robertson, 1976.
Crystal, D. Prosodic systems and intonation in English, Cambridge, C.U.P.,
1969.
Dore, J. (ms). On the acquisition of speech acts: a pragmatic description of
early language development.
Dore, J. ‘Holophrases, speech acts and language universals’, in Journal of
Child Language 2, 1975, 21-39.
Duncan, S. ‘Some signals and rules for taking speaking turns in conversa¬
tion’, in Journal of Personality and Social Psychology 23/2, 1972, 283-
292.
Duncan, S. ‘Towards a grammar for dyadic conversation’, in Semiotica 9/1,
1973, 29-46.
Duncan, S. ‘On the structure of speaker-auditor interaction during speaking
turns’, in Language in Society 3/2, 1974, 161-80.
Duncan, S. and Niederehe, G. ‘On signalling that its your turn to speak’, in
Journal of Experimental Social Psychology 10, 1974, 234^17.
Ervin-Tripp, S. ‘On sociolinguistic rules: alternation and co-occurrence’, in
Gumperz, J. J. and Hymes, D. (eds.), Directions in Sociolinguistics, New
York, Holt, Reinhart and Winston, 1972, 213-250.
Ervin-Tripp, S. ‘The comprehension and production of requests by children’,
in Papers and Reports on child language development, Stanford, Stanford
University, 1974.
Ervin-Tripp, S. ‘Wait for me Roller-skate’, in Mitchell-Kernan, C. and
Ervin-Tripp, S. (eds.), Child Discourse, New York, Academic Press (in
press).
Ferguson, C. ‘Diglossia’, Word 15, 1959, 325-340.
Ferguson, J. Interruptions in spontaneous dialogue’, paper delivered to BPS
conference, Stirling, 1975.
Fillmore, C. J. ‘The case for case’, in Bach E. and Harms, R. T. (eds.),
Universals in Linguistic Theory, New York, Holt, Reinhart and Winston,
1968, 1-90.
Fillmore, C. J. ‘May we come in’, in Smith, D. M. and Shuy, R. W. (eds.),
Sociolinguistics in crosscultural perspective, Washington, Georgetown
University Press, 1972, 97-115.
 REFERENCES 185

irth, J. R. 1935. ‘The technique of semantics’, in Papers in Linguistics 1934-

1951, London, O.U.P., 1957, 7-33.
landers, N. A. Analysing Teaching Behaviour, Reading, Mass., Addison-
Wesley, 1970.
orguson, L. W. ‘Locutionary and Illocutionary acts’, in Warnock, G. J.
(ed.), Essays on J. L. Austin, Oxford, Clarendon Press, 1973, 60—185.
orsyth, I. J. ‘Patterns in the discourse of pupils and teachers’, in The space
between . .. CILT reports and papers 10, London, CILT, 1974, 77-99.
oster, M. K. ‘When words become deeds: an analysis of three Iroquois
longhouse speech events’, in Bauman, R. and Sherzer, J. (eds.), Explora¬
tions in the Ethnography of Speaking, London, C.U.P., 1974, 354-367.
rake, C. O. ‘Struck by speech: the Yakan concept of litigation’, in Gumperz,
J. J. and Hymes, D. (eds.), Directions in Sociolinguistics, New York, Holt,
Reinhart and Winston, 1972, 106-129.
iallagher, J. J. and Aschner, M. J. ‘A preliminary report on analyses of
classroom interaction’, in Merrill-Paimer Quarterly 9, 1963, 183-194.
iarvey, C. ‘Requests and responses in children’s speech’, in Journal of
Child Language 2, 1975, 41-63.
ibcrtz, C. The Religion of Java, New York, The Free Press, 1960.
ilendenning, E. H. English in Mechanical Engineering, London, O.U.P.,
1973.
Jordon, D. and Lakoff, G. ‘Conversational Postulates’, in Papers from the
seventh regional Meeting of the Chicago Linguistic Society, Chicago,
C.L.S., 1971, 63-84.
Jrice, H. P. ‘Logic and conversation’, in P. Cole and J. L. Morgan (eds.),
Syntax and Semantics III: Speech Acts, New York, Academic Press, 1975,
4l~58- . . .
Jumperz, J. J. ‘Linguistic and social interaction in two communities’, in
J. J. Gumperz and D. Hymes (eds.), ‘The ethnography of communication’,
American Anthropologist 66, 6, II, 1964, 137-154.
Jumperz, J. J. and Herasimchuk, E. ‘The conversational analysis of social
meaning: a study of classroom interaction’, in Shuy, R. W. (ed.), Socio¬
linguistics: current trends and prospects, Washington, Georgetown Univer¬
sity Press, 1972.
Jumperz, J. J. and Hymes, D. (eds.), Directions in sociolinguistics. New
York, Holt, Reinhart and Winston, 1972.
lalliday, M. A. K. ‘Categories of the theory of grammar’, in Word 17, 1961,
241-92.
lalliday, M. A-. K. Intonation and Grammar in British English, The Hague,
Mouton, 1967.
-lalliday, M. A. K. Explorations in the Functions of Language, London,
Arnold, 1973. j
lalliday, M. A. K. ‘A sociosemiotic perspective on language development’,
Bulletin of the School of Oriental and African Studies 37/1, 1974.
lalliday, M. A. K. Learning How to Mean, London, Arnold, 1975.
lalliday, M. A. K. and Hasan, R. Cohesion in English, London, Longman,
1976.
lalliday, M. A. K., McIntosh, A. and Strevens, P. The Linguistic Sciences
and Language Teaching, London, Longman, 1964.
Harris, Z. ‘Discourse analysis’, in Language 28, 1952, 1-30.
luddleston, R. D., Hudson, R. A., Winter, E. O., and Henrici, A. Sentence
and Clause in Scientific English, O.S.T.I., Report No. 5030, 1968.
Hymes, D. ‘Introduction: towards Ethnographies of Communication’, in
Gumperz, J. J. and Hymes, D. (eds.), The Ethnography of Communica¬
tion, American Anthropologist 66/6, 1964, 1-34.
186 REFERENCES

Hymes, D. ‘Sociolinguistics and the ethnography of speaking’, in Ardener, E.

(ed.), Social Anthropology and Linguistics, Association of Social Anthro¬
pologists Monograph 10, London, Tavistock, 1971, 47-93.
Hymes, D. ‘Models of the interaction of language and social life’, in
Gumperz, J. J. and Hymes, D. (eds.), Directions in Sociolinguistics, New
York, Holt, Reinhart and Winston, 1972, 35-71.
Hymes, D. ‘On communicative competence’, in Pride, J. B. and Holmes, J.
(eds.), Sociolinguistics, Harmondsworth, Penguin, 1972, 269-285.
Hymes, D. ‘Ways of speaking’, in Bauman, R. and Sherzer, J. (eds.), Explora¬
tions in the Ethnography of Speaking, London, C.U.P., 1974, 433-
452.
Jakobovitz, L. A. and Gordon, B. The context of foreign language teaching,
Rowley, Mass., Newbury House, 1974.
Irvine, J. T. ‘Strategies of status manipulation in the Wolof-greeting’, in
Bauman, R. and Sherzer, J. (eds.), Exploration in the Ethnography of
Speaking, London, C.U.P., 1974, 167-191.
Jefferson, G. ‘Side sequences’, in D. Sudnow (ed.), Studies in social inter¬
action, New York, The Free Press, 1972, 294-338.
Jefferson, G. ‘A case of precision timing in ordinary conversation: over¬
lapped tag-positioned address terms in closing sequences’, Semiotica 9/1,
1973, 47-96.
Jefferson, G. ‘Error correction as an interactional resource’. Language in
Society 3/2, 1974, 181-99.
Joos, M. The Five Clocks. I.J.A.L. 28, part V, 1962.
Johns, T. F. (ms). ‘Seminar discussion strategies: problems and principles
in role simulation’, University of Birmingham, mimeo.
Jupp, T. C. and Hodlin, S. Industrial English, London, Heinemann, 1975.
Keenan, E. O. ‘Again and again: the pragmatics of imitation in child lan¬
guage’. Paper presented to annual meeting of American Anthropological
Association, Mexico City, 1974.
Keenan, E. O. ‘Conversational competence in children’, in Journal of Child
Language 1/2, 1975, 163-183.
Keenan, E. O. and Klein, E. ‘Coherency in Children’s Discourse’, in Journal
of Psycholinguistic Research 4/4, 1975, 365-380.
Kendon, A. ‘Some functions of gaze direction in social interaction’, in Acta
Psychologica 26, 1967, 22-63.
Kendon, A. ‘Some relationships between body motion and speech: an analy¬
sis of an.example’, in Siegman, A. and Pope, B. (eds.), Studies in Dyadic
Interaction, New York, Pergamon Press, 1972, 177-210.
Kendon, A. ‘The role of visible behavior in the organisation of face-to-face
interaction’, in von Cranach, M. and Vine, I. (eds.), Social Communication
and Movement: Studies of Interaction and Expression in Man and Chim¬
panzee, London, Academic Press, 1973, 29-74.
Kendon, A. and Ferber, A. ‘A description of some human greetings’, in
Michael, R. P. and Crook, J. H. (eds.), Comparative Ecology and Be¬
havior of Primates, London, Academic Press, 1973, 591-668.
Kliebard, H. M. ‘The observation of classroom behaviour’, in Hitchcock, C.
(ed.), The Way Teaching Is, Washington, Association for Supervision and
Curriculum Development and Centre for the Study of Instruction,
1966.
Labov, W. ‘The reflection of social processes in linguistic structures’, in
Fishman, J. (ed.), Readings in the Sociology of Language, The Hague,
Mouton, 1968, 240-51.
Labov. W. ‘The logic of non-standard English’, in Georgetown Monographs
on Language and Linguistics 22, 1969, 1-31.
 REFERENCES 187

Labov, W. ‘The study of language in its social context’, in Studium Generate

23, 1970, 30-87.
Labov, VV. ‘Rules for ritual insults’, in Sudnow, D. (ed.). Studies in Social
Interaction, New York, Free Press, 1972, 120-169.
Labov, W. and Fanshel, D. (ms). Therapeutic Discourse: Psychotherapy as
Conversation.
Lakoff, R. ‘Language in context’, in Language 48/4, 1972, 907-27.
Laver, J. ‘The production of speech’, in Lyons, J. (ed.), New Horizons in
Linguistics, Harmbndsworth, Penguin, 1970, 53-77.
Leech, G. N. A Linguistic Guide to English Poetry, London, Longman,
1969.
de Long, A. J. ‘Kinesic signals at utterance boundaries in preschool chil¬
dren’, in Semiotica 11/1, 1974, 43-73.
Lyons, J. An introduction to theoretical linguistics, London, C.U.P., 1968.
McNeill, D. The Acquisition of Language: The Study of Developmental
Psycholinguistics, New York, Harper and Row, 1970.
Maclean, J. English in Basic Medical Science, London, O.U.P., 1975.
Medley, D. M. and Mitzel, H. E. ‘Measuring Classroom Behaviour by Sys¬
tematic Observation’, in Gage (ed.), Handbook of Research on Teaching,
Chicago, Rand McNally, 1963.
Meux, M. D. ‘Studies of learning in the school setting’, in Review of Educa¬
tional Research 37, 1967.
Miller, W. and Ervin, S. ‘The development of grammar in child language’,
in Bellugi, U. and Brown, R. (eds.), The Acquisition of Language, Mono¬
graphs of the Society for Research in Child Development 29/92, 9-34.
Mitchell, T. F. ‘The language of buying and selling in Cyrenaica’, in
Hesperis 44, 1957, 31-71.
Mountford, A. English in Workshop Practice, London, O.U.P., 1975b.
Discourse analysis and the simplification of reading material for English
for Special Purposes. Unpublished M. Litt. thesis, University of Edin¬
burgh, 1975a.
Muir, K. (ed.), Othello, Harmondsworth, Penguin, 1968.
O’Connor, D. J. and Arnold, G. F. Intonation of Colloquial English, Lon¬
don, Longman, 1959.
Pearce, R. D. The structure of discourse in broadcast interviews. Unpub¬
lished M. A. thesis, University of Birmingham, 1973.
Piaget, J. The Language and Thought of the Child, New York, Harcourt
Brace, 1926.
Polgar, S. A. ‘Biculturation of Mequaki teenage boys’, in American Anthro¬
pologist 62, 1960, 217-235.
Reeder, K. (ms). ‘On young children’s discrimination of illocutionary force: a
report on work in progress’.
Reisman, K. ‘Contrapuntal conversations in an Antiguan village’, in
Bauman, R. and Sherzer, J. (eds.), Explorations in the Ethnography of
Speaking, London, C.U.P. 1974, 110-124.
Ross, J, R. ‘On declarative sentences’, in Jacobs, R. A. and Rosenbaum, P. S.
(eds.), Readings in English Transformational Grammar, Waltham, Mass.,
Ginn, 1970, 222-272.
Ross, J. R. ‘Where to do things with words’, in Cole, P. and Morgan, J. L.
(eds.), Syntax and Semantics Vol. 3: Speech Acts, New York, Academic
Press, 1975, 233-256.
Sacks, H. Mimeo lecture notes. (1967-71).
Sacks, H. ‘On the analysability of stories by children’, in Gumperz, J. J. and
Hymes, D. (eds.). Directions in sociolinguistics, New York, Holt, Reinhart
and Winston, 1972a.
188 REFERENCES

Sacks, H. ‘An initial investigation of the usability of conversational data for

doing Sociology’, in Sudnow. D. (ed.), Studies in Social Interaction, New
York, The Free Press, 1972b, 31—74.
Sacks, H. (ms). Aspects of the sequential organisation of conversation.
Sacks, H., Schegloff, E. A. and Jefferson, G. ‘A simplest systematics for the
organisation of turn-taking for conversation’, in Language 50/4, 1974,
696-735.
Sadock, J. M. ‘The soft interpretive underbelly of generative semantics’, in
Cole, P. and Morgan, J. L. (eds.), Syntax and Semantics Vol 3: Speech
Acts, New York, Academic Press, 1975, 383-396.
Salmond, A. ‘Rituals of encounter among the Maori: sociolinguistic study
of a scene’, in Bauman, R. and Sherzer, J. (eds.), Explorations in the
Ethnography of Speaking, London, C.U.P., 1974, 192-212.
Scheflen, A. E. ‘The significance of posture in communication systems’, in
Psychiatry 27, 1964, 316-31.
Schegloff, E. A. ‘Sequencing in conversational openings’, in American
Anthropologist 70/6, 1968, 1075-95.
Schegloff, E. A. ‘Notes on a conversational practice: formulating place’, in
Sudnow, D. (ed.), Studies in social interaction, New York, Free Press,
1972, 75-119.
Schegloff, E. A. and Sacks, H. ‘Opening up closings’, in Semiotica 8/4, 1973,
289-327.
Schlesinger, I. M. ‘Production of utterances and language acquisition’, in
Slobin, D. I. (ed.), The Ontogenesis of Grammar, New York, Academic
Press, 1971, 63-101.
Searle, J. R. ‘What is a Speech Act?’, in Black, M. (ed.), Philosophy in
America, Ithaca, Cornell University Press, 1965, 221-239.
Searle, J. R. Speech Acts, London, C.U.P., 1969.
Searle, J. R. ‘Indirect Speech Acts’, in Cole, P. and Morgan, J. L. (eds.).
Syntax and Semantics Vol. 3: Speech Acts, New York, Academic Press,
1975, 59-82.
Shatz, M. ‘The comprehension of indirect directives: can two-year-olds shut
the door?’ Paper presented the summer meeting of the LSA. Amherst,
Mass., 1974.
Sherzer, J. ‘Namakke, Summakke, Kormakke: three types of Cuna speech
event’, in Bauman, R. and Sherzer, J. (eds.), Explorations in the Ethno¬
graphy of Speaking, London, C.U.P., 1974, 262-282.
Siegman, A. W. and Pope, B. (eds.) Studies in dyadic communication, New
York, Pergamon, 1972:
Sinclair, J. McH. ‘Taking a poem to pieces’, in Fowler R. (ed.), Essays on
Style and. Language, London, Routledge & Kegan Paul, 1966, 68-81.
Sinclair, J. McH. ‘Lines about Lines.’, in Kachru, B. B. and Stahlke,
H. F. W. (eds.), Current Trends in Stylistics, The Hague, Mouton, 1971,
251-261.
Sinclair, J. McH. ‘Linguistics in Colleges of Education’, in Dudley Journal of
Education, 1973, 17-25.
Sinclair, J. McH., Forsyth, I. J., Coulthard, R. M. and Ashby, M. C. The
English Used by Teachers and Pupils. Final Report to S.S.R.C. Mimeo,
Birmingham University, 1972.
Sinclair, J. McH., and Coulthard, R. M. Towards an analysis of discourse,
London, O.U.P., 1975.
 REFERENCES 189

Straker-Cook, R. H. ‘A communicative approach to the analysis of ex¬

tended monlogue discourse and its relevance to the development of
teaching materials for English for Special Purposes’, unpublished M.
Litt thesis, University of Edinburgh, 1975.
Strawson. ‘Intention and Convention in Speech Acts’, in Philosophical Re¬
view 73, 1964, 439-460.
Sweet, H. A primer of phonetics, Oxford, Clarendon Press, 1906.
Taba, H., Levine, S. and Elsey, F. F. Thinking in Elementary School Chil¬
dren, U.S. Dept, of Health, Education and Welfare, Office of Education.
Cooperative Research Project No. 1574, San Francisco State College,
1964.
Thorne, J. P. ‘Stylistics and generative grammars’, in Journal of Linguistics
1, 1965, 49-59.
Warnock, G. J. ‘Some types of performative utterance’, in Warnock, G. J.
(ed.), Essays on J. L. Austin, Oxford, Clarendon Press, 1973, 69-89.
Watson, O. M. and Graves, T. D. ‘Quantitative research in proxemic be¬
havior’, in American Anthropologist 68, 1966, 971-985.
Weick, K. E. ‘Systematic observational methods’, in Lindzey, G., and
Aronson, E. (eds.), Handbook of Social Psychology, Vol. 2, Reading,
Mass., Addison-Wesley, 1968.
Widdowson, H. G. An applied linguistic approach to discourse analysis. Un¬
published PhD. thesis, University of Edinburgh, 1973.
Widdowson, H. G. ‘The deep structure of discourse and the use of
translation’, in Corder, S. P., and Roulet, E. (eds.), Linguistic insights in
applied linguistics. Second Neuchatel Colloquium in Applied Linguistics,
Brussels, AIMAV, Paris, Didier, 1974.
Widdowson, H. G. ‘An approach to the teaching of scientific English dis¬
course’, in RELC Journal 5, 1974, 27^10.
Widdowson, H. G. ‘EST in theory and practice’, in Jones, K. and Roe, P.
(eds.), English for Academic Study. ETIC Occasional Paper, London,
British Council, 1975.
Wilkins, D. ‘An investigation into the linguistic and situational content
of the common core in a unit/credit system’, Strasbourg, Council of
Europe, mimeo, 1972.
Wilkins, D. ‘Grammatical Situational and Notional Syllabuses’. Paper
presented to the third International Congress of Applied Linguistics,
Copenhagen, August 1972.
Wilkins, D. ‘Notional Syllabuses and the concept of a minimum adequate
grammar’, in Corder, S. P. and Roulet, E. (eds.). Linguistic Insights in
Applied Linguistics, Brussels, AIMAV, Paris, Didier, 1974.
Williams, S. A sociolinguistic analysis of the general practice interview,
University of Birmingham, unpublished M.A. thesis, 1974.
Winter, E. O. ‘A causal-relational approach to English texts’, special issue
of Instructional Science, 6, 1, 1977.
 Index

Abbs, B., 138, 142-143 Braine, M„ 154, 155

Abercrombie, D., 170 Brazil, D. C„ 115, 117, 119-137,
acknowledgement, 167, 168, 169 158
act, 101, 102, 104, 105, 106 Bricker, V. R., 38
accept, 106 Brown, R., 154, 155
acknowledgement, 167,168,169
bid, 105 Candlin, C. C. 143, 146, 150
comment, 106, 113 category bound activities, 81
cue, 105 chaining rule, 71
directive, 104, 107, 111, 112, Chomsky, N„ 2, 9, 30, 154, 155
162, 163, 164, 165 codeswitching, 32-33, 46, 47
elicitation, 104, 105, 107, 111, Cohen, L. J., 21
112, 113, 131, 133, 134, 136, coherence, 9, 10, 63, 116
146 cohesion, 9, 10, 75, 116
informative, 107, 111, 112, 113, command, 97, 103, 107, 109, 110,
131 111, 112, 113, 166
nomination, 105 comment, 106, 113
starter, 112, 168 communicative act, 139, 147, 148-
adjacency pair, 53, 70-71, 73, 96, 150
106 classification, 149
Albert, E. M„ 30, 44, 57 definition, 149
Allen, J. P. B., 147, 148, 150-153 explanation, 147-149
alternation, 36 generalisation, 149
answer, 64, 65, 70, 104, 158 observation, 148
summons-answer, 84 communicative competence, 30
Arewa, E. O., 42 communicative function, 138, 140
Arnold, G. F„ 116-117 competence, 2, 3, 30, 31
Ashby, M. C., 105 completion, 55
Aschner, M. J., 93 see also: possible completion
Austin, J. L., 8, 11-21, 24, 25, 27, conclusion, 102
39, 67, 98, 104, 116 constative, 11, 15, 16
constitutive rule, 22-23
back channel, 62 context of situation, 1, 3
Barnes, D., 93-95, 96 co-occurrence, 36
Bellack, A. A., 96-98, 99 Coulthard, R. M., 105
Bellugi, U„ 154 Crystal, D., 116
Bereiter, C., 50 cue, 105
Biddle, B. J., 92
Blom, J. P„ 32, 33, 47, 52 directive, 104, 107, 111, 112, 162,
Bloom, L., 155 163, 164, 165
Bloomfield, L., 1,2 indirect directive, 25-26
192 INDEX

discourse, 1, 6, 9, 10, 153 123, 157, 159-162, 166, 170

Dore, J., 156-158, 160, 161 Harris, Z., 3, 4, 63
Duncan, S., 61-62, 92 Hasan, R., 75
Dundes, A., 42 hearer’s maxim, 81
hesitation, 9, 49
elicitation, 104, 105, 107, 111, 112, Hodlin, C., 143
113, 131, 133, 134, 136, 146 Huddleston, R. D., 36
Engelmann, S., 50 Hymes, D„ 7, 23, 30-49, 52
Ervin-Tripp, S., 36, 48, 154, 162,
163 illocution, 9, 17-18, 21, 147, 148,
evaluation, 106 150, 152, 156
exchange, 100, 102, 103, 104, 106, illocutionary act, 17-23, 25, 26,
128-131, 132-137, 143, 146, 104, 151
176 illocutionary force, 19-22, 29,
boundary exchange, 103 96, 152, 158
informing exchange, 103 illocutionary potential, 27, 29
directing exchange, 103 implicit performative, 28
eliciting exchange, 103-106 indirect speech act, 25, 28
indirect requests, 25, 72
Fanshel, 66, 67, 72, 111 informative, 107,111,112,113,131
feedback, 103, 104 initiation, 95, 96
Ferguson, C., 34 intention, 19, 20, 21
Ferguson, J., 57, 61 interpretive rules, 31, 39, 47, 51,
Fillmore, C. J., 140, 154 64-69, 93, 108-111, 115
Firth, J. R., 1. 3, 9 interruption, 23, 57, 58, 60, 61
Flanders, N. A., 93, 95-96, 98, 99 intonation, 61, 86, 112, 116-137,
focus, 102, 103 158
Forguson, 21 Irvine, J. T„ 44, 45, 52
Foster, M. K., 42
Jakobovits, L. A., 141-142
Fox, J. R., 31
Jefferson, G., 8, 32, 39, 52, 55-56,
Frake, C. O., 42, 44
69, 73-74, 91, 92
frame, 101, 103, 129
Johns, T. F„ 143-146, 150
function indicating device, 22, 23
Joos, M., 34
Jupp, T. C., 143
Gallagher, J. J., 93
Garvey, C., 165, 168-169 Keenan, E. O., 157, 158, 166-168
gaze, 60-61 Kendon, A., 59, 60-61, 106, 107
Geertz, C., 34-35, 42, 46, 52 key, 45-16, 120, 124-131, 132,
genres, 38, 39, 41, 42, 46, 52 133-137
elementary genres, 38 Klein, E., 166, 167
complex genres, 38 Kliebard, H. M., 92
gesture, 45, 59, 60, 163
Glendenning, E. H., 149, 150 Labov, W„ 7, 8, 30, 32, 34, 40-41,
Gordon, D„ 29, 141-142, 154 50, 51, 52, 62, 64-68, 69, 72, 98,
grammar, 1 99, 108, 111, 121, 154
grammaticality, 3, 30 Lakoff, G„ 3, 141-142, 154
Graves, T. D., 49 Lakoff, R., 3, 29, 154
greeting, 23, 32, 40, 44, 45, 53, 70, latent patterning, 37
84, 140, 141, 146, 158 Laver, J., 120
Grice, H. C., 65, 171, 173 Leech, G. N„ 170
Gumperz, J. J., 32, 33, 47, 52 level, 1,6, 100
locution, 9, 18, 20, 21, 147
Halliday, M. A. K„ 34, 75, 100, locutionary act, 17,18, 21, 22
105, 117-119, 120, 121, 122, de Long, A. J., 59-60
 INDEX 193

Lyons, J., 9 primary performative, 20, 21, 25

performative verb, 22
Maclean, J., 150 perlocution, 18, 21
McCawley, J. D., 3, 154 perlocutionary act, 17-20
McIntosh, A., 34 perlocutionary object, 19
McNeill, D., 155 perlocutionary sequel, 19
Medley, D. M., 93 pivot + open clause, 154-155
membership, 82, 119 possible completion, 54, 56, 57, 59,
membership categorisation de¬ 60
vice, 80, 81 post-completor, 56
see also: category bound activi¬ posture, 45, 59, 106, 107
ties prayer, 38, 39
metastatement, 101, 103 pre-closing, 86
focus, 102 possible pre-closing, 59, 86, 129
conclusion, 102 presequence, 71, 82-83, 168, 169,
Meux, M. D., 92 172
Miller, W„ 154 proclaiming tone, 121-124, 126,
misfire, 12-13 127, 131, 132, 136
Mitchell, T. F., 3, 4, 5, 6 promise, 14, 22-24
Mitzel, H. E., 92 promising, 22-24, 25
monologue, 113, 114
Mountford, A., 114, 115, 147, question, 8, 64, 65, 66, 70, 97, 99,
148-149, 150, 152 103, 108, 110, 111, 112, 113,
move, 8, 68, 69, 97-98, 101, 102, 123, 129, 131, 140, 163, 164,
103,104, 105, 106, 133, 134 166, 169, 171-172
follow-up move, 106 question-answer pair, 4, 7, 53,
initiating move, 106 54, 64, 70, 71, 72, 74, 91, 103,
reacting move, 97 173, 174
responding move, 97, 106 questioning repeat, 73
soliciting move, 97 types of teacher question, 93
structuring move, 97
Muir, K., 173 rank,99-102
reason for call, 76-77, 85
newsworthiness, 76 Reeder, K., 164-165
Niederehe, G., 62 referring tone, 121-124, 126, 127,
norms of interaction, 31, 32, 48- 131, 132, 136-137
51, 139 register, 34, 36
noticeable absence, 70 regulative rule, 22-23
notional syllabus, 139-140 Reisman, K., 32
O’Connor, D. J., 116-117 repetition, 9, 37, 104, 135, 136,
Othello, 170-180 157, 168
request, 8, 24, 40, 70, 71, 97, 99,
overlaps, 53, 54, 55, 56, 70
104, 140, 141, 168
participant, 5, 13, 41, 42, 43-44, for action, 66-67
45, 46, 47, 53, 115 for clarification, 73
pause, 52 for confirmation, 66, 131, 136
Pearce, R. D., 105, 114 indirect request, 67-69
Polgar, S. A., 49 response, 95, 96, 107, 113, 132,
performance, 2, 39 134, 135, 136
performative, 11-17, 19, 27, 28,98, ritual insults, 32, 40-41, 43
99 Ross, J. R„ 3, 27, 28, 29, 111, 154
explicit performative, 19, 20, 21, rules
26 consistency rule, 81, 83
implicit performative, 28 constitutive rule, 22-23
194 INDEX

rules—cont. 44, 46, 47, 68, 104, 138, 157,

economy rule, 81 162, 166
essential rule, 24 primitive speech act, 156-157,
interpretive rules, 31, 39, 47, 51, 158, 160, 161
64-69, 93, 108-111, 115 see also: act
preparatory rule, 23 speech community, 31, 32-34, 40,
productive rules, 93 47, 48, 51, 139
propositional content rule, 23 speech event, 39^48
regulative rules, 22-23 components of speech events,
sequencing rules, 7, 63, 93, 125, 41-47
130, 133 channel, 42, 46
sincerity rule, 23 key, 45-46
rules of speaking, 3, 32 message content, 46-47
rule breaking, 47-48 participants, 42, 43-44, 45, 46,
47, 53
Sacks, H., 8, 19, 32, 39, 40, 45, 52- purpose, 42, 44-45
57, 69, 70-72, 74, 75-92, 96, 98, setting, 42-43, 44, 46
99, 101, 102, 103, 104, 106, 107, topic, 32, 36, 42, 46, 47, 75-79,
116, 166, 168, 172, 176, 177, 178 80, 85, 86, 96, 101, 102, 103,
Sadock, J. M., 29 105, 106, 167, 168
Salmond, A., 42, 47, 52 speech styles, 31, 34-38, 39, 46
Sankoff, G., 9 significant speech styles, 37
Scheflen, A. E., 106 statement, 11, 27
Schegloff, E. A., 8, 32, 39, 52, 56, stories, 79-81
59, 69, 72, 73, 82, 83, 84, 85, 86, story preface, 79
87, 91, 92, 102 Straker-Cook, R., 105
Schlesinger, 155 Strawson, 14, 18, 20, 21
Searle, J. R„ 8, 17, 20, 21-27, 98, Strevens, P., 34
104, 148, 156 stylistic mode, 37-38,39, 41,42, 46
self-selection, 54, 56, 57 stylistic structure, 37-38, 39, 41,
sentence, 7, 9, 10, 54, 63, 100 42, 46
sequence, 71, 98, 129 Sweet, H., 120
insertion sequence, 72, 73, 74,
91, 133, 169 Taba, H., 93
misapprehension sequence, 9, tactics, 107, 111-113, 115
73-74 text, 9, 10, 153
presequence, 71, 82, 83, 168, Thorne, J. P„ 27, 170
169, 172 tone, 117-118, 120-124, 126, 127,
side sequence, 73-74, 92 128, 131-132, 136-137
shared knowledge, 65, 66, 108, tone group, 118-120, 121, 125-
121, 175, 176 127, 131, 132-137
see also: referring tone tonic, 118
Shatz, M., 162-164, 165 tonality, 119
Sherzer, J., 43, 52 tonicity, 118-119
silence, 52, 54, 55, 56, 57, 61 topic, 32, 36, 75-79, 80, 85, 86, 96,
Sinclair, J. McH., 6, 8, 25, 37, 40, 101, 102, 103, 105, 106, 167,
69, 98-115, 116, 143, 168, 170 168
situation, 5, 39, 50, 107-111 topicality, 77
skip-connecting, 78-79, 88, 178 topic bounding, 86
speaker, 43 topic change, 77-78
speaker state signal, 62 topic conflict, 78-79, 88
cues for speaker change, 61-62 topical coherence, 80-83
see also: turn-taking topic bounding, 86
speech act, 7, 9, 11-29, 39, 40, 41, see also: stories
 INDEX 195

transaction, 101, 102, 103, 106 utterance incompletor, 57

turn, 69, 70, 77, 92
turn-taking, 52-62,70, 79,93,105, Warnock, G. J., 13
113, 114, 166 Watson, O. M., 49
turn signal, 61-62 ways of speaking, 31
self-selection, 54, 56, 57 Weick K. E., 92
tying, 74-75, 77 Weisbrod, 61
Widdowson, H. G„ 9, 115, 141,
uptake, 20 146, 147-148, 150-153, 154
usage, 9 Wilkins, D., 8, 138, 139-140, 141,
use, 9 143
utterance, 8, 9, 10, 68, 100, 128 Winter, E. O., 152
mmm
mmm
mmw
,Longman
The new Applied Linguistics and Language Study series is intended to
reflect an increasing awareness by language teachers of the contribution
to classroom teaching of linguistics, sociolinguistics and psycho¬
linguistics. The volumes in the series cover areas of pedagogical
application of such theoretical and descriptive studies and thus define
more clearly the scope of applied linguistics and language study for
the more informed language teacher and student.

Discourse analysis is a comparatively new discipline which concerns

itself with the study of language function and supra-sentential structure.
This book traces the origins of discourse analysis in linguistics,
philosophy, sociology and anthropology, presents some of the major
findings and discusses some of the applications in the fields of language
teaching, psycholinguistics and stylistics.

9780582550872
10/03/2019 6 28-3