leech principles of pragmatics
DESCRIPTION
Leech principles of pragmaticsTRANSCRIPT
LONGMAN LINGUISTICS LIBRARY
General editors
R. H. Robins, University of London
G. N. Leech, University of Lancaster
TITLE NO 2 G
eneral linguistics A
n Introductory Survey Third Edition R. H. ROBINS
6 A Short History of Linguistics
Second Edition R. H. ROBINS
10 Studies in English Adverbial Usage SIDNEY GREENBAUM
12 Phonetics in Linguistics A book of R
eadings EDITED BY W
. E. JONES AND J. LAVER
13 Structural Aspects of Language Change JAM
ES M. ANDERSON
14 Philosophy and the Nature of
Language DAVID E. COOPER
15 Semantico-Syntax
FRANS LIEFRINK
16 From Signs to Propositions
The Concept of Form
in Eighteenth-C
entury Semantic
Theory STEPHEN K. LAND
17 Spatial and Temporal Uses of
English Prepositions A
n Essay in Stratificational Sem
antics DAVID C. BENNETI
18 The English Verb . F. R. PALM
ER
19 Principles of Firthian Linguistics T. F. M
ITCHELL
20 Problems in French Syntax
Transformational-G
enerative Studies NICOLAS RUW
ET TRANSLATED BY SHEILA M
. ROBINS
21 Text and Context
Explorations in Semantics
and Pragmatics of D
iscourse TEUN A. VAN DIJK
22 The Evolution of French Syntax A C
omparative APproach
MARTIN HARRIS
23 Modality and the English M
odals F. R. PALM
ER
24 Grim
m's G
randchildren C
urrent Topics in Germ
an Linguistics THOM
AS HERBST, DAVID HEATH HAN$-M
ARTIN DEDERDING
25 Explanation in Linguistics The logical problem
of language acquisition EDITED BY NORBERT HORNSTEIN AND DAVID LIGHTFOOT
26 Introduction to Text Linguistics ROBERT-ALAIN DE BEAUGRANDE AND W
OLFGANG ULRICH DRESSLER
21 Spoken Discourse A m
odel for analysis W
ILLIS J:DMONDSON
28 Psycholinguistics Language. m
ind. and world
DANNY D. STEINBERG
29 Dialectology
W. N. FRANCIS
30 Principles of Pragmatics
G.N.LEECH
Principles of Pragm
atics G
eoffrey N. Leech .
t>l=lt> Q
QJ:l
J:l c:=
:7 LO
NGM
AN LONDON AND NEW
YORK
LONGMAN GROUP LIMITED Longman House, Burnt Mill, Harlow
E8118x CM20 2JE, England Associated companies throughout the world
Published in the United Ststes of America by Longman Inc., New York
© Longman Group Limited 1983
AU 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 1983 ISBN 0 582 55110 2 Paper
British Library Cataloguing in Publication Data Leech. Geoffrey
Principles of pragmatics. - (Longman linguistics library; 30)
1. Pragmatics 2. Language - Philosophy I. Title
401 P99.4.P72 ISBN 0-582-55110-2
Library of Congress Cataloguing in Publication Data Leech, Geoffrey N.
Principles of pragmatics. (Longman linguistics library; no. 30)
Bibliography: p. Includes index. l
1. Pragmatics. I. Title. II. Series. P99.4.P72L43 1983 410 82-22850
ISBN 0-582-55110-2 (pbk.i J Set in 1 0/11 pt Linotron 202 Times
Printed In Singapore by Selector Printing Co Pte Ltd
To Tom and Camilla
Contents
Preface X
A note on symbols xiii
1 Introduction I I. I Historical preamble I 1.2 Semantics and pragmatics 5
1.2.1 An example: the Cooperative Principle of Grice 7
1.3 General pragmatics IO I ·4 Aspects of speech situations 13 I. 5 Rhetoric 15
2 A set of postulates 19 2.1 Semantic representation and pragmatic
I9 interpretation 2.2 Rules and principles 21" 2.3 Convention and motivation 24 2.4 The relation between sense and force 30 2.5 Pragmatics as problem-solving 35
2.5.1 The speaker's task, viewed in terms of analysis 36
2.5.2 The addressee's task, seen in terms of heuristic analysis 40
2.6 Conclusion 44
3 Formalism and functionalism 46 3.1 Formal and functional explanations 47 3.2 Biological, psychological, and social varieties of
functionalism 48
viii CONTENTS
3·3 The ideational, interpersonal, and textual functions of language 3. 3. I A process model of language 3.3.2 An illustration 3.3.3 The textual pragmatics
3·4 The ideational function: discreteness and determinacy
3·5 Examples of 'overgrammaticization' 3.6 Conclusion
4 The interpersonal role of the Cooperative Pri.ncip!e 4.1 The Cooperative Principle (CP) and the Politeness
Principle (PP) 4.2 Maxims of Quantity and Qualitr .
4.2.1 Implicatures connected wtth defimteness 4 3 Ma."im of Relation · . 4:4 The Hinting Strategy and anticipatory illocut10ns 4.5 Maxim of Manner .
4·5· r The obliquity and uninformatlveness of negation
5 The Tact Maxim 5.1 Varieties of illocutionary function 5.2 Searle's categories of illocutionary acts 5.3 Tact: one kind of politeness. 5 ·4 Pragmatic paradoxes .of pohteness . 5.5 Semantic declaratlves,
interrogatives and 1mperattves 5.6 The interpretation of impositives 5·7 Pragmatic scales . s.8 Tact and condescension
a A survey of the Interpersonal Rhetoric 6. I Maxims of politeness .
6. r. I The Generosity Max1m 6.1.2 The Approbation Maxim 6. r .3 The Modesty Maxim 6. I .4 Other maxims of
6.2 Metalinguistic aspects ot politeness 6.3 Irony and banter 6.4 Hyperbole and litotes 6.5 Conclusion
7 Communicative Grammar: an example
70 73 76
79
79 84 90 93 97 99
100
104 104 105 107 IIO.
II4 II9 123 127
131 131 133 135 136 138 139 142 145 149
CONTENTS
7. 1 Communicative Grammar and pragmatic force 7.2 Remarks on pragmatic metalanguage 7·3 Some aspects of negation and interrogation in
English 7·3·1 Syntax 7.3.2 Semantic analysis 7·3·3 Pragmatic analysis
7·3·3·I Positive propositions 7-3·3·2 Negative propositions 7·3·3·3 Ordinary yes-no questions 7·3·3·4 Loaded yes-no questions
7-4 Implicatures of politeness 7 ·5 Conclusion
8 Perlormatives 8. I The Pcrformative and Illocutionary-Verb Fallacies 8.2 The speech act theories of Austin and Searle
8.2. I Declarations 8.3 Illocutionary descriptive and
non -descriptive approaches 8.4 Illocutionary performatives and oratio obliqua 8.5 The pragmatics of illocutionary performatives 8.6 The perforrnative hypothesis 8. 7 The extended performative hypothesis 8.8 Conclusion
3 Speech-act verbs in English 9· r Locutionary, illocutionary, and perlocutionary 9.2 A survey of speech-act verb classes .
9.2. I Illocutionary and perlocutionary verbs 9.2.2 Classifying illocutionary verbs 9.2.3 Problems of classification and their solution 9.2-4 Phonically descriptive and
content-descriptive verbs 9·3 Is there a separate class of performative verbs? 9-4 A semantic analysis of some illocutionary verbs 9·5 Assertive verbs 9.6 Conclusion
·t 0 Retrospect and prospect
References
Index
I 57 I 57 I 59 164 !04 I65 I65 I66 169 171
174 !74 I75 179
I8I I84 189 192 I93 195
I98 199 203 203 205 207
212 2I3 2!6 223 225
229
234
243
Preface
Pragmatics can be usefully defined as the study of how utterances have meanings in situations. In this book I present a comp-!ementarist view of pragmatics within an overall programme £or studying language as a communication system. Briefly, this means studying the use of a language as distinct from, but wmp-!ementary to, the language itself seen z.s a formal system. Or more briefly still: grammar (in its broadest sense) must be separ-ated from pragmatics. To argue this, it is not sufficient to define pragmatics negatively, as that aspect of linguistic s1udy which can-not be accommodated in linguistics proper. Rather, onto: must det velcp theories and methods of description vihich are peculiar to pragmatics itself, and show thai these have to be different from those which are appropriate to The domain of prag-matic<; can then be defined so as to delimit it fr.::;m grammar, at tbe same tirne to shO"'.' how the two fields combine within an inte-grated framework for studying language.
Up to now, the :;trongest influences on those developing a pragmatic paradigm have been the formulation of a view of meaning in terms of illocutionary force by Austin and Searle, and of a view of meaning in tenns of conversational implicature by Grice. These have also been the strongest influences on the ideas I present here. But my approach to pragmatics is by way of the thesis that communication is problem-solving. A speaker, qua communicator, has to solve the problem: 'Given that I want to bring about such-and-such a result in the hearer's consciousness, what is the best way to accomplish this aim by using language?' For the hearer, there is another kind of problem to sol\1-e: 'Gi\'en that the speaker said such-and-such, what did the speaker mean me to understand by that?' This conception of communication
PREFACE xi
leads to_ a rhetorical. approach to pragmatics, whereby the speaker ts seen as trymg to achieve his aims within constraints imposed by principles and maxims of 'good communicative be-
In this not only Grice's Cooperative Principle, but other pnnc1ples such as those of and Irony play an important role. sum up: pragmatics dtffers from grammar in that it is
evaluative. I hope that, through this onentatwn, th1s book will help to bring about a new rapproche-ment L"Ctween grammar and rhetoric.
Chapter I sketches the historical and intellectual background to the present and proposes a set of postulates which are enlarged upon m Chapters 2 and 3· Chapter 3 advocates a com-bination _of and viewpoints in the philos-ophr of Chapter 4 begms a more descriptive part of the hnox, wh1ch develops the application of the maxims of Grice's Cooperative Principle within the more general framework of an Inte:Personal RheJoric. Chapters 5 and 6 concentrate on other n:ru:1ms of the Rhetoric, notably maxims of polite-
framework whtch has been elaborated in these three chapters 1s then put to descriptive use in an account of how a lim-ited of grammar - the grammar of certain negative
sentence types - is pragmatically implemented m Enghsh. The title 'Communicative Grammar' has been reason-<ibly appl!ed to a linguistic description which, like this one, relates
forms to their various pragmatic utilizations. After UlS demonstratwn, Chapters 8 and 9 return to more
They that :'1 rhetorical view of prag-!flatlcs. reqmres us to take a dtfferent vtew of perfonnatives and of dlocutwnary acts from that which is familiar in the 'classical'
formulations of Austin and Searle. The view is put for-ware. that Searle's taxonomy of illocutionary acts should be rein-terpreted as a semantic taxonomy of speech-act verbs.
I have benefited from many discussions of these issues with colleagues and with audiences both in Britain and overseas. I am particularly grateful to a group of postgraduate students at Lan-caster who discussed the first draft of this book with me: they
Susan Andrew McNab, Dilys Thorp, and last out not least, Jenmfer Thomas, to whom I am also indebted for subsequent discussions and criticisms. A colleague at Lancaster, R: L. V. done me the favour of casting a searching but fnendly pht_losophtcal eye over several chapters, and suggesting a
of My co-editor of the Longman Linguis-!1C'> Ltbrary,. Professor R. H. Robins, has also kindly given me the oenefit of h1s c.omments on the final manuscript. The customary
xii PREFACE
disclaimer that I alone am responsible for the shortcomings of this book i! particularly appropriate here:· on a subject so controversial u the present one, even my most benevolent critics have found -: and no doubt will find - enough cause for disagreement.
I acknowledge with thar.ks the permission granted by John Benlamins, 8, V., to reprint, as part of Chapter 7 of the present work, part of the paper 'Pragmatics and conver-sational rhetoric' which I contributed to Herman Parret, Marina Sbisa; eds, Possibilities and Limitatiorns of Prqgrtu!tics_, Benjamins, 1981.
Umrn,?ity;of Lancasfef' Mayi<¥12 G.N.L.
A note on symbols
The symbols s and h are used throughout the book to symbolize 'speaker(s) or writer(s)' and 'hearer(s) or reader(s)' respectively. A subscript added to one of these symbols indicates that the son referred to is a participant in -the primary speech situation, secondary speech situation, etc, FoE' example, s1 means 'primacy speaker', s2 means 'secondary !Speake('.
The symbols t s, t h, ! s, and interpreted as follows: t s = 'desirable for the speaker' "! h = 'desirable for the addressee'
s = 'undesirable for the speaker' · lll = 'undesirable for the addressee'
Additional abbreviations are: CP = 'Cooperative Principle' PP = 'Politeness P"rinciple' IP = 'Irony Principle'
Chapter 1
Introduction The aim of science is increase of verisimilitude
[Karl R. Popper. Objective Knowledge, p. 71]
In a broad sense, this book is about the nature of human lan-guage. In a narrower sense, it is about one aspect of human language, which I believe is important for understanding human language as a whole. This aspect I shall call GENER.i<,.L PRAGMATICS.
1.1 Historical preamble The subject of 'pragmatics' is very familiar in linguistics today. Fifteen years ago it was mentioned by linguists rarely, if at all. In those far-off-seeming days, pragmatics tended to be treated as a rag-bag into which•recalcitr!ll..t data could be conveniently stuffed, and where it could be equally Conveniently forgotten. Now, many would argue, as I do, that we cannot really understand the nature of language itself unless we tanguage is used in communication:
How has this change come a6'out?1 In part, the whole of the recent history of linguistics can be described in terms of successive discoveries that what has gone headlong into the rag-bag can be taken out again and sewed and patched into a more or less pre-sentable suit of clothes. To the generation which followed Bloomfield, linguistics meant phonetics, phonen-Jcs, and if one was daring-morphophonemics; but syntax was considered so ab-stract as to be virtually beyond the horizon of discovery AU this changed after Chomsky, in the later 1950s, discovered the cen-trality of syntax; but like the structuralists, he still regarded meaning as altogether too messy for serious contemplation. In the earlier 196os (for by this time the pace of linguistic advance had quickened) Katz and his collaborators (Katz and Fodor 1963; Katz and Postal 1g64; Katz 1964) began to find out how to in-
2 INTRODUCTION
corporate meaning into a formal linguistic theory, and it was not long before the 'California or bust' spirit led to a colonization of pragmatics. Lakoff, with others, was soon arguing (1971) that syntax could not be legitimately separated from the study of lan-guage use. So pragmatics was henceforth on the linguistic map. Its colonization was only the last stage of a wave-by-wave expan-sion of linguistics from a narrow discipline dealing with the physical data of speech, to a broad discipline taking in form, meaning, and context.
But this is only part of the story. First, all the names men-tioned in the preceding paragraph are American, for it describes the progress of mainstream American linguistics. It is probably more true of linguistics than of other subjects that its dominating influences have been American; but we should not forget that many influential scholars, both in the USA and elsewhere, have continued to work outside the 'American mainstream'. We should not overlook independent thinkers such as Firth, with his early emphasis on the situational study of meaning, and Halliday, with his comprehensive social theory of language. And equally important, we should not overlook the influences of philosophy. · When linguistic pioneers such as Ross and 4koff staked a cla!m in pragmatics in the late 196os, they encountered m-digenous breed of philosophers of language who had been qwetly cultivating the territory for some time. In fact, the more lasting influences on modern pragmatics have been those of philosophers; notably, in recent years, Austin (1962), Searle (1969), and Grice (1975).
The widening scope of linguistics involved a change in the view of what language is, and how linguistics should define its subject. The American structuralists were happiest with the idea that lin-guistics was a physical science, and therefore did their best to rid the subject of appeals to meaning. 2 But by accepting ambiguity and synonymy as among the basic data of linguistics, Chomsky opened a door for semantics. Subsequently, Chomsky's dis-affected pupils in the generative semantics school went a stage further in taking semantics to be base for their linguistic theories. But once meaning has been admitted to a central place in lan-guage, it is notoriously difficult to exclude the way meaning varies from context to context, and so semantics spills over into prag-matics. In no time the generative se!fianticls!s found they had bit-ten off more than they could chew. There is a.justifiabte tendency in scientific thought to assume that an existing theozy or paradigm works until it is shown to fail. On this basis, the generative semanticists tried to apply the paradigm of generative grammar
HISTORICAL PREAMBLE 3
to problems - such as the treatment of presuppositions and .of illocutionary force - which most people would. now regard as volving pragmatics. The attempt faded: not m the way in which theories are supposed to on of a cructal falsifying observation, but 10 the way m whtch to happen in linguistics, through a slowly accumulatmg wetght of adverse arguments.3 .
I should explain that I am using the term PARADIGM roughly m Kuhn's sense, not as a synonym for 'theory', but a gen-eral tCffm to set of background assumptl(),!!§ makes about. ...
digm term 'generative grammar' in pract1ce refers to a whole. set of theories which share certain assumptions: that language IS a mental phenomenon, that it can be studied. through the algor-ithmic specifications of rules operating according to certain con-ventions, that the data for such theories are available through intuition, that languages consist of sets of sentences, etc.
While the generative semanticists were exploring the outer limits of this paradigm in semantics and pragmatics, Chomsky himself, with others of similar views, was interested in a narrower definition of the scope of this paradigm, that of the so-called Ex-tended Standard Theory, which then evolved into a narrower Re-vised Extended Standard Theory. These versions of generative grammar have maintained the of syntax; has been relegated to a peripheral position 10 the model, and nas to some extent been abandoned altogether.5 Pragmatics does not enter into the model at all, and indeed Chomsky has strongly maintained the independence of a grammar, as a theory of a 'mental organ' or 'mental faculty', from consideration of the use and functions of language.6 · . • • • •
This more limited definition of the scope of hngutsttc theory ts, in Chomsky's own terminology, a 'competence' theory rather than a 'performance' theory. It has the advantage of maintaining the integrity of linguistics, as within a walled city, away from con-taminating influences of use and .. Bu! many. grave doubts about the narrowness of thiS paradtgm s definttiOn of lan-guage, and about the. high degree of abstraction and idealization of data which it requires. .
One result of this limitation of generative grammar to a stnct fonnalism is that, since about 1970, it has been progressively losing its position as the dominant paradigm of More and more linguists have found their imagination ad intellect eli-
4 INTRODUCTION
gaged by approaches more than those allowed for in generative grammar. These approaches do not yet add up to an integrated paradigm for research, but they have had the effect collectively of undermining the paradigm of Chomsky. Socio-
has entailed a rejection of Chomsky's abstraction of the 'ideal native speaker/hearer'. and artificial
place emphasis on a 'process' model of human Iaii-guage abilities, at the expense of Chomsky's disassociation of linguistic theory from psychological process. I.ext linguistics and
analysis. have refused to accept the limitation of linguis-tics to sentence grammar. Conversational ·analysis has stressed the primacy of the social dimension of language study. To these developments may be added the attention that pragmatics - the main subject of this book - has given to meaning in use, rather· than meaning in the abstract.
Cumulatively these approaches, and others, have led to a re-markable shift of direction within linguistics away from 'com-petence' and towards 'pedormance'. This shift is welcome from many points of view, but the resulting pluralism has meant that no comprehensive paradigm has yet emerged as a successor to generative grammar. A unified account of what language is has, I believe, been lost. Hence the purpose of this book is to argue in favour of a fresh paradigm. This does not mean that the ideas I shall present are highly original: paradigms emerge over a period · and decay over a period, and the ideas put forward here have seemed to me to be 'in the air' il} a way which makes it difficult to pin down their origin in particular authors. 7 Neither will this book attempt an overall account of language: instead, it will con-centrate on arguing the validity of a particular view of the dis-tinction between grammar and pragmatics. This argument, however, will have fundamental implications for the way one looks at language. In essence; the claim will be that grammar (the ab-
formal system of language) and pragmatics (the principles of hnguage use) are complementary domains within linguistics. We cannot understand the nature of language without studying both these domains, aod the interaction between them. The con-sequences of this view include an affirmation of the centrality of formal linguistics in the sense of Chomsky's 'competence', but a
. recognition that this must be fitted into, and made answerable to, a more comprehensive framework which combines functional with formal explanations.
At this point, I shall merely state the major postulates of this 'formal-:functional' paradigm. In the next chapter I shall examine them and argue their prima-facie plausibility; in the remaining
SEMANTICS AND PRAGMATICS 5 chapters, I shall try to justify them in more detail tbrough analy-sis of particular descriptive problems. The postulates are:
PI: The semantic representation (or logical form) of a sentence is distinct from its pragmatic interpretation.
P2: Semantics is rule-governed (= grammatical); general prag-matics is principle-controlled ( = rhetorical). .
P3: The rules of grammar are fundamentally conventional; the principles of general pragmatics are fundamentally non-conventional, ie motivated in terms of conversational goals.
P4: General pragmatics relates the sense (or grammatical mean-ing) of an utterance to its pragmatic (or illocutionary) force. This relationship may be relatively direct or indirect.
Ps: Grammatical correspondences are defined by mappings; pragmatic correspondences are defined by problems and their solutions.
P6: Grammatical explanations are primarily formal; pragmatic explanations are primarilyiunctional.
P7: Grammar is ideational; pragmatics is interpersonal and textual.
P8: In general, granu,:tar is describable in terms of discrete and determinate categories; pragmatics is describable in terms of continuous and indeterminate values.
The effect of these postulates is to define two separate domains, and two separate paradigms of research, making up a single 'com-plex' paradigm for linguistics. Arguments in favour of this para-digm are based on the simplicity and naturalness of the expla-nations it offers. There is no clear way of testing the validity of scientific paradigms: they exist on a more abstract plane than the scientific method which Popper described as 'the method of bold conjectures and ingenious and severe attempts to refute them'. Nevertheless, by exploring, formulating, and refining paradigms of research, we are determining the background assumptions on which the search for truth abOut language will proceed with in-creased understanding.
1.2 Semantics and pragmatics In practice, the problem of distinguishing 'language' (langue) and 'language use' (parole) has centred on a boundary dispute be-tween semantics and pragmatics. Both fields are concerned with meaning, but the difference between them .can be traced to two different uses of the verb to mean:
6 INTRODUC110N
[I] What does X mean? [ 2] What did you mean by X? Semantics traditionally deals with meamng as a dyadic relation, as m [I 1: while pragmatics deals with meamng as a tnadic rela-tion, as m [2]. Thus meanmg m pragmatics ts defined relative to a speaker or user of the language, whereas meanmg m semantics ts defined purely as a property of expresstons m a gtven language, m abstraction from particular sttuations, speakers, or bearers. Tbts IS a rough-and-ready distinction whtch bas been refined, for par-ticular purposes, by philosQphers such as Mortis (1938, 1946) or Camap ·(1942).8 I shall redefine pragmatics for the purposes of lingwstics, as the study of meamng m relation to speech Situations (see 1.4 below).
The vtew that semantics and pragmatics are distinct, though complementary and mterrelated fields of study, ts easy to appreCI-ate subjectively, but is more difficult to JUStify m an obJective way. It is best supported negatively, by pomting out the failures or weaknesses of alternative vtews. two clear alterna-tives are eossible: it may be clatmed that the uses of meanmg shown in l I] and [ 2] are both the concern of semantics; or that they are both the concern of pragmatics. The three vtews I have now mentioned may be diagrammed and labelled as shown m Fig. I.I.
Semantics f r---------. I I
Pragmatics Pragmatics
'Semantiasm' 'Complementansm' 'Pragmatiasm'
FIGURE I.l
Because of difficulties of termmology and definition, it 15 hard to pm down clear cases of semantiasm and pragmatici.sm. In prac· tice, one notices a preference of a semantic a pragmatic one, or VIce In a modiied thexefore, the labels ·•semarttiast' and may ·Qo · ·• ·tO who as much of;tllf: iif. · as
SEMANTICS AND PRAGMATICS 7 of abstract mental entities such as concepts, and who have m one way or another assimilated semantics to pragmatics. For example, Searle (I969:17) argues for an approach wbtcb vtews the theory of meam.ng (and m fact the whole of language) as a sub-part of a theory of action; thus meanmg m defined m terms of what speech acts speakers perform relative to hearers. On the other hand m generative semantics m the earlier I97QS, there was an effort to assmrilate pragmatics to semantics, particularly by argumg for the PERFORMATIVE HYPOTHESIS (Ross 1970), m terms of Wbtcb a sen-tence, m its deep structure or semantic representation, is a per-formative sentence such as I state tQ you that X, I order you to Y In this way, ·the illocutionary or pragmatic force of an utterance was encapsulated n'l its semantic structure. 9
The two opposed positions of Searle (1969) and Ross (1970) appear to be very close together, because of the significance they both attach to performative sentences (see 8.2, 8.6). But m fact, they are at opposite poles, as one can see by reading Searle's critique of the perfonnative hypothesiS (Searle 1979:162-·79). The contrast can also be studied. in two contrasting approaches to mdirect illocutions such as Ctm. you pass the salt: the approach taken by Searle (1979 [I915b]:30-57), and that taken by Sadock (1974, esp. 73-:95).
The third vtewpomt, that of complementmsm, tS the one I shall support. The arguments for thts pasition v.ill take the fol-
form. Any account of ,meamng m language must (a) be faithful to the· facts as we observe them, and (b) must be as sim-ple and generalizable as possible. If we approach meanmg entire-ly from a pragmatic pomt of view, or entirely from a semantic pomt of vtew, these requrrements are not met; however, if we approach meamng from a pomt of vrew winch combmes seman-tics and pragmatics, the result can be a satisfactory explanation m terms of these two criteria.
8 INTRODUCTION
ments of this book, I cite it again here. Under this principle, four categories of MAXIMS are distinguished:
The Cooperative Principle (abbreviated to CP)
QUANTITY: Give the right amount of information: ie I. Make your contribution as informative as is required. 2. Do not make your contribution more informative than is
required.
QUALITY: Try to make your contribution one that is true: ie 1. Do not say what you believe to be false. 2. Do not say that for which you lack adequate evidence. RELATION: Be relevant.
MANNER: Be perspicuous; ie I. A void obscurity of expression. 2. Avoid ambiguity. 3. Be brief (avoid unnecessary prolixity). 4· Be orderly.
(Adapted from Grice 1975)
The kind of constraint on linguistic behaviour exemplified by Grice's CP differs from the kind of rule normally formulated in linguistics, or for that matter, in logic, in a nu.nber of ways (see below, 2.2). (I \shall not, for the present, distinguish between 'principles' and 'maxims', since the latter are simply, according to Grice's usage, a special manifestation of the former). (a) Principles/maxims apply variably to different contexts of lan-
guage use. . (b) Principles/maxims apply in variable degrees, rather than in an
ali-or-nothing way. (c) Principles/maxims can conflict with one another. (d) Principles/maxims can be contravened without abnegation of
the kind of activity which they control. The last of these statements amounts to a claim that, in Searle's-terminology (1969:33ff), conversational principles and maxims are 'regulative' rather than 'constitutive'. The rules of a language (eg rules for in English) normally as an Integral part of the definition of that language, but maxuns do not. Hence if one tells a lie in English, one breaks one of Grice's maxims (a Maxim of Quality); but this does not mean that one any way to speak the English language. In fact, it has been c · ed that part of the definition of human languages is
SEMANTICS AND PRAGMATICS 9
that they can be used to deceive or misinform (see Lyons 1977:83 -4; Thorpe I972:33). On the other hand, if one breaks the rule for tag-questions by saying We've met before, isn't it? rather than We've met before, haven't we? one thereby fails in some particular to speak the English language.
One element, although it is part of the everyday interpretation of the terms 'principle' and 'maxim', has been carefully omitted from the above definition. This is the implication that such con-straints are of a moral or ethical nature. The requirement to tell the truth might, indeed, be regarded as a moral imperative; but the reason for including it in a scientific account of language is de-scriptive rather than prescriptive. The maxims form a necessary part of the description of ling\listic meaning in that they explain how it is that speakers often 'mean more than they say': an ex-planation which, in Grice's terms, is made by means of pragmatic implications called CONVERSATIONAL IMPLICATURES. For exam-ple, in strictly logical terms,
[3] Many of the delegates opposed the motion is not inconsistent with the proposition that
[4] All the delegates opposed the motion. In most contexts, however, it will normally be interpreted as ex-cluding that possibility, on the grounds that if the speaker knew that all of the delegates opposed the motion, the first Maxim of Quantity ('Make your contribution as informative as is required') would have obliged him to be informative enough to say so. In this sense, (3] CONVERSATIONALLY IMPLICATES [5):
[5] Not all the delegates opposed the motion. But the implicature only holds if certain premisses or 'enabling assumptions' hold. These include first the assumption (which may or may not be supported by cont;;:xtual evidence) that the speaker knows whether all the delegrnl!s opposed the motion, and second-ly the assumption that th:.. speaker is observing the CP. It is quite open to the speaker to opt out of the CP, for example purpose of wantonly deceiving the hearer. <?ne can tell bes 1!-1 English or in any other language, but the potnt about the CP ts that if speakers told lies randomly and indiscriminately, we should no longer be able to oommunicate by means of language.
In saying that people normally follow the CP, then, one is by no means taking a moral stance. But one thing that cannot be denied is that principles introduce communicative values, such as truth-fulness, into the study of language. Traditionally, linguists have
10 INTRODUCTION
avoided referring to such values, feeling that they undermine one's claim for objectivity. But so long as the values we consider are ones we suppose, on empirical grounds, to be operative in society, rather than ones we impose on society, then there is no reason to exclude them from our inquiry. ·
The distinction I have just drawn between 'principles'/'maxims' and 'rules' will be developed and modified as this book proceeds. As it stands, it is too absolute. Rules are not always so in their application as has been implied, nor can the comp· lementary roles of semantics and pragmatics be made so precise as it appears in examples [3] and [4]; but the CP is exemplary in that it shows a division of labour between the SENSE of an utter-ance and its FORCE. Taking this simple example, we may see in it the semantics/pragmatics distinction as a special case of Sans-sure's distinction between langue and parole ( 1959 [ 1916]: II-13), or Chomsky's distinction between competence and performance (1965:3-9). Both these distinctions, however, have been used to define language, for the purposes of linguistics, in an abstract way, effectually excluding the data of language use from con-sideration. For this reason and for others, I shall avoid these fam-iliar dichotomies; instead, I shall refer to the formal language system as GRAMMAR, in opposition to PRAGMATics, which may also, in a more general sense, be regarded as part of the domain of linguistics. This use of grammar, although broad, actually corre-sponds to a current use of the term to mean the study of the whole language as in 'transformational grammar'.
1.3 General pragmatics I have mentioned that my principal subject in this book is GENERAL PRAGMATICS. By this term I mean to distinguish the study of the general conditions of the communicative use of lan-guage, and to exclude more specific 'locp.l' conditions on language use. The latter may be said to belong to :a less abstract field of soCIO·PRAGMATICS, for it is clear that the Cooperative Principle
the Politeness Principle operate variably in different cultures Oi' communities, in different social situations, among different social classes, etc. 11 One has only to think. of the school-boy taboo against'telling (ie the inopportune telling of the truth!); or the way in which politeness is differently interpreted in {say). Chinese, Indian, or American societies, to realize that pragmatic descriptions ultimately have to be relative to specific SQcial conditions. In other words, socio-pragmatics is the socio-logical interface of pragmatics. Much of the work which has taken
GENERAl. PRAGMA11CS II
place in conversational analysis has been limited in this sense, and has been closely bound to local conversational data. 12 The term PRAGMALINGUISTICS, on the other ·hand, can be applied to the study of the· more linguistic end of pragmatics - where we con-sider the particular resources which a given language provides for conveying particular illocutions13 (Fig. I .2).
General pragmatics
I [Grammar J Pragmalinguistics
f ted to i L-------------1
i Socio-pragmatics (Sociology]
l_-- J FIGURE 1.2
In contrast, general pragmatics, as studied here, is a fairly ab-stract study. Of course, we do need detailed pragmalinguistic studies which are language-specific, and detailed socio-pragmatic studies which are culture-specific. At the same time, we also need studies at the more general level as a necessary stage of abstrac-tion between the study of language in total abstraction from situ-ati.on, and the study of more socially specialized uses of language.
My definition of 'general pragmatics' will be further restricted to the study of linguistic communication in terms of conversational principles. It will be limited, that is, to a RHETORICAL model of pragmatics. This means that certain topics which may justly be considered part of pragmatics will be put into the background. First, I shall take little account of what Grice has called CONVENTIONAL IMPLICATUitES, ie pragmatic implications which are derived directly from the meanings of words, rather than via conversational principles.14 (For example, in the sentence She was poor, but she was honest, the word but carries the implicature that for a person to be poor is a good reason for supposing him not to be honest. The meanings of 'pragmatic particles' such as de in Classical Greek, ja in German, and sita in Finnish may be placed in the same category. )15 Another exclusion is the tudinal function of intonation, and of non-verbal commum-cation through gestures and paralanguage. More relevant, but still peripheral to my present concern, is the study of what may be· callea REFERENTIAL PRAGMATICS, ie the assignment Of reference to referential expressions in a given utterance: these include in-dexical elements such as personal pronouns and the tense of the verb.
One further field which might be included in pragmatics, and
I2 !NTRODUCflON
which is only touched on here, is the study of relatively perma-nent parameters of situation in to language choice, such as those which have _been included by Halliday (eg 1978) and o·;.hers under the headmg of REGISTER, and which still others have
the heading of STYLE (Crystal and Davy 19f>9). Tne between pragmatics and register corrresponds to on;;:
that has drawn by Argyle and Dean (1965) for non-verbal commumcatwn, between DYNAMIC and STANDING features of communication. That is, there are some features which tend to undergo. and modification during discourse (such as i!locutwnary force m Austin's sense, 1962:roo); but there are also other features, such as formality of style, which tend tc remain stable over fairly long stretches of time. It is nevertheless not always easy to separate these two types of condition. Polite· ness, for instance, is often a function of both: standing features such the social distance between participants interact with dynam1c features such as the kind of illocutionary demand the speaker is making on the hearer (request, advice, command, etc.) to produce a degree of poHt-::ness appropriate to the situation (see 5-7).
narrowed.the field in this way, I shall now present a dia-gram (Ftg. destgned to capture the distinction implied be· tween semantics (as part of the grammar) and general pragmatics (as of the use of the I am taking for granted here a famthar and well-established tripartite model of the language system (grammar), consisting of semantics, syntax, and pho-nology. These levels can be regarded as three successive coding systems ":hereby 'sense' is converted into 'sound' for the purposes of encodmg a message (PRODUCTION), or whereby 'sound' is converted into 'sen.se' for the purposes of decoding one
Figure L3 shows that the grammar interacts With pragmatics via semantics. This view, although a useful start-
Phonology
Syntax
s{mantirs
I + Pragmatics FIGURE 1.3
I 1
Grammar
I j
AS?ECIS OF SPEECH SITUATIONS
ing-point, is not the whole story; we may note, as an exception, that pragmatically related aspects of phonology ( eg the polite use of a rising tone) interact directly with pragmatics, rather than in-directty, via syntax and semantics.
.4 Aspects of speech situations The question inevitably arises: how do we know we are dealing with pragmatic, rather than with semantic phenomena? Since pragmatics studies meaning in relation to speech situation, refer-ence to one or more of the following aspects of the speech situ-ation will be a criterion.
(i) Addressers or addressees Following the practice of Searle and others, I shall refer to addressers and addressees, as a matter of convenience, as s ('speaker') and h ('hearer'). These will be a shorth:md for 'speaker(s)iwriter(s)' and 'hearer(s)ireader(s)'. Thus the use of the abbreviations s and h does not restrict pragmatics to the spoken language. A significant distinction can be made (cf Lyons 1977:34) between a receiver (a person who receives and inter-prets the message) and an addressee (a person who is an intended receiver of the message). A receiver, that is, might be a by-stander or an eavesdropper, rather than an addressee. This dis-tinction is relevant to the present inquiry, in that the analyst of pragmatic meaning is best thought of as a receiver: a proverbial 'fly on the wall' who tries to make sense of the content of a dis-course according to whatever contextual evidence is available. The use of the symbol h, however, will always signify one or more addressees, or persons to whom the utterance is addressed bys. ·
(ii) The context of an utterance CONTEXT has been understood in various ways, for example to in-clude 'relevant' aspects of the physical or social setting of an utterance. I shall consider context to be any background knowl-edge assumed to be shared by s and h and which contributes to h's interpretation of what s means by a given ·utterance.
(iii) The goal(s) of an utterance I shall often find it useful to talk of a goal or function of an utterance, in preference to talking about its intended meaning, or s's intention in uttering it (see further 2:3·3.1). The term goal is more neutral than intention, because it does not commit its user
14 INTRODUCTION
to dealing with conscious volition or motivation, but can be used generally of goal-oriented activities, The term intention can be misleading on this score.
(iv) The utterance as a form of act or activity: a speech act Whereas grammar deals with abstract static entities such as sen-tences (in syntax) and propositions (in semantics), pragmatics deals with verbal acts or performances which take place in par-ticular situations, in time. In this respect, pragmatics deals with language at a more. concrete level than grammar.
( v) The utterance as a product of a verbal act There is another sense in which the word 'utterance' can be used in pragmatics: it can refer to the product of a verbal act, rather than to the verbal act itself. For instance, the words Would you please be quiet?, spoken with a polite rising intonation, might be described as a sentence, or as a question, or as a request. How-ever, it is convenient to reserve terms like sentence and question for grammatical entities derived from the language system, and to reserve the term utterance for instances of such entities, identified by their use in a particular situation. Hence an utterance may be a sentence-instance, or sentence-token; but strictly speaking, it cannot be a sentence. In this second sense, utterances are the el-ements whose meaning we study in pragmatics. In fact, we can correctly describe pragmatics as dealing with utterance meaning, and semantics as dealing with sentence meaning. However, there is no need to assume that aU utterances are sentence-tokens. We may wish to isolate as an utterance a piece of language which is either too short or too long to be classified as a single sentence.
The meaning of utterance in (iv) and the meaning of in (v) can be easily confused: there is a difference, but not a particu-larly marked one, between describing Would you please be quiet? as an utterance (as in (v) ), and describing the act of uttering Would you please be quiet? as an utterance (as in (iv)).16 Fortu-nately, the confusion can be alleviated, since it is generally con-venient to say that 'utterance' in the sense of (iv) corresponds to 'speech act', or more precisely to ILLOCUTIONARY ACT, in the sense of that term employed by Austin (I9()2:10o). This means we can use illocutionary act or illocution for the utterance-action as described in (iv), and can keep the term utterance for the lin-guistic product of that act When we try to work out the meaning of an utterance, this can be thought of as an attempt to recon-struct what act, considered as a goal-directed communication, was it a goal of the speaker to perform in producing the utterance.
RHETORIC
Thus the meaning of an utterance, in this sense, can be called its LOCUTIONARY FORCE. (Austin in fact distinguished illocutionary
:Cts from other kinds of acts, notably locutionary and per!ocu-tionary acts. But the other kinds of act !urther 9.1) ran be: largely discounted in an ac.count of pragmahcs). __
From the above-mentioned elements of (I) addresser and addressee, (ii) context, (iii) goals, (iv) iliocutionary act, and (v) utterance we can compose a notion of a SPEECH SITUATiON, cmnprisin'g aU these elements, and perhaps other elements .as w"'H such as the time and the place of the utterance. Pragmatics is from semantics in being concerned with meaning in relation to a speech situation.
1.5 Rhetoric Earlier I characterized the present approach to pragmatics as 'rhetorical'. This use of the term 'rhetorical' is very traditional,
.· referring to the study of the effective use of language com-··· munication. But whereas rhetoric has been understood, m par-
ticular historical traditions, as the art of using language skilfully for persuasion, or for literary expression, or public speaking, I have in mind the effective use of language m Its most general sense, applying it primarily tp conversation, and ',?!llY secondarily to more prepareJ and pubhc uses of 1he point about the term rhetoric;, in is focus It on a goal-oriented speech s1tuat1on, m whtch s uses language m order to produce a particular effect in the mind of h.
I shall also use the term RHETORIC as a countable noun, for a set of conversational principles which are related by. their tions. Using a distinction familiar in the work of Halliday, I snail distinguish tWO rhetorics, the INTERPERSONAL and the TEXTlJ:"'L rhetoric (Fig. 1.4) (see also 3.3). Each of. the two (whose functions will be explained later) conststs of a set of pnn-ciples, such as the and the PP already The prin.ci-ples, in tum, constst of a set of mrunms, m accordance ,w1th Grice's terminology. Grice's 'maxim', however, I shall call a sub-maxim', thereby introducing another level _into the hierar.chy. But I do not wish to insist too rigidly on th1s four-level hierarchy, since it is not always clear to what level a given precept belongs. For example, of Grice's two Maxims of qualitr (which ,I call sub-maxims), the second seems to be a predtctab!e extenston of the first:
Maxim I: Do not say what you believe to be false.
I6 INTRODUCTION
Maxim of Tact
Inter- /Maxim of Generosity personal ---Politeness rhetoric Principle of Approbation
Textual rhetoric
FIGURE 1.4
Maxim of Modesty
\Irony Principle < .... . -.... .
\ ......... . _,./End-focus Maxim
Processibility ?-E d 'gh M . Principle ----...._-- n -wei t . ax1m
Clarity Principle <:::::::: : : Economy Principle <::::::: : : : (maxims)
Expressivity Principle <::,: : ·.
Maxim 2: Do not say that for which you lack adequate evidence.
If we say something for which we lack adequate evidence, we do not know whether what we say is true or false. Therefore Maxim 2 simply says 'Do not put yourself in a position where you risk breaking Maxim I'; and both can be summarized in the precept 'Avoid telling untruths'. The taxonomic lay-out of Fig. 1.4 is merely a way of sketching out in advance some of the territory to be covered in this book (especially in Chs 4-6). It is not meant to be definitive.
The rhetorical principles socially constrain communicative be-haviour in various ways, but they do not (except in the case of 'purely social' utterances such as greetings and thanks) provide
NOTES I7
the main motivation for talking. Cooperation and politeness, for instance, are largely regulative factors which ensure thaL. once conversation is under way, it win not follow a fruitless or di;;rup-dve p;:;.th. It is therefore necessary to distinguish between iHocu· tionary goals a.'ld social goals, or equivalently between the iilocu-tionary force of an utterance and its RHETORICAL FORCE- ie meaning it conveys regarding s's adherence to rhetorical dples (eg how far sis being truthful, polite, ironic). Together, the iHocntio:1ary force and the rh•etorical force of an utterance mak'e up its PRAGMATIC FORCE.
The distinction between SENSE (meaning as semantically deter-mined) and FORCE (meaning as pragmatically, as well as seman-tically determined) is essential to this study. But it is also essential to realize the bond between the two: force includes sense, and is also pragmatically derivable from h in a wwy to be •elucidated Chapter 2.
Notes· I. See Newmeyer (19&P) for a readable book-length account of the
events to which this paragraph alludes. 2. Bloomfield's famous dismissal of meaning as the 'weak point in l&n-
guage study' (!933/35:140) was followed by attempts to exclude meaning altogether from the techniques of iinguistics - see especia!iy Harris (1951:7).
3· The rise and decline of the performativt: hypothesis (see 8.6) is par-ticularly instructive from this point of view. The controversy over pre-suppositions also raged through the I970s, with Wilson (1975), Kempson (1975), a.nd Gazdar (1979) against the logical (ie semanti-cist) account of presupposition which was current in linguistics in the late 196os.
4· Kuhn's 'paradigm' (11}62), and Lakatos's 'research 1978) are in this respect more illuminating, as a scientific
model for linguistics, than the doctrine of the cn1cial experiment. 5· See Chomsky (1976:Jo6), and Chomsky and Ronat (1979:56-7). 6. See Chomsky's controversy with Searle on the functional ba<>is of
language (1976:ss-77). Among many influences and pan•.Hel developments, I should men-tion the authors cited in note 3 above, as well as Harni§h (1976) Holdcroft (1978), Bach and Harnish (1979), Giv6n (1979), and Edmondson (1981).
8. For an account of these earlier philosophical approaches, see Lyons (197PI5).
9· The generative semantics approach did not distinguish betweea the deep syntactic structure and the semantic structure of a sentence. See especially McCawley (r¢8) and Lakoff (1971).
18 INTRODUCTION
10. Grice (1915:47) himself notes the importance of politeness as a fac-tor in account of conversational meaning. Earlier accounts of politeness in terms of rhetorical principles and maxims are to be found in Leech (198o:9-30, 79-n6).
1 I. Thus socio-pragmatics would involve the assignment of variar:t values tq the principles and maxims. But are the CP and the PP um-versal? It would be surprising if they did not operate in some form or other in all societies. Counter-examples have been reported - eg Keenan (1976) cites the case of .Malagasy speakers disregard the Maxim of Quant1.ty) - but as Harrush out (1976:J4o, note 29), these are not true smce there is no claim that the CP is observed categoncally.
12. Eg in the work of Gumperz, Schegloff and Sacks, Labov and FansheL A useful survey of research in this area is Corsaro (1981).
13. I am indebted to Thomas (1981) for formulating the tic/socio-pragmatic distinction (also Thomas, forthcommg).
14. Grice (1975:44). For further discussions of conventional implicature, see Hamish (I976:331-340) and Karttunen and Peters (1979).
15. On ja, doch and other 'modal particles' in German, see Bublitz (1978); on sitii, see Hakulinen (1975).
16. The two senses of utterance are distinguished in French by the terms enonciation and enonce. For discussion, see Lyons (1977:26).
Chapter2
A set of postulates The most useful thing about a principle is that it can always be sacrificed to expediency
[Somerset Maugham, The Circle, Act ll
To darify the distinction that is being proposed between seman-tics and pragmatics, I shall give in this and the following chapters some explanation of the postulates listed on page 5. These pos-tulates will be repeated in small capitals under· each heading of this chapter. Although I shaH argue that the postulates have prima-facie plausibility, more detailed arguments in their support will be presented later, in the descriptive chapters.
2.1 Semantic representation and pragmatic interpretation PI: THE SEMANTIC REPRESENTATION (OR LOGICAL FORM) OF AN
UTTERANCE IS DISTINCT FROM ITS PRAGMATIC INTER-PRETATION.
My face-value reason for accepting this complementarist view (see p 6) is that contrary views have in the past led to implaus-ible, if not absurd, accounts of language.
Let us look first at an example of implausibility in the SEMANTICIST position: an attempt to assimilate pragmatic phenomena to semantics. According to the PERFORMATIVE HYPOTHESIS put forward by the generative semantics school in the early 1970s (see further 8.6), every sentences Sin a language is in its deep or semantic structure a performative sentence roughly of the form I state/declare/ask/etc. S. By this means the illocution-ary force of an utterance was given a place in its semantic repre-sentation (which, since this school did not distinguish semantic representations from syntactic ones, was also its deep syntactic representation). The most absurd manifestation of this approach I have met was S. R. Levin's proposal that the deep structure of every poem (for example, of every one of the 154 Sonnets of
20 A SET OF POSTULATES
Shakespeare) begins with a deleted performative which reads: l imagine myself in and invite you to conceive of' a wm:ld in which ... (1976:150). But without considering such exotica as this, one can appreciate the implausibility of the performative hypothe-sis, by merely considering that one of its consequences is the ing. In a piece of exp0sitory written prose (say, an encyclopaedia artide), every single sentence will have, 'in its underlying struc-ture, a performative preface such as I state that. . . which has undergone deletion; so that a article will presum-ably have the same prefixed dause (or a similar one) repeated 100 times. Moreover, to make it more implausible, thh; perform-ative will indude a reference to the writer (1), even in formal expository prose cf the kind in which first person reference is avoided for stylistic reasons. 1 The perfonnative hypothesi., was an apparently inadvertent attempt to 'grammaticize' pragmatic phenomena (ie illocutiona..ry forces), and could, I believe, only have been entertained by those for whom the grarnmatical paradigm of generative grammar was considered all-sufficient. Only in this way could they have tried to ignore the obvious: viz that language takes place in situations? (1 shall elaborate this view of the per-formative hypothesis in 8.6).
On the other hand, there have been rather implausible attempts to state everything abcut meaning, if not about language in general, in terms of speech act:-: and speech situations" This attempt reached something of an extreme in Alston's proposal to define word meaning in terms of a word's contribution to mo-cutionary force:
A meaning of W1 is W2 = df. In most sentences in which W2 occurs, W1 can be substituted for it without changing the inocutionary-act potential of the sentence.
{Alston
Alston defines synonymy of words (Wt. W2 , etc) as a matter of the similar illocutionary-act potential of the sentences in which they occur; but since words do not normally constitute illocution-ary acts in isolation, he has no way of explaining how words have meanings in themselves. The meaning of cow, for him, will not depend on any idea or definition of a cow as a bovine milk-giving animal, but will be a function of all the illocutionary acts we can perform using the word cow.
A better-known example of pragmaticism is Searle's speech-•11Ct theory (1969); summarized in his suggestion that 'a theory of language is part of a theory of action' (1969:17). In addition to illocutionary acts, Searle envisages grammatical acts of various
RULES AND PRINCIPLES .
k . d . 21 . m s; In other words hen f II tem into the y translates the grammatical sys-
nous speech acts: · .. for c.ertain purposes one mi h . utterance acts into phonetic t :sh break up what I _have called And, of course tor m""'t u 'p. _onemic morphemic acts etc , v;,prposesmthe . .,. '. necessary to speak of acts at aU o' '. o. lmguistics it is not morphemes, sentences, etc. . ne cal. Just dtscuss phonemes,
Th , [Searle I969·2sJ e And, of course ... ' acknowled . . .
a matter too slight to hold d' - hke a hand wavmg away paradigm 'language = action'up IscussJon, the limitation of the paradigm in his treatment presses with this m fact, a half-truth: there are th' mgh The paradtgm embodies,
but that does not mean :::t t at can ?e done with lan-domg. I shaH consider Se I .. ' language ts aU a matter of Chapters 8 and 9. ar "' s speech-act theory critically in
2.2 Rules and prim:iple!l\ P2: SEMANTICS IS RULE-GOVERNED '
PRAGMATICS IS PRINCIPLE CONT GENERAL I have already associated the
of 'constitutive rules', and t;e o. Wlth con-hts concept of 'regulative ruies' A pnnaples of pragmatics with logic, it is scarcel necessa . . s regards rules m grammar and tinction further.J. general the first. part of this dis-
Under condition(s) X y is ts: X, Y, or Z is sometimes ,I . denvable from Z (where for example the rule of do · n English transformational syntax
' -support may be stated as follows: '
a verb, replace the
[cf Akmajian and Heny 1975:124]
In propositional logic the rule of Mod p From A and A :::> B d . B us onens says: S h . , enve .
u.c rules etther apply or the do no 1 . . . of rules being applied to a Y . t app.y. There ts no question with another, of one rul-' extent, of one nlle conflicting. variable factors of Th another, etc., according to· governed' in this wav will b e •. ea that grammar. is !see 3-4), since there" are .. on· .he other hand, I shall assum . cu tt-..s m this Vtew. On
e, as.ts assumed by all who attempt
A SET O
F POSTU
LATES
22 .t
mm
ars that such a view
of gramm
ar is substantially tow
n e gra '
correct d
· · 1
be The . division of labour betw
een rules an pnnclp es
. .
· E glish
The rule of changmg an
by passtv:. m
it defines a set of
achve structure mto. a. passtve on
h that if those changes are
which is
and has the same sense as the structure W
I w
started (see, how
ever, 3.4): M
artha killed the fly -The fly w
as killed by Martha.
1 c one fails to appiy one part of this rule (for. example, the in::ri
tion of by in front of the the result ts
omething that lS no longer part o
. • fl
killed Martha. The m
axims
Tex-guage.
e J·Y ,
lat;v, They mclude
tualRhetoric (see 3·3·31 are,_ how
ever, if the rules of
h M
·m
of End-focus, which recom
men s
. t e l ax!
l'ow I.t
part of a clause which contam
s new
the anguage a 1
, •
• •
information should be placed at .he end.
The fly was killed by M
AR
THA
. .
In this afn
the conditions under which tt ts
orme ,
d" f
the present position and
by Searle (I¢9:33-42). Searle r:gards speech acts, including illocutionary acts, as defined by rules.
the hypothesis of this book is that speaking a langua_ge a mal tter
· · · '
d. t
systems of constttutw
e rues. of perform
ing speech acts accor mg o
. .
[ibid.' p 38]
. .
f him
'counts as' a prom
ise/order/request/etc. A
n or
' f
1 lassified as propositional con-
accordmg to a defined set o ru es, c
d f al rules
t ry rules sincerity rules, an
essen I .
tent rules, 1969·67) by the follow
ing rules: Thus a w
armng IS
•
Pro ositional content: Future event or etc. E_·
Pre:aratory: (I) h [sic]. has, to beheve E w
ill occur and is not m
h s mterest.
. (2) It is not obvious to both s and h that E w
ill occur.
Sincerity· s believes E is not in h's best interest. h
F. • ot
Essential: Counts as an undertaking to the effect t at
IS n in h's best interest.
RU
LES AN
D PR
INC
IPLES 23
Such rules assume a taxonom
ic decision: either an utterance counts as a w
arning, or it does not. 3 For example, there is a clear
distinction between a w
arning (in which the future event, as en-
visaged by s, is not in the interests of h) and a piece of advice (w
here the future event, as envisaged by s, is in the interests of h). It is true that such clear-cut interpretations occur in a few
cases, for exam
ple in performative utterances beginning I prom
ise you ... , I w
arn you ... , I advise you .... But on the w
hole, they represent an unrealistic and unsubtle view
of what com
muni-
cation by means of language is like. A
ny account of illocutionary force w
hich defines it in terms of rules like this w
ill present a lim-
ited and regimented view
of human com
munication. By this
view, all hum
an comm
unication boils down to perform
ing certain action-categories. It is alm
ost as if rituals such as baptizing a baby, sentencing a crim
inal, or naming a ship w
ere somehow
typical of the w
ay human beings interact w
ith one another. I w
ould like to quote the following utterance as a counter-
balance to that impression, and as in som
e ways a m
ore represen-tative specim
en of human speech behaviour:
Considering that I am
a hostage, I should say that I have been treated fairly.
This highly ambivalent utterance is reported to have been said by
an Am
erican hostage in Iran in 198o, when he w
as due for early release. It w
as said to reporters who ·asked him
how he had been
treated. The utterance was presum
ably intended to supply some
information w
hich the reporters could treat as news; at the sam
e tim
e, to tell the truth; to reassure the public in the USA
(no doubt including the speaker's fam
ily) that he had not been ill-treated; to avoid saying som
ething that might give offence to his
captors and so delay his release. The w
ay in which such m
otives interrelate and conflict in the function of the utterance m
akes any classification of it in term
s of 'declaring', 'reporting', 'acknow-
ledging', etc; as a naively simplified account of w
hat kind of com-
munication w
as involved. A better m
odel might be som
ething nearer to a linguistic juggling act, in w
hich the performer has to
simultaneously keep several balls in the air: to fulfil a· num
ber of goals w
hich compete w
ith one another. Of these goals, obeying
the CP (giving the required am
ount of telling the
truth, speaking relevantly) must be considered only a part. This
may be an extrem
e case, but it manifests an am
bivalence and m
ultiplicity of function that is far from unhsual as an exem
plar of w
hat language can do. The indeterminacy of conversational utter-
ances also shows itself in the N
EGO
TIAB
ILITY of pragm
atic factors; that is, by leaving force unclear, s m
ay leave h the opportunity to
24 A
SET OF PO
STULA
TES
choose between one force and another, and thus leaves part of
the responsibility of the meaning to h. For instance,
If I were you, I'd leave tow
n straight away.
can be interpreted acCOrding to context as a piece of adviCe, a
warning, or a threat. H
ere h, knoV.ing som
ething s's likely
intentions, may interpret it
threat, and on tt as
s w
ilf always be able to claim
that h was a p1ece of
gwen .
from the friendliest of m
otives. In this way,
'rhetonc of versation' m
ay show itself in s's ability to have h1s cake and eat tt.
2.3 Convention and m
otivatioll P3:
TH
E
RU
LES O
F G
RA
MM
AR
A
RE= FU
ND
AM
EN
TA
LL
Y
CO
NV
EN
· TIO
NA
L; TH
E PR
INC
IPLES O
F PR
AG
MA
TICS
AR
E FU
ND
A·
MEN
TALLY
N
ON
-CO
NV
ENTIO
NA
L, ie M
OTIV
ATED
IN
TER
MS
OF C
ON
VER
SATIO
NA
L GO
ALS.
Searle's account of illocutionary acts also conflicts with P3. Searle
says that the rules for performing
.interpreting -
acts are conventional. For example, It ts a m
atter of convent10n ... that the utterance of such and such expressions under conditions counts as the m
aking of a promise' (1969=37). H
ence tf w
e ask why a sentence such as I'll pay _you back tom
orrow,
spoken by someone w
ho has just borrowed som
e money, co?nts
as a promise in Searle's definition, t.he only
must be:
cause the rules say so.' But w
e can g1ve tn
.of mott\-
ated discourse, as to why such a proposttlon,
some
action in the future by s, will be understood
a -
as a m
eans of assuring h, that is, that the action will
out, and hence, in effect; of putting oneself under an
to sure that outcom
e, I would argue from
the dtrect10n
from Searle: that a. prom
ise is recognized as a. not
means of rules (except in so far as rules are reqm
red. tn detenrun-irig sense), but by m
eans of a recognition of s's Searle's rules apply only to the extent that
specify condtttons w
hich will normally follow
from that recogm
tlon. W
hat is conventional is the semantic fact that a sentence of the
syntactic form I'll pay you back tom
orrow expresses a proP?sition
describing a particular future act by the speaker. That ts, the sense is conventional, in that it is deducible from
the rules of gram
mar (am
ong which I here includ.e lexical
but the force is arrived at by m
eans of mouvated pnnctJ?les
as the C
P. The CP w
ill imply, that is, that unless
Is !he
Maxim
of Quality, s w
ill make sure that the action w1ll be earned
CO
NV
ENTIO
N A
ND
MO
TIVA
TION
out; and that unless sis breaking the Maxim
of s's state-
ment of the undertaking to pay back the m
oney has some rel-
evance to the present speech situation, in which the m
onev is being borrow
ed. Thus if one ·
· (a) know
s the sense of the utterance, (b) know
s the conversational principles that apply to it, (c) know
s the context, (d) is able to em
ployinformal com
mon-sense reasoning to (a).
(b), and (c), one w
ill easily arrive at the conclusion that I'll pay you back tom
orrow is intended as a prom
ise (see further Ch. 7). In this
way w
e recognize a 'division of labour' between convention and
motivation in language. H
owever, the m
atter is not that simple. I have here taken
'conventional' to mean the sam
e as 'arbitrary' in the Saussurian. sense. Saussure's notion of the arbitrariness of the linguistic sign (1959 [1916]:67ff)"is one of the com
er-stones of modem
linguis-tics, and in general, linguists have taken it for granted that linguistic categories and linguistic rules are arbitrary: that is, they are in no w
ay predictable from, or deducible from
, extralinguistic realities. W
hile I take this to be true in a fundamental sense of
gramm
ar, I shall also argue that there are two levels of expla-
nation of gramm
ar. The rules of a gramm
ar (that is, of the gramm
ar of a particular language) are arbitrary; but there is also a gram
mar': an explanation of the typo!ogical or universal character-
istics of gramm
ars in general. I hold that at this level we can
reasonably attempt to give explanations of w
hy gramm
ars have the rules that they have, and that such explanations m
ay appeal to pragm
atic motivation. For instance, it is w
ell known that ·in
many languages, including English, it is possible (and in fact
usual) to delete the subject of an imperative sentence: (You)
come here! This irregularity clearly has a pragm
atic motivation: in
the majority of cases w
e can predict (imperatives having a horta:
tory function) that you will be the understood subject, and therec
fore there is nothing to be lost by its omission. (See the Principle
of Economy, p 67.)
· .
To clarify m
atters, let me distinguish betw
een two kinds of
conventionality. There is the absolute conventionality of the rule, for exam
ple, that in English the word designating the m
ale of the hum
an species is pronounced /mren/. This one has to learn as a
bare fact when learning E11glish (either as a native or as a foreign
learner), and no motivation can be found for it. (H
istorical ex-planations can, of course, be given as to w
hy the word has com
e to have its present pronunciation, but these w
ill also be ultimately
A SET O
F POSTU
LATES
arbitrary, in so· far as they derive the word from
an earlier trary form
.) There is, M
O!IV
A!E
D
o£ a rule for which som
e motlvatlon ts evtdent, but w
htch over-determ
ines the choice of linguistic behaviour which w
ould be predictable fr_om that m
otivation. For such cases, two kinds of
statement are required: the first states the rule as
matter
vention and the second states that given that th1s rule ex1sts, 1t
is a on extralinguistic grounds, that it does so.
One case of such m
otivated conventionality is the following:
[I] Good luck! =
'I wish you good luck'
{ 2] Bad luck!
= 'I regret your bad luck' The fact that [ 1] cannot m
ean 'I regret your good luck' and 2] cannot m
ean 'I wish vou bad luck' follow
s from the PP, the pnn-
ciple which (as it applies to language) m
eans that people on the w
hole prefer to express polite rather than impolite
pp 81 -2). But at the sam
e time, it is a
of convention. m
English that (for example) G
ood luck functiOns as
of good wishes, rather than as a m
eans of congratulatmg h on h s
good luck; and that although bad luck and misfortune
the sam
e sense, we can com
miserate w
ith someone by saym
g Bad luck!, but not by saying M
isfortune! C
onsider another case where gram
matical possibilities are
limited by pragm
atic principles: [3] C
an you post these letters? Yes, I can.
Will she post these letters?
Yes, she w
ill. Shall I post these letters?
?*Yes, you shall.
The
paradigm relating questions to elliptical
is broken in the case of shall. If w
e interpret Shall/ as meanm
g Is it your w
ish that I ... ?', the question is polite,. and is used for m
aking an offer. The reply, how
ever, 1s not polite, be-cause the relevant sense of you shall im
plies the imposition of s's
will on h (see 5.4). In this case w
e could give a gram
matical explanation f?r the
of you shall, vzz that shall is obsolescent m
present-day Engl1sh, and does not occur (in general) w
ith second-person w
e also
give a pragmatic explanation:
you_ sh::'ll1s a_n tmpohte, if not
positively imperious reply to gtve. It ts likely, m
that both
explanations are partly correct: we can argue that It ts on
of the pp that shall in modem
English syntax has a defective paradigm
. '
. There is therefore a need for tw
o levels of statement m
gram-
mar: one conventional and the other non-conventional. T
he basic
CON
VEN
TION
AN
D M
OTIV
ATIO
N
statements about gram
matical rules are conventional, w
hereas the m
etagramm
atical explanations are not. In contrast, constraints in pragm
atics are primarily m
otivated, and only secondarily. if at all, conventional. Part of the essence of G
rice's CP, for exam
ple, is its extralinguistic m
otivatiqn in terms of social goals. G
rice postulates, as a m
eans of explaining conversational implicature,
that speakers usually make their contribution to a conversation
'such as is required by the .accepted purpose'-Or direction. ;of the
talk exchange'. He points out that his m
axims apply ooth. to
linguistic and non-linguistic behaviour: for example, one can
violate the Maxim
of Relation not only in w
hat one says, but in w
hat one does:
I expect a partner's contribution to be appropriate to imm
ediate needs at each stage of the transaction; if I am
mixing ingredients
for a cake, I do not expect to be handed a good booki or an
oven cloth .... [G
rice I97Si47]
Similarly, one can show
politeness not only' in one's speech, but (say) by opening a door for Som
eone, rather than·slamm
ing it in som
eone's face. ·
In so far as gramm
ar is motivared, it is m
otivated at least in part by pragm
atic considerations. For example, gram
mar is to
some extent adapted to such needs as .perform
ing mocutionary
acts while at the sam
e time being polite and cooperative. It w
ould not be surprising to discover, in m
ore ways than have so far been
suggested, that gramm
ar is like it is because iris useful. Pragm
atic adaptations of gramm
ar can presumably be studied,
in the long term, as a m
atter of historical change and evolution, 5
or synchronically; by noting exceptions to rules in the present state of the language. Such exceptions m
ay either be PRA
GM
ATIC
R
ESTRIC
TION
S or PRAG
MA
'f.IC EXEM
PTION
S: that is, some restric-
. tion may be m
ade on an existing rule, or some exception m
ay be m
ade to it. An exam
ple of the former w
i_.ch, like shall above, is from
the field of modality is the suffix -ooli'' in M
alayalam. This
suffix has the meaning 'perm
ission'.• but lis restricted to optative sentences, and to questions w
ith first-person prvnouns. 6 Hence
this gramm
atical form is lim
ited to illocutions of n;; 1uesting and' granting perm
ission. An exam
ple of exemption is:
[4] Would you m
ind if I smoke? (cf the m
ore regular Would
you mind if 1 sm
oked?)
which com
bines the unreal past: would , m
ind with the present
s1r.oke.·This is an·exception to the normal rule that conditianal
28 A
SET OF PO
STULA
TES
sentences have concord of mood betw
een the main and the con-
ditional clauses. The pragm
atic reason for this exception is evidently that sentences such as [4] are PR
AG
MA
TICA
LLY SPEC
IAL-
. IZE
D for use as polite requests. The hypothetical past in such
sentences is in origin a device of polite evasion; but this device has becom
e so conventionalized in formulae such as _W
ould you,. m
ind ... ? that its hypothetical force has atrophied. Consequent-
ly, the non-hypothetical verb smoke is irregularly substituted for
the gramm
atically regular smoked. These and other pragm
atic in-fluences are characterized as exceptions to m
les. But this does
not prevent them from
being regarded themselves as rules: new
ruies entering the gram
mar norm
ally begin as exceptions to other rules. The idea that one rule states a condition on, or exception to, another m
ore general rule is standard in gramm
ar. The process w
hereby pragmatic constraints becom
e tionalized' 7 in the pragm
atically specialized features of gramm
ar provides an explanation of how
, over a long time-scale, gram
mar
itself becomes adapted to pragm
atic constraints. In this, these features show
a resemblance to other partly conventionalized fea-
· tures of language, and of other sem
iotic systems. O
ne such phe-nom
enon is metaphor. M
etaphors vary in their degree of conven-tionalization, from
the totally unconventional 'poetic' metaphor,
to the totally assimilated 'dead' m
etaphor. Another such feature
is the iconicity associated with onom
atopoeic words. The English
word pipe, for exam
ple, is less onomatopoeic now
than it was
when, before the G
reat Vow
el Shift, it was pronounced
(Intonation patterns may be partially conventionalized in a sim
i-lar w
ay.) The m
ain reason for mentioning these .parallels in the present
context is to bring out the gradience of conventionalization in pragm
atics, a gradience which is particularly noticeable in the
case of utterances
used indirectly
as requests
(see Sadock
ll974:97ff). At the non-conventional end of the scale are indirect
illocutions such as: [5] A
re you able to repair this watch?
Although it w
ill be pragmatically interpreted in the right context
as carrying the force of a request ('I want _to know
if you can m
end this watch, and if so, I w
ant you to do so'), this sentence is not gran'llatically specialized for that purpose. Part!y conven-tionalized are requests like
[6] Can you repair this w
atch? w
hich are specialized to the extent that they are gramm
atically
CO
NV
ENTIO
N A
ND
MO
TIVA
TION
29
assoc!ated with item
s and structures which resem
ble those of im-
than of questions (cf Sadock 1974:73-95); eg the ... m
edtal please m [6a], and the can you tag-question in [6b]:
[6a] Can you please repair this watch?
[ cf Please repair this watch.] [ 6b] R
epair this watch, can you?
[cf Can you repair this watch?] A
t the most conventional end of the scale, am
ong other tures, are gram
matical form
ulae such as oaths and greetings (H
ow
do?) .the .w
ord please itself, which has lost w
hat It once had, and is just treated as a
particle of pohteness. ,.
forms such as [6] have another
to metaphors ana m
tonation patterns in that they are oom
et1mes,
not always, translatable into
equive alent fonns m
other !anguages. For example, [6] can be straight·
forwardly translated m
to Portuguese:.
[7] pode consertar este rel6gio?
(You) can m
end this watch?
there are types of construction w
hich cannot be trans-lated fr.om
Er.gbsh mto Portuguese or vice versa:
[8] Seni {
} este relogio? It is that you w
ould/could mend this w
atch?
!his request w
ith the verb to be has no m
English. m
ore cases like please, m
eaning-are unlikely: please contrasts sem
antically m
th equtvalent polite formulae in ether languages, such as Portu·
gues_e por favor {'as a favour'), Arabic m
in fadlak ('out of your graaousness').
... '
by emphasizing the
between gram
mar
· has cannot sim
ply G
nunmar
iS conventiO
nal, pragm
atics 1s
non-conventi9R'ill. • .N
evertheless, such demarcation difficulties are m
et all
of language: one might cite, for exam
ple, the difficulty of-· w
hether sequen?es like on to and cannot consist of one •1 tw
o words.
phenomena are sym
ptoms of the ·
language ts :W evolving system
, and do not the
in conventionality, betwoon
and difference m
ay be SUIT'wlarized m
· s,mm
mar w
pnmarily conventional fu'"H! secondarily
30 A
SET OF PO
STULA
TES
motivated; pragm
atics is primarily m
otivated and secondarily conventional.
2.4 The relation between sense and force
P4: G
ENER
AL PR
AG
MA
TICS R
ELATES T
HE
SEN
SE (O
R G
RA
MM
ATI-
CA
L ME
AN
ING
) OF A
N U
TTERA
NC
E TO ITS PR
AG
MA
TIC FO
RC
E. TH
IS R
ELATIO
NSH
IP M
AY
B
E R
ELATIV
ELY
DIR
ECT
OR
IN
DIR
EC
T.
It has already been claimed that sem
antics and pragmatics de-
scribe the meaning of an utterance in different w
ays. The task of
pragmatics is to explain the relation betw
een the.se two
meaning: the sense (w
hich has often been descnbed as the ht-eral' or face-value m
eaning) and the (iilocutionary) force. I assume,
along with m
any others, that the sense can be described by means
of a SEM
AN
TIC
REPR
ESENTA
TION
S in some form
al language Of
notation. The force w
ill be represented as a set of implicatu;::es.
'Implicature' is here used in a broader sense than G
rice's, I
follow G
rice in believing that 'the presence of a conversataonal im
plicature must be capable of being w
orked out' (1975:.so), by m
eans of the type of informal reasoning_ referre.d to in 2.?.2. This
is a corollary of the claim that pragm
attcs studtes behavtour that is m
otivated, in terms of conversational goals. I m
ust stress, how
ever, that in the present account. all impiicatures are prob-
abilistic. We cannot ultim
ately be certru.n of what a speaker m
eans by an utterance. T
he observable conditions, the utterance and context are determ
inants of what s m
eans by the utterance U; tt is the
of h to diagnose the most likely interpretation. Since,
as I have already pointed out, utterances are liable to illocution-ary indeterm
inacy, it is not always
for h, a
reasonable diagnostician, to rome to a defim
te concluston about w
hat s tneans. Interpreting an utterance is ultim
ately a matter of guessw
ork, or (to use a JJtore dignified term
) hypothesis I shall
illustrate this with an exam
ple of the type which G
nce uses to exem
plify the CP:
[9] A: W
hen is Aunt R
ose's birthday? B: ·It's som
etime in A
pril. . T
he sense of the reply is simply a proposition to the effect that
Aunt R
ose's birthday oCcurs in April (it could have been
ed in. the form of a long disjunction It's either on the first o
f April,
or on the second of April, or ... ). B
ut A will derive from
it an piece of m
eaning: viz. that B doesn't know the exact
THE RELA
TION
BETWEEN
SENSE A
ND
FORCE
31 date of the birthday (ie B doesn't know
whether A
unt Rose's
birthday is on the first, the second, the third ... or the thirtieth of A
pril): How
is the extra meaning, or im
plicature, arrived at? The follow
ing are the three mai:n stages:
(i) 'f!le prima-facie observation is that there is som
ething 'up' W
ith B's reply. B does not give the right am
ount of in-form
ation for A's needs. That is, B has apparently violated
.. the CP_(.specificaH
y, the Maxim
of Quantity).
(u) 1s, how
ever, no to suppose that B is being
debberately uncooperative. Therefore, A
can reasonably assum
e that B IS observing the CP, and that this apparent
breach of the Quantity M
axim is due to A
's wish \.) uphold
the CP at another point. Therefore w
e must look for a
reason why the C
P should cause B to give less information
than A needed. ·
· (iii) This reason can be. supplied o.n the grounds that B w
as trying to uphold the M
ax1m of Q
uahty. Suppose that B is observing the C
P and that B does not know w
hen Aunt R
ose's birth-day is, except that it is in A
pril. Then B will not random
ly state that
is (say) on I April, or on 6 A
pril, or on 1.9 A
pnl, to do so, although it w
ould uphold the M
axtm of Q
uantity, would be to violate the M
axim of Q
ual-ity (ie to risk telling a lie). Therefore, to be on the safe side B w
ill merely say that the birthday is in A
pril. In the of any ::>ther
this explanation will be accepted, as consiStent W
ith the CP. Therefore it w
ill be concluded that B does not know
which day in A
pril is Aunt Rose's birthday.
The stages of this inference are (i) rejection of the face-
as inconsistent with the C
P; (ii) search for a new
mte!Pretat1on consistent w
ith the CP; (iii) finding a new
in-and checking that it is consistent w
ith the CP. The
new
an implicature I, on the grounds that
the that Its necessary in order to m
ake s's sayin& U
consJStent W
ith the CP.
'!Ills is not a formalized deductive logic, but an inform
al rational
strategy. It consists in (a) formulating
most likely
hypothesis, then (b) testing it, and, if it fads, (c)
the next most likely available hypothesis,
and so on .. ThiS kind is a general strategy em
ployed by hum
an bemgs. for. solvm
g mterpretative problem
s. It is found on the one band m
htghly abstract and complex scientific theorizing 9
and on the in. hom
ely examples, such as the follow
ing. If an eleanc bght fads w
hen the switch is turned 'O
n, the first and
32 . A
SET OF PO
STULA
TES
most likely hypothesis is that the bulb is broken; if the
is re-placed, and the light still does not go on, the next m
ost likely guess is that the lights have fused, or perhaps that the connection is faulty. T
he process goes on until a solution (ie a hypothesis consistent w
ith the observed facts) is found. A
nother point to notice about the Aunt R
ose example is that
one maxim
of the CP is upheld at the expense of another. In this
case, as is normally the case, s is assum
ed to have given ence to the M
axim of Q
uality over the Maxim
of Quantity. B
ut there are cases w
here the preference is reversed. It has been re-ported to m
e that in certain places (eg in parts of Italy and Brazil)
a stranger asking a question (eg asking the way to a destination)
will characteristicaU
y be given a false answer in preferenee to a.'1
uninformative one.
Further, the elaborate spelling-out of a rational process, as in above, should not, of
be taken to imply that such
processes necessarily happen laboriously and consciously in the m
ind of the interpreter. The purpose of such a com
mentary is,
first, to that . the pragm
atic force is motivated by
general principles of rational and social behaviour (in the sense of 2.3), and second, to give a rough outline of the postulated pretative process w
hich may, how
e\ler, be to a greater or lesser degree autom
atized. Searle (1979 [I975b]: s6-7) aptly describes the problem
of interpreting an indirect iUocution such as Can you
pass the salt? as follows:
·
The problem
to me som
ewhat like those problem
s in the epistem
ologica.l ana1ysis of perception in which one seeks to explain
how the percdve11 l'<iO
Ogcl:res an object on the basis of im
perfect sensory input. T&-se qu<e!lti.on, H
ow do I know be bas m
ade a request w
hen he ha& o:r.:i; asked :me a question about my abilities? m
ay be like the question, How do I know it W<'lliS a car w
hen all I perceived a
flash going past me on the highw
ay?
Just as one may autom
atically condense a complex set of visual
and auditory ilnpresskms and inferences into a single observation
'That w
as a car•, so one may condense a long argum
ent into a single unheeding respcm
sce 'A wan1s m
e to pass the salt'. ·
Can you pass the salt is an example (a stock exam
ple, as it hap-b
pens) of an indirect illocution, ie in Searle's definition, a case •m w
hich one act is perform
ed mdirecdy by wa:y of per-
forming another'. I follow
Searle (1979 [1975bl) in proposing for such illocutions an explanation m
odelled on Gricean im
pli-al!ture. G
rice's conversational implicatures w
ere de·11ised to explain 0
sentences in which a speaker appears to m
ean more than he says.
!HE
RELATIO
N B
E1WEEN
SENSE A
ND
FORCE
33 It is. a natural
of this type of explanation (which he
applied to declarative sentences) to apply it to other, non-
deda.ratrve examples of 'm
eaning more than one says'.
But
theory t<Jm
, naturally enough, to treat m
dJ.rect tllocutions as definea by the same kinds of ru!es as appiv
kinds of illocutionacy act. My ow
n position on dlocut1ons (to be developed m
Ch. 7) differs from Searle's in tw
o respects.
· I ta_ke it
illoc":tionary rules are required for tndlrt'lCt dlocut1on1;; m
du:ect illocutlonary force is stated sim-
ply by <;'f a set of im
plicatares. Such imp!icatures are
attnbutmg m
ental states ( eg propositional attitudes such as belief) to s. In the case of [9] for
the imp!icatures asooci-
ated with B's reply It's som
etime in A
pril (a§Suming that B is
observing the CP) include the follow
ing:
[ ma] B believes that A
unt Rose's birthday is in A
pril (v'" M
axim of Q
uality). ·
· """
[mb] B_ is not aw
are of which day in A
pril is Aunt R
ose's bL."i:hday (via r.-w
Jms of Q
uantity and Quality).
Although the. interpretation of an indirect illocution like Can you
pass the sa.lt IS m.or.-.. e c:omnlex than this
for ex"'..,l'll,P, h
pp n
II" e•
uu
•r·-· as w
eu as the <;P), .the general method of description is
same, and the
tmphcature (corresponding to Searle's 'in-
dJiect speech act') Is that s h.to !?ass the
(seep uo). .
I do not draw a rustm
ct1on ootween drrect and indirect
AU
illoeutions are 'indirect' m that their force is de-
implicature. T
here is, however, a great deal of variation
m then degree of indirectftess. T
he most direct iH
ocutions are sim
ply those to which, in the absence of contrary evidence, eve
may apply. the m
ost likely pragmatic hypothesis, ie w
hat may be
cal!ed the default interpretation. If, for example, the
of B m
[9] had been It's ore m April, the default interpretation
would have applied:
[na] B believes that Aunt R
ose's birthday is on 10 April (via
Maxim
of Quality).
Grice applies the term
'conversational impli{'.ature' to cases such
i]S !IOb], but I shall extend it a!w
to defuult interpretations such as poa] and [uaJ. 'Ibis extension is a consequence of the that sense and force are tw
o distinct kinds of meaning. It is nor-
!hat is, for an to have both a sen.se and. a force,
m cases (such ...
statements of inform
atiot") w
he.e an utterance s meanm
g-m-oontext set;m
s to follow au _,.
34 A
SET OF PO
STULA
TES
matically from
its sense. 10 The advantage of this position, which
is the one I shall adopt, is that it capitalizes on two very different
sets of insights about pragmatic m
eaning -those of G
rice's theory of conversational im
plicature, and those of Searle's speech-act theory. In fact, the latter is reinterpreted in term
s of a general-ized version of the form
er. For example, Searle's sincerity rules
are treated simply as cases of s obeying the M
axim of Q
uality (see 2.5.2).
. I have said that pragm
atic force is specified by means of state-
ments attributing som
e mental state to s. A crucial point, how
ev-er, is that statem
ents such as [10a] and [xob] are not direct claim
s about what is going on in s's bead. R
ather, they are state-m
ents about what s m
eans to convey by the utterance in question. This conclusion depends on the assum
ption that s is observing the C
P, and perhaps other rhetorical principles. Despite appear-
ances, then, pragmatics is a study in w
hich only the meaning that
is publicly available for interpretation is analysed. In pragmatics,
as .elsewhere, the linguist is interested in m
aking publicly confirm-
able observations about language, and in constructing theories to explain such observations.
' Since pragm
atics is about meaning in speech situations, w
e clearly cannot m
ake any pragmatic claim
s about what is going on
privately in someone's head. If s, for exam
ple, says It's six o'clock, w
e cannot 'take it for granted that s believes that it is six o'clock. For all w
e know, s m
ay be an inveterate clandestine liar. B
ut we can say that the im
plicature that s believes that it is six o'clock is part of the m
eaning, or the force, of that utterance. To be m
ore precise, then, [IOa] and [wb] should have been prefaced
as follows:
I ·
{It's sometim
e in April }
B implicates, am
ong other n saym
g It's on 1
0 A
pril
things, that ...
But m
eaning is tw
o-sided: as Searle (1969:43), paraphrasing
Grice (1957:385) puts it: In speaking I attem
pt to comm
unicate certain things to my hearer by
getting him to recognize m
y intention to just those
things. I achieve the intended effect on the hearer by get!\ng him to
recognize my intention to achie1e that effect:
In other words, 'm
eaning' as it is used in pragmatics (ie in the for-·
mula's m
eans F by u; where s = speaker, F = force, and U
= utterance) is characterized as a R
EFLEXIV
E INT
EN
TIO
N, ie an m-
pRAG
MA
TICS AS PROBLEM
-SOLV
ING
35
tention whose fulfilm
ent consists in its recognition by h.U As
Bach and H
arnish (1979:15) point out, however, this reflexive in-
tention is executed only by virtue of. what they call 'the Com
-m
unicative Presumption', ie the m
utual belief, shared by s and h, that w
hen someone says som
ething to else, it is done
with som
e , illocutionary goal in mind. Such a presum
ption, in-deed, m
ay be said to be follow, as a lim
iting case, from the
Maxim
of Relationas interpreted in 4·3 below
: 'A
n utterance U. is relevant to a speech situation if U can be in-
terpreted as contributing to the conversational goal(s) of s or h.'
To this may be added yet another presum
ption about the nature of pragm
atic force.: that if s means F by U, then s intends h to rec-
ognize the force F by way of the sense (ie: the gram
matical m
ean-ing) of U.
. .
. This does not m
ean, of course, that mtscom
mum
catiOn does
not take place. It means that pra¥m
atics is concerned w
ith publicly conveyed m
eaning, and does not. tak.e account of
miscom
munication, or of secret com
mum
cation. Thus tf s says My aunt has a villa in Vladivostoki, m
eaning by that that s has three aces and tw
o kings up his sleeve, this is no concern of pragm
atics, because the meaning conveyed in that case has
ing to do With the sense of the utterance. 12 The factors w
htch can lead to failure of com
munication are varied, and not all of
them are in the pragm
atic domain: for
sand h may not
share the same linguistic know
ledge; the physical channel be-tw
een s and h may be im
paired; s ma:1 not be observing rhetorical
principles; s and h may
di:ffereQ1 values. B
ut in so far as w
hat s means by U
is from
what h
stands by U, it is not part of
about such dif-ferences. A
t the same tim
e, it has to be" admitted, as I have
already indicated, that what s m
eans by U
to some extent
indeterminate, so giving h opportunity, w
ithin1imits,_to
or decide the force of U (see further 7. I). T
hus pragmatics m
ust be concerned, quite centrally, w
ith indeterminacy.
. Having em
phasized the reciprocal .participation of and h
the illocutionary act, I must now
pomt out an essentutl w
ay m
which their roles are different.
2.5 Pragmatics as problem
-solving Ps:
GRA
MM
ATICA
L A
RE D
EFINED
BY
MA
P-
A SET O
F POSTU
LATES
PING
S; PR
AG
MA
TIC
CO
RR
ESPON
DEN
CES
AR
E D
EFINED
B
'J PR
OB
LEMS A
ND
THEIR
SOLU
TION
S. Prag.'!latics involves PR
OB
LEM-SO
LVIN
G both from
s's and from h's
' point of view. From
s's point of view, the problem
is one of rung: 'G
iven that .I want the m
ental state of the hearer to change or ·to rem
ain unchanged in such and such ways, how
do I proo duce an utterance w
hich wm make that result,m
ost likely?' From
it's point of view, the problem
is an interpretative one: 'Given
that s has·said U, w
hat is the most likely reason for s's saying U
?' T
he problem-solving procedures are very diff'!rent in. these cases.
The latter case, that of interpretation, m
ay in fact be considered a 'm
eta-problem-solving' procedure, since the problem
h has to solve is 'W
hat was the c.om
w>:m
irative problem s w
as trying to solve w
hen s said U?'
Unlike the sense--sound m
aypings and the sound-sense pings of gram
mar, these pr<'blem
-solving procedures cannot be defined by algorithm
s. They involve general human intelligence
assessing alternative probabllities on the basis of contextual evidence.
The speake:r•s task, view
ed m:term
s of means-ends
Briefly, the problem
-solving strategy of s can be viewed as a fm
m
of means-ends analysis. 13 This ;:m
alysis represents a problem and.
its solution. in. the form of a directed graph, w
ith initial states and final states (see Fig. 2.1).
FIGU
RE 2.1
:;: = initial state (individual feels cold).
2 = final state (individual feels warm
). G
= goal of attaining state 2 (getting warm
). a
= act.ion (sw
itching heater on).
This is the very simplest graph possible. T
he diamond represents
the initial state, and the square the finai state. The solid arrow
represents an action taken by some individual in order to · fuffil
the goal. The broken arrow
represents the goal (possessed by the individual at state I) of attaining the final state.
There is a natural extension of this m
odel to include intermedi-
ate states (which are both objects of subsidiary goals, and con-
ditions for the fulfilment of the lli'1al goal). Further extensions
introduce multiple goals, sim
ultaneous states, negative goals (ie the
J?AAGMATICS AS PRO
BLEM-SO
LVIN
G
37 m
aintenance of status quo), etc. The sim
plest case of a means-
ends analysis involving language is one like Fig. 2.1, in which a
represents a speech act. In this way we m
ay interpret the fulfil-m
ent of the reflexive intention described in 2.4 above as follows:
x = initial state (s means h to understand F by U)
2 = final state (h understands F by U)
G = goal of attaining state a
B = speech act (act of uttering U) A
slig..litly more com
plicated linguistic example of m
eans-ends analysis is that in w
hich stakes for granted that h's understanding of the m
essage will lead h to perform
a required action (Fig. 2.2).
Switch or. the heater!
fiGU
RE 2.2
I = initial state (s feels coid)
1 = interm
ediate state (h understands that s wants the
beater on) 3
== final state (s feels warm
) G
-= goal of attaining state 3 (getting warm
) a
= s's action of telling h to s1i'<ik:h on the heater b = h's action iu sw
itching on tbe heater
T'ne oox of the followi..ng shape D
is used to :represent ·an ffi, ae:rm
ediate state which is (i) the fuffilm
ent of a subsidiary goal, (H
) a condition for the attaimnent of the fu!al goaL It is thus
a final state with respect to an im
mediate goal, and an initial state
'•liitb I:espect to an ulterinr goat Com
:r;ared with the action represented in Fig. 2
.1. that of
Fig. 2.2 may be thought to represent an
fuffilment of a·
goat But by this standard, all purpooeful uses of language '''ill be
regarded as indirec:t: 14 that is, whenever w
e use language as a n;eam
of bringing about some end, this im
plies oome chain of ao-
such as is found in Fig. 2.2 -1:mt often the chain is longer
ana more com
plex. In fact, such iUoc-1.1tions as that of Fig. 2,2 have been called D
IREC
T SPEECH
.ACTS Of D
iREC
T ILLOC
UTIO
NS by
who have contrasted them
SPEEC
H A
CTS or
A SET O
F POSTU
LATES
IND
IREC
T ILLOC
UTIO
NS (see Searle I979 [1975b); Sadock 1974).
Searle defined indirect speech acts as 'cases in which one illoc-
utionary act is performed indirectly by w
ay of performing
(1979 [1975b]:6o). That is, for Searle, an indirect speech act can
be regarded as a means of perform
ing a drrect speech act. But in
the present means'-ends fram
ework, even a 'direct' speech act
such as the imperative Sw
itch on the heater! is an indirect means
of achieving some goal, in that it is directed at a subsidiary goal.
Therefore 'indirect illocutions' are simply iHocutior!S w
hich are m
ore indirect than others; and indirectness is a matter of degree.
The scale of indirectness can be notionally .repreJ>ented in terms
of the means-ends analysis by the length of the m
eans-ends chain co!"..necting the speech act to its goal.
· I shall present one further diagram
(Fig. 2.3) to show how
the
" /
'J( /
.,;;r //
r;PP_ .;-"" .1.1
.;"""' /,.
...... ,'
_________ g __________ C
old in hoe, isn'tUF
c
b
x = initial state (s feels oold)
2 = interm
ediate state (h understands that s is aware that it is
cold) 3
= intermediate state (h understands that s w
ants the heater on)
4 = final state (s feels w
arm)
G
= goal of attaining state 3 (getting warm
) GPP
= goal of observing the Politeness Principle G
' = further goal(s) (unspecified)
a = s's action of rem
m:king that it is cold
[b = s's action of telling h to
on the heater-see below]
c = h's action in sw
itching on the beater
pRAGMATICS. AS PRO
BLEM-SO
LVIN
G
39 goal that w
as achieved in the examples represented in Figs. 2. I
and 2.2 can be achieved by a more indirect illocution. Figure 2.3
represents Searle's that an indirect speech act ( = action a) is
a means of perform
ing another speech act (action b). The dia-
gram suggests, how
ever, that Searle's concept of a direct speech act underlying an indirect speech act is an unnecessary construct, resulting from
Searle's way of looking at iH
ocutionary acts as de-fined by conventional rules, rather than as defined by their func-tion in a m
eans-ends analysis. 15
There is only one speech act in question here: the uttering of C
old in here, isn't it? Searle's analysis, however, requires that this
one iHO;..--ution actually instantiates two illocutions, one being the
means to the other. Searle appears to claim
that (i) there are.two
speech acts, (ii) that one is performed by m
eans of the other, and (iii) they take place sim
ultaneously, both being performed in the
same utterance. B
ut this analysis makes the concept of a speech
act quite mystical and abstract. A
better way to interpret a dia-
gram like Fig. 2.3 w
ould be to say that b is an action performed
not by s, but by h, and that this action is the act of interpreting the utterance C
old in here, isn't it? as having the implicature that
s wants h to sw
itch the heater on. That is, w
e should replace the statem
ent regarding bin brackets under Fig. 2.3 by: b =
h's action in inferring that s wants h to sw
itch the heater on.
This is not the only interpretation of the sentence Cold in here,
isn't it?, but it is a probable on,e, given certain circumstances.
Such a remark about the tem
perature could be, alternatively, a piece of chit--chat, of pbatic com
munion, w
ithout ulterior goal cept the m
aintenance of social relations (see 6.2). And we should also note this as a case of potential indeterm
inacy; it could be that s uttered C
old in here, isn't it? partly in order to maintain
friendly social relations, and partly in the hope that h 'WJ.l do
something to alleviate the cold: in w
hich case, it is up to h to in-terpret the illocution as having an im
positive or coercive force, but only if he so w
ishes. It is with this in m
ind that I have repre-sented s in fig. 2.3 as having other p<>tendal goals (m
arked G').
The goal of upholding the PP (G") is included in Fig. 2.3 for
another purpose. It may be taken as a general principle of goal-
directed behaviour,.. that individuals adept the most direct COUrse
of action that is judged to be consistent with the fulfilm
ent Of their goals. (This is one w
ay of interpreting the Maxim
of Man-
ner.) Hence if an s, as in Fig: .2.3, em
ploys an indirect strategy to fulfil a goal, the reason for this is likely to be that. s w
ants to
40
A
SET OF l'O
S'I'ULA
TES
other goal, in a,ddition to G. This is the justification
for postt!ng the extra goal G· P of preserving the PP and thereby
maintaining good social relations. T
he CP, the PP, ;nd other rhe-
may, in
be seen as regulative goals w
inch perstst as part of the background against which all other.-'
goals must be considered. O
r perhaps it is betteli' to think of them\
as !legative the goals of avoiding uncooperative and im
-polite behavtour. O
ther goals may conflict or com
pete with these
most , ob":'iously' an im
positive goal {one requiring one s w
tll on someone else) runs counter to the
pohteness. Thus the !lnaking of an _•innocent', non=
nnpo:ut1ve .r7mark s11ch. as C
old in here, isn't it? b.eoomes a gam
bit for
goals: for evading a breach of polite-ness w
hile still atmm
g at an impositive goal. Providing, in accord-
ance with G
ricean conversational implicature, that the forf'.e can
be 'worked out', s can rely upon h's ow
n compliance w
ith the PP as a reason for h's perform
ance of the desired action. T
here is likely to be one dissatisfaction with the above inter-
of linguistic in term
s of a means-ends analy-.
sts. It :s that analysts
to regard all uses of language as bavm
g an runctu:m
.. Surely, it might be m
:gued, w
e cannot treat all as m
ot:avated by the goal of bdnging about som
e result m the m
ental or physical condition of the addressee? W
hile one cannot, of course, rule out the occurrence of
uses of. language (eg purely expressive speech), 1t iS
my C<!ntent10n that; broadly interpreted, the
analys1s apphes to comm
unicative uses of language m
general. How
ever, the term 'goal' is slightly restrictive and the
term 'i.ntention'
more so, in suggesting a degree of c'onscious
or debberate planmng of discourse w
hich the model does not
imply. O
ne can, needless to say, uphold the CP or the PP in one's behaviour w
ithout being conscious of the l,';)xistence of such principles; and the sam
e applies even to more specific goals.
The
of 'goal', in analysis, should be applicable to the
phat1c use of language (see o.z), the avoidance of taboo subiects and. taboo vocabulary, etc .• and other cases w
here ··although the pa:tern of linguistic .behaviour m
ay be clear, .few people .vioold that tbe user 1s aw
are of the goals that motivlite this beG
haivtour. In short, the term goal is used in the neutral A
rtifical o
f' a Stare;'wbicb the behaviour of the
m such a w
ayas.to fa,iUtate a given <n1tcom
e.
2:.5.:i!\ se2n
terms of heuristic
kina of task w
hich an addressee faces m
pRAG
MA
TICS AS PRO
BLEM-SO
LVIN
G
41
terpreting an utterance may be described as H
EUR
ISTIC. A
heuris-tic strategy consists in trying to iden1:ify the. pragm
atic f<:'rce of 'utterance by form
ing hypotheses and checkmg
agamst avail-
able evit!ence. If the test fails, a new
1s T
he w
hole process be repeated
until a hv'UOthesis w
hich is successful, in that 1t does not conflict wtth .evt-
is arrived at. Tne w
hole be represented, m
. a rather oversim
plified way, as show
n m F1g. 2.4. ·The problem
, m
this case, is a problem of interpretation. From
the sense of what
is said; together with
(about background assum
ptions (that s ts .observmg the usual pnn-
dp1es) h forms a hypothesis about the goal(s) of the utterance. For
exarn,ple, if s says It's Aunt M
abel's birthday next Monday, the
most likely hypothesis is: s m
eans [h to be aware [that A
unt Ma.bel's birthday is next
Monday]]
Pro
blem
-2. H
ypothesis -3·
i I
I
I I
\ I
I test sueceeds
1 J
(default inte1-pretation) • \ . I
test fails
11>e whole thing can be sim
plified if we use P as a sym
bol for the sense of U
(in this case Pis a proposition). Then the sense of
utterance can be taken as given for the purpose of the problem-
solving.process, and can be stated as fullows:
A. s says to h [that P]
Tne hypothesis about the force of P can likew
ise to generalized as follow
s; B. s m
eans [h to be aware [that P]]
This· amounts to· a claim
·that U is an inform
ation-giving utter-ance; w
hat Searle and others have .called an .assertion. claim
or hypothesis can now
be checked oy 1t ts
sistent (assuming the relevant pragm
attc pnnctpies) Wltn the evi-
dence of context:
42 A SET OF PO
STULA
TES
C. s believes [that P] (M
a'l!:im of Q
uality) D
. s believes [that his not aware [that P]](M
axim of Q
uantity) E. s believes [that it is desirable
[that h be aware [that P]]]
(Maxim
of Relation)
Once the hypothesis has been form
ed, certain conditional conse-quences, such as C
, D, and E, follow
from the assum
ption that (i) the hypothesis is correct, and (ii) that s is observing the C
P. C follow
s, since if C were not the case, s w
ould be telling a lie, and breaking the M
axim of Q
uality. D follow
s, since, if D w
ere not the case, s w
ould be saying something w
hich so far ass is aware
has no information value for h, and· in that cases w
ould be break-ing the M
axim of Q
uantity (by giving too :tittle information, ie
none at all). E follows since, if it w
ere not tlte case that E, then s w
ould be !>aying sometbjng
t,q the situation, and would
thus be breaking the Maxim
of Relation.·.·. "
. (The M
axim of R
elation bas been m\iQ
zed on the ground that its vagueness m
akes it almost vacuous. m
a pragmatics em
ploying m
ea:Ds,....ends anaty.,. however, •&
.televant' has a relatively clear m
eaning: it mew
'Make your 'conversational contribution
one that will advance ·the goals of yourself or of your
addressee' -see further 4.3).
If all these consequences are consonant with the evidenc-e of context, the hypothesis w
ill be accepted. H one or m
ore. of them
conflict with context, the hypothesis w
ill be rejected, and a new
set of probabilities must be considered. The next hypothesis to 1Je/
tried will be the
which appears m
ost likely. in the lighf of eVi-dence already noted. 1be acceptance of the initial and m
ost likely interpretatidn w
ill be Called the DEFA
ULT IN
TERPR
ETATIO
N. That
is, it is the interpretation that is accepted in default of any evi-
dence to the contrary. The statem
ent regarding the goal(s) of s (B above) W
ill be called. the MIN
IMU
M IL
LO
CtrriO
NA
RY
ASSU
MPTIO
N,
and the implicaturesderived from
that (eg: C, D, E) w
ill be called CO
RROBO
:RATIV
E CO
ND
mO
NS.
-In a general sense (m
ore general than that of Grice) all these
statements (except A
) may be calle_d im
plicatures. They are pro-"isional pragm
atic implications, w
hich can, be cancelled
if they are inoonsistenf with G
riee .applies the term
implicature principally to cases w
here the default interpret-: ation is rejected, due to an apparent flouting of a m
axim; but thist
in the present model, is a special case of a m
ore general inferen• . tial pattern. ·
The earlier example·ofC
old m here, isn't it? illustrates bow
a or default interpretation m
ay have to .be rejected; Sup-
pRAGMATICS AS PRO
BLEM-SO
LVIN
G
43
se on the one hand, that this utterance is spoken (as is quite in a context w
here his fully aware of the coldness of the
temperature. Then the im
plicature D
fails,. and quently E fails as w
ell, unless some
. kind of illocutiOnary
force (eg phatic remark about the w
eather) can be postulated. suppose, on the other hand, that C
old if! here, is_n't it? is ?ttered in .circum
stances where the
very In thts
implicature C
would fail, and S?J:ne J,Iew hypothests
that s ts being ironical) is. sought. By thts
a pragmatic .m
terpret-ation m
ay be amved at m
ore or m
drrectly, to
number of
steps for h to
at a factory solution (for further
see 4.4). In this heunstlc analysis, there is a scale of indirectnesS •.
t<:> that already observed in the m
eans-ends. (Ulaiys1s of s s planmn.g of
the utterance. The more indirect s's iliocution, the m
ore indtrect ish's inferential path in reconstructing it.
This account of pragmatic interpretation can be m
isconstrued unless it is understood in a sufficiently abstract w
ay. First, 'prob-lem
-solving' suggests a deliberate puzzling-out of meaning; but in
the case of pragmatic interpretation, t.lte process m
ay well be
highly automatized; there is no im
plication that are
scious or are anived at as a result of expliCit cerebratiO
n. second, one should not expect that default interpretations are sam
e in different contexts. The expectations of addressees will
vary according to situation! so that m
ay be a inter-
pretation in one w
tll not be. so .m anothe!. F<:>r the default interpretation of a question m
many situatiO
ns Will .be
'an information-seeking illocution', carrying the corroborative
condition that s does not know the answ
er to the question con-cerned. B
ut this is presumably not the m
ost likely interpretation in an exam
ination paper .or in a legal cross-examination. Thirdly,
in the account I have assumed, there is a deductive ordering of
implicatures, w
hereby the hypothesis is formed first of all, and
the consequences are derived from it
In though, it m
ay well be that som
e of the corroborative conditions are registered before the hypothesis is form
ulated; in other
words,
a partially inductive m
ay. be The
account I have given may be som
etbmg of an tdealizauon, depart-
ing at various points, from the actual intellectual processes of the
hearer. At this point, how
ever, I shall merely stress the general
plausibility of a heuristic procedure in showing ·how
sational im
plicatures are (in Grice's term
s) 'capable of bemg
worked out', and
by an argument' .. In this w
ay, a general m
ethod of showm
g bow force can be denved from
sense w
ithout resort to arbitrary convention can be demonstrated.
44 A
SET OF PO
STULA
TES
Another look at the set of statem
ents A-E above w
ill show
that they correspond remarkably closely to Searle's speech-act
rules. Searle's rules for assertions (1969:6s) in fact run as follows:
Propositional content: Any proposition, P.
Preparatory: (I) s has evidence (reasons, etc.) for the truth of P. (2) It is not obvious to both s and h that h know
s .
(does not need to be remindeqof, etc.) P.
' Sincerity: s believes P.
· · Essential: C
ounts as an undertalr.ing to the effect that P repre-sents an actual state of affairs.
Of these, the propositional content rule corresponds to .A ('s says
to h that P'). The second preparatory rule corresponds to B ('s m
eans [h to be aware [that P]]'). The sincerity rule and the first
preparatory rule correspond to C ('s believes [that PJ'). T
he lack of correspondence is in E (w
hich has no equivalent in Searle's rules), and in Searle's essential rule, w
hich states I! 400nvention' by w
hich an· assertion will be interpreted, but is arguably redundant.
Without going into the m
erits of this or that set of :rules, I wish to
argue that Searle's speech-act rules, with one category of excep-
tion, Ca!l .an be replaced by impl.icatures. T
he exception is the propositiohal content rule, w
hich in my account corresponds to
a statement of the sense of the utterance. T
he sense alone is con-ventional; all other 'ruies', in m
y acoount, are non--conventional im
plicatures, derived (by means of the heuristic strategy) from
the sense and from
general conversational principles such as the C
P.
2.6 Conclusion
This chapter has explained five out of the eight postulates listed ·o
n p. 5· The rem
aining three postulates will be dealt with in the
next chapter, the main topic of w
hich will be a functional account
of the relation between gram
mar and pragm
atics.
Notes
1. Ross (1970), in his fam
ous article 'On declarative sentences'
ring forward the perform
ative hypothesis, actually points out this difficulty.
. 2. Sim
ilarly, Ross (1970:254) pointed out the advantage of an alterna-
tive 'pragmatic hypothesis', w
hich would have m
ade use of speech situations, but w
hich, since it involved postulating extra-gramm
atical entities, he regarded as incapable of form
ulation.
45
3 . To be fair to Searle, be does recognize that there is a great deal of
unclarity as to what counts as one kind of m
ocutionary act, and what
counts as another (see Searle 1979 [1975a]: 29: ' ... the illusion of lim
itless uses of language is engendered by an enormous unclarity
about what constitutes the criteria for delim
iting one language game
or use of language from another').
4· Searle himself (1969:44-9) criticizes the approach, w
hich be at-tributes to G
rice, of defining m.eani."'.g in terms of 'intending to perform
a pedocutionary act', B
ut I find unconvincing Searle's claim that
rome illocutionary acts, such as greetings and prom
ises, are not associated, in term
s of their meaning, w
ith perlocutionary effects. 5· I know
of no historical studies of the influence of Interpersonal Rhe-
toric on tbe evolution of gramm
ar; but regarding Textual Rhetoric,
the views put forw
ard by Bever (I'f76) and Bever and Langendoen (I975) are relevant.
. .
6. 'This example is from
Veloo (198o: 54--5).
1· See Bach and H
amish (I979:I95-202) on 'standardized indirection'.
Bach am
i Haptish distinguish standardization, as a short-circuiting
of the process of working out m
ocutionary intent, from conven-
tiomility.
· •
· 8. O
n semantic representation, see C
lark (I976:12.-'-I4), Leech (1981 [
9· Cf Popper's form
ula representing the hypotbetico-deductive method
of science (1972:119, 242-3, 297). For an analogue in Artificial In-
telligence problem-solving algorithm
s, see New
ell (I9'J3:12/f). 10. Exceptionally, the force of an uttetim
ce is determined by conven-
tion: see 8.2.1 on 'declarations'. u
For /various versions of the 'reflexive intention' definition of mean-
ing, see Grice (1957), Searle (r¢9: 44-9), and Bach and H
amish
(I979:IJ-I5)· n.. C
f Searle's critici1lm of G
rice for disregarding the role of convention in determ
ining meaning (Searle 1969:43-4).
13. On m
eans-ends analysis in
artificial intelligence, see W
inston (1977:130-42); m
eans-ends analysis is applied to pragmatics by
Parisi and Castelfrancbi (Ig8I): ·
. ·
14. Cf B
loomfield's fam
ous parable of 'normal speech' (1933135:22-7),
the story of Jack, Jill, and the apple; Bloom
field, l'!owever, w
as not . concerned w
ith illocutions, but with stim
uli and responses .'\vitbin a behaviourist paradigm
, ie with perlpcutionary effects.
IS. The sense of 'indirect' I am using is different from
that of Searle, and hence does not in itself im
ply a conflict with his positi,on on in-
direct speech acts. Searle's position has the further difficulty that arises in cases such as Can you pass the Tim
es?, where it iS--ciearJ..1:
absurd that a question about h's weightlifting ability is intended, and
therefore the only illocutionary force one can attribute to the utter-ance is that of a request.
Chapter3
Formalism
and functionalism
For use almost can change the stam
p of nature. The Form
remains, the Function never dies.
!Ham
let, 1!1. ivj
[Wordsw
orth, The River Duddon)
As t"-No approaches to linguistics, form
alism and functionalism
tend to be associated w
ith very different views of the nature ,of
language. 1 ·
(a) Formalists ( eg C
homsky) tend to regard language prim
arily as a m
ental phenomenon. Functionalists (eg H
alliday) tend to regard it prim
arily as a societal phenomenon.
(b) Formalists tend to explain linguistic universals as deriving
from a com
mon genetic linguistic inheritance of t}le hum
an species. Functionalists tend to explain them
as deriving from
the universality of the uses to which language is put in hum
an societies.
(c) Formalists are inclined to explain children's acquisition of
language in terms of a built-in hum
an capacity to learn lan-guage. Functionalists are inclined to explain it in term
s of the developm
ent of the child's comm
unicative needs and abili-ties in society.
(d) Above
all, form
alists study language as an autonom
ous system
, whereas functionalists study it in relation to its social
function.
On the face of it, the tw
c approaches are completely opposed to
one another. In fact, however, each of them
has a considerable am
ount of truth on its side. To take one point of difference:
it would be foolish to deny that language is a psychological
phenomenon, and equally foolish to deny that it is a social
phenomenon. A
ny balanced account of language has to give atten-tion to both these aspects: the 'internal' and 'external' aspects of language. M
ore generally, my conclusion w
ill be that the correct approach to language is both form
alist and functionalist.
FORMAL A
ND
FUN
CflO
NA
L EXPLA
NA
TION
S 47
3.1 Formal and functional explanations
The approach I am taking is sum
marized in P6:
P6: G
RA
MM
ATIC
AL
EXPLA
NA
TION
£ A
RE PR
IMA
RILY
FO
RM
AL;
PRA
GM
ATIC
EXPLA
NA
TION
S ARE PR
IMA
RILY
FUN
CTIO
NA
L.
This postulate overlaps with P3 (2.3). T
o the extent that gram-
matical rules are conventional, the theory or m
odel of gramm
ar w
hich explains them w
ill be formal. In so far as the principles of
general pragmatics are m
otivated or goal-oriented, the theory w
hich explains them w
ill be functional. B
roadly, a formal gram
matical theory such as transform
ational gram
mar (C
homsky I96s:I5-18) defines a language as a set of
sentences. These sentences have meanings (senses) and pronun-
ciations, and so in effect the gramm
ar has to define a set of map-
pings whereby particular sepses are m
atched with particular
pronunciations (Chom
sky 1965:15-18). The centrallevel of syntax,
at which each sentence is represented as a string of w
ords or form
atives, is an essential component of this com
plex mapping.
Three levels of representation -sem
antics, syntactic, and phono-logical -
are therefore assumed, and the justification of dis-
tinguishing these levels is that there are many-m
any mappings
between them
. In addition to mapping rules, there are rules of
weH
-formedness, specifying w
hat is a well-form
ed or gramm
atical representation at each level.
Such a model is intended to represent w
hat native speakers im-
plicitly know to be the case about their language. N
ative speakers of English, for exam
ple, know that That girl w
ashed himself is
semantically ill-form
ed (ie nonsensical); that The purse was stole a burglar is syntactically ill-form
ed; that /dva/ and /xlep/ are un-phonological sequences in English (though not in Polish); that the sentences It is possible that not all the plates were broken and Som
e of the plates may not have been broken are capable of ex-
pressing the same sense; that W
e need more public schools is
ambiguous betw
een the two readings 'W
e need a larger number
of public schools' and 'We need schools w
hich are more public'.
A form
al theory must account for innum
erable facts of these kinds, providing an account of our linguistic know
ledge in the form
of a set of rules and categories determining the form
of linguis-tic representations at different levels. T
he requirements m
ade of this theory, as of any theory, are those of consistency, predic-tive strength, sim
plicity, and coverage of data. In this formal
sense it explains the facts of a
speaker's knowledge of his
language. Explanation in pragm
atics goes further than this, and yet is in
FORM
ALISM
AN
D FU
NC
llON
AU
SM
.•· ,·
a sense a weaker form
of explanation. It is weaker because
principles impose w
eaker constraints on language be-haviO
ur than gramm
atical rules: they can only be predictive in a probabilistic sense. O
n the other hand, it answers the question
'Why?' in a w
ay that goes beyond the goals of formal gram
mati«
cal theories. It explains that X occurs rather than Y because .X is m
ore in accord with the w
ay language functions as a comm
unia tative system
. Formal explanati!)n will alw
ays leave something un-
explained, and hence, if a functional explanation is available, we
should not hesitate to use it. Let it be added that the predomi-
of a to language study up to the present
time has led to m
appropnate attempts to fit pragm
atic phenom-
ena into theories of grmnm
ar (see 3.5). Hence functionalism
m
ay be brought in to redress a balance which has tipped in favour
of formalism
.
3.2 Biological,
and social \farieties of furectionslism
W
hat is meant by a functional explanation? It m
eans explaining why a given phenom
enon occurs, by showing w
hat its contri-bution is to a larger system
of which it is
a sub-system. A
s far as language is concerJ\ed, a functional theory is one w
hich defines language as a form
of comm
unication, and therefore is concerned w
ith showing how
language works w
ithin the larger systems of
human society. Talk of purposes, ends, goals, plans, also pres!!p-
poses functionalism. W
hen we discuss illocutions or m
eanings in term
s of intentjons (as is comm
on among philosophers -
see G
rice 1957, Searle 1969:42-50) or in terms of goals (asl did in
2.5.1), we are indulging in a functional explanation. In discussing
of language, it is better to use the term 'function', be-
cause It leaves open how far the attainm
ent of goals is due to con-scious states of the individual, or for that m
atter, whether the-
goal. is an attribute of the individual, the comm
unity, or the spectes.
Even so, functionalism is a problem
, because it appears to re-quire a non-em
pirical, teleological explanation. There is a major
exception to this: in biology, we are entitled to use functional ex·
planations in a scientific context, on the grounds that the
theory of natural selection, Darw
in 'showed that it is in principle
possible to reduce teleology to causation' (Popper 1972:267). I shall go further in sketching how
such a reduction can be made
also for linguistic functionalism, adapting for this purpose Pop-
per's evolutionary epistemology (I972:Io6ff), together w
ith his functional theory of language (1972 [I963]:134-5, 1972:II9-22).
VAtUBTIES OF FU
NcnO
NA
LISM
49
Evolutionary theory explains why a species, through elim
in-ation of its less successful stock, alw
ays ends uo more or less w
ell adapted to its environm
ent. Equally, an com
munication
system is successful in a biological sense in so far as it prom
otes the survival of the species w
hich uses it. But this biological func-
tionalism will not get us very far w
ith human language. A
lthough the language-using ability is no doubt to a considerable extent geneticaH
y inherited, linguistic behaviour itself is something that
is learned by each individual, and is passed on by cultural trans-m
ission. Other kinds of functional explanation -
and social -are required to account for the successful develop-
ment of rich and com
plex linguistic behaviour patterns in the in-dividual, and in society.
Here Popper's epistem
ological theory of the three worlds is
helpful. He argues (1972:106) that the follow
ing three worlds are
distinct domains of hum
an knowledge:
first, the world of physical objects or physical states; secondly, the
world of states of consciousness, or m
ental states, or perhaps behavioural. dispositions to act; and thirdly, the world of objective contents o
f thought, especially of scientific and poetic thoughts and works of art. A
major concern of Popper's is justifying the existence of 'the
third world' of 'objective know
ledge', or knowledge 'w
ithout a know
ing subject'. This involves showing how
language itself has been the channel w
hereby the biological level of evolution be-cam
e the basis for a more rapid and pow
erful kind of evolution, the evolution of know
ledge: 'the linguistic formulation of theories
enables us to criticise and to eliminate them
without elim
inating the race w
hich carried them' (1972:70). A
n essential part of this explanation is postulating a progression. from
lower to higher
functions in the evolution of human language. W
hereas in more
primitive com
municative system
s the expressive and signalling functions of language (corresponding to the .interpersonal func-D
Argumentative function
CD
.. t tu'
. escnptzve
nctzon
· B·s· tz· tfi
· zgna m
g unctlon
t A Expressive function
. FIGU
RE
3.1
(using language to present and evaluate argum
ents and explanatio11s) (using language to describe things in the external w
orld) (using
language to com
municate
infor-m
ation about internal states to other indi-viduals) (using language expressing internal states of the individual)
50 FO
RMA
LISM A
ND
FUN
CTIO
NA
LISM
tion of human language) are upperm
ost, Popper attributes the accelerated evolution of know
ledge w 'the tremendous biological
advance of the invention of a descriptive and argumentative lan-
guage'. The functional stages in the evolution of language from
non-linguistic com
munication m
ay be represented as shown in
Fig. 3.1 (although Popper does not himself put the functions in a
strict order). These functions form
a hierarchy, in that a higher function must
coexist with all functions low
er than itself, whereas a low
er func-tion does not necessarily im
ply the presence of higher functions. There is also, how
ever, feedback from higher to the low
er func-tions, in that once a com
munication system
has progressed to the higher functions, these functions can give rise to m
ore sophisti-cated behaviour at the low
er levels. For example, a descriptive
language permits one to describe one's internal states, and there-
by to express them m
ore explicitly than would otherw
ise be poss-ible. This hierarchy m
ay be postulated, I would add, not only phy-
logenetically, in the linguistic development of the hum
an race or species, but ontogenetically, in the developm
ent of the individual child. In Popper's philosophy of science, the evolution of theories (through the argum
entative function of language) is an analogue, on the higher plane of the third w
orld, of the Darw
inian principle of natural selection w
hich operates in biological evolution in the w
orld of physical phenomena •.
The question that Popper needs to answ
er is: How
do the evol-utionary 'jum
ps' from one level to a higher one take place? H
e offers the follow
ing illustration of how language and other form
s of leaJ;ned, socially useful products of behaviour m
ay arise from
unintended causes:
How
does an animal path in the jungle arise? Som
e animal m
ay break through the undergrow
th in order to get to a drinking-place. Other
animals find it easiest to use the sam
e track. Thus it may be w
idened and im
proved by use. It is not planned -it is an unintended
consequence of the need for easy or swift m
ovement. This is how
a path is originally m
ade .,. perhaps even by men-and how
language and any other institutions w
hich are useful may arise, and how
they m
ay owe their existence and their developm
ent to their usefulness. They are not planned or intended, and there w
as perhaps no need for them
before they came into existence. B
ut they may create a new
need, or a new
set of aims: the aim
-structure of animals or m
en is not 'given', but it develops, w
ith the help of some kind of feed-back
mechanism
, out of earlier aims, and out of results w
hich were or w
ere not aim
ed at. [Popper I9']2:II7]
VA
RIETIES OF FU
NC
TION
ALISM
5I
The following linguistic parallel m
ay be offered. The expressive behaviour of new
-born babies (eg crying) produces a reinforcing reaction on the part of the m
other . .But it is not dear at what
point the unintended behaviour patterns of the baby become
rudimentary intentions, and thence fully-form
ed intentions, so advancing the baby's com
municative role from
the involuntary ex-pressive stage to the deliberate signalling stage. 2
Popper does not claim that either his three w
orlds or his four language functions are exhaustive. I therefore do not feel I am
contradicting his fram
ework in proposing to extend his three
worlds to four. The m
issing link, in Popper's evolutionary epistem-
ology, is a world of societal facts (or w
hat Searle has called 'in-stitutional facts') intervening be.tw
een his second (subjective) and his third (objective) w
orlds. Thus Popper's objective 'third world'
becomes, in this redefinition, a 'fourth w
orld' (Table 3.1). The four w
orlds now ascend by a natural order of em
ergence to the objective w
orld of facts, and Popper's four functions of language provide the m
eans of transition whereby one w
orld could have em
erged out of another. The expressive function of behaviour m
ay evolve in a purely physical world, as a sym
ptom of som
e biological state: a bird's fluttering its w
ings, for example, as an in-
dication of alarm. B
ut once this symptom
is reinterpreted as an expressive sign of the internal state of the individual, w
e have ac-cess to a w
orld of subjective experience. The event may be the
same, but its interpretation as representing the internal state or
disposition of the animal opens up a new
world of possibilities.
The. next step on the ladder of comm
unication admits us from
a w
orld of psychologiCal to a world of societa,i phenom
ena. We
treat our pets as comm
unicating expressively when they exhibit
signs of hunger, excitement, pain, etc.; but at som
e undetermined
stage the same pets can, by an act of reinterpretation, be sup-
posed to 'signal' to us information about their inner states, and so
to act in a goal-oriented, comm
unicative way. O
nce this step is taken, entry has been m
ade to a world of social objects, states,
and events. This is no longer a subjective world, but rather an in-
tersubjective one, for, as the triangulation performed by m
ore than one observer can establish the location of som
e observed ob-ject, so a num
ber of individuals from the sam
e social group can m
utually confirm the m
eaning of some phenom
enon which is ex-
ternal to all of them. O
n the basis of such confirmed com
muni-
cative values there may arise social institutions such a" ow
nership, m
arriage, rights, obligations: these 'institutional facts' could not exist outside a w
orld in which the 'signalling' function of com
-m
unication has established a reality beyond the individual. The
52
.sa .. s:: ::s
0 !3--S'g 8.s
FORM
ALISM
AN
D FU
NC
TION
ALiSM
t>()
.!3 .§ .... ] 8
53 intersubjective w
orld {)f -social fact, in turn, becomes the pre-
requisite of the descriptive function of language. Concepts of refer-
ence, truth, and falsity could not exist outside a social world in
which individuals can share and com
pare their descriptions of reality.
The last w
orld, the objective world of W
orld 4, can in turn be explained -as evolving out of the descriptive function of language. T
he hypostatization of descriptions of the world through the
descriptive functioning of language can easily lead to a reinter-pretation w
hereby the appropriateness of 'true' and 'false' is judged not by direct m
eans (ie comparing a description w
ith the reality it is m
eant to describe), but by indirect means (ie judging
truth value on the basis of reasoning and argument). O
nce this step has been taken, w
e are able to postulate the eXistence of
facts which are independent of the observations by individuals, or
even by groups of individuals. This account of the developm
ent of language functions can be readily observed in children's language developm
ent, but cannot be observed in an evolutionary sense. O
ur knowledge of the ori-
gin and evolution of language is limited in large m
easure to that very brief and recent period of hum
an development for w
hich we
have historical records. Even so, it is perhaps not unreasonable to suppose that the argum
entative function of language could not have developed its full potential until the invention of w
riting . W
ithout the means to record linguistic m
essages, so that ad-dressers and addressees can be w
idely separated in time and space,
it is difficult to conceive of the existence of 'objective knowledge' in
Popper's sense, ie knowledge w
hich exists independently of any know
er. Popper's examples of such know
ledge (eg mathem
atical know
ledge, knowledge stored in libraries, scientific know
ledge) all presuppose a w
ritten medium
. Lines C, D
, and E of Table 3.1 show how
there are parallels of functional adaptation at the levels of the four w
orlds, in each of w
hich we can observe a principle of adaptation by som
e 'organ-ism
' to its environment through the transm
ission of information.
In the biological sphere, information is passed from
one gener-ation to another genetically and the unit of transm
ission here is the species. In the psychological sphere, the individual m
ember of
the species can transmit inform
ation to itself in the sense of learn-ing through positive and negative reinforcem
ent of previous behaviour patterns.
The adaptive
mechanism
is
not natural selection, but its psychological analogue, in the life history of the individual, of conditioning: a process w
hereby unsuccessful be-haviour patterns are abru;'doned and successful ones adopted. In
54 FO
RMA
LISM A
ND
FUN
CTIO
NA
LISM
the cultural or societal sphere, the social group a
small group such as a tribe, or a large
such as a clVlhzatH_m)
transmits inform
ation to new generations through
of its new m
embers. In this w
ay, a whole society m
ay be smd to_
learn from the experience of previous generations. In the
ot technology, for exam
ple, . the present generation can av?Id the necessity of having to invent the w
heel anew,
of un-
successful attempts at flight undertaken by-
avmtors of
the nineteenth century. Finally, the same parad1gf!l of
through error elimination is observed in the area of 1deas, as 1Hus:
trated pre-eminently in the
ot science. 'In this w
orld we can m
ake theoretical m
. a sim
ilar way to that in w
hich we m
ake ?iscovenes m
w
orld 1,' says Popper. For exam
ple, 'we dtscover pn?le num
bers, and Euclid's problem
of whether the sequences of p_nm
e nun:bers is infinite arises as a consequence' (1972:74). AU these m
ents are due in large measure to the fact
man, the speakm
g anim
al, can pass
on the fruits
of expenence by
means of
language. .
. .
Popper's hierarchy of worlds ts
for. understand:ng w
hat we are doing w
hen we study
Smce the fou. th
world incorporates the other three w
orlds, can subsum
e the study of physical, psychological, and social reality.
ques-tion therefore arises: w
hat kind of world are w
e studymg w
hen w
e study language? There is a hierarchy of corresponding to the four w
orlds. _The most ?asic type, wh1c_h
treats language as purely physical, 1e as to W
orld I, :s plainly inadequate; how
ever, we noted earher (p
that th1s view
has not been universal: that the post-Bloomfield1an
alists regarded linguistics at least by aspiration as a physical sci-ence. The second theory-type, w
hich treats language as a mental
phenomenon, is the type advocated by
and the generative gram
mar school. A
defect of this theory IS that cannot handle social facts about language; and a further defect IS that in consequence, it cannot ueyond the linguistic com
petence of the md1vtdual.
has been disguised by Chom
sky's claim to be dealm
g With the
knowledge of the 'ideal native speaker-hearer' -
an abstract and fictional version of the individual.
The third theory-type treats language as a social_ Exam
ples of such theories are those of Saussure, day. B
ut Saussure's concept of a_ language as a soctal w
hich exists apart from any particular m
embers of the hngm
sttc com
munity is
already half-way tow
ards fourth
and final theory-type, w
hich regards language as an mm
ate of the fourth
VA
RIETIES OF FU
NC
TION
ALISM
55
world -
the world of objective know
ledge. It is this which gives
point to Saussure's observation that language is 'outside the indi-vidual w
ho can never create nor modify it by him
self (Saussure 1959 [ 1916] : 14). Just as, in Popper's exam
ples, the knowledge
contained in books and logarithm tables m
ay exist outside the subjective know
ledge of any living individual, so a language may
exist apart from the speech com
munity to w
hich it belongs. The existence of Latin, for instance, does not depend on the survival of a Latin-speaking com
munity. Further, a language m
ay exist even though no one person exists w
ho can speak, read, or under-stand it. A
lthough at face value this seems absurd, there is noth-
ing odd about saying that the Etruscan language exists, even though the Etruscan language is at present pot know
n by anyone in the w
orld. In fact, it would be rather perverse to take the con-
trary view, and to claim
that when the com
munity of Etruscan-
speakers died out, the language thereby ceased to exist; for if scholars decipher Etruscan w
ritings in ten years' time,
their achievem
ent will not be in inventing the language, but rather in
rediscoveringit. It is in this sense that languages exist in an auton-om
ous world w
hich cannot be reduced to a World 3 of social
phenomena, a W
orld 2 of mental phenom
ena, or a World I of
physical phenomena. I therefore propose that a linguistic theory,
properly regarded, is a World 4 theory about a W
orld 4 phe-nom
enon. This position, although it has not been overtly adopted by C
homsky, is covertly adopted by him
when he argues that the
adequacy of gramm
ars (eg in representing the boundary between
gramm
atical and ungramm
atical sentences) can be directly check-ed against the intuitions of native speakers. It is reasonable to argue, indeed, against m
ore orthodox views of em
pirical con-firm
ation, that linguistic knowledge is public know
ledge, because native speakers, in sharing a language, also share a com
monality
of implicit linguistic know
ledge. Thus although our introspections are private, the data w
hich we obtain through introspection are
public and objective, being available for corroboration by the in-trospections of other people. (This does not m
ean that reports of intuitions are alw
ays clear and free from error.) C
homsky's m
en-talist position w
ould .be indefensible if he really claimed, as he
appears to believe, that the private and subjective judgements of
the native speaker are the basis for determining the descriptive
adequacy of a gramm
ar. 3 In practice, Chom
sky's position has been m
ore realistic. And in fact all those linguists, including
Saussure and Chom
sky, who have m
ade language, as a system
abstracted from particular speakers and hearers, the focus of
their studies, have unwittingly taken up a W
orld 4 position. If w
e now return to the distinction m
ade in Chapters I and 2
56 FO
RM
AU
SM A
ND
FUN
CTIO
NA
USM
r· between gram
mar and pragm
atics, it can be seen that the gram-
mat is a 'W
orld 4' phenomenon, and that linguistics is unique
·among scientific disciplines, in that it aim
s to provide WOI'ld 4
p/lmati(;Jns of World 4 phenom
ena. (This reflexive characteriz-ation of linguistics m
ay go a long way tow
ards explaining the peculiar difficulties of the discipline.) Pragm
atics, on the other hand, deals w
ith relations betw
een language as a W
orld 4 phenom
enon, and language as a World 3 (social) phenom
enon. G
ramm
ar, studying language as a thing in itself, provides formal
studying in relation to society
as a whole m
World 3, m
ms at a functional perspective.
· B
ut the hie!archy of worlds is not
matter of one-w
ay depen-dence. There ts the requirem
ent that we interpret higher levels as
being. realized in phenom
ena, and as evolving dia-.
chrontcally under thelf mfluence. A
nd there is also the require-m
ent that the lower levels be interpreted in the light of how
language
with m
ore general societal and mental dom
ains. in spite of the
validity of the postulate P6, we rec-
ogmze not only that functtonal explanations play a role in gram
-m
ar, but that formal explanations play a role in pragm
atics. 3.3 The Ideational, interpersonal, and textual functions of
language From
Popper's four language functions, I move on to those of
Halliday, to w
hich they bear a strong resemblance. H
owever,
whereas H
alliday treats all his fUnctions as being intrinsic to
gramm
ar, I interpret them
P7: G
RAM
MA
R IS
IDEA
TION
AL;
PRAG
MA
TICS IS
INTER
PERSO
NA
L A
ND
TEXTU
AL.
Halliday's three functions (see H
alliday 1970, 1973) are: (a) The ideational function: language .functioning as a m
eans of conveying and interpreting experience of the w
orld. (This function is subdivided into tw
o sub-functions, the Experien-tial and the Logical sub-functions.)
(b) The interpersonal function: language functioning as an ex-pression .of one's attitudes and an influence upon the atti-tudes and behaviour of the hearer.
. (c) The
function: language functioning as a means of con-
structmg a
text, ie a spoken or w
ritten instantiation of language.
Functions (a) and (b) subsume Popper's. four functions in the fol-
lowing w
ay. The ideational function is an amalgam
of two sub-
FUN
CTION
S OF LA
NG
UA
GE
57 functions w
hich Halliday calls 'experiential' and 'logical', and
which· correspond to Popper's descriptive
and argumentative
functions. The interpersonal · function corresponds to Popper's expressive and signalling functions, which are based on sim
ilar functions (Ausdruck and Appell) distinguished by K
arl Buhler
(1934). Halliday has explained that he finds it unnecessary to
maintain this distinction w
hich Buhler, Popper, and aiso Jakob-
son (1g6o) have drawn betw
een functions oriented towards the
speaker's and the hearer's ends of the comm
unicative process: he sees the expressive and signalling functions ·are being m
erged in a single interpersonal function. For the purposes of pragm
atics, I agree w
ith him: in 2.4 I have already suggested that there is no
point in distinguishing between s's m
eaning and h's meaning. The
third function of Halliday, the textual function, is of a very differ-
ent status from the others. H
alliday gives it the special status of. an 'enabling function', and says that it is instrum
ental to the other tw
o (1970:143, 165). I shall argue that although the textual organiz-ation of language plays an im
portant part in an overall functional account of language, it is m
isleading to call the textual function a 'function' at all: there is som
ething back-to-front about saying that language has the function of producing instantiations of it-self. It is not language that has the function of transm
itting itself · through texts, but texts that have the . function of transm
itting language.
My m
ain disagreement w
ith Halliday, how
ever, is over his w
ish to integrate all three functions within the gram
mar. I m
ain-tain, in contrast, that the ideational function belongs to gram
mar
(which conveys ideasto the hearer through a sense-sound m
ap-ping), and that the interpersonal function and the textual 'func-tion' belong to pragm
atics. From the· speaker's point of view
the Interpersonal R
hetoric and the Textual Rhetoric m
ay characterized respectively as 'input constraints' and 'output con-straints' on the gram
mar (Fig; 3.2). From
the hearer's point of view
, these constraints are reversed, so that the Textual Rhetoric
constrains the input, and the Interpersonal Rhetoric constrains
the output of the decoding process. Although H
alliday insists that the three functions are of equivalent status, he does drop one or tw
o hints as to the special importance of the iCieational function.
While he deprecates, for exam
ple, the popular view that language
is a vehicle of ideas, he concedes that 'the ideational function .... is a m
ajor component of m
eaning in the language system w
hich is basic to m
ore or less all uses of language' (1973:38-9). My ow
n opinion, in contrast to this, is that the popular view
of language is essentially correct: that it is the ideational function (subsum
ing,
FORM
ALISM
AN
D FU
NC
TION
ALISM
/ lnte""'"""al Rbetorie
(;nput oomtnrint.)
17agL -
{;deat;onai)
Phonology
Textual Jhetoric (output constraints)
FIGU
RE 3.2
as we have noted, Popper's descriptive and argum
entative func-tions) w
hich makes hum
an language what it is: an extraordinarily ·
powerful instrum
ent of thought and comm
unication. Without the
ideational component, the gram
mar, w
e could well be in the
same com
municative league as gibbons and chim
panzees. I have criticized H
alliday elsewhere (Leech 1980:22-5) for
what I regard as a tendency to 'overgram
maticization': that is, a
tendency to seek a gramm
atical explanation (in terms of rules and
categories) of the interpersonal and textual aspects It is notable that recently, how
ever, Halliday (1980:66-70) has
presented an altogether more flexible concept of gram
mar, in
which the interpersonal and textual functions are associated w
ith non-discrete types of structure, w
hich he calls 'prosodic' and 'periodic'. It is also notable that he has draw
n attention to a fac-tor shared by the interpersonal and textual functions -
namely,
the fact that they each have a speaker-oriented and a hearer-oriented aspect. In these respects, H
alliday seems to be m
oving closer to a conception of language in w
hich the ideational com-
ponent is gramm
atical in an orthodox sense (dealing in con-stituent structures, rules, and system
s), as distinct from the inter-
personal and textual components, w
hich are more pragm
atic in conception.
·
J.J.I A process m
odel of language .
To show how
the interpersonal and the textual pragmatics fit into
an overall functional view of language, I shall build on Fig. 3.2 so
as to represent both the speaker's and the hearer's ends of the com
municative
process. 4 In
the follow
ing diagram
, w
hich show
s linguistic comm
unication in terms of m
eans-ends analysis
FUN
CTION
S OF LA
NG
UA
GE
(ideational)
Spea H
earer Text
(textual)
FIGU
RE 3·3
59
(see 2.5.1), the three functions of Halliday form
a hierarchy of instrum
entality. In Fig. 3.3, a linguistic act of comm
unication (an utterance)
is described
as constituting a transaction on
three different planes: as (a) an interpersonal transaction, or D
ISCO
UR
SE;
as (b)
an ideational
transaction or
MESSA
GE-
TR
AN
SMISSIO
N; and as (c) a textual transaction or TEX
T. 5 But
these are ordered such that the discourse includes the message,
and the message includes the text. H
ence the whole utterance
may be described as:
·
. [DISC
OU
RSE
I by means of !M
ESSAG
E] by means of [TEX
T J The discourse is the w
hole transa.::don, regarded as an attempt to
convey a particular illocutionary force to the hearer. The goal w
hich s has in state I is fulfilled as the force of the discourse is understood by h. This successful outcom
e is signified by the final state 6. (The term
'discourse' is used in preference to 'illocution' or 'illocutionary aft' w
hich would also be an appropriate term
for the w
hole transaction. But 'discourse' suggests that the field of
activity in fact contains a sequence of illocutions. :i do not, that is, w
ish to limit the value of Fig. 3·3 by suggesting that it only ap-
plies to single utterances. On the other hand, I do not w
ish to em
bark upon the particular problems of analysing continuing dis-
course -a task w
hich is best left to discoui.'se analysts (see further pp 231-2). ·
In order to achieve the goal indicated by 6 in the diagram, s
must choose a sense (or ideational content) w
hich conveys the in-tended force. This stage (1-2 on Fig. 3.3) is the one at w
hich the rhetoric, including the C
P and the PP, imposes 'in-
put constraints' upon the message. A
ssumfug that the m
essage is correctly transferred to h, h m
ust go through the parallel stare
6o FORY..ALISM
AN
D FU
NCTIO
NA
LISM
(5-6) of working out the force. The m
essage itself has to be E
NC
OD
ED
(stage
2-3) syntactically
and phonologically
(or graphologically) as a text, w
hich is a linguistic transaction in actual physical fonp. (either auditory or visual). This stage (3-4) precedes the opposite process of D
EC
OD
ING
the text into its form
.as a MESSA
GE (stage 4-5).
The encoding stage (2-3) is essentially a gramm
atical process of m
apping the sense to an appropriate phonetic output, as sketched in Fig. 3.2. It is undertaken, how
ever, under the control of the principles of Textual R
hetoric, which help to determ
ine the stylistic form
of the text in terms of segm
entation, ordering, etc. Like the Interpersonal R
hetoric, the textual rhetoric is based on speaker-hearer cooperation, a textually 'w
ell-behaved' utterance being one w
hich anticipates and facilitates h's task iii decoding, or m
aking sense of, the text. The constraints of Textual Rhetoric
also operate to stage 4-5, which stands for h's phon9logical, syn-
tactic, and semantic decoding of the text. From
h's point of view
, they take the form
of.expectations which, so long as they are ful-
ease the decoding process. For example, h m
ay feel en-titled to expect that s w
ill observe normal
constraints, and w
ill avoid ambiguities. In conversation, these expectations
are often disappointed: s faces the problem of sim
ultaneously planning and. executing the utterance, and this is the unintended cause of m
any factors of 'normal non-fluency', such as false
starts, syntactic blends, and other gramm
atical or textual infel-icities. To produce a w
ell-behaved text is to coordinate a number of
complex skills, and it is not surprising if failure in the perform
-.9f these skills often leads to a rhetorically 'unhappy' utter-
ance.' Eqr this reason, it is in written language (w
here planning execution
be separated . in time) that the operation
01 the 1 extual Rhetonc can be observed m
ore directly. It w
ill be noticed that Fig. 3.3, in representing a functional m
odel of language, also provides a PRO
CESSIN
G m
odel, in which
the of linguistic production and interpretation are spelled
out in a means-ends chain. There is no harm
in construing the m
odel as such, so long as we rem
ember that there is no necessary
inference from 'm
eans-+ goal' to 'before'-+ after'. As an attem
pt , to
processing in real time, Figs 3.2 and 3·3
would fad m
fauly obvious ways.
Fig. 3.2 would suggest that
"we process a
semantically in its entirety before w
e .syntactic encoding, and that w
e process it syntac-tically ,an tts entirety before. w
e embark. upon phonological pro-
.eessing. But psycholinguistic evidence show
s that in both the -enoodin..g and decoding of utterances, the different levels of linguis-
FUN
CI10N
S OF LA
NG
UA
GE
tic processing, frequently (perhaps typically) are simultaneous-
ly in operation. 6 Similarly, Fig. 3·3 w
ould suggest that we plan the
utterance as a discourse before we tackle its. encoding as a text.
But again, it is com
mon experience that .we often start talking
without being sure of w
hat it is, in entirety, that we w
ant to say, and that w
e often change and modify our illocutionary goals in the
course of speaking. Nevertheless, I propose that the m
odel of Fig. 3·3 has m
ost of the design features we w
ould need to build into a real-tim
e model of language-processing. To. m
ake sucfr a m
odel more adequate, w
e should have to recogniZe tflat a text is in itself a phenom
enon which unfolds in tim
e, and that all the com
ponents of Fig. 3·3 can themselves undergo tem
poral pro-gression. Fis-3·4 (in w
hich sequence in time is indicated by
the variables a, b, c, ... ) gives a better approxim
ation to a real-rune language-processing m
odel:
Discourse
___________ ..,... L-.r....L...,....a.-.-"--· -·
(interpersonal} ·
FJGtl'ltE 3·4
The m
oden have adopted is 'functional' in an everyday sense:.. it show
s how the various elem
ents of gramm
ar and'. rhetoric con-tribute to the functioning of language in the service of goal-directed behavfour. Thus, although I have retained H
alliday's three functional labels; l
have construed jun&tion in a goal-oriented sense ·w
hich is not evident from H
alliday's use of the term
. Halliday's reference to the 'te:lttual fuD
ction' Of language as an •enabling functiO
n.-can be IJHlde sense of in this framew
ork, as the teXt functiO
ns as a means of transmitting the m
essage; just as the m
essage functions as a means of transm
itting the illoeutionary. force to the ddressee. H
owever, rather than to.say w
ith Halliday
that a language has a te:lttual function, it would m
ake better to say that a text has a linguistic function -
a function in the com-
mtm
icatioaoUinguistic m
essages. In this m
odel, the interpersonal function appears to be super-ordinate to the other tw
o. We should, how
ever, distinguish two-
FORM
ALISM
AN
D FU
NCTIO
NA
LISM
ways in w
hich .one function can be subordinate to another. It can be inferior to the other function in the chain of events in a m
eans-ends fram
ework (in this sense the ideational function is inferior
to the interpersonal). But it can also be subordinate in the sense
of being less organizationaUy developed, .or less im
pqrtant in its contribution to the total m
eaning. In this latter sense, the inter-personal function is often -
perhaps usually -:-subordinate to the ideational function. Its contributiop t9.
'Yhole m
ay· be great (?gin the case of indirect illocutions), putif m
ay .also be negli-gible, as in the case of straightforw
ard informational uses of lan-
guage. The m
ost 'desituationhlized'. qf language -those in
which interpersonal pragm
atics has least play-
are those (eg encyclopaedia articles) in w
hich the World 4 com
munication
of 'objective knowledge' is param
ount. ·
3.3.2 An illustration The follow
ing is a simple exam
ple of how the m
odel by Fig. 3·3 (or for that m
atter Fig. 3.4) works out in practice.
The situation I propose to examine is.that of ordering a m
eal in a restaurant.
I The. custom
er(s) wants to be served w
itp Steak Diane.
2 In order to attain the goal in I., s chooses a m
essage, an ideation of this illocution: ie s form
ulates the proposition 'I'd like Steak D
iane'. Note that s oouldi have form
ulated this speech act in m
any other ways: for exam
ple, more econ-
omically but brusquely ('Steak D
iane'), more politely ('I
would like the Steak D
iane, please'), more dictatorially
('Waiter, bring m
e some Steak D
iane'). The choice tw
een these is determined in part by the extent to w
hich the situation dem
ands politeness: eg choosing a dish l!t a restaurant is in this respect different from
choosing a dish at a private dinner party.
3 In order to convey the
s encodes itas, a text, and the phonation of the sentence I'd like Steak D
IAN
E (w
ith a falling nucleus on D
iane) results. (Note that s could have
chosen otper texts. to express the saine proposition: eg: Steak D
iAN
E I'd like. This utterance w
ould be less in accord w
ith Textual Rhetoric, how
ever, as it would violate
the end-focusmaxim
.) 4
The text is heard by h. .
. .
5 Then h decodes it into a m
essage, which (if the trans-
mission is S\lccessful) has the sam
e. sense as the original m
essage at 2. ·
6 Finally, h interprets the force of the m
essage, which (if
FUN
CTION
S OF LA
NG
UA
GE
transmission of the discourse is successful) is recognized as
s's giving an order for Steak Diane to h. The order is re-
latively· indirect, in that s sim
ply states his preference, leaving it to h to w
ork out what is intended. B
ut this force is determ
ined by well-established im
plicature. The Interpersonal R
hetoric enters into this example in the stages
1 -2 and 5.:-6, in that s clearly relies on h's deriving an im
plicature 's w
ants to be served with Steak D
iane', and that this derivation depends on h's assum
ption that s is observing the CP and (poss-
ibly also). the PP. The role of the Textual R
hetoric is not so obvious. B
ut to textual choices let us im
agine that .the w
aiter has returned with the dishes for both sands's fellow
-diners. A
t this points might say either (a) M
ine is the Steak J)iane, or (b) The Steak D
iane is mine. A
lthough they have the same sense,
there two .utterances w
ould be appropriate to different contexts: (a) w
ould be appropriate where the w
aiter was trying to rem
em-
ber which dish shad ordered (as contrasted w
ith what others had
ordered); (b) would be appropriate w
here the waiter w
as carrying the Steak D
iane, and trying to remem
ber to whom
that dish should be given. M
ore generally (a) would be appropriate w
here. M
ine (the fact of identifying s as the recipient) was given inform
-ation, and the Steak D
iane was new
information; w
hereas (b) w
ould be. appropriate where the given-new
relations were re-
versed. One m
ay assume here that the end-focus m
axim w
ould be observed, and that each utterance w
ould be pronounced with
the nucleus as indicated here: · (a) M
ine is the Steak DiANE. (b) The Steak D
iane is MlNE.
Suppose, however, that the nucleus w
ere placed on the subject (w
ith a subsidiary rise on the complem
ent): -
(a') MiNE is the Steak D
iane. (b') The Steak DiANE is mine. 7
In these cases the intonation would indicate a reversal of the
given-new relations as indicated in (a) and (b). The end-focus
maxim
would be violated, but in these cases probably not to an
extent to cause any difficulties for the hearer. The most w
e could say is that the decoding task w
ould be to some extent m
ore com-
plex in the cases of (a') and (b'), as'compared w
iu (a) and (b). This is an exam
ple of how the Textual R
hetoric fits into the total com
municative proces;>.
3·3·3 The textual pragmatics
The textual pragmatics has so far been chiefly illustrated only by
one maxim
: the Maxim
of End-focus. I shall now propose a
FOR
MA
LIS.r-AN
D FU
NC
TION
ALISM
scheme for the Textual R
hetoric which resem
bles that already proposed for the Interpersonal R
hetoric: there is. a set of four principles, and each principle can be
into maxim
s. T
he four principles are those outlined in Slobin (1975):
I. 'Be hum
anly processible in ongoing time';
2. 'Be dear';..
3· 'Be quick and
4. 'Be expressive'.
Slobin's reasons for postulating these principtes are somew
hat different from
the present one-He argues that these precepts are
observed by languages themselves, ·rather than by the users of
languages. Thus under conditions of change, languages will al-
ways tend to change in directions w
hich preserve these principles. This is
whether w
e examine diachronic language developm
ent (eg the drift tow
ards analyticity of Indo-European languages}, or the child's acquisition of language, or borrow
ing between lan-
guages in--contact, or the evolution of creoles from pidgins.
I have no reason to quarrel with the evidence Slobin presents
for arguing that these principles are actuaUy at w
ork in languages them
selves: such arguments are consonant w
ith the case for re-garding gram
mars (as I have suggested) as being under the func-
tional influence of pragmatics. H
owever, m
y present interest is in observing these principles in the exercise of styliStic preferenee in language use. I shall label Slobin's principles as foltow
s:
1. The processibility principle
This principle recomm
ends that the text should be presented in a m
anner which m
akes it easy for the hearer to decode in time. A
text (in contrast to a message). is essentially linear and tim
e-bound: thus in encoding w
e are on en presented witlf choices as to
(a) how to segm
ent the message fhto units; (b) how
to assign de-grees of prom
inence or subordination to different parts of the m
essage; and· (c) how to order the parts of the tnessage_ These
tfiree types of decisions are interrelated. For example,. the End-
focus Maxim
applies to tone-units and therefore its operation is dependent on Iogicaliy prior choices regarding the segm
entation of the utterance into tone-units; the segm
entation decision im-
plies a decision about focus (which part of the tone-unit w
ilt be signalled u
prominent by m
eans of the nucfear tone); and the End-focus M
axim im
plies that this in-
entails a cision about order. I asfl:um
e that the· End-focus-Maxim
-is func-tional-in that it facilitates phonological decoding of the m
essage. T
he way in w
hich it does th.is is not entirefy ciear, 8 but-the fact
FUN
CTIO
NS O
F LAN
GU
AG
E
t this principle is a universal or of langu.age
. and
I977:548) in itself prov1des a reason for bebev-'ng its 'functtonahty.
· 1
· The Processibility Principle applies not c:_nly to phonolog1ca , but also to syntactic and sem
antic aspects 01 the text. For le
re arding syqtactic ordering, we ma_y postulate for
a tx '·
g of End-weight w
hich (broadly) mduces a syntacttc struc-
h" h 'light' precede 'heavy' ones. H
ence the tu:re m
w IC
• f
. ht characteristic English
has a predomm
ance o.
ng -
branching over left-branching, and many
transfor-m
ations (eg the rule of ex:traposition) serve the Mrunm
of _End-w
eight by helping to ensure that complex constituents are placed
at the end of a clause or sentence. That Sim
on will resign is on the cards. It is on the cards that Sim
on will resign.
A ain the exact form
ulation of and motivation for this
g 't
l r
but that it exists in some form
or other for Enghsh are no c ea ,
) 1 ·
h rdly to be d other SV
O (subject-verb-object
anguages ts a
. :ubted.9 N
otice that the two m
axims, that ?f
and that of End-w
eight, although they. operate on dtfferent, levels of j_qg (phonological and syntactic), tend to be m
u!uatly supportmg.
a com lex constituent also tends to be a constituent w
htch con-tains :m
ajor focus of new information; and
are therefore likely to be tw
o reasons for wishing to place At at the end of the
sentence. Parallel to the End-weight M
axim
appears on the sem
antic level, an End-scope Thts
,hat operators such as a negative
or .a rather than follow
' the elements
other .ogtca w
hich are within their scope. Thts m
axtm w
ould ex-plain the preferred readings of [I} and [2]:
{I] Evecyone in the room know
s at least two languages. in the
12] . At . teast tw
o languages are known by everyone
room. 1Q
· P.rden:ed readillg-of!I]: --('lfX
X &
INR
OO
MX
) (3y"" 2 (L
AN
GU
AG
E Y &
KN
OW
X,
y)))
P.referred reading oflz]: (3}'""z (LA
NG
UA
GEJI &
('tfx ((PER
SON
X & IN
RO
OM
XJ ._(K
NO
W
.
66 FO
RMA
LISM A
ND
FUN
CTIO
NA
LISM
Although there has been considerable controversy over the inter-
pretations of sentences such as these, an explanation which broad-
ly fits the facts is that both readings are possible with both [I] and
[2], but that in [1] the reading in which the existential quantifier is
within the scope of the univers?! quantifier is greatly preferred,
whereas in [ 2] there is an equally strong preference for the read-
ing in which the scope-relations are reversed. This preference
may be regarded as a pragm
atic preference, and it follows from
a m
axim corresponding to the End-w
eight Maxim
. Just as the End-w
eight Maxim
gives preference to greater depth of bracket-ing to the right on the syntactic level, so the End-scope M
axim
gives preference to greater depth of bracketing towards the right
on the level of semantic representation. It can be surm
ised that the End-w
eight and· the End-scope Maxim
s are motivated by
similar restrictions on hum
an mem
ory capacity in the left-to-right parsing of tree-structures.
2. The clarity principle The C
larity Principle again applies to different levels of coding, but in general it m
ay be broken down into tw
o maxim
s, (a) a. Transparency M
axim, and (b) an A
mbiguity M
axim:
(a) Retain a direct and transparent relationship
seman-
tic and phonological structure
(ie betw
een message
and text). 11
(b) A void am
biguity. (O
n the overlap between these and G
rice's Maxim
of Manner,
see 4.5.) ·
To illustrate· (a) on the syntacti:: level: for clarity's sake, is a
. good idea for semantically adjacent item
s also to be syntactically adjacent. 12 It is for this reason that discontinuous structures in syntax tend to cause problem
s of understanding: [3] The m
orning came at last w
hen we were due to leave.
The separation of the modifying clause w
hen we were due to leave from
its head morning obscures the relationship betw
een argu-m
ent and predicate. The requirement to avoid am
biguity is close-ly connected w
ith transparency, but it can be important in its Q
WD
right. For exam
ple, ambiguity occurs notoriously w
ith pronoun anaphora:
[4] If the baby won't drink cold m
ilk, it should be boiled. w
e may extend the avoidance of am
biguity to include
FUN
CfiO
NS O
F LAN
GU
AG
E
ance of 'garden path' ambiguities, ie syntactic am
biguities which
are temporary, and are resolved by a later part of the sam
e sentence: 13
[5] Before we started eating the table was absolutely loaded w
ith delicacies.
• It could be argued that such cases do not ultim
ately lead to a loss of clarity ( eg the reading that som
eone was eating the table, in
[5], is soon ruled out by subsequent context). But the sam
e point can be m
ade about all ambiguity: the danger from
ambiguity is
not :.o much that it w
ill end by misleading h, as that it w
ill con-fuse and delay h's interpretation of the sentence. In this respect, the C
larity Principle might be regarded as subordinate to the Pro-
cessibility Principle. 3. The econom
y principle The Econom
y Principle ('Be quick and easy•) can be regarded as
a valuable precept not only for h but for s. If one can shorten the text w
hile keeping the message unim
paired, this reduces the am
ount of time and effort involved both in encoding and in de-
coding. As this description im
plies, the Economy Principle is con-
tinually at war w
ith the Clarity Principle. O
n the phonological lev.el, for exam
ple, economy favours elisions, assim
ilations, and other abbreviating and sim
plifying processes. But obviously to
maxim
ize the 'principle of least effort' in this way w
ould be to m
ake the text unintelligible. In practice, a balance has to be struck betw
een saving t_ime and effort, and maintaining intelligi-
bility. This balance will clearly depend in part on contextual fac-.
tors, such as the physical distance between sand h, and the social
predictability of the message .
Similarly, on the syntactic level, the Econom
y Principie has a contributory M
axim of R
eduction which m
ight be simply enunci-
ated as 'Reduce w
here possible'. But reduction should evidently
not be recomm
ended where it leads to am
biguity. The processes w
hich are subsumed under the heading of 'reduction' here are (a)
pronominalization, (b) substitution by other pro-form
s, eg: do, so, and (c) ellipsis (or deletion). For exam
ple, sentenfe [4] above is an exam
ple of injudicious pronominalization: in oriler to avoid
ambiguity in this case, s w
ould have to sacrifice economy .by re-
peating the noun milk:
[6] If the baby won't drink cold m
ilk, the milk should be
boiled. The sam
e considerations apply to other forms of reduction, for
68 FO
RMA
LISM A
ND
FUN
CTIO
NA
LISM
example to pro-form
substitution and ellipsis: [7a] Jam
es enjoys golf more than Jam
es enjoys tennis. [7b] Jam
es enjoys golf more than he does tennis.
[7c] Jam
es enjoys golf more than tennis.
Of [7a]-[7c], the longest sentence [7a] is the 'unhappiest' variant
and the shortest one [7c] is the 'happiest'. Thus far the Principle of Econom
y dictates the preferences. But if. the m
ost reduced form
, as in [8c] below, introduces am
biguity, then a less reduced but unam
biguous sentence, eg [8b}, will probably be preferred:
[8a] James likes M
ary more than D
oris likes Mary.
[8b] James likes M
ary more than D
oris does. [8c] Jam
es likes Mary m
ore than Doris.
The pragmatic. point about reduction is that it abbreviates the
text, and often simplifies its structure, w
hile maintaining the re-
coverability of the message. It is w
hen, for some reason, the
message's recoverability is im
paired that reduction comes into
conflict with the C
larity Principle.
4· Tire expressivity principle The fourth principle is m
ore diffuse and difficult. to define. It is easier to say w
hy we need it, than w
hat it consists in. If the three principles of Processibility, . Clarity, and Econom
y were· the only
pragmatic factors constraining the· form
of texts, language would
be limited to efficient, but pedestrian transactions. W
ith the Expressivity Principle. w
e are concerned with
in a broad sense· w
hich includes expressive and aesthetic· aspects of com
munication, rather than sim
ply with
For example,
anJconicity Maxim
(which invites the user, all other things being
equal, to mak-e the text im
it.ate aspects of the message) should be
included in it. 14 For the present, we m
ay-note the influence of the . Expressivity Principle in inhibiting reduction:
19] John Brow
n was guilty of the crim
e, and John Brown
would have to pay for it.
. f IO] They put in the best they had and
put .in the best we
had ..and we beat them
and beat them bad.
{Jody P.OweU: reported in tt!e,Gainesville Sun, is
1919). 15
J I I] She saw there an objec:;t. That OOjec-t w
as the gaHow
s. She w
as afraid of the .gallows.
. .
. . . .
[Joseph Conrad, 1.'-he SecretJ'!g.ent,C
h. Ilj In .each Q
f it w
ould be possibie 1o abbreviate the text:w
it.Mut causing-am
biguity.. Tne-:fact that the 'Economy Fnn-
FUN
CTIO
NS O
F LAN
GU
AG
E
cipre does not operate,. although it is not inbibik:d by ambiguity..-
suggests that some other principle is in play. W
e can reasonably argue·that these are cases of
RE
PE
nnoN, w
here tne· em
phasis of repetition has some rhetorical value. suclr as surpris-
ing, impressing, or rousing the interest ofthe addressee. Thus the
repetition ot John Brown in [9} seems to· carry the im
plicature: 'John Brow
n and no one other than John Brow
n would have t(}
pay for it! ·
What bas been said about the Textual R
hetorfc suggests strong parallels w
ith the Interpersonal Rhetoric. Thus textual m
axims,
fike the maxim
s of the CP and the PP,
(i} appfy differently to different contexts; (ii} apply to variable degrees; (iii} m
ay compete with-one another;
(iv) may be exploited for the purpose of im
plicature; (v} are regulative rather than constitutive;
(vi) are interpreted as goal-oriented, and as serving goals which
are comm
on between s and h.
Of these sim
ilarities, (iii) and (iv), which are central to G
rice,.s treatm
ent of the CP, m
ay need further explication. With regard to
(iii), we have already noted that w
hereas some m
axims m
ay tend to w
ork towards a com
mon end (eg the End-focus ana the End-
weight M
axims), others (such as, in general, the C
larity Principle and the Econom
y Principfe) tend to compete or conflict w
ith one another. A
further example of such com
petition arises between
the End-weight M
axim and the Transparency M
axim, in -the case
of a discontinuous constituent such as the modifying clause in
[:;] The morning cam
e at last when we were aue to leave. The discontinuity is an.infringem
ent of the Transparency Maxim
,... in the interests of the M
axim of End-w
eight. Note that [31 is
slightly less felicitous than [3a}: fJaJ The m
orning came w
ften.wi were due to leave.
and this appears to be due. to tfie greater claim of the M
axim of
End-weight in [3a]. That is,.. if the relative clause in [3a-} w
ere-not postponed to the end of the
the result-would be a very
strong violation of the Maxim
of End-weight (a com
plex subject follow
ed by a very simple predicate):
[3b] The morning when we
due to-leave came.
Hence the postponem
ent of the relative clause has stronger m
otivation in [38} than. in [31-The felicity of an utterance.-here as
70 FO
RMA
LISM A
ND
FUN
CTIO
NA
LISM
elsewhere, is a m
atter of balancing the. competing claim
s of differ-ent m
axims.
. The· exploitation of m
.axim violation for the purpose of im
pli-cature has already been illustrated above in connection ·with ex-pressive repetition. A
further example is the im
plicature which
arises from a violation of end-focus in cases like these:
[12} Is she .BADly hurt? . [ 13] Is SIRE badly hurt? [ 14] is she badly hurt?
In [I2], for instance, s implicates that s is already aw
are of the fact that 'she' is hurt. In [13], s im
plicates that sis already aware
that someone is badly hurt. In [ 14], s im
plicates that sis aware that
someone has claim
ed or believed her to be badly hurt. .
3.4 The ideational function: discreteness and determinacy
I have tried to show that the . 'textual function' of language (in
Halliday's term
s), like the 'interpersonal function', can inost appropriately be handled by pragm
atic rather than by gramm
atic1
cal description. This depends, however, on how
happy one can be_ w
ith the gramm
ar/pragmatics distinction for w
hich I argued in ,C
hapter 2. It is time to reconsiderthis question with: reference to
the last of the eight postuhitesi.listed on p.. 5: '·
',•··.
P8: IN
G
E:-l"ERA
L G
RA
MM
AR
iS D
ES(;cRIB.t\BLF; iN TER
MS .:OE: D
IS-C
RETE A
ND
"rs
DE"" SC
RIB
AB
LE. IN
TERM
S IN
DETER:M
mA
TE V
AL
UE
S. ;;
This postulate, .eV
en· t!lough· it 'is· hedged· by qualifications,
amounts to a claim
that gramm
ar is a much m
ore orderly affair than pragm
atics. On the w
hole, I believe this to be true; but it cannot just be takeO
:,ror T
he difficulty is that in -::-years, the assum
pti&s of diScreteness and determ
inacy which
bave characterized gramm
atical descriptions (particularly those of transform
ational gramm
arians) in the past have been challenged, particularly in sem
antics. Labov's well-know
n study of the mean-
. ing of cup (1973), and. Ross's w
ell-known study of 'squishes' in
syntax {1973), have am
ong the more influential papers
drawing attention to scalar phenom
ena in gramm
ar. Similar stud-
ies of gradience, emphasizing the fuzziness of gram
matidil catej6
ories, are those of-Bolinger (I<)6I), Q
uirk (1965), and others. . R
ecent developments bringing
question the· discreteness. of gram
matical
have also. taken place.in.socio!inguisttcs,,
THE ID
EATIO
NA
L FUN
CTIO
N: D
ISCRETENESS A
ND
DETERM
INA
CY
71
where variable rulc:s and im
plicative scales have been used to account for sociolectal variation in quantitative and scalar term
s. It w
o'!ld how
ever, to reject the postulate of gram
matical
on. the basis of such evidence. Firstly, I am
not that studies such as those m
entioned actually count as evidence 11gainst a categorical view
of gramm
arP and secondly, I believe that even in the face of a considerable am
ount ?f
the fuzziness of gramm
atical categor-tes, tt w
ould still be reasonable to assume that in gram
mar the
prhnary processes are discrete (categorical), whereas the secon-
dary processes are (non-categorical).
Scalar values might encro.ach upon gram
mar. in any of the fol-
l?wi.ng three w
ays. First, there may be cases of syntagm
atic con-tm
utty between tw
o segments ( eg the transitions betw
een vowels
and consonants in continuous speech). Secondly, there may be
cases ..of paradigmatic fuzziness betw
een two classes (this is again
illustrated by segmental phonology: in, for exam
ple, the articu-latory distinction betw
een [pJ and [bJ, which is gradual rather
than absolute). 18 Thirdly, a rule which m
akes reference to one or m
ore categories may operate indeterm
inately, to produce' sen-tences w
hich are gramm
atical only to a certain degree. If all three of these kinds of continuousness w
ere present at the same tim
e there w
ould no doubt be grave unclarity about the working of
.. But it is likely thatthe gram
matical system
can tolerate a constderable am
ount of w
ithout failing to operate as an essentially discrete system
. A partial analogy m
ight be drawn
with a digital com
puter in which certain tolerances are built in to
allow for fluctuations of voltage. So long as the fluctuations w
ere w
ithin certain limits, there w
ill be no serious indeterminacy in the
system, and there w
ill be no possibility of mistaking the m
achine for an analogue com
puter. 19 .
· .
. ·
The psychological notion of a C
ATEG
OR
Y, as investigated by
Rosch and her associates, is crucial to the understanding of the
categorical nature ofgramm
ar. 20 4tegories are defined
by to prototypes, or 'good exam
ples' of the category in questton (for exam
ple, a prototype, fish has a cigar-like shape, fins, scales, a tail, etc.; trout and haddock are close to the typic fish,
octopus, and barnacles are not). Another
way to regard a category is as a fuzzy set of defining features:
the category 'vehicle', for instance, may be defined by such fea-
tures as (i) mobility; (ii) naving w
heels; (iii) locomotion; (iv)
movem
ent along the ground; (v) carrying passengers; (vi) pro-pelled by;an engine, etc. B
ut certain of these features (eg (i)) will
be more Im
portant than others (eg (vi)). The most prototypic of
72 FO
RMA
LISM A
ND
FUN
CTIO
NM
.JSM
vehicles (perhaps,_in this modem
age, a car) will have m
ost or all of the m
ore important defining features. O
n the other hand, dis-agreem
ents can arise as to whether a specim
en remote from
the prototype befongs to a category or not. For exam
ple, is a child's scoeter or a helicopter a vehicle?
The idea of a prototypic category. is apparently applicable to
perceptual and cognitive processes in general; it is alS<> applicable to fm
guistlc or logical concepts such as agency, causation, ability.. N
o doubtthe same concept carr be applied to syntactic and phono-
logical categories: some verbs are m
ore 'verby"" than others, som
e consonants are more 'consonantal: than others,
But it is
important to observe that categories im
ply two stages of
perientiai processing: on the one hand, we-m
ust recognize the in-dividual features w
hich assign an entity to the category; and on the other Jiand, w
e must recognize the-category as a w
lroie, as a G
estalt. On the form
er level there may be a considerable am
ount of indeterm
in_acy: it may 'De unclear w
hether a given specimen be-
lOngs to a giV
en category (in experim
ent, there was con-
siderabte perfpheral uncertainty about what objects could be
called cups). But on the latter level, there is no-doubt that tw
o categories are distinct: that there is, for exam
ple, a distinction be--tw
een consonants and vowels, even though. the actual boundarie$
between the tw
o categories may in certain fustaw-es be unclear.
What I am
suggesting here is something that bas been general-
ly taken for grapted on the phonic level: tile continuousness and overlap of phodetfc param
eters has not prevented phonologists froni postulating, a11d w
ofldng with, diScrete segm
ents amf con-
The em
ie-etic distinction,_ or the f0m1-su6stance diStinc-
has been assumed to have tllis basis. O
rr the semantic level,
tbere has been less agreement. B
ut the same principle holds
good; the language handles experience in terms of categories, and
. n:tfle actual criteria m
embership-of these categories
·belong to the psychological theory of reference, rather than to the linguistic theory of sense. 21 O
n the syntactic lever, too, we are
aealing with categories: for exam
ple, word classes such as verb
and adjectiVe, and clause elem
ents-such as subject and object in-vite this
of treatment. It is probabfy true that there is less '
categorical indetermfuacy in syntax than in the 'outer' levels of
coding, semantics and phonology. This w
ould fit in with a reasoft(
able theory of the ideational function of language, in-which-
mar is regarded· as a categorical system
mediating betw
een two
domains_ of largely non--discrete data: the referential data of our
models of experienced reality, and the phonetic data of speech. M
y general conclusion, is that so long as
EXA
MPLES O
F 'OV
ERGRA
MM
ATICIZA
TION
' 73
cal' is defined in a sense consonant with R
osch's prototypic egories, w
e may rephrase P8 m
uch more sim
ply as follows: G
RA
M-
MA
R IS ESSE
NT
IAL
LY
CA
TEGO
RIC
AL; PR
AG
MA
TICS IS ESSEN
TIALLY
N
ON
-CA
TEGO
RIC
AL. N
ote that this statement does not com
mit m
e to denying som
e indeterminacy in the operation of gram
matical
rules; we can still maintain the idea that gram
matical rules oper-
ate in an ali-or-none fashion, and yet hold that the categories w
hich define conditions for the rules are fuzzy to a certain de-gree. Such an in-betw
een position will be consistent w
ith these tw
o observations: (a) that gradients between gram
matical cat-
egories do occur; and (b) that linguists have managed in the past,
and will probably continue to m
anage, to obtain good approxi-m
ations to the nature of language without abandoning assum
ptions of syntagm
atic and paradigmatic discreteness.
3.5 Examples of 'overgram
maticization'
Since I have argued, on this basis, the distinctness of the ideational com
ponent of language (the gramm
ar) from the interpersonal
and textual components (w
hich belong to pragmatics), I w
ill con-clude by underlining the advantage w
hich this distinction brings to the treatm
e!lt of gramm
ar. The tendency in the past (particu-larly in transform
ational gramm
ar) has been to 'overgramm
ati-cize', ie to treat gram
matically aspects of linguistic behaviour
which are m
ore suitable for pragmatic explanation. I have already
referred to this tendency in the treatment of the interpersonal
function ( eg in the gramm
atical treatment of illocutionary force
by means of the perform
ative hypothesis). It remains to point out
some exam
ples of 'overgramm
aticizing' of the textual function . A standard transform
ational treatment of discontinuous noun
phrases involves a condition whereby a postm
odifyin¥ clause may
be extraposed, but a postmodifying phrase m
ay not. 2 This results in a discrim
ination between !15], w
hich is supposed to be gram-
matical, and·{ 16}, w
hich is supposed to be ill-formed:
l 15] A jug_got broken w
hich was from India.
[ 16] *A jug got broken from
India.
Such a distinction is based on doubtful acceptab1lity judgements,
·and the restriction on the rule is unnecessary if we assum
e that the difference betw
een jiS] and li6}·is a matter of degree of
pragmatic acceptability, 'Tather than of gram
maticaiity. T
hus [-!6] is predictably less 'happy_, .than [IS} because the M
axim of End-
weight pr<W
ides a -stmnger m
otWation-for the extiaposition -in the
.case -otj 1.5} than in Cclse of I U
i·J._,.qere is .a _g.ood example
74 FO
RMA
LISM A
ND
FUN
CI10N
ALISM
where gram
mar im
poses a discrete distinction on data for which a
pragmatic solution, in term
s of the relative strength of one maxim
over another, is m
ore appropriate. A
further example involving the M
axim of End-w
eight is:
{ [17a] Don't leave out W
illiam.
[17b] Don't leave W
illiam out.
{ [18a] Don't leave out the boy w
ho scored two goals in the
match last Saturday.
[18b] Don't leave the boy w
ho scored two goals in the m
atch last Saturday out.
{ [19a] Don't leave out yourself.
[19b] Don't leave yourself out.
{ [2oa] Don't leave out him
. [2ob) D
on'tJeave him out.
[17a] and [17b] are routine illustrations of the need for a Particle-Postponem
ent rule which m
oves a particle such as out to the end of the sentence. A
standard transformational gram
mar form
u-lation of this rule (see Chom
sky 1957:II2; Akm
ajian and Heny
1975: 178) is to make it obligatory in just those cases w
here the ob-ject noun phrase is a pronoun. A
ccording to this formulation,
therefore, [2oa] is ungramm
atical. How
ever, it is obvious from
examples such as {18a] and [18b]
there can be strong rhe-torical reasons for preferring one ordering to the other. The appli-cation of the postponem
ent rule flagrantly violates the End-weight
Maxim
in [18b], and the result is an extremely 'unhappy' sen-
tence. Conversely, both the w
eight and the End-focus Max-
ims w
ill predict that where the object is a personal pronoun, as in
[2oa] and [2ob], there will be no.m
otivation at all for postponing it, and indeed there w
ill be clear reasons for not doing so. In other w
ords, the virtual unacceptability [1oa], like that of (18b], is predictable. from
pragmatic considerations, and there is no
need to exclude it as ungramm
atical. To make the argum
ent m
ore forceful, however, w
e may note that an exam
ple like [19a], in w
hichthe object reflexive pronoun, is som
ewhere on the
scale of acceptability between [17a] and [2oa]. M
oreover, a gram-
matical restriction is too strong: [2oa] is not unacceptable if, for
some contextual reason, there is a contrastive nucleus on H
IM:
[21] He's the best player w
e've got: you can leave o1.1t any of. the others, but for H
eaven's sake don't leave out HIM
. .In this exceptional case, the End-focus M
axim provides a m
otiv-ation for the postponem
ent, in defiance of the End-weight M
axim.
A final exam
ple of 'gramm
aticizing' ·textual pragmatics in-
EXA
MPLES O
F 'OV
ERGRA
MM
ATICIZA
TION
' 75
volves an obligatory transformation w
hich reduces a noun phrase: [22a] John Sm
ith; admires John Sm
ith; more than any other
politician. [22b] John Sm
ith admires him
self more than any other poli-
tician. A
gain, a standard transformational analysis w
ould be to treat the reflexivization of [22b] as obligatory, at least if the subject and object N
Ps are co-referential. Consequently, [22a} would either
be ungramm
atical, or would be gram
matical only on condition
that the two
Smiths are different people. B
ut this is clearly m
correct: [22a) IS not only apparently gramm
atical, but its more
likely interpretation is one in which the tw
o John Smiths refer
to one and the same person. A
pragmatic explanation
of the kind proposed on pp 68-9, is preferable: w
e note that s failed
t?. (even
there is no threat of ambiguity in the sub-
s.ttutlon of a reflexive pronoun), and therefore we interpret this
as 'expressive in this case for the purposes of irony.
The same
may be offered for cases w
here, according to transform
atiOnal gram
mars, the rule of Equi N
oun Phrase D
eletion opera.tes obligatorily (see Akm
ajian and Heny 1975:298
-300), thus ruhng out [23a] as ungramm
atical: [23a] John Sm
ith; would like John Sm
ith; to become the next
Prime M
inister. ·
[23b] John Smith w
ould like to become the next Prim
e Minis-
ter.
Once again, then, there is advantage in an appropriate division of
labour between gram
mar and pragm
atics. And the distinction be-
tween the discrete values of gram
mar and the continuous values
of pragmatics is strengthened in so far as exam
ples of gradable acceptability, such as those just discussed, can be convincingly show
n to have a pragmatic origin.
A rather different kind of overgram
maticization is found in the
functional gramm
ar of Halliday.
Regarding the interpersonal illocutionary force in term
s of discrete op-tions w
tthm sem
antic networks, and regarding the textual func-
tion, he handles intonational factors such as the segmentation of
the text tone-units and the placing of the tonic, in term
s of gram
matical system
s of discrete choice. 23 For Halliday, such
textual choices are often described in terms of 'm
arked' and 'un-m
arked' options. For example, w
hat I have described as 'end-focus' has been described by H
alliday as 'unmarked inform
ation focus', such that the choice betw
een, say,
FORM
ALISM
AN
D FU
NC
TION
ALISM
[24] Is she badly HURT? · and the variants w
ithout end-focus, given as [12]-[14] on p 70 w
ould be described as a gramm
atical choice between unm
arked and m
arked options. Similarly, the choice betw
een I love peaches and Peaches I love he describes as one of 'unm
arked theme' ver-·
sus 'marked them
e', the theme being the elem
ent (in unmarked
declarative cases, the subject) which is placed first in the clause.
A significant point is that H
alliday's definition of 'unmarkedness'
points to a pragmatic interpretation of this concept: he describes
it as the choice that is made under neutral conditions, 'unless
there are reasons to the contrary'. In pragmatic term
s, the un-m
arked term is the one w
hich is chosen by default, where there
are no factors (such as competing m
axims) to override it. In this
way, H
alliday's textual choices within the gram
mar readily lend
themselves to reinterpretation in term
s of a Textual Rhetoric.
3.6 Conclusion In C
hapters 2 and 3 my purpose has been to enum
erate some
essential differences between gram
mar and pragm
atics, and to develop these differences through discussion and exem
plification. I have argued for a form
alist account of gramm
ar, and for a func-tionalist account of pragm
atics. At the sam
e time, I have argued
for necessary interrelations between these tw
o ways of explaining
language. The formalist-functionalist view
of language I have been putting forw
ard can be summ
arized as follows:
'Language consists of gramm
ar and pragmatics. G
ramm
ar is an abstract form
al system for producing and interpreting m
essages. G
eneral pragmatics is a set of strategies and principles for achiev-
ing success in comm
unication by the use of the gramm
ar. Gram
-m
ar is functionally
adapted to the extent that it possesses
properties which facilitate the operation of pragm
atic principles.'
Notes I. See D
ik's discussion of 'two paradigm
s for the study of language' (1978:4-5). For a sam
ple of formalist and functionalist view
s of the basis of linguistics, see C
homsky (19'76) and H
alliday (1973, 1978). 2. Trevarthan (1977) reports research on the em
ergence of purposeful com
municative behaviour in infants.
3· For Chom
sky's views on introspection anq his rejection of objec-
tivity, see Chom
sky (1964:61, 79-81). For his views on descriptive
adequacy, see Chom
sky (1964:62-3). ·
n The treatm
ent of language in terms of com
municative prckess is
something w
hich pragmatics has iq com
mon w
ith psycholinguistic (see Q
ark: and Clark 1977:35-292) and som
e textlinguistic (see de B
eaugrande and Dressler I98I:31-47) approaches to language.
5· The use c0f
and 'discourse' as distinct levets in the analysis of the connected use of language is fam
iliar in -the work of W
iddowson
( 1975:6) and others.. ·
6. On m
ulti-level processing in language comprehem
iioo, see Clark and
Clark (1977=49); on sim
ilar processing in language prodUction see
Oark and C
lark (1917;292). '
7. exam
ples show a subsidiary risin; tone on tbe last w
ord: for this
of distribution of information, see Firbas (198o).
8. End-focus IS consonant with the principle of Functional Sentence
PerSpective, that comm
urJcative dynamism
, or weight increases to-
wards the end of a given textual segm
ent. The ooncept of FSP has been developed am
ong Czech linguistics, and especially by Firbas
(see eg Firbas 1C)So). 9· End-w
eight as a principle which facilitates syntactic processing is
in somew
hat different terms, by Y
ngve (1961), Bever
(1970, .. and Frazier (I97!PO
). For an explanation within a
neuroungwsttc fram
ework, see Luria (1976:Is8-9).
IO. These exam
ples are from C
homsky (I957:IO
o-I). Chom
sky argues that [I] and
·have different meanings, w
hereas Katz and Postal
(1964:72-3) and Leech (1969:52) maintain that both sentences are
in sam
e way. C
arden (1973), through informant test·
mg, finds a basts of support for both view
s. I believe that the 'rhetori-cal' explanation proposed here is the one w
hich best fits what w
ould otherw
ise be a rather puzzling set of observations. n. C
fthe 'Transparency Principle'. put forward by Lightfoot (1979:1:n-
40) as a means of explaining constraints on historical changes in
syntax. u. C
f the principles of Natural Serialization and N
atural Constituent
Structure (the latter attributed to R
. B
artsch) in V
ennemann
. (1973:40-1). Like Lightfoot's 'Transparency Principle' these are in-troduced to explain properties of gram
mars· but they have an
obvious rhetorical motivation.
' IJ. O
n 'garden paths', see Oark and O
ark (1977:8o-2). 14. O
n the .importance of iconicity in syntax, see B
olinger (198o: Ch. 3),
and (With reference to literary style) Leech and Short (1!)81:233-
42). IS.
in D
ressler (xg8x:x68). 16. Q
mrk s term
for the phenomenon of gradience is 'serial relation-
Ross,
this phenomenon by m
eans of a two-
dimensional array m
which m
stances are plotted against criteria. 17. There is considerable disagreem
ent about the status of variable according, to one view
, the variability is not part of gramm
ati-cal com
petence, and the rules themselves are categorical (for a sur-
vey, see Rom
aine 1981). Similarly, it is arguable that categoricaf
FORM
ALISM
AN
D FU
NCTIO
NA
LISM
rules do not exch.ide gradience. Contrast Labov's
of the_ term·
'categorical' (1973 ) with that associated w
ith prototypic categones on. the lines of R
osch (see note 20 below).
. .
. 1 s. C
fthe experiments of M
iller and Nicely (1_955), a?d the
these and other speech-perception m
_Clar an . .
ar ( 1977 : I 9 I _ 220 ). Essential to speech perception _Is the a?Ihty of the hearer to m
ake categorical decisions on the basts of contmuously
variable auditory cues. ·
1 · f
19 . Leech and Coates (Ig8o) argue, on the ?asis of
ana ys1s o_ m
odal auxiliaries that gradience and
can be a ref latively m
inor in
of because o
the establishment of a 'quantltattve stereotype .
. 20
The extensive research carried out by Rosch and her ass?ctates on
· the 'prototype' basis of conceptual and perceptual categones_ sam
pled in Rosch and Mervis (1975) and Rosch (1977). For
applications of this
research, see
Lakoff (1977),
Leec 9
[1974]:84-6). . .
21 This is argued m
Leech (1981 [1974).84-6). . .
fr N
P' 22 : See, for exam
ple, the formulation of the 'Extrapositton
om
transformation in B
urt (1971 :72). .
. )
23 . This criticism of H
alliday is elaborated m Leech (1980.22-6 ·
Chapter 4
The interpersonal role of the Cooperative
Principle It is undesirable to believe a proposition w
hen there is no ground whatever for
supposing it true. [Russell, Sceptical Essays, p. 1 J
Jack: Gw
er.dolen, it is a terrible thing for a man to find out suddenly that all his
life he has been speaking nothing but the truth. Can you forgive me?
[Wilde, The Im
portance of Being Earnest, Act Ill]
In the remaining chapters, I shall investigate the interpersonal
rhetoric iri greater depth than has so far been possible. In this w
ay, I shall be seeking answers to som
e major problem
s at the 'pragm
atic end of semantics', by seeking to apply the m
odel out-lined in C
hapters 2 and 3 to the description of English. I shall be considering, in particular, how
to deal with politeness phenom
-ena,
illocutionary force,
performatives,
indirect illocutions,
and the meanings of speech-act verbs. In this, I shall be treading
some fam
iliar ground, but the approach I shall use will be to
some extent unfam
iliar. For example, I shall be trying to show
exactly how
the CP and the PP interact in the interpretation of in-
directness. If I can show that both these principles are required to
account for pragmatic interpretations, I shall be on the w
ay to ex-plaining the need for a 'rhetoriC
', in the sense of a set of prin-ciples w
hich are observed in the planning and interpretation of m
essages.
4.1 The C
ooperative Principle (CP) and the Politeness Principle (PP)
So much has been w
ritten in general support of Grice's concept
of the CP that I m
ay take this principle to some extent for
granted. But it is necessary to give som
e explanation here of (a) w
hy the CP is needed, and (b) w
hy it is not sufficient, as an ex-planation of the relation betw
een sense and force. It will also be
necessary to consider the function, in the present model, of each
of its constituent maxim
s (see 4.2.5). This will be the task of the
present chapter.
Bo TH
E INTERPERSO
NA
L ROLE O
F mE
COO
PEilATIV
E PRINCIPLE
In brief:, the answers to questions (a) and (b) in the previous
paragraph are as follows. W
e need the CP to help to account (as
already explained) for the relation between
and and
this kind of explanation is particularly ·welcom
e where Jt
puzzles which arise in a truth-based approach to sem
antics. H
owever the C
P in itself cannot explain (i) why people are .often oo
in conveying what they m
ean; and w
hat is the re-lation betw
een sense and force when non-declaratiV
e types of sen-tence are being considered. G
rice himself, and others w
ho have invoked1he C
P have understandably reflected the logician's tra-ditional
with truth, and hence w
ith m
ean-ing; w
hereas I_ shall be more interested in a
psychologically oriented application of pragmatic pnnc1ples. This
is where politeness becom
es important.
There have also been objections to Grice's C
P on the grounds that it does not stand up to the evidence of real language use. For exam
ple it has been argued that conversational constraints such as those 'of the C
P do not work because the m
ajority of declara-tive sentences do not have an inform
ation-bearing function kin and O
'Malley 1973). It has also been argued that the
of the CP are not universal to language, because there are hngm
s-tic com
munities to w
hich not all of them apply (K
eenan 1974). M
y first observation on these is that they are not neces-
sarily so 'damning
they T
o reJeC! the .CP quan-
titative grounds would be to m
lStake maxuns for statistical norm
s-w
hicil they are not. And no claim
has. m
ade that the CP
applies in an identical manner
all soc1et1e.s. one of the
main purposes of socio-pragm
attcs, I
It, 1s to find out how
different societies operate maxtm
s m different w
ars, ex-
ample by giving politeness a higher rating than
m cer-
tain situations or by giving precedence to one maxtm
of the PP rather than
(see 6. 1.3). How
ever, it be
that the CP is in a w
eak position if apparent excepttons to 1t can-not be satisfactorily explained. It is for this reason that the PP can be seen not just as another principle to be added to t1t.e C
P,. but as a necessary com
plement, w
hich rescues the CP from
senous trouble.
. .
Two exam
ples where the PP rescues the C
P are the followm
g:
[1] A: W
e'll all miss Bill and A
gatha, won't w
e? B
: Well, w
e'll all miss B
ILL. [2] P: 3om
eone's eaten the icing off the cake. C: It w
asn't ME. In [1], B apparently fails to observe the M
axim of Q
uantity:
THE CO
OiiERA
TIVE PRIN
CIPLE AN
D TH
E POLITEN
ESS PRINCIPLE
81
when A asks B to confirm
A's opinion, B m
erely confirms part
of it, and pointedly ignores the rest. From this w
e derive an im-
plicature: 'S is of the opinion that we w
ill not all miss A
gatha! B
ut on what grounds is this im
plicature arrived at? Not solely on
the basis of the CP, for B could have added ' ... but not A
gatha' w
ithout being untruthful, irrelevant, or unclear. Our conclusion is
that B could have been more inform
ative, but only at the cost of being m
ore impolite to a third party: that B therefore suppressed
the desired itj>rmation in order to uphold the PP.
In [ 2], typtcally an exchange between parent P and child C,
there is an apparent irrelevance in Cs reply: C
seems to react as
if he needs to exonerate himself from
the evil deed in question. C
's denial is virtually predictable in such a situation, as if C were
being directly accused of the crime. M
y explanation of this appar-ent breach of the M
axim of R
elation is as follows. Suppose P is
not sure who is the culprit, but suspects that it is C. Then a sm
all step of politeness of P's part w
ould be to withold a direct accu-
sation, and instead to make a less inform
ative, but undoubtedly true assertion, substituting an im
personal pronoun someone for
the second-person pronoun you. Thus P's remark in [2] is inter-
preted as an indirect accusation: when C
hears this assertion,-C
responds to it as having implicated that C
may w
ell be guilty, denying an offence w
hich has not been overtly imputed. W
hat this suggests, then, is that the apparent irrelevance of C
's reply is due to an im
plicature of P's utterance. C responds to that im
pli-cature, the indirectness of w
hich is motivated by politeness, rather
than to what is actually said.
It is notable that the replies in both [I] and [ 2] would alm
ost certainly have a fall-rise tone, w
hich is an intonation often associated w
ith indirect implicature. A
more im
portant point, how
ever, is this: both examples illustrate how
an apparent breach of the C
P is shown, at a deeper level of interpretation involving
the PP, to be no such thing: in this way, the C
P is redeemed from
difficulty by the PP.
In its negative form, the PP m
ight be formulated in a general
way: 'M
inimize (other things being equal) the expression of im
-polite beliefs', and there is a corresponding positive version ('M
aximize (other things being equal) the expression of polite be-
liefs') which is som
ewhat less im
portant. 1 In [1) and [2], the sup-pressed im
polite beliefs are 'We w
on't miss A
gatha' and 'You
have eaten the icing off the cake'. Polite and impolite beliefs are
respectively beliefs which are favourable and unfavourable to the
hearer or to a third party, where 'favourable' and 'unfavourable' .
are measured on som
e relevant scale of values (see 6.1). It should
82 THE INTERPERSONAL ROLE OF THE COOPERATIVE PRINCIPLE
be stressed, again, that the real beliefs of s are not in question, but w
hat s PURPORTS to believe. H
ere we should consider the general social function of these tw
o principles, and the 'trade-off' relation between them
. The CP enables one participant in a conversation to com
municate on the
assumption that the other participant is being cooperative. In this
the CP has the function of regulating what w
e say so that it con-tributes to som
e assumed iilocutio11ary or discoursal goal(s). It
could be argued, however, that the PP has a higher regulative
role than this: to maintain the social equilibrium
and the friendly relations. w
hich enable us to assume that our interlocutors are
being cooperative in the first place. To put matters at their m
ost basic: unless you are polite to your neighbour, the channel of com
munication betw
een you will break dow
n, and you. will no
longer be able to borrow his m
ower.
· There are som
e situations where politeness can take a back
seat. This is so, for example, w
here s and h are engaged in a col-laborative activity in w
hich exchange of information is equally im
-portant to both of them
. But there are other situations w
here the PP can overrule the CP to the extent that even the M
axim of
Quality (w
hich tends to outweigh other cooperativem
axims) is
sacrificed. That is, in certain circumstances, people feel justified
.in telling 'white lies'. For exam
ple, s may feel that the only way
of declining an invitation politely is to pretend to have an alterna-tive engagem
ent. But w
e should distinguish 'white lies' such as
this, which are m
eant to deceive the hearer, and cases whiCh are only APPARENT breaches of tht;; CP. There is a difference betw
een politeness 'off the record' (cf Brow
n and Levinson 1978:I34ff) and politeness 'on the record' (for exam
ple when s says You
couldn't help me m
ove these tables could you?, and it is quite evi-dent that h couLD
move them
). N
otice that examples like [1] and [2], w
hich belong to the second type, can easily tip over into an ironic interpretation. Irony is in fact a second-order principle, w
hich builds upon, or exploits, the principle of politeness. 2 The Irony Principle (IP) m
ay be stated in a general form
as-follows:
'If you must cause offence, at least do so in a w
ay which doesn't
overtly conflict with the PP, but allow
s the hearer to arrive at the offensive point of your rem
ark indirectly, by way of im
plicature.'
Irony typically takes the form of being too obviously polite for
the occasion. This can happen if s overvalues the PP by blatantly breaking a m
axim of the CP in order to uphold the PP. For exam
-ple, in [ 1] there w
as an obvious breach of the Quantity M
axim,
.and in [3] there is an obvious breach of the Quality M
axim:
THE COOPERATIVE PRINCIPLE A
ND
THE POLITENESS PRINCIPLE
[3] A: G
eoff has just borrowed your car.
B: Well, I like THAT!
derived from the Irony Principle w
orks roughly ·as follow
s'm·th1s case:
'What B says is polite to G
eoff and is dearly not true. There-fore w
hat B really means is im
polite to Geoff and true.'
can put it in _Grice's ow
n terms as follow
s. In being polite one ts often faced w
tth a CLASH between the CP and the pp so that
ch_oose how far to 'trade off' one against the other; but
m bem
g tromc, one EXPLOITS the PP in order to uphold, at a re-
_level, CP .. A person w
ho is being ironic appears to be decelV
lng or m1sleadm
g _h, but in fact is indulging in an 'honest' form
of apparent deceptiOn, at the expense of politeness:
Telling 'white lies'
Ironic 'truthfulness' FIG
UR
E 4
.1a
FIG
UR
E 4
.1b
This has
in_dicated some dangers in the use
of the term
politeness . There 1s an unfortunate association of the term
with superficially 'nice', but ultim
ate1y insincere form
s of ?um
an and it is therefore tem
pting to wrlte off
(at least m som
e cultural environments) as being a triv-
Ial and. dispensable factor which is no m
ore than a 'garnish' on the senous use
language. In pointing out the importance of the
PP for _the explaming of
principles (the CP and the IP) I have tned to show
otherwise. W
hat tends to confuse the issue I
think, is a failure to betw
een ABSOLUTE and
POLITENESS. In general, in these chapters I shall
be dealmg w
tth absolute politeness, as a scale, or rather a set of _(see 5·7· 6.1), having a negative and a positive pole. Som
e dlocut10ns (eg orders) are inherently im
polite, and others (eg P?lite. _Negative politeness therefore con-
ststs m m
tmm
mng the tm
pohteness of impolite illocutions, and
THE IN
TERPERSON
AL RO
LE OF TH
E COO
PERATIV
E PRINCIPLE
positive politeness consists in maxim
izing the _of po!ite
· illocutions (which includes taking opportunities for'perfonrung
polite illocutions in situations where no speech m
ay be otherwise
called for). I shall be dealing with the strategies for producing
and interpreting polite illocutions, and placing them on a scale of
absolute politeness. A
t the same tim
e, I am aw
are that people typically use 'polite' in a relative sense: that is, relative to som
e norm of behaviour
which, for a particular setting, they regard as typical. The norm
m
ay be that of a particular culture or language comm
unity. For exam
ple, I have been seriously told that 'Poles/Russians/etc. are never polite', and it is com
monly said that 'the Chinese and the
Japanese are very polite in comparison w
ith Europeans', and on.
These stereotypic comm
ents are often based on partial ev!dence, and one of the tasks of w
hat I earlier called 'socio-pragm
atics' is to examine the extent to w
hich language com-
munities do differ in their application of the PP (see further
6.1.3). Such a study would soon lead us to another kind of norm
: the norm
of politeness for a particular For exam
-ple, the English language, in particular the
variant o_f it, is rich in indirect im
positives, which I shall exam
me later, m
the next chapter. This can certainly give the im
pression, to native speakers of other languages, that the British are excessively polite (and hence perhaps insincere), w
hen asking favours of others .. A
further norm is one for a particular category of person, accordm
g to sex, age-group, etc. For exam
ple, in Japan, the scale of polite-ness is
exploited differently by
wom
en than
by m
en, and
(apparently) more by people in the w
estern part of the country than by people in the eastern part of the country. 4 It is on the basis of such group norm
s that we judge individual people as
being 'polite' or 'impolite' in particular
('John w
as very rude to his mother', etc.). Relattve politeness ts there-
fore variable on many dim
ensions, according to the standard or set of standards under scrutiny. G
eneral pragmatics m
ay reason-ably confine its attention to
in the absolute .
Returning to the C
P, let us constder the grounds on whtch tts
various maxim
s have been proposed and jpstified, and how they
will fit into the present m
odel.
4.2 Maxim
s of Quantity and Q
uality "l:he first tw
o maxim
s can be considered together, since (as I indi-cated earlier) they frequently w
ork in competition w
ith one another: the am
ount of information s. gives is lim
ited by s's wish to avoid telling an untruth. For this reason, H
amish has even
MAXIM
S OF Q
UA
NTITY
AN
D Q
UA
LITY
proposed a combined m
axim as follow
s: .M
axim of Q
uantity-Quality: M
ake the strongest reievant claim
justifiable by your evidence.
ss
[Harnish 1976:362]
And he quotes (ibid.) a m
ore detailed version of the same m
axim
formulated by O
'Hair (1969:45): 'U
nless there are outWeighing
good reasons to the contrary, one should not make a w
eaker statem
ent rather than a stronger one if the audience is interested in the extra inform
ation that could conveyed by the latter.'
'Strength' here refers to the amount of inform
ation comm
uni-cated. The strictest w
ay to interpret it, in terms of tw
o proposi-tions P and Q
, is to say: 'If P entails Q and Q
does not entail P,
then Pis stronger than Q
.' On this basis, we can explain w
hy cer-tain im
plicatures arise in the interpretation of logical operators such as the quantifiers, not, and, and or. Consider first the quan-tifiers 'all' V
and 'some' 3:
[4] Jill ate some of the biscuits.
leads one to conclude the falsity of: [4a] Jill ate all the biscuits.
But this is not a logical inference (assuming a standard interpre-
tation of some and all as representing the existential and universal
quantifiers), but is due to implicature ie: h concludes that s, in
asserting (4], implicates that N
OT [4a]. This is evident from
the fact that the inference can be cancelled by the addition of contra-dictory inform
ation: 5
[4b] Jill ate some of the biscuits-in fact she ate all of them
.
The way to understand this is as follow
s. First, let 'all' and 'some'
be an ordered pair on a scale of 'operator strength', construing this to m
ean 'A proposition P containing "all" is stronger than an
otherwise equivalent proposition Q
containing "some"'. (Strictly,
this applies only when 'all' and 'som
e' are outside the scope of other operators). Then note, as a general rule, that T
HE
WEA
KER
PR
OPO
SITION
IMPLIC
ATES TH
AT S B
ELIEVES TH
E NEG
ATIV
E OF TH
E STR
ON
GER
PRO
POSITIO
N. In the present case, (4] im
plicates that's. believes that Jill did not eat all the biscuits'. The explanation of-this im
plicature is as follows:
{a) s has uttered a weaker proposition Q
where s could just as
easily and relevantly have uttered a stronger proposition P. (b) By the M
axim of Q
uantity-Quality, this, in the absence of
86 TH
E INTERPERSO
NA
L ROLE O
F THE CO
OPERA
TIVE PRIN
CIPLE
contrary information, m
eans that the evidence s has does not justify the assertion of P, but does justify the assertion of Q.
(c) This leads to the implicature that s believes P to be false, ie: s
believes that not-P. (There is, how
ever, a need to distinguish this POSITIV
E version -from
NEU
TRA
L version of the implicature at (c). The neutral ver-
sion is: 'S does not believe that Pis true, nor does s believe that
P is false.' This would be the conclusion if s sim
ply did not have .enough evidence to decide, as in:
[4c] Jill ate sOME of the biscuits (but I don't know w
hether she ate all of them
). W
hether the positive or the neutral implicature is assum
ed de-pends on the situation. O
ne context will suggest that s has w
itheld the inform
ation because of lack of knowledge, and another that s
has witheld it because of a definite belief to the contrary.)
Now
note that a converse relation of 'strong-to-weak' obtains
between the negations of [4] and [4a]:
[5] Jill did not eat any of the biscuits. (negation of (4])
[sa] Jill did not eat all of the biscuits. (negation of [4a])
In this case, the former proposition is stronger than the latter, ie
[s] is stronger than [sa] rather than [sa] being stronger than [s]. In fact, it is a general rule that if P
is stronger than Q, then not-Q
is stronger than not-P. H
ence when s utters not-P, the im
plicature (in its positive version) is that s believes the negative of not-Q
; or to sim
plify, that s believes that Q. Thus the im
plicature of [sa] 'Jill did not eat all of the biscuits' is that 's believes that [4], ie that Jill ate som
e of the biscuits'. W
hat is true of this particular example can be generalized as
follows:
[6] (PO
SITIVE V
ERSIO
N) If P
is stronger than Q, then
(i) s implicates by Q
that s believes that not-P, and (ii) s im
plicates by not-P that s believes that Q.
[7] (N
EU
TR
AL
VER
SION
) If pis stronger than Q, then
. (i) s implicates by Q
that s is not aware w
hether p or not-p
(ii) s implicates by not-P that s is not aw
are whether Q
or not-Q
. Im
plicatures therefore hold between the 'w
eaker' propositions, and are reciprocal, in the m
anner indicated by the arrow in Fig.
4.2 for 'all' and 'some'.
MAXIM
S OF Q
UA
NTITY
AN
D Q
UA
LITY
Positive
All
Not som
e Stronger
(=not any, no)
Some
Not all
Weaker
FIGU
RE
4.2
(For simplicity, I have again extended the use of 'strong' and
'weak' from
the propositions themselves to the operators in term
s of w
hich the propositions differ.) The generalizations [ 6] and [ 7] can then be illustrated by a set of logical pairs as show
n in Table 4.1 (cf G
azdar 1979=49-SO). The list could be extended. 6 In the case of 'becom
e' and 'remain' it is necessary that the event de-
scribed by 'become' should take place at a tim
e (t 0) anterior to the tim
e of the state described by 'remain' (t+).
TAB
LE4.1
POSITIVE
NE
GA
tiVE
P is stronger than Q
not-Q
is stronger than not-P p
Q
not-Q
not-P
A. A
ll Som
e_ N
ot any N
ot all
B. More than
(As m
uch/ N
ot (as much/
Not m
ore than n
many as) n
many as) n
n
c. H
ave to, B
e able to, N
ot be able to, N
ot have to m
ust can
cannot
D. B
e certain Think that
Not think that
Not be certain
that that
E. R
emain
Become
Not becom
e N
ot remain
[at t+] [at t']
[att] [at t+]
F. X andY
X
orY
not-(X or Y
) not-(X & Y
)
G. Succeed in
Try to N
ot try to N
ot succeed in
Using P for the stronger and Q
for the weaker term
(eg 'have to' = P and 'can' = Q
) we can derive im
plicatures of the follow-
ing kinds: not P 1
Q (eg 'not have to'
'can') Q
I
not-P (eg 'can' . 1
'not have to')
88 TH
E INTER
PERSO
NA
L RO
LE OF THE CO
OPERA
TIVE PRI.N
CIPLE
(The positive version of the implicature is chosen, and X
---
y is intended to be an abbreviation for: 's in asserting X impli-
cates that s believes that Y. ') Some exam
ples of the various kinds of im
plicature listed in Table 4.1 are given below; the im
pli-catures are bi-directional, as is signalled by the double-headed arrow
s:
BI,
'Nora has (as m
any as) three children.'-=---
'Nora has no m
ore than three children.' C
I 'Em
ployees do not have to retire at 65.' 1
'Employees C
AN
retire at 65.' D
I 'I think G
randpa is asleep.'-:.... 1-
'I am not certain that G
randpa is asleep. ' 7
EI
'Betty did not rem
ain ill.' 1
'Betty got w
ell.' FI
'Sue works at the office on Thursday or on F
riday
.·---'Sue does not w
ork at the office on both Thursday and Friday.'
GI
'Frank tried to open the door.'_.....__ 'Frank did not succeed in opening the door.'
By way of the 'Q
uantity-Quality M
axim', then, a large num
ber of inform
al inferences can be accounted for. This not only shows
the explanatory value of the CP, but also strengthens the gfam
-m
ar, by helping to show that standard logical analysis can w
ork quite w
ell on natural language, in spite of appearances to the con-trary. A
lleged indeterminacies and illogicalities have afflicted
attempts to interpret natural language in term
s of formal logic.
But there is a strong hope that, by m
eans of the CP, they can be
not to logic, but to pragmatics.
At the sam
e time, a pragm
atic account of conversational infer-ences such as those exem
plified in BI-G
I above leads to sim-
plification of standard logics. Thus the distinction which has been
drawn betw
een inclusive and exclusive disjunction is an embar-
rassment from
the point of view of natural language sem
antics, in that it involves ascribing to one connective (or) tw
o distinct logi-cal interpretations, w
hich are nevertheless so close that one in-cludes the other. (That is, all sentences w
hich are true on an exclusive-or interpretation are also true on an inclusive-or inter-pretation.) B
ut in a 'cornplementarist' account w
hich interrelates logic lind pragm
atics, or needs to be given only one meaning, that
of inclusive disjunction; then the exclusive interpretation is de-rived from
the inclusive one by implicature, as in FI above. 8
The 'Q
uantity-Quality M
axim'
therefore provides further illustrations of a type of explanation w
hich is by now fam
iliar: a
MAXIM
S OF Q
UA
NTITY
AN
D Q
UA
LITY
8g
pragmatic explanation of som
ething which, in gram
matical tem
1s, appears problem
atic or anomalous. This suggests that the division
of labour between sem
antics and pragmatics results in m
ore satis-factory solutions for both disciplines.
As a further exam
ple, consider the asymm
etry of: [8] *I ran fast and could catch the bus. [9]
I ran fast but couldn't catch the bus. [from Palm
er 1980:92] [8a] I ran fast and w
as able to catch the bus. [9a] I ran fast but w
asn't able to catch the bus. Tne unacceptability (ungram
maticality?) of could in [8] is the
problem. W
e may accept, w
ith Palmer (1977, xg8o), that could
contrasts with w
as able to in terms of actuality. That is,. can/could
represents the state of being capable of doing something, but
without the actuality of the action, w
hich is additionally indicated by be able to. In this w
ay was able to and could form
a strong/ w
eak pair of the type exemplified in Table 4· I above. W
hereas could signals just the state of potentiality, was able to signifies both the potentiality and the perform
ance. By the argument that
applies in Table 4· I, speakers will prefer to use the stronger
alternative (was .able to) w
here evidence justifies it, so that could is used to im
plicates's lack of belief in the performance, as in [9].
Hence the oddity of [8] in contrast to [8a]. That this is a prag-
matic m
atter is supported by the observation that in a different context, the sentence w
ith could is acceptable: 9
[10] I could just/almost reach the branch: [from
Palmer I98o:95]
In such sentences as this, it is precisely the state of potentiality, as opposed to the actual act of perform
ance, that is given promi-
nence. This is only a partial explanation (which does not account
for the acceptability of the 'weaker' negative w
asn't able, to) but the outline of the argum
ent is clear: could is associated non-
performance because the C
P urges that if a stronger statement
can be truthfully made, then m
ake it we w
ill. The t>ame reasoning does not apply to the present tense form
, can, because the actual perform
ance arising fiom a potentialiy in the present is m
uch less likely to be know
n, lying as it does in the future. H
ere is a further example of asym
metrical acceptability, from
the syntactic dom
ain: '
[IIa] My sister is m
arried, and her husband works for N
ASA
. [ub] ?*M
y sister's husband works for N
ASA
, and she is
· m
arried.
90 THE INTERPERSONAL ROLE OF THE COOPERATIVE PRINCIPLE
The oddity of [ub] in con,t,}'ast to [ua] is a puzzle for someone
working entirely w
ithin a 'Sentence-gramm
ar framew
ork. Since both sentences have
sense, and since they both appear to be syntactically w
ell formed, there is no sensible w
ay of accounting for this asym
metry by gram
matical rule. B
ut as soon as w
e look at the pragmatics of [ub), w
e notice that its second clause violates the M
axim of Q
uantity: in fact, its second cla2se is entirely redundant, in that it provides no inform
ation other that w
hat has already been presupposed in [I 1 a]. This illustrates how
Maxim
of Quantity m
ust sometim
es be interpreted in a way
which applies to one part of a sentence, but not to another: w
hat is inform
ative at the beginning of a sentence is not necessarily so half-w
ay through it.
4.2.1 Implicatures connected w
ith definiteness A pair of w
ords which m
ight have been added to the 'strong-w
eak' pairs in Table 4.1 is the and a(n): as with the other pairs, if
the definite article is substituted for the indefinite article in some
proposition, the result is a proposition which entails the orig-
inal one. For instance (assuming w
e luiow w
hat is by the
secretary): 'Sally is the secretary' entails 'Sally is a secretary'.
Similarly:
'Sally is not a secretary' entails 'Sally is not the secretary'. B
ut the articles differ from the other pairs w
e considered, be-cause the basic contrast betw
een them is of a pragm
atic nature. The elem
ent of definiteness expressed by the (and also, inci-dentally, by m
any other words such as personal pronouns and
demonstratives) conveys s's understanding that there is som
e ref-erent that is to be identified uniquely in the contextual know
ledge shared by s and h. Thus w
hen SOIIleone uses the phrase the X, w
e infer from
this that ·
[ 12] There is some X that can be uniquely identified as the
same X by s and h.
Since this is essential to the meaning of the, rather than derived
from that m
eaning by means of conversational principles, [12]
should be called a coNVENTIONAL implicature (seep n
) rather than a CONVERSATIONAL one. 'U
niquely' means that w
e should be able to select the one X concerned from
all other Xs, (or, if X
is plural, that we should be able to select the one set o
f Xs con-
cerned from all other sets of X
s). Since the decision to use the
OF QU
AN
TITY A
ND
QU
ALITY
91
rather than a(n) is a matter of appropriateness to situation,
the reference of the X is likely to vary from one situation to
another. O
n the other hand, the indefinite article, as its name suggests,
may be m
ore negatively defined by the absence of this feature of 'definiteness' in the m
eaning of the. A table, for example, w
ill be em
ployed in conditions where the shared contextual know
ledge m
entioned in [12] does not obtain, and where there is therefore no
situational basis for the use of the. It is for this reason that the use ·of a(n) (and for that m
atter, of other indefinite determiners such
as some, few, and severall is associated w
ith denotata previously unm
entioned. I won a prize today implicates that h cannot be ex-
pected to know w
hich prize is intended. This, indeed, is the nega-tive im
plicature corresponding to [12) above, and it can be
arrived at via the Maxim
of Quantity, by reasoning as follow
s: since s avoided using the m
ore specific and informative expression the
prize, s does not believe that h has enough knowledge to identify
uniquely the prize concerned. But the C
P has more w
ork than this to do in the interpretation of the articles. C
onsider the fol-low
ing example, w
hich is given by Clark and O
ark (1977:122), follow
ing the model of G
rice (1975:s6): ·
Steven: Wilfrid is m
eeting a wom
an for dinner tonight. Susan:
Does his w
ife know about it?
Steven: Of cbuRSE she does. The w
oman he is m
eeting ls his w
ife.
Susan will norm
ally be justified, following the C
P, in assuming
that the wom
an mentioned by
not Wilfrid's w
ife. This is b(tause, again, a wom
an tends to implicate that s does not have ..
enough kll()wledge to infer w
hich wom
an is meant. Since anyone··
who knows W
ilfrid can be expected to know that he has a w
ife, Steven has broken the M
axim of Q
uantity in using a relatively uninform
ative expression (a woman) in preference to a m
uch m
ore informative one (his wife). H
e has, in fact, blatantly (and·" m
ischievously?) the M
axim of Q
uantity without breaking
the Maxim
ofj)ti'ality:. t4is is a good example of a proppsition
which is true·1'fom
the logical point of view, but is yet very m
is-leading in a pragm
atic way.·
. can also result indirectly ftom
the .l'Se of
Although the X w
ill normally be used in a context
where h 'is aw
are of which X is meant, there are som
e cases w
here, by a kind1'of filii accom
pli, s·causes h to adopt an assump-
tion of unique reference which h probably did not hold before s's
utterance. We m
eet this in sentences like: ·
92 TH
E INTER
PERSO
NA
L ROLE O
F THE CO
OPERA
TIVE PRIN
CIPLE
[ 13) Would you like to see the postcard I got from
Helen last
'week?
from w
hich h may infer, if h did not know
it before, that there is a
postcard such that s received it from H
elen last week.
We can say that [13] entails [14] s got a postcard from
Helen last w
eek
but also that [ 13] impiicates (because of the uniqueness im
pli-cature associated w
ith the) that
[ 15] There exists only one such postcard.
A similar uniqueness by 'fiat' is established in public notices such
as Mind the step and Bew
are the dog (see Haw
kins 1978:112, 121).
To go one step further, w
e may observe that the kind of im
-plicature w
hich Clark and H
aviland (1974, 1977) call a 'bridging assum
ption' may also be explained by the C
P. For example, the
sentence We w
ent into the garden and sat by the fish-pond re-quires, for its norm
al interpretation, the bridging assumption that
'the garden contained a fish-pond'. First, however, look at an ex-
ample of anaphoric reference w
ithout such assumptions:
[16] A: In the end, w
e got through the back door. [17] B: W
as the door locked? In [17] w
e draw the expected conclusion, consonant w
ith the M
axim of Q
uantity, that the door in [17] refers to the same object
as the back door in [16]. This implicature com
es from the sim
ple fact that the door m
ust be referring to some door w
hich is uniquely determ
ined in the context, and that in,. this case there is only one Cloor w
hich is in question, viz the back door. But to m
ake sense of [I8]<m
d [19], h has to use a slightly longer train of inference:
[18] A: In the,end, w
e got through the back door. [19} B: D
id you have to break the lock? ·w
e conclude that the lock mentioned in [19] is the lock of the
back door, but this implicature is arrived at not just through
shared knowledge that the back door is the only door in question,
but also from a piece of general know
ledge, viz that doors often have locks, from
which in tum
the likely inference is drawn that
this particular back door had a lock. This is the bridging assump-
tion from w
hich we are able to reach the conclusion that in [19],
B refers to the lock of the back door. This implicature, derived by
MAXIM
OF RELA
TION
93
means of the bridging assum
ption, saves B from a violation of the
CP. A
s yet a further example of an im
plicature derived from the
use of the articles, consider the double occurrence of a diamond
ring in: M
ary: I've lost a diam
ond ring.
B 'll W
11 1 1 . .
{ a diamond ring}
. l
: e
, u te w
as weanng
one thts m
orning.
By using the indefinite article (or equally the substitute form
one), Bill refuses to. com
mit him
self to whether the ring he saw
w
as the same one that M
ary lost. By pointedly avoiding an im
-plication of co-reference, he avoids incrim
inating Julie. Hence
(a) the Maxim
of Quantity is superficially "V
iolated by refusal to co-refer;
(b) this violation can be interpreted as reticence to avoid impol-
ite accusation; (c) but in fact, this reticence is counterm
anded by the fact that B
ill's remark w
ill be irrelevant unless he is suspicious of Julie.
·
Thus the Maxim
of Quantity, the Politeness Principle, and the
Maxim
of Relation help us to an interpretation of B
ill's remark as
something like an indirect accusation.
Before m
oving on to a closer look at the Maxim
of Relation
itself, I wish to point out that these exam
ples show a close
inter-relationship between
REFER
ENTIA
L pragmatics (p
I I) and IN
TE
RPE
RSO
NA
L pragm
atics. The determ
ination of what objects
are referred to by definite and indefinite expressions is in part determ
ined, as we have seen, by the C
P and even by the PP.
4.3 Maxim
of Relation · T
he Maxim
of Relation 'B
e relevant' has received various inter" pretations, som
e of which treat it as 'a special kind of inform
-ativeness'. Sm
ith and Wilson ( 1979: 177) give an inform
al definition of relevance as follow
s: A
remark P
is relevant to another remark.Q
if P and Q, together w
ith background know
ledge, yield new inform
ation I)gt derivable from
either P or Q, together w
ith background Such a definition m
eans that the connection betvPeen A's and lJ's rem
arks e&'l be show
n to. be one of relevance not cmly in of replies such as [:;w]:
94 TH
E INTERPERSO
NA
L ROLE O
F THE CO
OPERA
TIVE PRIN
CIPLE \;-
[20] A: W
here's myb6¥'<>f chocolates?
B: It's in your robm.
but in more oblique cases such as [ 21]
[21] A: W
here's my box of chocolates?
B: The children were in yo1;1r room
this morning.
[Smith and W
ilson 1979: 175) B's rem
ark in [21] can be made relevant to A
's question on the grounds that, supposing B does not know
the answer to the
tion, B's reply will nevertheless help A to discover the answ
er, by im
plicating that the children may have eaten the chocolates, or at
least that they may know
where they are. O
nce again, the . assum
ption that s and h are observing the CP (together w
ith background know
ledge) enables the implicature to be w
orked out. A
superficial failure in informativeness leads to a conclusion
that B's reply is relevant in contributing to the Maxim
of Quan-
tity at a more indirect lever.
How
ever, rather than see the Maxim
of Relation as subser-
vient, in this way, to the M
axim of Q
uantity, I would like to in-
terpret the relevance of one utterance to another (as in [2o] and [:u]) as part of a broader conception of relevance: the relevance of an utterance to its speech situation. In this broader sense, rel-evance m
ay be defined as follows:
'An utterance U
is relevant to a speech situation if U can be in-
terpreted as contributing to the conversationat goal(s) of s or h.'
Conversational goals m
.ay include both social goals ( eg observing politeness) and personal goals (such as finding one's box of ctiocolates). The personal, illocutionary goal of A in [2o] is to find out w
here the chocolates are. In B's reply, B adopts A's goal, and
supplies the information required. B
ut this goal is fulfilled by way
of another, goal w
hich is adopted by B: the maintaining of
the CP. In fact, in cooperative and socially motivated conver-·
sation, it is normal for one participant to adopt to som
e extent the assum
ed goal or goals of the other. There are, how
ever, examples w
here this is not true. Consider
this additional example given by Sm
ith and Wilson (1979:174):
[22] A: W
here's my box of chocolates?
B: I've got a train to catch. W
e should not call this a very cooperative reply, since it does not advance A
's quest for the chocolates. But B's rem
ark does
MA
XIM
OF R
ELATIO
N
95
become relevant if it is understood as an explanation of w
hy B cannot answ
er A's question. In this function, its contribution to
conversational goals is rather negative: it enables B to conclude the conversation w
ithout (too much) im
politeness. It does not con-tribute, in this case, to A
's goal, but to B's.
Returning to the standard question-and-reply sequence of [20],
the contribution of B to A's illocutionary goal m
ay be repre-sented, in term
s of means-ends analysis, as show
n in Fig. 4·3· Note: The area x, y, z indicates B
's contribution to A's goal. The dou-
ble-shafted arrows (
) symbolize the m
otivating relation between
a goal and an action.
CP
---------
I Initial state: A wants to know
where the chocolates are.
[a] A asks B where the chocolates are.
2 B is aware that A. w
ants to know w
here the chocolates are. [b) B tells A w
here the chocolates are. 3 Final state: A know
s where the chocolates are.
FIGU
RE 4·3
Question and answ
er 20
This is the simplest and m
ost direct case of a means-en,ds analy-
sis of dialogue. Taking the goal-oriented concept of relevance further
however, it is interesting to exam
ine a means-ends
of a more oblique reply such as that of [21] (Fig. 4.4).
(Both Figs. 4·3 and 4·4 are
as alm
ost inevit-able in any m
eans-ends analysts mvolvm
g complicated processes
such as the use of language.) In this case the goal symbolized by
[e] is not achieved: A still has to find out what has happened to
the ch6colates, and this is indicated by the dotted arrow connect-
ing 5 and 6 on Fig. 4·4· Also, the greater com
plexity of this exam
ple is m
Fig. 4·4· by of
Inter-personal Rhetc:iric at stages b and d, m
the pragmatic planm
ng and
96' TH
E INTERPERSO
NA
L ROLE O
F THE CO
OPERA
TIVE PRIN
CIPLE
Note: again B's contribution to the conversation is represented-by the
shaded area.
I Initial state: A w
ants to know w
here the chocolates are. [a] A asks B w
here the chocolates are. 2 B is aw
are that A wants to know
where the chocolates are.
[b] B plans a reply consistent with the C
P and the PP. 3 B is ready to transm
it the message of [b] to A ..
(cJ. B tells A that the children were in A
's room, this m
orning. 4 A ts aw
are that the children were in A
's room this m
orning. [ d] A w
orks out the force of [ b]. 5 A know
s something w
hich will help A to reach state z.
6 Final state: A knows w
here the chocolates are.
FIGU
RE 4·4 Q
uestion and oblique answer 21
interpretation of the utterance. Adm
ittedly these stages are pre-sent. even
the simplest of utterances (see 3·3·1), but they are
partlcula.rly in
of obliquity such as [21). B's reply
m, b k of Ftg. 4-3)-ts show
n to be motivated by the C
P, ie: B
reply 1s mtended to be .rele.yant to A
's conversational goal. In th1s ca..e, how
ever, there 1s an arg11ment for saying that the pp
also plays a role. The reason is thiS':'--l;t chooses to make an itl-
dir. ect reply in preference to a more
such as The chil· m
ay tt_Zken
The most likelY m
otive for this IS pobte reticence in referring to a possibly sinful act
by Instead of accusing the children, B m
akes a seem
mgly m
nocent statement about the w
hereabouts of the chil-' dren! leaving A to com
e to the impolite conclusion. Even this,
b,owever, m
ay not be the whole story. B's apparent politeness
With
to the m
ay be just a piece of archness, and an uom
: mterpretatu::m
may be intended. B m
ay be making
the dehberately ?btuse, but w
ithout intending to prevent A oom
mg to an
THE H
INTIN
G STRA
TEGY
AN
D A
NTICIPA
TORY
ILLOCU
TION
S 97
4.4 The Hinting Strategy and anticipatory illocutions
This example has show
n how the interpretation of 'indirect illoc-
utions' depends heavily on the Maxim
of Relation. This depen-' dence m
anifests itself in what I have elsew
here called a HIN
TING
sTR
ATEG
Y (Leech
198o [1977a]:II2-I4). In polite requests, for exam
ple, it is comm
on to ask a question about h's willingness or
ability to perform an action X as a 'hint' that you w
ant h to do X
: 10 Will you answ
er the phone? Could you answer the phone?
etc. The strategy consists in uttering an iUocution w
hose goal is interpreted as a subsidiary goal for the perform
ance of another illocution. Thus the exchange:
[23] A: C
an you answer the phone?
B: O
.K.
can be thought of as short-circuiting a more elaborate dialogue
such as: [24] A
: Can
answer the phone?
B: Yes.
A: In that case, please answ
er it. B: O
.K.
That is, the fulfilment of the inform
ation-seeking illocution Can
you ... ? is a piece of information w
hich, strictly, A has to possess if A is to know
if the conditions are appropriate for the perform-
ance of the desired
action by the
addressee. The
Hinting
Strategy, however, ensures that the first illocution in [24] does
service for a second, unspoken one. This strategy exploits the M
axim of R
elation in that in the context envisaged for [23], a question about h's ability to do X w
ill only conceivably be rel-evant aS a m
eans of h's eventual performance of X
. It is often helpful, in this w
ay, to think of a conversational ex-change as a com
pressed dialogue. The Hinting Strategy m
erely illustrates the m
ore conventionalized end of a 'scale of relevance' in term
s of which an utterance can
be interpreted as an AN
TIC
I· PA
TOR
Y ILLO
CU
TION
which prepares the w
ay for subseql:..!nt illoc-ution. In practice, of course, s olten assum
es the answer to the
question to be 'Yes'. B
ut even a can-question like Can you sing
more loudly? m
ay function both as an information-seeking ques·
tion and (more indirectly) as a conditional reqq.est. Its force m
ay be approxim
ately: 's wants to know
if h can sing any louder, and the reason for this is that a w
ants h to sing louder'. It does not therefore cease to be a genuine question because it has an ul-. terior purpose to w
hich the question is only an initial step. 11
One m
ay also allow the term
'anticipatory' to apply to con-
THE IN
TERPER
SON
AL RO
LE OF TH
E COO
PERATIV
E PRIN
CIPLE
ditional illocutions such as those performed by A in the follow
ing: [25] A
: Have you got any m
atches? B: Y
es. Here you are. (G
ives matches.)
[26] A: D
o you sell paper clips? B: . Y
es. you like large or sm
all? [ 27] A
: Have you seen m
y address book? B: Y
es, I think it's in the drawer.
[ 28] A: W
hat have you done with the new
spaper? .
B: I haven't done anything with it. It's there in the porch.
[29] A: D
o you happen to know w
hen the next bus leaves? B: Y
es: 5.20. [30] A
: Would you like rom
e more coffee?
B: Thanks. [31) A
: Did you eat all those m
uffins? . B: Y
es, I was so hungry.
In all these cases A's question is m
ore directly a means of eliciting
information., but m
ore remotely a m
eans to some other purpose:
The ulterior illocutionary goal is recognized by the fact that B (in the italicized sections) responds cooperatively wit.h the aim
of helping A to fuifil this goal. In [26], for exam
ple, B answers the
question, but also takes it for granted, in asking 'Would you like
large or small?', that A w
ould like to buy some. In [30] A asks a
question (with the force of an offer of coffee), and B replies as if
the coffee has already been given. In (31], A asks a question with
implications of reproach, and B replies as if to excuse him
self from
the imputed blam
e. The relative importance of the tw
o illo· cutionary goals -
the preparatory and the .ulterior one -varies
from case to case.
Perhaps the best illustration of the gap between the antici-
patory and the ulterior goal is the bizarre result of mistaking the
one for the other: [32] A
: Do you drink?
B: Of course. A
ll humans drink.
There is another type of implicature failure w
here the ulterior goal is m
isinterpreted. This is a fruitful source of stock jokes: [33] A
: Would you like to dance?
B: Sure. Do you know
anyone else who'd like to?
[34] Lecturer: You should have been here at nine.
'Student:· W
hy? What happened?
[35] Custom
er:Tbere's a fly'in my soup!
Waiter:
Don't m
ake a fuss, sir-they'll all want one.
MA
XIM
OF M
AN
NER
99
[36] Lecturer: Who w
asn't in class today? Student:
George W
ashington and Moby D
ick. The failure of B in all these exchanges is, of course, a failure to understand the relevance of A
's remark, ie a m
isunderstanding of how
A's rem
ark is meant to contribute to som
e conversational goal.
It now
clear that relevance, like informativeness and
truthfulness, is not a yes-or-no quality, but a matter of degree. In
some
like the reply It's in your room in [2o], the relevance
is strong and very .clear. A
t the other extreme, there are
cases.· where the relevance is unclear and indirect, like the reply
I've 'got a train to catch in [22]. Relevance is negatively associated
with directness (see 5·7), and correlates w
ith the length of the m
eans-ends chain that has to be constructed to represent the illoc-utionary force of the rem
ark. The Hinting Strategy in [3o] w
as easy to follow
:
[30] A: W
ould you like some m
ore coffee? B
: Thanks. '
Bu.t.one can imagine an even m
ore indirect exchange in which the
relevance of the reply is no longer clear:
[37] A: D
o you like coffee? [38} A
: Are you thirsty?
B: Thanks. B: Thanks.
·
Both these questions could be construed as prelim
inaries to an offer of a drink. B
ut they are too indirect to act readily as indirect offers. N
evertheless, we can im
agine contexts in which the force
of such questions would be m
ade clear by non-verbal R)eans ( eg if A w
ere brandishing a coffee pot at the same tiine), and w
here ex..: changes like [37] and [38] could reasonably occur; H
ere, as else-w
here, it is important to observe, that pragm
atic descriptions involve scales and indeterm
inacies: To take account of this, the
definition of relevance I proposed .earlier should be rephrased in a relative w
ay as follows:
-'A
n utterance U is relevant to a
situation to the extent that U
can be interpreted as contributing to the conversational goal(s) of s or h.'
4.5 Maxim
of Manner
The Maxim
of Manner ('B
e perspicuous') appears to be the Cin-derella of G
rice's four categories: others have folloied Grice in
mentioning it last, and it rarelyfigur:es.m
explanations of conver-c
• -.
• ••
:."
IOO
TH
E INTERPERSO
NA
L ROLE O
F THE CO
OPERA
TIVE PRIN
CIPLE
sational implicature. G
rice himself see.s
as in SOJI_le_ sense less im
portant than (eg) the Maxun o._ Q
uality,. and. as dif-fering from
the others in 'relating not ... to what
rather, to HO
W w
hat is said is to be said' (1975:46). Thts m1ght be _
taken as a clue that the Maxim
of Manner belongs not to the
CP
-and therefore not to the Interpersonal Rhetoric at
but the rhetoric of text. In fact, in the outline of the Textual R
hetonc given in 3·3·3· I introduced the C
larity Principle as one <:fits con-stituent principles. And the difference betw
een pers-
picuous' and 'being clear' is, to say the least, _not persptcuo':ls. N
one the less I believe that Grice w
as nght to reoogmze the
Maxim
of Mann;r as one of the elem
ents of his CP, and that the
charge to 'be clear' is placed on language users as of the In-
terpersonal Rhetoric, as w
ell.as of the. are
two kinds of clarity. O
ne kind conststs m m
aking unambtguous
use of the syntax and phonology of the _in
to. con-struct a clear TEX
T. Another type of clanty conststs. m
a
clear MESSA
GE, ie a m
essage which is perspicu<?us or m
telhgtble m
the sense of conveying the intended illocutionary to the
addressee. What this im
plies is that exchanges such as should be rather rare-as indeed, they probably are,
JOke books. Perspicuity in this sense is o6viously hand in_ glove w
tth evance; both the M
axim of M
anner and the of
will favour the m
ost direct comm
unication of one s point, and both, for that reason, w
ill militate
the obhqutty of the H
inting Strategy. It is presumably. for
that addressees w
ill normally assum
e the most direct m
terpretatton as the 'default interpretation', and w
ill seek indirect interpretations only w
hen direct interpretations are blocked. ··
If the Maxim
of Manner's only function w
as to support the M
rucim of Relation in this w
ay, there would be
for doubting its inclusion as. part
the CP.
I believe that negative sentences provtde evtdence for the m
dependent role of this m
axim.
-
4·S·I The obliquity and uninfonnatlveness of There are tw
o reasons for prop95ing that negative sentences pragm
atically less favoured than positive ones. first
ts that negatives, all things being equal, are less m
formattve than
their positive counterparts: [39] A
braham Lincoln was not shot by
Mazeppa.
[40] Abraham
Lincoln was shot by John Wilkes B
ooth. T
he ·world's population of negative facts is far greater than its
MAXIM
OF M
AN
NER
IOI
population of positive facts: for example, the num
ber of people w
ho did not shoot Lincoln is many m
illion times greater than the
number of people w
ho did shoot Lincoln, and it is for this reason that [39] is m
uch less informative than [40], although both state-
ments are true. T
he sub-maxim
of NEG
ATIV
E UN
INFO
RM
ATIV
E-N
ESS, as we m
ay call it, when Com
bined with the M
axim of
Quantity, im
plies that a negative sentence will be avoided if a
positive one can be used in its place. Moreover, it w
ill imply that
when negative sentences A
RE used, it w
ill be for a special pur-pose. In fact, the C
P will predict that negative sentences tend to
be used precisely in situations when they are not less inform
ative for a given purpose than positive ones: and this w
ill be when s
wants to deny som
e proposition which has been put forw
ard or entertained by som
eone in the context (probably the addressee). N
egative uninfonnativeness therefore provides an explanation of w
hy negative propositions are, in pragmatic term
s, denials of positive propositions w
hich are in some sense 'present in the
context'. This generalization is not, how
ever, convincingly applied in all cases. If w
e consider a negative sentence like Our cat is not m
ale, this is just as inform
ative as the contrasting positive statement
Our cat is m
ale. Moreover, in this case there is a positive sene
tence which is for all practical purposes synonym
ous with the
negative one: Our cat is fem
ale. But the negative sentence still
strikes one as being 'marked', and as requiring special interpret-
ation as a denial of what som
eone else has asserted. Since the M
axim o
f Quantity cannot explain this case, the M
axim of M
ana ner m
ay be suggested as an alternative means of explanation. The
explanation runs as follows: a negative sentence (as psycholing-
uistic research has shown -
see Oark and O
ark I977:107-IO)
takes longer to process, and is presumably m
ore difficult to pro-cess, than a positive sentence. Therefore, by choosing a negative sentence in preference to a positive one, s causes the utterance to be m
ore oblique and obscure than it need be. Therefore s violates the M
axim of M
anner. He m
ust be doing this for some reason -
and the most obvious reason for using the nega.tive sentence is to
deny its positive counterpart. t<_,
There are also some exceptions to the generaliZation that
nelative sente · .
e more 'm
arked' than positive ones, and carry im
plications o -:at The exceptions
to be negative ex-pressions of em
otion or attitude: I don't like Kenneth;_ He. doesn't believe in
We don't agree; etc. The negative ts often
preferred to
the syntactically
equivalent (J
dislike K
enneth, etc,) as a form. cf under:>tatem
ent. Negation here
l ,
.I ,
• •
'. ...
• '-'
,..; •
apparelil\t y a .1:-teugmg or a!:iv.lCe,
ffio .. ,v!h10n !i:J!
102 TH
E INTERPERSO
NA
L ROLE O
F UIE
COO
PERATIV
E PRINCIPLE
which m
ay be politeness or simply euphem
istic reticence in the expression of opinion and attitudes (see 6.1.2). Such cases are in-dependently explicable, and do not detract from
the general point about negation being pragm
atically interpreted as denial. . This diScussion of negation, therefore, supports the case for a
Maxim
of Manner as an independent part of the C
P, in spite of the overlap of its function w
ith the Maxim
of Relation and the
textual Principle of Clarity.
Notes
I. Earlier treatments of politeness w
ithin a linguistic framew
ork are Lakoff (1973),
Brow
n and
Levinson (1978),
and Leech (19&
[1977a]). The 'positive' and 'negative' aspects of politeness derive from
Brow
n and Levinson's distinction between positive and nega-
tive face (1978:64), and their consequent distinction between positive
and negative politeness (1978: passim)
2. Grice (1975) treats irony as a special kind of im
plicature or impli-
cative strategy, rather than as a principle in its own right. There is no
necessary conflict between this and m
y own treatm
ent of irony as a second-order principle. Such a principle m
ay, in fact, be regarded as a highly institutionalized strategy w
hereby ;speakers square their lan-guage behaviour w
ith more basic principles such as the C
P and the PP.
3· By relative politeness I m
ean politeness relative to context or situ-ation. In an absolute sense. [x} Just be quiet is less polite than [2] W
ould you please be quiet for a mom
ent? But there are occasions w
here [I] could be too polite, and other occasions where [2] w
ould not be polite enough. There are even som
e occasions where [2]
would strike one as less polite than [x]; w
here, for example, [1] w
as interpreted as a form
of banter, and where [2] w
as used ironically. It is only in a relative sense that w
e can talk of overpoliteness and underpoliteness ..
4· See Miller (1967:183-90) on politeness phenom
ena i_n Japanese. 5. O
n the cancellability of implicature, see G
azdar (1979:131 -2). His
notion of satisfiable incrementation ('A
ll the news that fits') accounts
for the cancellation of implicature w
here a conflict arises between
the Maxim
s of Quantity and Q
uality. 6. 'Strong' and 'w
eak' are interpreted here as terms of a sem
antic opposition of inverseness (Leech I9(i9:s6, 200). For treatm
ents of logical operators on the lines illustrated here, see H
orn ( 1976) and G
azdar (I 979). 1· Strictly, the tw
o statements of D
I are abbreviations for: 's believes ·
that s thinks that Grandpa is asleep', and 's believes that s is not cer-
tain that. Grandpa is asleep'. There is a principle of Transitivity of
Reflexive :Belief (see further p 190) w
hich allows these to be sim
-plified
tJtjnks that is asleep' and 's is not certain that
NOTES 103
.is asleep', the
that s cannot coherently claim to
believe himself to be m
beltef-state B, unless he is indeed in belief-state !J· A
t its most general, the principle states that any proposition
's that s P
R?P
(P)', w
here PROP is a belief predicate, allow
s to m
fer proposition 's
PROP (P)'. This applies
no. to positive belief-states, but also to negative ones, suGb as
uncertainty. 8. The traditional logical distinction betw
een inclusive and exclusive or like that betw
een inclusive and exclusive negation, bas been recently under attack from
more than one quarter, See B
arrett and Stenner . (1971) and K
empson (1977: 126-8).
9· points out (198o:92-3), could is also acceptable in a posi-
tive habitual sense: ... my father could usually lay hands on what he
wanted. 10. Earlier explanations of indirect illocutions are those of G
ordon and Lakoff (1971), Sadock (1974), and Searle (1979 [1975b]). See Leech (xgSo {1977a]: 87-9, n2-14) for a discussion of these, and their re-lation to the H
inting Strategy. n
. Searle [1975b) m
akes this point, and thereby improves on the
gtven by Gordon and Lakoff (1971), w
bo regard an indirect tllocution as contextually am
biguous between 'direct' and 'indirect'
rather than as conveying the 'indirect' interpretation vza the 'direct' one. In the present account illocutionary force is rep-resented by statem
ents about s's volitional attitude. The dual illoc-
utionary force ofCan you sing m
ore loudly? is reflected in the fact that its illocutionary force description contains tw
o volitional attitude statem
ents, one of them being im
plicated via the other.
ChapterS
The Tact Maxim
'Tis a m
axim trem
endous but trite [Lewis Carroll. The Hunting of the Snark)
Far. from being
·matter of 'being civil', politeness is
an Important m
1ssmg link betw
een the CP and the problem
of how
to relate sense to force. I have already emphasized the role of
politeness in pragmatics
chapt.ers, but in this and the chapter I shall exam
me how
1t Works m
more detail. W
here-as m
Chapter 4 I concentr:ated on the productive strategies of
means-ends analysis,
I shall now
concentrate more on the
heuristic strategies of interpretation, looking at politeness from
the addressee's rather than from the speaker's end.
5.1 Varieties of illocutlonary function
Different kinds and degrees of politeness are called for in differQ
ent situations. At the m
ost general level, illoeutionary functions m
ay be classified into the following four types, according to how
they_ relate to the social goal of establishing and m
aintaining conuty.
(a) coMPETITIVE: _The illocutionary goal com
petes with the social
goal; eg ordenng, asking, demanding, begging, etc.
(b) CONVIVIAL: The illocutionary goal coincides w
ith the social .' goal; eg offering, inviting, greeting, thanking, congratulating.
(c) T
he illocutionary goal is indifferent to the .
soetal goal; eg asserting, reporting, announcing, instructing. (d) CONFLICTIVE: T
he illocutionary goal conflicts with the social
goal; eg threatening, accusing, cursing, reprimanding"
Of .these,_ the first tw
o types are the ones whlcl:. chiefly involve l_X)liteness. VVhere the m
ocutionary function is (:'}MPE!I1'iVE (a)
politeooss is of a negative characier, and its purpose is to 'n::
SEARLE'S CATEGORIES OF ILLOCUTIONARY ACTS 105
duce the discord implicit in the com
petition between w
hat s wants
to achiev_e, and ·what is 'good m
anners'. Com
petitive goals are those w
h1ch are DISCOURTEous, such as getting som
e-one to lend you m
oney . 1 (For the sake of clarity, I shall make a
terminological distinction in applying 'courtesy' and 'discourtesy'
to goals, and 'politeness' to the linguistic or other behaviour in . w
hich someone engages in as m
eans to those goals.) The PP is · therefore required to m
itigate the intrinsic discourtesy of the goal. T
he second type, that of coNVIVIAL functions (b), is, on the contrary, intrinsically couR
TEous: politeness here takes a more
positive form of seeking opportunities for com
ity. Positive polite-ness m
eans observing the PP in that, for example, if you have an
opportunity to congratulate h on his tOOth birthday, you should do s?. In the third. categ?ry are. COLLABORATIVE illocutionary functions (c), for w
h1ch politeness l.S largely irrelevant. Most w
rit-ten discourse com
es into this category. And in the fourth categ-
ory of politeness. is out of the question,
because confhcttve Illocuttons are, by theu very nature, designed to cause offence. T
o threaten or curse someone in a polite m
an-ner is virtually a contradiction in term
s: the only way to m
ake sense of the idea is to suppose that the speaker does so ironically (see 6.3). Presum
ably in the course of socialization children learn to replace conflictive com
munication by other types (especially
br t_he type), and this is one good reason w
hy con-fhcttve dlocutlons tend, thankfully, to be rather m
arginal to hum
an linguistic behaviour in normal circum
stances. H
ence, in considering polite and impolite linguistic behaviour,
we m
ay confine our attention mainly to com
petitive and convivial illocutions, w
ith their corresponding categories of negative anJ positive politeness.
5.2 Searle's categories of illocutionary acts T
he above classification is based on functions, whereas Searle's
classification of illocutionary acts (1979[1975a]) is based on varied criteria. 2 B
efore proceeding, however, w
e will find it useful to re-
late the two classifications, and show
how politeness affects
Searle's categories. Roughly speaking,. Searle's categories are-
defined as follows (for further
see 9.2.4): I. ASSERTIVES com
mit s to the truth of the expressed propo-
sition: eg stating, suggesting, boasting, complaining, claim
ing, reporting. Such illocutions tend to be neutral as regards polite-ness, ie they belong to the collaborative
(c) above.
100 . THE TACf M
AXIM
But there are som
e..exceptions: for example, boasting is gener-
ally considered to be impolite. Sem
antically; assertives are propositional.
2. DIRECTIVES are intended to produce som
e effect through action by the hearer: ordering, com
manding, requesting, advis-
ing, and recomm
ending are examples. They frequently belong
to the competitive· category (a), and therefore com
prise a category of illocutions in w
hich negative politeness is import-
ant. On the other hand; som
e directives (such as invitations) are intrinsically polite. To avoid confusion in using the term
'directive' in relation to. 'direct and indirect illocutions', I have preferred to· use the teim
.IMPOSITIVE for com
petitive illocu-tions in this class.
3· coMM
ISSIVES comm
its (to a greater or lesser degree) to some
future action; eg promising, vow
ing, offering. These tend to be convivial rather than com
petitive, being performed in the in-
terests of someone other than the speaker.
4· have
the fum;tion
of expressing, or m
aking know
n, the speaker's psychological-attitude towards a state of
affairs which the illocution presuPJ)O
Ses; eg thanking, congra-tulating,
blaming,. praising, condoling, etc. Like
the comm
issives, they tend t-o be ctinvivial, and therefore intrin-sically polite. The reverse is true, how
ever, of such expressives as 'blam
ing' and 'accusing'. ,.(!
5· DECLARATIONS are illocutions whose 'successful p¢form
ance ... brings about the correspondence betw
een the propositional content and reality'; eg
resigning, dism
issing, christening,
naming,
excomm
unicating, appointing, sentencing, etc.
In this, these actions are, as Searle says (18-19) 'a very special category of speech acts': they are perform
ed, normally speak-
ing, by someone w
ho is especially authorized to do so within
some institutional
(Oassical exam
ples arejudges. sentencing offenders, m
inisters of religion christening "'babies, . dignitaries nam
ingships, etc.) As institutional rather than per-
sonal acts, they can scarcely be said to involve For -
example, although sentencing a person is an unpleasant thing
. to do, the judge h;w com
plete authority in doing so, and can scarcely be . said to' sentence som
eone 'impolitely'. M
oreovef, politeness is not relevant to
because they do not have an addressee in the sense that {8pplies to personal dis-course: the person w
ho makes a declaration uses language as
an outward·sign that som
e institutional (social, religious, legal, etc.) action is perform
ed. It would thus be totally out of place,
and1 w
ould undermine the force of the declaration, if (say) the
TACf: ONE KIND OF POLITENESS 107
priest baptizing were to hedge his w
ords with politeness;
changing 'I baptize yoti ... ' into 'Could I baptize you ... ?',
etc. The same applies, to a lesser extent, to m
ore private dec-larations, such as resigning in a gam
e of chess, or bidding in a gam
e of bridge.
Although there are som
e cases not covered by the generalizations above, it is w
orth making the point that, as far as Searle's cat-
egories go, negative politeness belongs pre-eminently to the
DIRECTIVE class, while positive politeness is found pre-em
inently in the COM
MISSIVE and EXPRESSIV£ classes.
5.3 Tact: one kind of politeness Let us now
relate illocutionary acts more precisely to the kinds of
politeness with w
hich they are associated. I have said that politeness is essential asym
metrical: w
hat is polite w
ith respect to h or to some third party w
ill. be impolite
with respect to s, and vice versa. The justification for the m
axims
of politeness is precisely that they explain such asymm
etries, and their consequences in term
s of indirectness. I shall first of all ex-plain this w
ith reference to what is perhaps the m
ost important
kind of politeness in English-speaking society: that w
hich is covered by the operation of the TACT M
AXIM.
The Tact M
axim applies to Searle's directive and com
missive
categories of illocutions, which refer, in their propositional con-
tent X, to som
e action to be performed, respectively, by the
hearer or the speaker. This action may be called A
, and be
evaluated in terms of w
hat s assumes to be its cost or benefit to s
or h. 3 On this basis, X ('you w
ill peel those potatoes',_etc.) may
be placed on a cosT:-BENEFIT SCALE, as in the following exam
-ples:
[I] Peel these potatoes. [ 2] H
and me the new
spaper. [3] Sit dow
n. [4] Look at that . [5] Enjoy your holiday. [ 6] H
ave another
cost to h less polite
benefit to h m
ore polite A
t some rather indeterm
inate point oq this s le (depending on
the context) the relevant value becomes 'benefit
h' rather than 'cost to h'; but clearly, if w
e keep the imperative
od constant,
!08 TH
E TACT M
AX
IM
there is a general increase in politeness (other factors being equal) betw
een [I] and [6]. A
nother way of obtaining a scale of politeness is to keep the
same propositional content X (eg: X
= 'Y
ou will peel these pota-
toes') and to increase the degree of politeness by using a more
and more indirect kind of illocution. Indirect illocutions tend to
be more polite (a) because they increase the
of optional-ity, and (b) because the m
ore indirect an illocution is, the more
diminished and tentative its ferce tends to be.
[7] Answ
er the phone. [8] I w
ant you to answer the
phone. [9] W
ill you answer the phone?
[ w] C
an you answer the phone?
(u] Would you m
ind answering
the phone? [12] C
ould you possibly answer the
phone?
indirectness less P.Olite
etc. m
ore polite O
ne of the things pragmatics should explain is: W
hy do some in-
direct illocutions function as impositives, w
hile others do not? For exam
ple, [13] is an offer, rather than an impositive-it im
-plies that sitting dow
n is to h's benefit: ·
[13] Won't you sit dow
n? [14] C
an't yousit down?
[15] Wouldn't you m
ind sitting down?
On the other hand, [ 14] typically has an im
positive force, whereas
. [15] does not seem to be usable in either a com
missive or an im
c positive function. O
ther things which need explaining here are
these: (i) Why does the use of an indirect strategy, such as .the
addition of negation in [13] and [14], in the one case, that of[13], lead to greater politeness, and in the other, that of [14], w
ith its overtone of im
patience, lead to less politeness? And (ii) W
hy do different indirect illocutions have different em
otive or attitudinal im
plications which are not reducible to the sim
ple matter of de-·
gree of politeness? For example,
[ 16] You w
ill be silent. [17] C
an't you shut up? [18} I'd keep m
y mouth shut (if I w
ere you). are all, in the right context, im
positives, whose goal is the silence
TACT: O
NE K
IND
OF PO
LITENESS
I09 of h; but the w
ay they are formulated suggests a very different
strategy on part of the sin each case.
oth [16], which sug-
gests the severity of a military instruction,
d [17], which sug-
gests extreme irritation w
ith h's behaviour, are · polite; w
hereas [ 18] is m
ore like a friendly piece of advice for h's enefit. Thus it
'is quite insufficient to note, in [7]-[12] and [16]-[ ], the corre·
lation between indirectness and politeness: w
e must be able to say
not only how polite a given illocution is, but w
hy a particular de-vice of indirectness contributes to a particular illocutionary goal. For exam
ple in [7]-[12], the degree of indirectness correlates w
ith the degree to which h is allow
ed the option of not perform-
ing the intended action, answering the phone. Indeed, the point
of the strategy of indirectness, here, is to bias the impositive
more and m
ore towards the negative choice, so that it becom
es progressively easier for h to say no. In this w
ay, negative polite-ness (ie serving the avoidance of the cost to h) is increased.
It may seem
strange here to describe politeness, as I did earlier, as 'm
inimizing im
polite beliefs'. But on reflection,. it is
reasonable. The propositional content of all these sentences is discourteous to h in so far as it attributes som
e effort, trouble, or cost to h. B
y using the imperative in [1] and [7], s expresses the
belief that h will perform the action. (The use of the im
perative does not allow
that h has any choice in the matter, w
hereas the question form
of [9], for example, expresses doubt as to w
hether h w
ill do A.) B
ut as the element of doubt or negative bias is in-
troduced and increased in examples [9]-[12], so the expression of
the belief that h will perform
the action is weakened.
There are two sides to the Tact M
axim, a negative side 'M
ini-m
ize the cost to h', and a positive side, 'Maxim
ize the benefit to h'. The second is less im
portant, but is a natural corollary of the first. It m
eans, for example, that in proposing som
e action ben-eficial to h, s should bias the illocution tow
ards a positive outcome,
by restricting h's opportunity of saying 'No'. Thus an im
perative, w
hich in effect does not allow h to say 'N
o' is (in an informal con-
text) a positively polite way of m
aking an offer:' Help yourself;
Have another sandw
ich; etc. The positive bias can even be in-creased by the persuasive em
phasis of: Do have another sand-
wich!; You MU
ST have another sandwich! In this case, the m
ore indirect form
s of [9]-[12] are if anything less polite than the most
direct form: W
ould you mind having another sandw
ich? would
suggest that h would do s a positive favour by accepting -
and therefore perhaps that the sandw
iches were stale, inedible,· or
poisoned! The reason for this reversal of polite strategies in im-
positives and comm
issives is fairly obvious, and has to do with the
110 TH
E TACf MA
XIM
asymm
etry of politeness: what m
ust be expressed strongly by one participant as a 'polite belief m
ust be played down equally by the
other participant as an 'impolite belief. So increasing the positive
polite.ness of an offer means anticipating and counteracting the
negative politeness of the recipient. This helps to clarify w
hy the negative form of the question,
Won't you help yourself, etc., is polite in an offer. The negative
question is (as I shall propose later, 7·3-2), a question about a ?1
negative proposition, which itself (see 4-5) im
plies the denial of a positive proposition. The sense can be spelt out literally as fol-low
s: 'I hope and expect you to help yourself, but now it appears
that you will not help yourself; is this really so?' 4 In effect, it pays
h the complim
ent of holding a polite belief, and at the same tim
e politely (from
s's point of view) expresses disbelief in that belief,
and so invites h, in spite of ostensible reluctance, to accept the offer. In this w
ay the question is biased towards a positive out-
come. For the opposite reason W
ould you mind helping yourself
is polite as an impositive. The sense of m
ind in this construction indicates a negative expectation of the action A
, in that Would
you mind is sem
antically equivalent to Would you dislike . . . or
Would you object to ... In this respect it bas a built-in negative
bias, and contrasts with W
ould you like ... , which is m
ore natu-rally interpreted as introducing an offer. It is logically speaking a negative answ
er to this question which expresses h's ·com
pliance: (N
o, I wouldn't m
ind ... ) but even this is a non-comm
ittal re-sponse, m
erely carrying the sense 'I would not object', w
hich confirm
s that h is not unwilling rather than that h is w
illing to do A
. The introduction of a further negative into this strategy makes
no sense; hence the unacceptability of Wouldn't you m
ind ... ?
5.4 Pragmatic paradoxes of politeness
One m
ay argue that in 'ideally polite' circles, the determination ·
of the two participants in the discourse to be as polite as each
other leads to an infinite regress in the 'logic' of conversational behaviour. Suppose a and b are tw
o participants, and that A is a courteous action w
hich a wishes to perform
forb; for example, a
may be m
aking the following offer:
[19] Let me carry those cases for you.
Suppose, in addition, that both a and b are obeying the Tact M
axim to its fullest extent. Then the follow
ing might represent in
outline the first two stages of the infinite regress:
PRAGMATIC PARADOXES OF POLITENESS
(i) An offer is m
ade by a: (I) a is observing the· pp (2) A is favourable to b THEREFORE: (3) a (politely) im
plicates 'a wants A to occur'
III
(given) (given)
(ii) b declines the offer made by a:
(4) b is observing the PP
(from I and 2, and PP)
(5) A is unfavourable to a THEREFORE:
(given) (given)
(6) b (politely) implicates 'b does not w
ant A to occur' (from
4, 5, and PP) The tw
o implicatures of (3) and (6) am
ount to what w
e may call a
PRAGMA!I.c
an attribution of incompatible attitudes to
the I?arttapants m a dialogue. H
owever, on the assum
ption that a m
terpret the force. b's rem
ark (see 2.4), it is possible for a to tnfer from
(6) that It ts because b wants to obey theP
P that b
that b does not want A to occur. In other w
ords it is . pos_stble for a to infer that ( 6) is sim
ply implicated for the sake of
and therefore that b DOES want A to occur. Therefore
1t IS pohte for a to renew the offer m
ore strongly. But by the sam
e b
a's offer that the implicaturetbf (3)
ts for tt m
ay be tmplicated only for politeness' sake. So it is
polite for b. to m
ore. This tug-of-war lio)f m
utual deference
contmue until one of the participants yields to the
greater politeness of the other. (i) and (ii) above, lines (1-6) m
ay be thought of as repre-se!ltm
g parts a m
eans--ends analysis. But from
an interpret-t;>Otnt of vtew
, o:ay also form
part of a heuristic analysis, m
w_htcb case the w
orks from the opposite direction.
Fo_r from
the tmphcature of (3), h m
ay infer (I)!Hhat sis
bemg poht,e.
the politeness implicature
may be a
meta-tm
phcature whtcb .ttself contains a reference to
another implicature. Thus from
an offer like [9] Let me carry
those cases for you, a relatively direct implicature will be:
[20] s wants to carry h's cases.
But since?
by backgr?und knowledge, it is assum
ed already that carrym
g of cases ts unfavourable to s this w
ill lead to the im
plicature: '
[21] sis being polite and m
ay lead more indirectly to the m
eta-implicature:
H2
TH
E TAC
f MAXIM
[22] It is only because s is.being polite that s implicates that's
wants to carry h's cases'.
A further inference from
[ 22] is that s may therefore be violating
the Maxim
of Quality, ie that the m
ost direct implicature 's w
ants to carry h's cases' m
ay be false. The question we face here is one
of the sincerity or seriousness of polite behaviour. It is possible to infer [21] w
ithout taking the further step to [22], with its im
pli-cation that s m
ay not be sincere. That is; it is possible for a polite utterance to be inte7,reted as genuinely • polite, or sim
ply as 'polite on the record'.
The latter interpretation will obviously be
favoured if all the evidence indicates that 12o] is false. Therefore, if [20] is false, his able to discount [2o] as a
for s's offer as a m
atter of politeness, and the way ts open for a pohte
refusal, as indicated in (ii) above. This refusal in tum m
ay be re-jected by s as non-serious, and the gam
e of conversational ping-pong m
ay proceed again. W
e may observe in the , pragm
atic paradoxes of politeness a com
edy of inaction: it is as if two people are eternally prevented
from passing through a doorw
ay because each is too polite to go before the other. Sim
ilar paradoxes of behaviour are in
certain cultures in which an offer has to be repeated and dechned
n times before it is accepted. It is just as w
ell that in practice, no one is ideally polite.
The question is: Why does politeness m
anifest itself in this be-havioural or pragm
atic paradox? The answer to this appears itself
somew
hat paradoxical: that the paradoxes of politeness function as an antidote to a m
ore dangerous kind of paradox. This more
dangerous pantdox is a violation of the logic of action; that is, a state in w
hich two individuals, a and b, have m
-com
patible goals. It is epitomized in a situation, the
of that just visualized, in w
hich each of the two persons w
tshmg to
go through the doorway attem
pts to go before the other, with-
the result that they collide in the doorway! Such paradoxe&
clear-ly lead to direct conflict, and are socially, not to perilous. They m
ay be placed on a scale of decreasmg gravtty as
follows: 6
I. AC
TUA
L CO
NFLIC
T (strongest) .
{m.akes
}b d A
but b {tries to}not do A.
a tnes to make
0 '
does 2.
DISO
BED
IENC
E a tells/orders b to do A
, but b does not do A.
3· W
ILL
FLOU
TING
.
a comm
unicates to b that a wants b to do A
, but b does not do A.
1 PRA
GM
ATIC PA
RAD
OX
ES OF PO
LITENESS
4· W
ILL
INC
OM
PATIB
ILITY (w
eakest) a com
muirlcates to b that a w
ants b to do A, but b com
mu-
nicates to a that b does not want to do A
. C
orresponding to these types of dangerous situation, there are four m
ore types in which the positions of positive and negative
actions are reversed: Ia.
AC
TUA
L CO
NFLIC
T (strongest)
{ stops }
. { tries to do }
a tries to stop
b domg A
, but b does
A.
2a. D
ISOB
EDIEN
CE
a forbids b to do A, but b does A
. Ja.
WIL
L FLO
UTIN
G
a comm
unicates to b that a wants b not to do A
, but b does A.
4a. W
ILL
INC
OM
PATIB
ILITY (w
eakest) a com
municates to b that a w
ants b not to do A, but b com
-m
unicates to a that b wants to do A
. W
e can compare these conflict situations by im
agining a simple
drama in w
hich Ann ( = a) has the goal of getting Bill ( = b) to
give her £so. In the gravest case (1), Ann takes the m
oney by force, and Bill tries to prevent her; or else Ann tries to take the m
oney, and Bill prevents her. (If we leave out 'tries to', these
statements do indeed becom
e logical paradoxes or contradictions: eg 'A
nn took the money, but Bill prevented her from
doing so'.) In the least grave case (4), A
nn expresses her wish that Bill
should give her the money, and Bill expresses his w
ish not to do so. T
he outcome here is not direct conflict, but a volitional dis-
cord which (as w
e all know) is often a stepping-stone to a m
ore serious breach of com
ity. A
gainst this background, the function' of the Tact Maxim
is a negative one: it is a m
eans of avoiding conflict. All the conflict
situations detailed above involve an incompatibility of the general
form:
a VOL [X
], b VO
L [not-X]
where VO
L is a volitional predicate such as want, intend. (This is true m
anifestly of the weakest case 4 and 4a, and also, by im
-plication, of the others.) The Tact M
axim, in its m
ost absolute form
, prevents such incompatibilities from
arising, since 'Mini-
mize the cost to h' carries the im
plication 'Do not (express the
wish to) do w
hat h does not want'. I" both parties observe this
maxim
, there will be no m
ore conflict; but on the,other hand, the avoidance strategy is, as w
e have seen, a recipe for inactivity. This description, how
ever, polarizes a situation which is really
Il4 TH
E TACT M
AX
IM
a matter of degree. The Tact M
axim is observed to a certain ex-
tent, and this_means that on the one hand, conflict is not alw
ays avoided, and on the other hand, that inaction does not alw
ays re-sult. In im
positives; the action of the Tact -Maxim
is rather to cause us to suppress, to play dow
n, to hedge, beliefs which are
costly to h. And w
e have seen that the chief means of doing this
is to weaken the belief by biasing the illocution tow
ards a negative outcom
e. Alsa relevant is this general law
governing indirectness: that the m
ore indirect an implicature, the w
eaker its force.
5.5 Semantic representation of declaratives, interrogatives,
and imperatives
Bearing the above two points in m
ind, we will proceed, in 5.6, to exam
ine some of the strategies for m
aking polite and impolite im
-positives. It is already obvious, from
such straightforward sen-
tences as Sit down!, You will sit down, and W
ill you sit down?,
that gramm
atically, impositives m
ay take the form of any of the
three major sentence types. A
ccordingly, before going further, it is advisable to consider how
declarative, interrogative, and im-
perative sentences are represented on the semantic level.
The terms D
ECLA
RA
TIVE, IN
TERROG
ATIV
E, and IMPER
ATIV
E are typically used for syntactic categories, and I shall follow
this usage, in treating them
as basic sentence-types. They are conven-tionally distinguished from
corresponding semantic or speech-act
categories, referred to by such terms as 'assertion', 'question',
and 'comm
and'. If one accepts the complem
entarist position of this book, how
ever, one has to make a further differentiation be-
tween categories on the sem
antic and pragmatic levels. The
English language is, unfortunately, rather unfavourable to the com
plementarist view
in not providing us with a satisfactory
terminology for all three levels. It has consequently been easy
to assimilate sem
antic vocabulary to pragmatic vocabulary, or
vice versa -for exam
ple, to treat a 'question' as both a semantic
(logical) and a pragmatic entity. 7 I shall distinguish term
s on the three levels as follow
s: 8
SYN
TAC
TIC:
declarative interrogative
im)erative
t .. L
SEMA
NTIC
: questiO
n m
and
'asseLon' t
. 1 .. PR
AG
MA
TIC:
'asking' tm
postbve Thus the sense of a declarative sentence, of an interrogative sen-
SEMA
NTIC REPRESEN
TATIO
N
IIS tence, and of an im
perative sentence is respectively a proposition, a question, and a m
and. The link between the sem
antic and pragm
atic categories, however, is less clear-cut, as we w
ould ex-pect. It has been seen, for exam
ple, that a proposition or a ques-tion can have the force of an im
positive, and (more generally)
that strategies of indirectness ensure that every semantic type can
be matched w
ith a variety of pragmatic types. The quotation
marks I have put round 'assertion', and 'asking', in fact reflect
the uncertainty I feel as to whether these are useful tem
1s, given the non-categorical nature of iU
ocutionary force (seep 23). To be at ali useful, they should be defined in w
ays which m
ake them
less general than the corresponding gramm
atical terms. Perhaps
these are reasonable first approximations:
ASSER
TION
: an utterance whose illocutionary goal is to cause h
to be aware that [Y] (w
here [Y] is some proposition).
ASK
ING
: an utterance whose illocutionary gqal is to get h to
causes to be aware that [Y] (w
here [Y] is some proposition).
These definitions propose that assertions and askings are to do w
ith the passing of information betw
een s and h. This means that,
for example, an exam
ination question or a rhetorical question w
ould not have the force of an asking. A
nother terminological deficiency is the lack of a recognized
term for the general class of logical entities of w
hich propositions, questions,
and m
ands are subcategories.
There is
comm
on ideational content that m
ay be shared by propositions, questions, and m
ands, and which has been variously described as a 'propo-
sitional content', 'predication', or 'sentence radical'. 9 For instance, You will sit down, W
ill you sit down, and (You) sit dow
n! all share a com
mon propositional content X
, describing a sitting dow
n of h in the future. They differ in terms of logical form
, but I shall w
ant to use a single term, PR
OPO
SITION
AL, to apply to ali
three types of sentence sense. I shall indicate propositionals by the use of square brackets. A
proposition is the most fully spe-
cified type of propositional, and may be represented by a predica-
tion X, Y, ... within the scope of a positive or negative operator:
fneg (X)} or {pos (X
)}. A yes-no question can then be repre-
as a 'propositional function', ie a propositional in which
the question mark is a free variable ranging over pos and neg. 10
For example:
[23] Mary opened the door.
[24] Mary didn't open the door.
[25] Did M
ary open the door?
[pos (X)]
[neg (X)]
\ [? (X)]
\
II6 TH
E TAC
T MA
XIM
The free variable is in effect a gap in the sense of a proposition,
and so a yes-no question is characterizable as a defective osition, from
which one specification is m
issing, viz the polarity sign pos or neg. This departs from
standard logic in introducing a 'positive' as w
ell as a 'negative' operator, but such an addition is reasonable, and is like prefixing a'+
' sign to a numerical
ion to mark it as positive, rather than negative. In term
s famtbar
to linguists, the arithmetkal negative sign is (like the negative
operator in propositions) the marked term
, which is obligatorily
signalled, whereas L.;e positive is norm
ally omitted. T
he most
palpable advantage of this analysis is that it permits a unitary
logical of questions, since w
h-questions can be similarly
characterized as propositional functions. The free variable is in this case the unspecified argum
ent represented by the wh-w
ord w
ho, what, w
he'!, etc.; eg the variable x in the question: .
[26] Where does T
om w
ork? fpos (Tom
works at placex)]
A logically w
ell-formed answ
er to such a question is any prop-osition w
hich specifies a value for the variable x; eg for [26], '(Tom
works) in London', '(Tom
works) at the post office', etc. Sim
ilar-lv
a logically well-form
ed answer to a yes-no question is a prop-
O";ition which fills in the m
issing polarity pos or neg. 11
If questions are underspecified in comparison w
ith propo-sitions, m
ands are even more underspecified, since in contrast to
propositions and questions they have no contrasts of . or
modality. They do, how
ever, have the contrast between
and negative, and so may take the form
[pos (X)], [neg (X
)], JUSt like propositions. M
ands constitute a more general category than
what w
e normally understand by a 'com
mand': they reflect a
comm
on element of m
eaning shared by three moods, those of the
imperative, infinitive/and present subjunctive, in the traditional
description of the gramm
ar of English and of many other lan-
guages, All m
ay be described as NO
N-IN
DIC
AT
IVE
in that they do not describe som
e actual state of affairs, but rather invoke or conjure up som
e state of affairs which is envisaged as unfulfilled.
It is not coincidence that these non-indicative constructions, in English, are all represented by the base form
of the verb, without
inflexion: 12 , ·
I
[27] Go hom
e. (2nd person im
perative) [28] Let's all go hom
e. (1st person im
perative) [29] Everyone go hom
e. (3rd person im
perative) [30] They w
anted to go home. ·
(infinitive) [3 I] I vote that everyone o
f us go home. (present subjunctive)
SEMA
NTIC REPRESEN
TATIO
N
II7
Indicative and non-indicative propositions will be signalled by the
symbols I and N
respectively. .
Apart from
the imperative, m
ost form
s occur in subordinate clauses, w
here they are governed semantically by
particular predicates (most obviously by predicates underlying
verbs such as want, hope, decide, etc. -
see 9,2.3). This leads to the observation that prow
sitionals, whether propositions, ques-
tions, or mands, m
ay be included in other propositionals: , [32] lthink that you are m
istaken. I [ ... I [pos (X
)]] [33] D
o you know w
hether they won?
I [ ... 1 [? (X)]]
[34] They told me not to com
plain. I [ ... N
[neg (X)]]
Notice that non-indicative questions also occur, but are som
ewhat
restricted:
[35] I am not sure w
hether to apply for the job. I[ ... N [.?(X
)]] C
ompare sim
ilar examples of non-indicative w
h-questions: [36] She told m
e what to do.
[37] Why not leave straight aw
ay?
There are also, in English, archaic or formt,tlaic subjunctive m
ain clauses such as G
od be praised. These, like first-and third-person
imperatives, m
ay be described as mands; but overw
helmingly the
most com
mon type \pf m
and realized by a main clause is the
second-person w
hose pragmatically specialized func-
tion (see p 28) is recognized in the fact that its subject you is optional, and generally om
itted. W
e have already seen that the imperative cannot be associated
with any particular illocution such as an order, nor even w
ith a general illocutionary type such as im
positives. Any pragm
atic. generalization about the use of im
peratives has to be. broad enough to cover such utterances as: H
ave a good time (good
wishes); H
elp yourself (offer); Make yourself at hom
e (invita-tion); B
e whole {faith-healing); G
o to hell (curse); Say that again, and I'll hit you (threat); as w
ell as the standard impositive com
-m
and Stand still, etc. The comm
on ground . that these share is this: they all, in som
e respect or other, present the propositional content as a candidate for fulfilm
ent by h. There are important
issues which cannot be pursued here, particularly the qu.estion of
how far the relation of questions and m
ands to the illocutions they typically perform
(viz askings and impositives) is conven-
tional, rather than determined by Interpersonal R
hetoric. Kem
p-son (1975:147) opts for a conventional m
apping of one on to the other set of categories, w
hereas I prefer (in 7·3·3-I-4 below) to
II8 TH
E TA
Cf M
AX
IM
go the whole pragm
atic hog, and attempt an explanation entirely
in terms of Interpersonal R
hetoric. For the present, it is enough to add to the indicative/non-indicative distinction one further logical
contrast re!evant
to im
positives: the
distinction be-
tween
REA
L and U
NR
EA
L (or counterfactual or hypothetical)
propositions. T
he unreal mood, particularly as m
anifested in the past-tense m
odals would, could, and m
ight is, like the non-indicative mood
of the imperative, pragm
atically specialized to particular func-tions. Logically, unreal propositionals m
ay be assigned to the in-dicative category, for unreal conditional propositions can be true or false. 13 Thus the follow
ing are respectively necessarily false and necessarily true.
[38] If I were taller than m
y sister, she would be taller than
me.
[39] If I were taller than m
y sister, she would be shorter than
me.
But no truth value can be assigned to an unreal proposition if it
lacks a condition:
[40] Many people w
ould love a house like that. [41] Y
ou could open these letters. T
o interpret such sentences logically, we have to add som
e impli-
cit protasis ' ... if you wanted to', etc. It is therefore reasonable
to say, as many have assum
ed, that whenever an unreal prop-
osition occurs without a protasis, it is logically underspecified, and
needs to be supplemented by an im
plicit condition. Both non-
indicative and unreal propositionals have the following prop-
erties: they occur more generally in em
bedded positions, where
their occurrence is determined by the selectional restrictions of
predicates such as want, if, etc. 1\ and w
hen they do occur in non-em
bedded positions, they are pragmatically restricted to certain
illocutionary functions. These two properties are in fact aspects of
the same general characteristic, w
hich is that these types of prop-ositional (as the traditional gram
matical nam
e 'subjunctive' im-
plies) are not logicaliy independent. When they occur indepen-
. dently, it is therefore with the im
plication that they are in some
respect deficient, as compared w
ith real propositions. This analy-sis to som
e extent justifies a view, w
hich accords with the conven-
tional preoccupation of logic with propositions and truth values,
that propositions are the minim
al logically complete units of lan-
guage. Nevertheless, other propositionals share w
ith propositions the elem
ent of propositional content, and it is possible on this basis to com
pare them pragm
atically with one another.
TJ:iB INTER
PRETA
TION
OF IM
POSITIV
ES
6.6 The interpretation of impositives
In show
ing how
t.lte
interpretation of impositives
is graded
according to tact, I shall begin with the im
perative, as the most
direct form of im
position. An im
perative impositive is tactless in
that it risks disobedience, which is a fairly grave type of conflict
situation: [42] s: T
ake me hom
e. s intends [h to takes hom
e].
It is not clear to me w
hether this implicature is conventional or
conversational, ie is it just a matter of convention that m
ands w
ith second-pers<m subjects express s's intention to get h to do
something,, or is it that this interpretation is the m
ost 'default interpretation', given that s has uttered a m
and descnb-ing som
e future action by h? I shall not pursue this further. One
stage more tactful than [42] is [43]:
[43] s: I want you to take m
e home.
, IM
PLIC
AT
UR
E: s w
ants [h to be aware that [s w
ants [h to takes hom
e]]].
In [43] s is the Tact M
axim in uttering a proposition,
rather than a direCt mand. This is because an assertion does not
require any action as its most direct response, and so h is left w
ith a choice of w
hether,to carry outs's wishes or to ignore them
. But
if h is also observing the Tact Maxim
, then h will carry outs's
wishes, Thus, in so. far as s 'banks on' h's observing the Tact
Maxim
, [43] assumes the force of an im
positive. This m
eans that [43] violates the Tact Maxim
at one remove.
For if h is observing the Tact Maxim
, then h has no choice but to do w
hat s wants. Thus by uttering [43], s forces h EITH
ER to take
s home
OR
to break the Tact Maxim
. In either case, a violation takes piace, since by constraining h to do w
hat s wants, on pain_of
breaking the Tact Maxim
, sis himself breaking the Tact M
ax1m
by imposing his w
ill on h. So to the Tact Maxim
we add the fol-
lowing 'm
eta-maxim
': D
O N
OT PU
T h IN
A P9SIT
ION
WH
ERE EITH
ER S O
R h H
AS TO
B
RE
AK
TH
E TA
CT M
AX
IM.
It could be said that in [43] s the Tact M
axim for per-
sonal advantage. But the follow
ing req'!lest forms are som
ewhat
more polite:
[ ] { W
ill you . .
} take me hom
e?15
44 A
re you wdhng to
120 [ ] { C
an you }
? 45
Are you able to
take me hom
e.
THE TA
CI' MAXD4
The question form of [44] is felt to be m
ore tactful, because a yes-no question overtly gives h freedom
of response, ie freedom
to say yes or no. Moreover, by asking h about h's w
ishes, sis overtly putting him
self in the deferential role. Even so, by means of im
-plicature, this question can be seen
have the force of a request: IM
PLICA
TUR
E via the CP:
(a) s wants [h to take s hom
e]. (Hinting Strategy, M
axim of
Relation)
IMPLIC
ATU
RE via the pp and (a):
(b) By avoiding a direct im
perative, s observes the Tact Maxim
. (c) In that [44] is intended as an im
positive, s assumes that h is
observing the Tact Maxim
. (d) In assum
ing that h will interpret [44] as an im
positive, s assum
es that h assumes that s is observing the Tact M
axim.
(Otherw
ise, there would be no m
otive for s to adopt the H
inting Strategy.) G
oing one stage further, we notice that the question about h's
ability [45] is more tactful, as an im
positive, than [44]. This is be-cause [44],'construed via its im
plicated meaning as an im
positive, resem
bles [42] and [43] in allowing h no freedom
to refuse. If h answ
ers No, I w
on't to [44], h is valuing his own w
ishes above w
hat he understands to be those of s, which is im
polite. There-fore, one further im
plicature is made if one adopts the even m
ore oblique anticipatory illocution of (45]: (e) Iri assum
ing that [44] is an impolite im
positive, in that h under condition (d) has no choice but to take s hom
e, s assum
es that h assumes that s assum
es that h is observing the Tact M
axim.
The question about h's ability to do A avoids this impoliteness at
third remove, because it gives han 'out': his able to decline to do
A on grounds of being unable to do so. No,one can be blam
ed for a failure to do som
ething if the failure is due to inability. For exam
ple, h might say: I am
willing to do A
, but I can't, and might
justify himself by saying:
'Unless I am
able to do A, I cannot be responsible forfailing to
bring it about.' The adoption of the unreal form
s of [44] and [45] is, of course,
THE IN
TERPRETATIO
N O
F IMPO
SITIVBS
I2I
yet a further stage in the avoidance of comm
itment:
[46] Would you take m
e home?
[47] Could you take m
e home?
By replacing will and can by w
ould and could, s gives h another excuse for not com
plying with the request: the past-tense m
odals signify a hypothetical action by h,.and
in reply, h give a positive reply to the quest10n
cm.nm
:ttmg
to anything in the real world. To m
ake the tmphcatlon of an 1m-
positive even more rem
ote, tentative or negatively biased }terns can be added:
·
[48) Could you possibly take m
e home?
[49] Would you m
ind taking me hom
e? B
ut one point to notice about the hypothetical forms is that lack-
ing an overt protasis (see 5.5) they cannot (unlike questions with
can, for example), except in rather unusual contexts, be under-
stood as serious questions. They are therefore pragmatically high-
ly specialized towards the function of indicating 'on the record'
politeness. Other transparent politeness m
arkers are, of course, courtesy adverbs such as please and kindly.
If we turn to declarative sentences, here too there is a scale of
tact similar to that noted in the case of questions, w
ith can being m
ore indirect than will, and the hypothetical form
being more in-
direct than the real form:
[so] You w
ill take me hom
e. (ALso: You m
ust take me hom
e.) [51] Y
ou can take me hom
e. [52] Y
ou could/might take m
e home.
The parallel between propositions and questions cannot, how
-ever, be extended to w
ould: unlike Would you take m
e home?, You
would take m
e home does not have an im
positive force at all. Tnis lack of correspondence has the follow
ing explanation. The state-m
ent You will ... in [50] is if anything m
ore impolite than the
direct imperative:
it cannot therefore be combined w
ith the politely 'hedging' effect of the hypothetical form
would.
To m
ake the point more explicitly: propositions such as [so]-
[52) are noticeably less tactful than their matching question form
s in [44]-[47]. They state or im
plicate, the addressee's ability to act, and therefore deny him
the opportunity to decline what is offered
by the interrogative. But in addition, the You w
ill form of [so] is
interpreted most directly as a prediction (a kind of assertion
about a future happening), and therefore expresses s's confident belief that the future event A w
ill take place:
122 TH
E TACT M
AXIM
's wants h to be aw
are that h will takes hom
e.' T
he possibility of disobedience associated with a com
mand is
effectively ruled out by the use of a proposition beginning will.
As though by utterly unassailable authority, s claim
s to guarantee the future com
pliance of h. It is because [50] in this w
ay strengthens the impositive rather than tones it dow
n that You w
ill cannot be made m
ore' tentative through the change to w
ould. You will says 'I am
absolutely sure you will obey', but the
unreal would says 'I am
uncertain as to whether you w
ill obey'. In this
way,
their pragmatic forces
are incompatible
with
one another.
In contrast, the statement beginning You can is an appropriate
means of softening the effect of an im
positive. It can best be re-garded as a tentative version of You m
ust: by pointing out the ability of h to do the task, s in effect (by the H
inting Strategy) proposes that h do it.
It mitigates the force of You w
ill ... because, as w
e have seen (4.2, Table 4.1), You can ... carries the im
plicature 'You do not have to', and so offers h, on the face of
it, a pretext for jgnoring the hint. Since it is formulated as a prop-
osition rather than as a question, however, [51] does not overtly
offer h the choice of saying 'No', and is in this respect less tactful
than the corresponding question. Its politeness derives, in con-trast, from
its ambivalence: its sense allow
s it just as easily to be a recom
mendation or a piece of advice (an illocution for the benefit
of h) as to be an impositive. Since You can ... is itself a softened
variant of must, there is no reasonw
hy it should not-be hedged further by adoption of the hypothetical form
You cauld ... T
he Interpersonal Rhetoric can explain another lack of paral-
lelism betw
een propositions and questions. A negative question
with can m
ay have an impositive force, as w
e noted earlier, w
hereas there is no comparable force of a negative proposition:
[17} Can't you shut up?
[17a] You can't shut up.
As a negative proposition such as [17a] expresses the im
possi-bility of h's doing A
, it obviously cannot be used as a means of get-
ting A performed. (If
it would have the opposite effect
of dissuading h from doing A
.)But the negative question [17] can
indeed have the force of getting h to be quiet. In this respect it has a very different effect from
that of the negative question [ 13) W
on't you sit down, despite the sim
ilarity of their positive coun-terparts. A
s we have seen, [ 13] m
akes an offer more polite, w
hile [17] m
akes an impositive less polite. T
he explanation is simple
enough if we analyse a negative question, as before (seep uo),
PRAGMATIC SCA
LES
as a question about a negative proposition. The sense of [I7] can be roughly spelled out, on this basis, as follow
s (see 7.3.2):
'I have assumed that you cannot shut up. I now
doubt whether
this is true. I asky..ou to resolve my doubt.'
We m
ight reconstruct an ironic train of thought behind this ques-tion as follow
s: ·
·
A. 'Y
ou have been making too m
uch noise.' (impolite
vation) -
B. 'The only w
ay to reconcile this with politeness is to assum
e that you
help making too m
uch noise.' (polite tion)
C. 'Therefore I m
ake the polite assumption tl'oat you cannot shut
up.' D
. 'How
ever, everyone knows that people are able to be quiet
when they w
ant to.' E. 'Therefore m
y polite assumption at (C
) must be false.'
F. 'Therefore there is a conflict betw
een what I believe and w
hat it is polite for m
e to believe. I ask you to confirm m
y belief.' The result of
ironic negation is thus to bias the interpretation tow
ards a positive reply, and hence towards :om
pliance. There-fore this question is less polite than the moc·e neutral question Can you ... ? O
ppositely, Won't you sit dow
n biases the answer
in favour of the polite assumption 'Y
ou will sii. low
n' in the case of an offer, and is therefore m
ore polite than the corresponding neutral question W
ill you sit down?
5. 7 Pragmatic scales
The preceding sections have identified three scales with a bearing
on the degree of tact appropriate to a given speech situation. These are: I. The C
OST
-BE
NE
FIT .SC
ALE (p 107) on w
hich is estimate(, f1e
cost o:r benefit ofthe proposed action A to s or to h. 2. T
he O
PTION
ALITY
SC
ALE on w
hich illocutions are according to the am
ount of choice which s allow
s to h (p 109). 3· T
he IND
IREC
TNESS SC
ALE on W
hich, from s's point of view
, illocutions are ordered w
ith. respect ·to the length of the path (in term
s ofmeans-ends analysis) connecting the iU
ocutionary act to its illoc:uti()nary goal.
The i,ndirectness scale can also be formulated from
h's point of
THE T
AC
f MAXIM
view, in term
s of the length of the inferential path by which the
force is derived from the 'sense. Therefore strictly speakhig, there
are two scales of indirectness: one for the speaker, and one for
the hearer. Since, however, h's inferential strategy is a step-by-
step reconstruction of what h understands to be s's illocutionary
strategy, we can assum
e a close correspondence between them
. There is generally no need, in discussing indirectness, to dis-tinguish betw
een s's and h's points of view.
The cost-benefit scale is also, strictly speaking, made up of
two distinct scales: cost/benefit to s and cost'benefit to h. In
general, these two m
easures vary inversely; but it is possible for them
to vary independently. For example, s m
ay propose a course of A
ction which is, in s's estim
ation, at a cost to himself,
and beneficial to h. This is appropriately described as an offer (see Table 9.2, p 217); eg:
[53] Would you like to use m
y electric drill? ( j h, t s) O
n the other hand, s may
a course of action which s re-
gards'as beneficial to h, but which i-s not costly in any w
ay to s:
[54] I'd use an d:::ill if I w
ere you. ( j h)
Tht. w
ould be more appropriately described as a piece of advice.
(T e arrow
s indicate 'beneficial to' ( j) and 'at a cost to' ( t ). , e m
ight add, though, ·that there is a strong association be-tw
een these two scales, because im
positives and comm
issives typically propose an action w
hich involves a transaction between s
apd h: that is, where s gets som
ething done for h or vice versa. In such typical cases, it is unnecessary to-distinguish betw
een the speaker's and the hearer's cost-benefit scales, since a positive position on one w
ill inevitably imply a negative position on the
qther, ie whatever is t s is t h; and w
hatever is t sis t h. l This brings to m
ind a useful analogy between m
any illocutions (or, m
ore directly, the actions which are the subject of those
illocutions) and comm
ercial transactions. 16 In an impositive such.
as Would you m
ind cleaning the windows? there is implied a
tt:ansfer of 'goods', or more often of 'services', from
h to s; while
a comm
issive such as Would you like m
e to clean the windows? t&
re is implied a transfer in the opposite direction. Som
e ex-Jl13Y sim
ilarly Cafry im
plications of a transfer of goods or/ services· in th(f past: if you thank som
eone, you presuppose a transfer of goods or services from
h to s. The comm
er-cial m
etaphor does not have to be restricted, moreover, to 'bi-
lateral' such as these. C
onsider the expressive illoc-utions w
hich we call A
POL
OG
IE' and PA
RD
ON
S. Apologies express
pRAG
MA
TIC SCALES
125
regret for some offence com
mitted by s against h -
and tbere is no im
plication that s has benefited from the offence. N
everthe-less; an apology im
plies a transaction, in that it is a bid to change the
of the relation between sand h. If the apology
is successful, it will result in h's pardoning or excusing the of-/
fence. Significantly, if we oom
mit an offence against som
eone, we
talk of owing that person an apology, thereby treating the apol-ogy as in som
e sense an expiation of the offence. The metaphor
whereby deeds m
ake us 'debtors' or 'creditors' of one another ap-plies not only to good deeds (favours), but also to bad deeds (of-fences), so that apologizing, like thanking, can be regarded as an acknow
ledgement of an im
balance in the relation between s and
h, and to some extent, as an attem
pt to restore the equilibrium.
The mercantile m
etaphor, in fact, is more than a superficial
similitude. It m
ight be argued that the mercantile w
orld, on the contrary, is a special case of a social w
orld in which the standing
of one person k relative to another person l can be measured in
terms of w
hat k owes l or l ow
es k. Such an account of human re-
lationships is needed quite generally; without it w
e should not be able to explain the sense of a speech-act verb such as pardon, w
hich does not involve any exchange of goods or services, but nevertheless im
plies the cancellation of a debt. Without such an
account, moreover, there w
ould be difficulty in explaining the m
eanings of such speech-act verbs as beg, petition, beseech, all of w
hich resemble request, but have the additional im
plication that s is in som
e way acknow
ledging the debt that would result from
h's · perform
ance of the designated action: [55] Jim
lbegged me to lend him
my bicycle.
[56] Jim asked m
e to lenq him m
y bicycle. [57] Jim
demanded that I lend him
my bicycle.
On the other hand, dem
and in [57] implies that s does not ac-
knowledge that any debt w
ill result from h's action. In this respect,
beg in (55] demand in [57] represent opposites, w
hile ask in [56] is neutral betw
een them. The sam
e point could be made, but
with a change of vocabulary, if w
e said that Jim in [55] regards
the loan of the bicycle as pl3cing him under an obligation, w
hile , in [57) he regards it as a m
atter of right, 17 .
The cost-benefit scale therefore brings with it an im
plicit -bal-:-1
ance-sbeet of s's and h's relative standing, and there also seems
to be a taat assumption that a m
aintenance of equilibrium is
desirable. 18 The goal of som
e speech acts, such as thanks and apologies, .can then be seen as the restoration of equilibrium
, or at least the reduction of disequilibrium
, between sand h.
Cl). u c; !:!
Horizontal
r-------------· .::!
distance t::
FIGU
RE 5· I
Social distance
THE TA
CT M
AXIM
In addition to the scales already m
entioned -the cost-
b:nefit scale, the opttonality scale, and the indirectness scale _ t ere are tw
o further scales which are highly relevant to polit •
ness. These are the scales which, in B
rown ?nd G
ilman's
(I 960), determine
choice between fam
iliar an
respectfu, pronouns of address m m
any European hin ua es (for exam
ple, !he _choice between tu and vous in
-r:e be
vtsuahzed as
a tw
o-dimensional graph
as show
n Fig. 5 ·I· The vertical axis m
easures the degree of distance in term
s of 'pow
er' or A
UTH
OR
ITY of one participant over
an?ther. Thts IS an m
easure, so that someone in auth-
onty may use a fam
dtar form of address to som
eone who, in re-
turn, uses the respectful form. T
he horizontal axis, on the other. hand, m
easures what B
rown and G
ilman call the 'solidarity' factor
or what I shall prefer to regard, from
the opposite point of view'
as 19 T
he overall degree of respectfulness, fo; : g,ven speech SituatiO
n, depends largely on relatively permanent
actors of status, age, degree of intimacy, etc., but also, to som
e on
temporary role of one person relative to another 20
ecturer m1ght feel it reasonable to say to a student G
et that essay to m
e by next week, but not M
ake me a cup o
f coffee. In the case he w
ould be exercising his legitimate authority over ·
t .e aca.dem
ic behaviour; but in the latter case, he would
be recognized role. A
gain, rights and duties . are Im
portant m defim
ng the standing of participants Lo relation to one another. . In
society, there is no pronoun distinction to signa
t .Is type of social relationship, but, as the atXlve exam
le already
the amount of tact requited in an im
positive $ru be determ
med m
part by the degree of respect i"""licit in ·, stance tow
ards h. ·
· · · '""'!"
· s s
TAC
f f.ND
CO
ND
ESCEN
SION
!27
We can now
summ
arize the way these various param
eters influence tact as follow
s: (i) the greater the cost of A to h, (ii) the greater the horizontal social distance of h from
s, (iii) the greater the authoritative status of h w
ith respect to s, (iv) the greater w
ill be the need for optionality, and correspond-ingly for indirectness, in the expression of an im
positive, if sis to observe the Tact M
axim.
5.8 Tact and condescension T
he final clause (iv) of the above generalizatio'l is not without
exceptions, for although optionality implies indirectness, indirect-
ness does not imply optionality. There are som
e impositives
where indirectness does not contribute to tact, and even som
e ;,vhere indirectness m
ilitates against tact. One exception already
noted is a negative question such as Can't you be quiet?: the nega-
tive question is more indirect than the positive question, but is
less tactful, since it reduces the amount of choice im
plied. This exam
ple also leads on to another kind of exception, such as Must
you make all that noise?, in w
hich indirectness leads to an ironic interpretation. M
ost interesting, perhaps, is a type of example
not yet considered, where s uses a form
of utterance which looks
like a comm
issive, and which is nevertheless intended and inter-
preted as an impositive. A boss m
ight say, with apparent indul-
gence, to his new secretary:
[58] Would you like to type these letters?
Another exam
ple, from an earlier era, is the perm
issive use of m
ay in giving au order; for example, in a V
ictorian schooiboy-story, the headm
aster might dism
iss a boy by saying:
[59] You m
ay go now, Sm
ith. O
n the face of it, the utterers of [58] and'[59] are being polite in offering h a chance to do som
ething gratifying. But in fact, a
different a.'l.alysis is required, for there is little doubt that [58] w
oUld be follow
ed by the typing of the letters, and [59] by the im-
mediat¢ departure of the boy. It is often felt that a sentence like
[58] is objectionable precisely because it trades on the authori-tative status of s, Si!lce s's position is such that h cannot but ac-know
ledge his. m:rthority, h feels obliged to accept the 'offer', and
hence s is free to enjoy the pleasure of condeseension. Here
excessive of s d6es not m
ake use of the Irony Principle.
128 TH
E TACI' MAXIAf
Rather' the H
inting Strategy comes into force·
(i) s is_ apparently being polite in offe . .
. f("") A
' m a. m
anner which suggests
h lthe chotce of doing B
ut A ts not pleasant.] IS Peasant.
(tu) How
ever s's rem
k ·
. A
. '
ar can only be relevant if s w
ants h to do (tv) A
nd, since s has authority over h h. .
. ,
IS requued to do A Th1s strategy on the part of
· suppose the existence of a
reasonable only if we again
By appearing to give h a choice a
between s and h.
balance in his own favour Th.
' appears. to m
crease the credit h recognizes that s's tact..
. IS .a vantage :IS spoiled, however if
makes an
danger is that s m
cases of iron;y being too
rt condescensJon. In these cases as
, po l e can m
ean being impolite.
,
Notes
I. See Brow
n and Levinson's d" .
.., ( =
threatening acts). lscussion (I978:71 -3) of intrinsic F
fAs
-· In his 1979 revision of tb .
acts', Searle changes the arttcle 'A
taxonomy of illocution
use the terminology of the :
e
to 'assertive'. Her!1
are those of Austin
v)erstodn. Other, related taxonom
ies 55): .
u, , an B
ach and Harnish (I979:4I-
3· should be m
ade of the . .
and duties as factors e of socially assum
ed w
eightiness of s's imposition
mto the evaluation of the
4· On these im
plications of ·
e further note 17 below
· .
negative quest" _,
· questions to negative pr
. . mns, anu the relation of
20). oposttions, see Leech (I974:3I8-
5-C
f Brow
n and Levinson's , 1 S)
. . .
and 'off the record' in :stm
ctton between 'on the record'
6. An earlier account of
h ca ve acts.
[I977aJ:xo8). sue
pragmatic paradoxes is given in Leech
7· Thts account differs from that of
question as a type of illocuf Searle ( I9(i9:66), w
ho defines a separate 'real questions'
act, and then finds it necessary to 'exam
questions' (where s
;:nts to find out the answer) from
follow
Levinson (I978) in show;. s tb answ
er It is better to
they _have a comm
oD basis
at although Jogically
' pe orm a W
ide range of prag!!latic 8. O
n the concept of IY.O
nd cf L .
. .. statem
ents in that their tro ."Y?ns
'Mands differ from
rather than "it is so" , L .
..... ts to be mterpreted as "so be it.,
Hare I9JO} betw
een distinC!ion (borrow
ed ' -""'Y•c, and neusttc only parti:illt cor-
NOTES 129
responds to the threefold distinction between syntactic, sem
antic, and pragm
atic abstractions that I have presented here. 9-
'Propositional content' is Searle's (1969) term. K
empson (1975:43-
4) adopts Stenius's ( 1967) term 'sentence-radical'.
10. On this approach to questions, cfH
udson (1975) and Leech (1981). Searle (1969) applies the concept of propositional function only to w
h-questions. I 1. It is necessary
to distinguish betw
een a 'logically w
ell-formed
answer' in this sense, and a 'pragm
atically appropriate response' to a question. Exam
ples [21] and [22] from Sm
ith and Wilson on p 94
above illustrate the latter, but not the former.
12. Support for this point of view com
es from the argum
ents of Bolinger
(1977:152-!h) and Dow
nes (HfT/) in favour of the syntactic and
semantic identity of the im
perative and the infinitive. 13. U
nreal or oounterfactual conditional propositions and their truth values are discussed in M
cCaw
ley (198I:3II-26). 14. For factuality conditions im
posed by predicates, see Leech (1980 [1977bJ) and Leech (Ig8I [1974]:301-6).
15. The tW
o sentences [44) are not pragmatically equivalent, nor are the
two sentences [45]. The sentences W
ill you ... and Can you ... are m
ore institutionalized as impositives than their 'paraphrases'. In
addition to this, Sadock (1974:78) also argues that the pairs are clearly distinct on syntactic grounds. H
e thus claims that in one sen-
tence the request mocution is
in the gramm
ar' whereas in
the other sentence it is not. While acknow
ledging that such differ-ences do exist, I am
interested here in focusing on the equivalence in sense, and to som
e extent in force, between these sentences, rat.lter
than emphasizing their dissim
ilarity. See 8.7 below for further refer-
ence to Sadock's theory of indirect illocutions. 16. The com
mercial analogy has also been exploited by B
rown and
Levinson (1978) in their interpretation of comm
unicative acts as in-volving exchange of 'goods and services'. A
comm
ercial model is
also supported by the observation that some iliocutions such as
promises and bets require, for their successful perform
ance, a con-tractual relation betw
een sand h. See Fotion (1982). 17. The im
portance of rights and obligations in t!le assessment of cost or
benefit, and hence of politeness, is evidenced by an example I ow
e to Jennifer Thom
as. Suppose a passenger p asks a driver d to stop the bus at a bus-stop. V
ery little politeness is required for this speech act, because it is the driver's job (ie his occupational duty) to let pas-sengers get off at bus-stops. B
ut now suppose that p asks d to stop
the bus outside p's house, where there is no bus-stop. Jn this
a great deal of politeness, as weH as other redressive bebayiour such apologizing and explaining, m
ay be required. In both cases the am
ount of trouble or effort on the part of ti!e driver is the same; but
the imposition is far greater in the case w
here the driver is asked to do som
ething 'as a personal favour'. -
18. Presecving equilibrium also seem
s to im
portant aspect of other
130 TH
E TACT M
AX
IM
kinds of human
behaviour. See Argyle and D
ean on the preservation of equilibrium
in the distancing of one participant
another in interactive behaviour. 19·
and. Gdm
an's scales of power and solidarity have been a _
phed to pobte?ess by Brown and Levinson (1978=79-89).
p 20·
and (196o:261) cite as a 'favourite exam
ple' the shift 0 d
mountam
eers from vous to tu after a 'certain critical alti-
tu e' IS re::ched.
Chapter&
A survey of the Interpersonal Rhetoric
Lady Brackne/1: ... Is this Miss Prism
a female of repellant aspect, rem
otely connected w
ith education? Chasuble {som
ewhat indignantly): She is the m
ost cultivated of ladies. and the verypicture of respectability.
Lady Brackne/1: It is obviously the s11me person. !W
ilde, The Importance o
f Being Earnest, Act Ill]
In describing the Interpersonal Rhetoric I have so far concen-
trated on the Cooperative Principle and one m
axim of the Polite-
ness Principle, viz the Tact Maxim
. By restricting the range of
discussion in this way, I have tried to show
, through illustration, the explanatory value of a rhetoric of w
hich Grice's C
P is just one com
ponent. I have also mentioned an Irony Principle (4oi); but it
is. pow necessary to consider w
hat other principles and maxim
s m
ust be postulated in order to explain the relation between sense
and force in human conversation. In short, I shall try. to fill in
some gaps in m
y introductory diagram of the
Rhete
oric (Fig_. 1.4). This will take m
e on to more speculative ground,
but will nevertheless establish, in outline, quite a rich system
of principles and m
axims.
6.1 Maxim
s of politeness A
part from the Tact M
axim, there are a num
ber of maxim
s deal-ing w
ith polite behaviour. Before describing them
, I shall note, as a general point, that politeness concerns a relationship betw
een tw
o participants whom
we m
ay call self and other. In conver-sation, self w
ill normally be identified w
ith s, and other will typically
be identified with h; but speakers also show
politeness to third parties, w
ho may or m
ay not be present in the speech situation. 1
The label other m
ay therefore apply not only to· addressees, but to people designated by third-person pronouns. T
he importance
of showing politeness to third parties varies: a key, tactor is
whether or not the third party is present as a bystam
fe!'; another is w
hether the third party is felt to belong to s's or to h's sphere
132 A
SURV
EY O
F THE IN
TERPERSON
AL RH
ETORIC
of influence. To take a clear case: s has to be m
ore polite in re-ferring to h's spouse than in referring to s's ow
n spouse. Even in this area,
however,
there are
cross-cultural variations:
in som
e societies, a m
an discussing his wife w
ill treat her as 'self', and therefore feel free, perhaps even obliged, to denigrate her; but in other societies, he w
m treat her as 'other'.
The m
axims of the PP tend to go in pairs as follow
s:
(I) TACT M
AX
IM (in im
positives and comm
issives) (a) M
inimize
cost to other
[(b) :Maxim
ize benefit
to other]
(II) G
ENER
OSITY
:tliAX:IM (in im
positives and comm
issives) (a) M
inimize benefit to se/f[(b) M
aximize cost to selfl
(III) APPR
OB
ATIO
N M
AX
IM (in expressives and assertives)
(a) Minirrlize dispraise of other [(b) M
aximize praise of
other] (IV
) MO
DESTY
MA
XIM
(in expressives and assertives) (a) M
inimize praise of self [(b) M
aximize dispraise .of
self] (V
) AG
REEM
ENT M
AX
IM (in assertives)
(a) Minim
ize disagreement betw
een self and other · [(b) M
aXim
ize agreement betw
een self and other] (V
I) SYM
PATH
Y M
AX
IM (in assertives)
(a) Minim
ize antipathy between self and other
[(b) Maxim
ize sympathy betw
een self and other] The statem
ent of these maxim
s has been simplified for con-
venience: strictly, (I)( a), for example, should rei.d: 'M
inimize the
expression of beliefs which express or imply cost to other', and the
other maxim
s should be similarly expanded. In that they rec-
omm
end the expression of polite rather than impolite beliefs, all of
them com
e under . the PP. The first four m
al{ims go in pairs be-
cause they deal \vith bipolar scales: the cost-benefit and praise-dispraise scales. The other tw
o maxim
s deal with unipolar scales:
the scales of agreement and. sym
pathy. Although there are var-
ious connecting links between the scales, each m
axim is distinct
in that it refers to an evaluative scale which is distinct from
the scales referred to by the others. W
hereas (I) and (II) respectively concern the cost of benefit of future action to other and to self, (III) and (IV
) respectively concern the degree to wl:rlch s's re-m
arks cqnvey some good or bad evaluation of other and of self.
For ie, the A
pprobation Maxim
is exemplified in the intrin-
sic cop.rtesy of congratulations, and the Modesty M
axim in that of
apologies.. A
lthough categories of illocution (5.2) correlate only
\ 4
MA
XIM
S OF PO
LITENESS
133
roughly with different types of politeness, I have associated one
or more of these categories w
ith each maxim
. Since his fifth category of 'declarations'
does not involve
politeness (5.1}, how
ever, this category is excluded from the list. In justifying
these maxim
s I shall draw on various kinds of evidence: the evi-
dence of pragmatic asym
metries betw
een s and h, of itnplicatures, and of indirectness generally. (The Tact M
axim has already been
fully illustrated in s.J-s.8.) N
ot ali of the maxim
s and sub-maxim
s are equally important.
Of the tw
inned maxim
s (1)-(IV), (I) appears to be a m
ore pow
erful constraint on conversational behaviour than (II), and (III) than (IV
). This, if true, reflects a more general law
that politeness is focused m
ore strongly on other than on self. More-
over, within each m
axim, sub-m
axim (b) seem
s to be less important
than sub-maxim
(a), and this again illur.trates the more general
law that negative politeness (avoidance of discord) is a m
ore w
eighty consideration than positive politeness (seeking concord). O
ne further difference in importance should be noted, although it
is not reflected in the form of the m
axims: politeness tow
ards an addressee is generally m
ore important than politeness tow
ards a third party.
Once m
ore we should bear in m
ind that these maxim
s are observed 'up to a certain point', rather than as absolute rules. It is particularly im
portant to remem
ber this with the w
eaker sub-m
axims, those in square brackets, such as 'M
aximize dispraise of
self'. A person w
ho continuaiiy seeks opportunities for self-denigration quickly becom
es tedious, and more im
portantly, will
be judged insincere. In this way the C
P (Maxim
of Quality) re-
strains us from
too modest, just as in othercircum
stances it restrains us from
being too tactful.
. 6.1.1 The Generosity M
axim
MIN
IMIZE B
ENEFIT TO
SELF: MA
XIM
IZE CO
ST TO SELF
I have already mentioned the bilateral aspect of im
positive and com
missive speech acts. B
ilaterality means that in practice, there
is little need to distinguish the 'other-centred' Maxim
of Tact from
the 'self-centred' Maxim
of Generosity. T
he asymm
etry of [1] and [2] or of [3] and [4], for exam
ple, can be explained in term
s of either of these maxim
s:
{ [I] tyou can lend me your car. (tim
polite) [2] I can lend you m
y car.
{ [3] You m
ust come and have dinner w
ith us. [4] tw
e must com
e and liave dinner with you. (tim
polite)
134 A
SURV
EY O
F THE IN
TERPERSON
AL RH
ETORIC
·(Note: T
he daggert indicates that this utterance is markedly less
acceptable, in terms of absolute politeness,
the w
ith which it is paired; rem
ember that w
e are still concerned wtth
absolute rather than relative politeness.)
The offer {2] and invitation [3] are :?resum
ed to be polite for two
reasons: firstly, because they imply benefit to h, and secondly and
less crucially, because they imply cost to s. B
ut in [I] and [4], the relation betw
een s and h on both scales is reversed. On the other
hand, sometim
es the iliocuti9n is such that the Tact Maxim
alone is relevant: a piece of advice such as You can get them
for less than half the price at the m
arket is meant to be beneficial to h, but
does not imply any cost to s apart from
the verbal effort to giving the advice itself.
In yet other cases, tb.e G
enerosity Maxim
appears to apply w
ithout the Tact Maxim
: for example, a request
for a second helping is slightly more polite if h's role as potential
benefactor is suppressed: Could I have som
e more X? M
arginally still greater politeness is achieved if reference is
to s as beneficiary: Is there som
e more X
? But the hypothesis that the
Generosity M
axim is less pow
erful than the Tact Maxim
is sup-ported by the observation that an im
positive can be softened, and thereby m
ade more polite, by om
ission of reference to the cost to h. This restricts the description of the action A to s's benefit from
the transaction:
[5] Could I borrow
this electric drill? is m
arginally more polite than C
ould you lend me this electric
drill? And (say)
[6] I wouldn't m
ind a cup of coffee is m
arginally more polite than C
ould you spare me a cup o
f cof-fee? This is because the iH
ocutionary goals of [5] and [6] overtly , com
pete with the G
enerosity Maxim
, but not with the Tact
Maxim
. There is a converse tendency to suppress s's part of the transaction in com
missives:
[7] You could borrow
my bicycle, if you like
(cfi could lend you my bicycle, if you like).
[8] Would you like these pencils sharpened?
. (cfW
ould you like me to sharpen these pencils?)
The playing down of s's beneficent role here is a m
irror image of
the strategy in [5] and [6]. The idea is t!'1at it is m
ore polite, in an offer, to m
ake it appear that the offerer makes no sacrifice, so
that in turn it can become less im
polite for h to accept the offer.
MA
XIM
S OF PO
LITENESS
135
6.1.2 The Approbation M
axim
MIN
IMIZ
E
OF O
THER
; MA
XIM
IZE PRA
ISE OF O
THER
. A
n unflattering subtitle for the Approbation M
axim w
ould be 'the Flattery M
!fxim' -
but the term 'flattery' is generally reserved
for insincere approbation. In its more im
portant negative aspect, this m
axim says 'avoid saying unpleasant things about others, and
more particularly, about h'. H
ence whereas a com
pliment like
What a m
arvellous meal you cooked! is highly valued according to
the Approbation M
axim, tw
hat an awful meal you cooked! is
not. Similarly, it is acceptably polite to say (referring to the per-
formance of a m
usician):
[9] A: H
er performance w
as outstanding! B: Yes, w
asn't it! B
ut suppose that B is the performer:
[:w] A: Y
our performance w
as outstanding! B: tyes, wasn't
it! In this case, B falls foul of the M
odesty Maxim
, to which w
e shall turn in the next section.
Since dispraise of h or of a third party is impolite, it is under-
standable that, as in the
case of the Tact M
axim,
various strategies of indirectness are em
ployed in order to mitigate the
effect of criticism:
[II] A: H
er performance w
as magnificent, w
asil't it! B: Was
it? e;'
Assum
ing that both A and B listened to the performance, B's re-
ply is evasive and implicates an unfavourable opinion. By ques-
tioning A's statem
ent, B implicates that he is not sure w
hether A
's judgement is correct. The im
polite implicature derives from
the unlikelihood that B
's question is simply a request for infor-
mation, and from
the fact that if B had been able, sincerely, to agree w
ith A, B w
ould (by the PP}have done so. .
In this case, B apparently ·violates the CP (M
axim of Q
uan-tity). G
rice gives another example of an uninform
ative reply: that of a person w
ho writes a reference for a student applying for a
philosophy job: [12] 'D
ear Sir,"''Mr. X
's colT..mand of English is excellent, and
his attendance at tutorials has been :regular. Yours, etc,'
[Grice 1975:52]
In explaining the implicature of thls violation of the M
axim of
Quantity. G
rice adds that s ' ... must ... be w
ishing to impart in-
formation that he is reluctant to w
rite down. The supposition is
136 A
SURV
EY OF TH
E RH
ETORIC
tenable only on the assumption that be thinks M
r. X is no good at
philosophy.' I would add, to supplem
ent Grice's gloss, thats's re-
luctance to declare his opinion is due to the Approbation M
axim. In other cases, the reluctance to criticize m
anifests itself in in-stitutionalized form
s of understatement:
[13] You could be m
ore careful. [ 14] H
er performance w
as not sa good as it might have been.
[15] A: D
o you like these apricots? B: I've tasted better.
With reference to som
e scale of value, these sentences in effect say 'a higher position on the scale is possible'. B
ut where the
Approbation M
axim is in force, a failure to com
mit oneself to a
favourable opinion implies that one cannot (truthfully) do so. In
other words, the lack of praise im
plicates dispraise.
6.1.3 The M
odesty Maxim
M
INIM
IZE
PRA
ISE OF SE
LF: M
AX
IMIZE D
ISPRA
ISE OF SELF.
The Modesty M
axim, like the other m
axims of politeness, show
s itself in asym
metries:
[16] A: They w
ere so kind to us. B: Y
es, they were, w
eren't they.
[17] A: Y
ou were so kind to us.
B: tyes, I w
as, wasn't I.
[ 18] How
stupid of me!
[ 18a] tHow
clever of mel
[19] tHow
stupid of you! [19a] H
ow clever of·you!
[20] Please this sm
all gift as a token of our esteem.
[21] tplease accept this large gift as a token of our esteem
. A
s [16] shows, it is felicitous to agree w
ith another's comm
enda-tion except w
hen it is a comm
endation of oneself. Similarly [18]
shows how
self-dispraise is regarded as quite benign, even when it
is exaggerated for comic effect. In [20), the understatem
ent of-one's generosity is show
n to be quite normal, and indeed, con-
ventional, in contrast to.cthe exaggeration of one's generosity. As
-[17] and [21] illustrate, Ul break the first submaxim
of Modesty is
to comm
it the social transg(ession of boasting. -
The following con·,,,ersation betw
een two Japanese w
omen
(quoted from M
iller 1967: 289-90) exemplifies how
a pragmatic
'paradox' can arise through the conflict of the Modesty and
Approbation M
axims, just as it can (see 5.4) through the
of the Tact and Generosity M
axims w
hen an offer is declined. In this case, the pragm
atic paradox takes the form of·
repeatedly denying the truth of a complim
ent: A
A m
a, go-rippa na o-niwa ae -
My, w
hat a splendid garden
MA
XW
S OF PO
LITENESS
gozamasu w
a ne. shibafu ga hirobiro to shite ite, kekko de gozam
asu wane.
B iie, nan desu ka, chitto m
o teire
ga yukitodokim
asen m
ono de
gozaimasu
kara, m
o, nakanaka itsumo kirei ni
shite oku wake ni w
a mairi-
masen no de gozam
asu yo. A a, sai de gozaim
asbo ne. kore dake
o-hiroin de
gozamasu
kara, hitotori o-teire asobasu no ni datte taihen de gozai-m
asho ne. demo m
a, sore de m
o, itsumo yoku o-teire ga
yukitodoite irasshaimasu w
a. itsum
o honto
ni o-kirei
de kekko de gozam
asu wa.
B iie, chitto
mo
sonna koto
gazamasen w
a.
137
you have here -the law
n is so nice
and big,
it's certainly
wonderful, isn't it!
B Oh no, not at all, w
e don't take care of it at all any m
ore, so it sim
ply doesn't alw
ays look as nice as w
e would like
it to.
A Oh no, I don't think so at an
-but since it's such a big gar-
den, of course, it m
ust be quite a trem
endous task to take care of it all by yourself; but even so, you certainly do m
anage to make it look nice
all the time: it certainly is nice
and pretty any time one sees
it. B N
o. I'm afraid not, not at all
It appears that in Japanese society, and more particularly am
ong Japanese w
omen (see M
iller, ibid: 290), the Modesty M
axim is
more pow
erful than it is as a rule in English-speaking societies, w
here it would be custom
arily more polite to accept a com
pli-m
ent 'graciously' (eg by thanking the speaker for it) rather than. to go on denying it. H
ere English-speakers would be inclined to
find some com
promise betw
een violating the Modesty M
axim and
violating the Agreem
ent Maxim
. .
There is an obvious trade-off between different m
axims of the
PP, just as there is between the m
axims of the C
P. The Modesty
Maxim
sometim
es comes into conflict w
ith some other m
axim, in
which case w
e have to allow one m
axim to take priority over the
other. In [17], for example, B adheres to the A
greement M
axim
at the expense of the Modesty M
axim, but in this situation the
Modesty M
axim plainly carries greater w
eight. In the Japanese conversation above, A partially agrees w
ith B over the work en-
tailed by the garden, but then reiterates her complim
ent. In ing food to a guest, a Japanese m
ay say 'Please [have] one'), thereby apparently m
mJJlllZm
g generosity': B
ut thiS may be seen as a result of attaching greater i;.1portance
A SU
RVEY
OF TH
E INTERPERSO
NA
L RHETO
RIC
to mo(ksty: to offer m
ore than one is to suggest that one's food is w
orth eating. In contrast, an English-speaking host might w
ell be considered niggardly if he passed round the peanut-bow
l with the
words: H
ave a peanut! It is normally considered to be m
ore polite to offer a large quantity: H
ave as many as you like. The greater
value attached to the Modesty m
axim in Japanese culture is indi-
cated further by the greater degree of understatement em
ployed in giving presents. W
hereas an English person may (as in [2o]}
call his gift 'small', the Japanese m
ay go further, and say 'This is a gift w
hich will be of no use to you, but .. .' A host m
ay even go to the extrem
e of denying the existence of the food he is offering:
Nani m
o (meshiagaru m
ono wa) ari-m
asen ga, dozo 'T
here is nothing (to eat), but please .. .' In this w
ay, a maxim
of politeness may overrule the M
axim of
Quality.
6.1.4 Other m
axims of politeness
Although there is less evidence for other m
axims, it is w
orth noting, for exam
ple, that there is a tendency to exaggerate
agreement w
ith other people, and to mitigate disagreem
ent by expressing regret, partial agreem
ent, etc. We m
ust therefore talk in term
s of a Maxim
of C
ompare the rudeness
of the reply in [22] with the replies in [23)-[25]:
[22] A: It w
as an interesting exhibition, wasn't it?
B: tNo, it w
as very UIJinteresting.
[23] A: A referendum
will satisfy everybody.
B: Yes, definitely.
" .. [24] A
: English is a diffic11lt1anguage to learn. ..
B: T
rue, but the grafunjar is quite easy. [25] A
: The book is trem
endously well w
ritten. B: Y
es, well w
ritten as a whole, but there are som
e rather boring patches, don't you think?
As [24J.and [25] show
,-partHd diSagreem
ent'is often preferable to com
plete disagreement. W
e mayalsoadd a M
axim of Sym
pathy, w
hich explains why congratuhitibris and conaolences are cour-
teous speech acts, even thOugh condolences express beliefs w
hich are negative w
ith regard to the hearer:
{ 2{\} I'm terribly sorry to hear thatyo;ur cat died,·.
This is pofite;-:in contrast, for example, W
ith trm terribly pleased
to hear thllt your: (:(zt died. There is nevertheless some reticence
about expression of condolences, since to refer to the propo:-
META
UN
GU
ISTIC A
SPECTS O
F POLITEN
ESS 139
sitional context X is in fact to express an impolite belief (see 7.4)
in the sense of a belief unfavourable to h. Hence it m
ight be prefer-able to say, instead of [26]:
· [27] I'm
terribly sorry to hear about your cat. Such is the pow
er of the Sympathy M
axim that, w
ithout further inform
ation, we interpret [27) as a condolence, ie as an ex-
pression of sympathy for m
isfortune, and [ 28] as a congratulation: [28] I'm
delighted to hear about your cat.
That is, w
e assume that the event alluded to in [27] is unfortunate
(such as a death), and that in [28] is fortunate (such as the win-
ning of a prize in the cat-show). T
hus, the following exchange
would be, to say the least, atypical of hum
an converSation:
[29] A: I'm
delighted to hear cat.
· B: W
hat do you mean? H
:e's just died. A
; Precisely.
6.2 Metaiinguistic •spom
of politeness Politeness is m
anifested not only in the content of conversation, but also in the w
ay conversation is managed and structured by its
participants. For exam
ple, conversational behaviour such
as speaking at the w
rong time (interrupting) or being silent at the
wrong tim
e has impolite im
plications. Consequently w
e some-
times find it necessary to refer to the speech acts in w
hich we or
our interlocutors are engaged, in order to request a reply, to seek perm
ission for speaking, to apologize for speaking, etc.: [3o] C
ould you tell me w
hat time the bus leaves, please?
[31] May I ask if you're m
arried? .
[32] I must w
arn you not to discuss this in public. [33] W
e regret to inform you that the aspidistra stands are no
. longer obtainable.
Such utterances are 'metalinguistic' in that they refer, in the
mode of oratio obliqua (8.3, 8.4), to illocutions of the current
conversation. Sentences such as [31]-[33] ha-ve, in fact, been know
uas hedged performatives, 3 since they m
ay be regarded as polite m
itigaticns of utterances such as I warn you that X, We in-
form you that X
. The avoidance of a direct-speech utterance can
be one more exam
ple of a strategy of polite obliquity: [31 ], for exam
ple, is a more tactful variant of A
re you married? Like other
indirect impositives, [3o] and [31] em
ploy the Hinting Strategy:
140 A
SURV
EY O
F THE IN
TERPERSON
AL RH
ETOIU
C
obtaining the conversational cooperation of h, w
hich is the
ostensible function of [3o], is a preliminary goal leading to the
ultimate goal of obtaining inform
ation. ·
Why does a speaker som
etimes find it advisable to use such
metalinguistic strategies? Part of the reason is that speech acts are
like other kinds of action in involving .some cost of benefit to s or
h. Minim
ally, for example, answ
ering a question involves some
cooperative effort 9.!1 the .. part of the person addressed; and in addition, som
e questions, such as How
old are you? or Are you
married? m
ay be felt · to Be---a. serious imposition in that they
threaten the privacy of h. 4 In this lightJ3 I Lis felt to be to some
extent a genuine request for permission:
to allow
ed to intrude on the private territory of h, and even· thouglrh w
otl,ld find it difficult to reply w
ithout giving the information desired,
the May I ask you ... ? is m
ore than a vacuous formality. It is
understandable, then, that 'hedged performatives' are used as
devices of politeness, especially when h is a person of a m
ore authoritative status than s. W
e cannot automatically assum
e the right to engage som
eone in conversation, let alone to use that conversation as a m
eans to our ow
n ends. Even some ostensibly
'polite' illocutions, such as giving advice, •y
be judged to be im-
positions, requiring a preface such as. Could I suggest ... ? , or
Might I just give you a w
ord of advice?,Tbe
for regarding advice as hnpolite, of course, is that
the recomm
ended action A
may be considered beneftijl to the
the actual speech act of advising m
ay offend both tl?e Modesty and
Approbation M
axims, because it tai:es for granted that s is super-
ior in knowledge, or experience, or judgem
ent, etc. to h. Un-
ambivalently
polite illocutions, on the other hand, m
ay be
introduced by a 'hedged performative' w
hich stresses the desir-ability or even inevitability of the illocution:
I want to thank you ...
We are delighted to announce ...
I must tell you how
much I adm
ire your ... B
ringers 1 of bad tidings may f'.nd it advisable to express both the
distasteful and the unavoidable nature of their task: I'm
sorry to have to tell you ... W
e regret to have to inform you that ...
I must w
arn you that ...
To engage a person in con-;er8ation, particularly if that person is a stranger or a superior (ie an addressee w
ho is 'distant' in terms
of the horizontal and vertical scales discussed in 5.7), may itself
META
LING
UISTIC A
SPECTS OF PO
LITENESS
141 l
. be regarded as an act of presum
ption, for conversation implies
cooperation on the part of h as well as of s. This probably ex-
plains in part why, in the historical developm
ent of some lan-
guages (such as Italian, Portuguese, and Germ
an), polite forms
of address have developed out of third-person pronouns and verb form
s, as if s can only politely enter into conversation with h
through the evasive tactic of pretending that h is a listener, but not an addressee. It m
ay also explain in part the tact in English of indirect askings such as I w
onder if you would lend m
e your coat. The im
plicature here seems to be that s does not feel entitled to-
ask h a question, and therefore expresses interest in knowing the
answer to the question in a m
anner which suggests that it is no
part of h's responsibility to provide it. The polite and hppolite im
plications of silence must also be
considered. 5 The atlage on which the children of a previous age
were reared, 'D
on't speak unless you are spoken to', reminds us
that silence may be the only polite form
of behaviour available to som
eone of little status. But if one has been engaged in conver-
sation by someone else, silence is a sign of opting out of a social
engagement to observe the interpersonal rhetorical principles, and
is hence in many circum
stances a form of hnpoliteness. These
contradictory implications of silence som
ethnes lead to a prag-m
atic impasse w
hen· two or m
ore people· engaged in 'conversation are joined by an outsider. The new
comer m
ay feel it rude to inter-tbe conversation, but the
may feel it rude not to
gtve the newcom
er a chance to JOID
The result m
ay be an un-com
fortable hiatus in the conversation. '
The problem of how
to end a conversation politely is familiar
to every competent language user, and m
akes us aware of the
close connection between politeness and-the activity of talking
merely to preserve sociability, the type of behaviour w
hich Mali-
nowski (1930) nam
ed PH
ATIC
C
OM
MU
NIO
N. W
e may, indeed,
argue for an additional maxim
of the m
etalinguistic 'Phatic M
axim' w
hich may be provisionally form
ulated either in its negative form
'Avoid silence' or in its positive form
'Keep
talking'. 6 It is the need to avoid silence, with its hnplication of
opting out of comm
unication, which accounts, at a rather trivial
level, for the discussion. of stock subjeCts such as the weather, and
less trivially, for the occurrence of uninformative statem
ents such as You've had your hair cut! A
s such remarks patently violate the
Maxim
. of Quantity, here is yet another case w
here an apparent breach of the C
P is to be explained in terms of another m
axim -
in this instance the Phatic Maxim
. How
ever, it is inadequate to ·describe phatic com
munion as shnple avoidance of silence. M
ore positively, such conversation, if it has no other illocutionary
point, serves to eJttend the comm
on ground of agreement and ex-
perience shared by the participants. Hence the choice of subject
matter tends to· be non-controversial, and to ccncentrl;lte on the
attitudes of the speakers, rather than on matters of fact. In this
context, even You've had your hair cut! makes its contribution to
the progress of the conversation, . by making h aw
are that s has noticed som
ething of which s is already aw
are, and .by giving h an opportunity to elaborate on personal experience. in a new
direc-ticm
. On the assum
ption that such exploration of comm
on ground of experience and attitude is alw
ays possible, we might therefore
treat avoidance of silence as a special case of. the Agreem
ent and Sym
pathy Maxim
s (6.1.4). But w
hether or not the Phatic Maxim
is to be subsum
ed under these other maxim
s, it seems reasonable
to argue that in this case, as in others, the apparent uninforma-
tiveness of language is to be attributed to other conversational principles, and is not to be regarded as evidence against the valid-ity of the C
P.
6.3 Irony and banter The previous discussion of irony (4. I) has
thai the Irony Principle (IP) takes its place alongside the C
P and ilie PP in the Interpersonal R
hetoric. 7 This principle, however, is parasitic
on the other two, in the follow
ing sense. The CP and the PP can
be seen to be functional by direct reference to their role in pro-m
oting effective interpersonal comm
unication; but the IP's func-tion can only be explained in term
s of other principles. The IP is a 'second-O
£der principle' which enables a speaker to be im
polite w
hile to be polite; it does so by superficially breaking the
CP, but ultim
ately upholding it. Apparently, then, the IP is dys-
functional, if the PP promotes a bias tow
ards comity rather than
conflict in social relations, the IP, by enabling us to bypass polite-. ness, prom
otes the 'antisocial' use of language. We are ironic at
someone's expense,
scoring off others by politeness that is obviously insincere, as a substitute for im
politeness. T
he insincerity may be m
ore or less obvious; it may take the
form of a breach of the M
axim of Q
uantity (as on pp 8o, "82), or m
ore often of a breach of the Maxim
of Quality:
[34] That's all I wanted!
[35] With friends like him
, who needs enem
ies? [36] Bill w
anted that news like he w
anted a hole in the head. U
sed ironically, [34] is taken to mean 'That's exactly w
hat I did not w
ant'. Here the falseness of the statem
ent will probably be
IRON
Y A
ND
BAN
TER 143
made clear by. a contradictory tone of utterance, m
ore appropri-ate t? the
?f '!hat's last. straw
'. In [35] and [36], the M
axim of Q
uality 1s mfnnged by Im
plicature rather than by direct B
ut _the insincerity of s's purported opinion is clear · from
Its absurdity. Thus [35] purports to express a belief that e!lem
ies are a thing, and [36] presents a siinilatly favourable
vtew of holes m
the head. An indirect violation of the M
axim of
Quality is also present in im
peratives such as Don't m
ind ME (w
ill addressed to a pt;rson w
ho has just, say, rudely barged m
to the speaker; or DO
help yourself (won't you?), said
to someone w
ho is only too obviously helping himself already. A
com
mand, to be felicitous in a goal-oriented m
odel, requires that the addressee has not ·yet com
plied with it. This condition is
violat7d in the above examples. In a rather sim
ilar way, the
question Do you have to spill ash on the carpet? im
plicates the speaker's belief that h m
ay be incapable of avoiding the fault in question. In each of these cases, therefore, s appears to m
ake an assum
ption which is patently untrue, and by that m
eans 1m
pbcates that the opposite assumption, w
hich is impolite, is
true. T
he ironic force of a remark 1s often signalled by exaggeration
or understatement, w
hich make it difficult for h to interpret the
remark at its face value. The speaker of [34], for exam
ple, pre-tends to take up an extrem
e position in saying that That's all (ie the only thing) I w
anted. The ironic force would not have been
registered if s had simply said the opposite of the truth, as in
That's what I w
anted. This instance of exaggeration may be coun-
terposed to an understatement such as Som
e of his w
ords were not Sunday school w
ords (Mark Tw
ain), which is also ironic, but
for the opposite reason. In this case it is the Maxim
of Quantity,
the_ Maxim
?f Quality that is m
ost directly violated. By nega-
tive umnform
atlveness (see 4·5-I) s politely implicates that 'one
all of his words. to be Sunday School w
ords'. But clearly
th1s ts a false expectatiOn. Therefore w
e infer as before
the .
. '
' oppostte state of affarrs to be the true one: that the m
an was
given to using bad language. Unlike [34], this is not an untrue
proposition, but merely a highly uninform
ative one. Ironic under-statem
ent typically, py negation, makes a claim
which is m
anifest-ly far w
eaker than a claim that could be m
ade. Irony varies in force from
the comic irony of M
ark Twain to
the more offensive
of sarcastic comm
ands such as Do help
yourself. It appears to be dysfunctional, in providing a
bei_ng to
the IP may w
ell have a posi-tive function m
perm1ttm
g aggresston to manifest itself in a less
A SU
RVEY
OF m
E IN
TERPERSON
AL RH
ETORIC
dangerous verbal form than by direct criticism
, insults, threats, etc. W
hereas an insult can easily lead to a counter-insult, and hence to conflict, an ironic rem
ark is less er<>y to answer in kind.
It combines the· art of attack w
ith an apparent innocence which is
a form of self-defence. The function of irony m
ay thus be tenta-tively explained as follow
s. If the PP breaks down, it is liable to
break down on both sides: direct accusation leads to counter-
accusation, threat to counter-threat, and so on. But because irony
pays lip-service to the PP, it is less .easy to break the PP in one's response to it. H
ence the IP keeps aggression away from
the brink of conflict.
If we acknow
ledge the existence of an Irony Principle, we
should also acknowledge another 'higher-order principle' w
hich has the opposite effect. W
hile irony is an apparently friendly way
of being offensive (mock-politeness), the type of verbal be-
haviour known as 'banter' is an offensive w
ay of being (m
ock-impoliteness).
· ,'-.... T
he Banter Principle, as w
e may call it, is ·clearly of m
inor im
portance compared w
ith other rhetorical principles we have
examined. B
ut it is manifested in a great deal of casual linguistic
conversation, particularly among young people. 8 For exam
ple, in a gam
e of chess, one person may say jokingly to another: W
hat a m
ean cowardly trick! referring to a particular clever gam
bit. Or
two friends m
ay greet one another with rem
arks such as Here
comes trouble! or Look w
hat the eat's brought in! This principle m
ight be expressed as follows:
'Ill order to show solidarity w
ith h, say something w
hich is (i) obviously untrue, and (ii) obviously im
polite to h.' Like irony, banter m
ust be clearly recognizable as unserious. Since overpoliteness, as w
e have seen (p 82), can have the effect of signifying superiority or ironic distance, underpoliteness can have the opposite effect of establishing or m
aintaining a bond of The reason is this. A
low value on the scales of auth-
ority and social distance (5.7) correlates With a low
position on the scale of politeness; that is, the m
ore intimate the relationship,
the less important it is to be polite. H
ence lack of politeness in it-self can becom
e a sign of intimacy; and hence, the ability to be
impolite to som
eone in jest helps to establish and maintain such a
familiar relationship. The im
plicature derived from the B
anter Principle is just the opposite of that derived from
the IP (see p 83):
'What s says is im
polite to h and is clearly untrue. Therefore w
hat s really means is polite to h and true.'
HY
PERBOLE A
ND
LITOTES
145 W
e might go so far as to call the B
anter Principle a 'thirf:l-order principle', because it m
ay itself exploit B
anter could be de-scribed as m
ock-irony in cases like A fine friend YOU
are!, said jokingly (say) to a partner w
ho has given away an advantage in a
card game. The interpretation of this utterance requires a double
reversal of values: (i) Y
ou are a fine friend. (face-value) (ii) By w
hich I mean that you are
NO
T a fine friend. (Irony Principle)
, (iii) B
ut you A
RE
my friend, and to show
it, I am being
impolite to you. (B
anter Principle) The 'higher-order' principles, in that they rely upon the im
pli-catures of
principles, involve greater indirectness>in the w
orking-out of the force of the utterance, and are therefore less pow
erful in their effect. For this reason, we can place the PP,
the IP, and the Banter Principle in a hierarchy of im
portance, corresponding
to the
order in
which
they have
just been
mentioned.
6.4 Hyperbole and litotes
Two w
ays of violating the CP w
hich deserve separate consider-ation are H
YPER
BO
LE (overstatement) and LITO
TES (understate-m
ent). Nam
ing these devices by their classical names brings to
mind their role in traditional rhetoric, and pertinently recalls the
continuity between 'rhetoric' as it is understood here and 'rhet-
oric' as it has been variously understood sincifthe time ofA
ristotle. To understand these pragm
atic strategies, we first have to
appreciate that truthfulness is not always a m
atter of making a
straightforward choice betw
een truth and falsehood. Just as truth conditions are often to be represented in term
s of values on -a scale, so telling the truth m
ay itself be judged as a matter of de-
gree, according to how accurate s is in representing such scalar
values. 'Hyperbole' refers to a case w
here the speaker's descrip-tion is stronger than is w
arranted by the state of affairs described, and 'litotes' refers to the converse of this. A
hyperbole such as It m
ade my blood boil constitutes a violation, in som
e degree, of the M
axim of Q
uality, and a litotes such as I wasn't born yester-
day constitutes in some degree a violation of the M
axim of ·
Quantity. B
ut as with irony, the.violation·of the C
P is only a su-perficial m
atter: we w
ould not apply these rhetorical terms to
utterances in which overstatem
ent or understatement w
as actual-, ly used to deceive the addressee. M
oreover, as with
the-best safeguard against deceit is to m
ake sure the utterance is so
A SU
RVEY
OF TH
E INTER
PERSO
NA
L RHETO
RIC
much at variance w
ith context that no one could reasonably be-lieve it to be 'the w
hole truth, and nothing but the truth'. Hence
hyperbole and litotes are further of the by
mar pattern of conversational im
plicature: we reach the m
dtrect force of s's rem
ru.k by means of an obvious face-value violation of
the CP. 9
We have already m
et cases where the justification for hyper-
bole and litotes is politeness. There will naturally be a preference
for overstating polite beliefs, and for understating impolite ones:
while an exaggeration such as
That was a delicious m
eal! is favoured in praising others, an uninform
ative denial-a typical de-vice of understatem
ent -is frequently used in criticism
: I wasn't
overimpressed by her speech. The understating of praise will
mally be directed tow
ards s rather than towards h:
[37] That wasn't such a bad m
eal that I cooked. [38] That w
asn't such a bad meal that you cooked.
The grudging complim
ent of [37 J is relatively acceptable as a form
of self-CC>'lgratulation; but [38] is glaringly impolite as a
complim
ent to a hostess on her cuisine, more especially because
the negative statement im
plicates that it was to be expected that
the meal w
ould be bad (see 4-S.I). B
ut not all cases of hyperbole and litotes can be explained by reference to their role in enhancing politeness. The frequency of overstatem
ent in ordinary conversation has its testimony in m
any idiom
atic expressions, as in Her eyes nearly popped out o
f her head; It m
akes my blood boil; H
e was all ears; That'll cost the earth; I've been w
orking my fingers to the bone, etc.; also in the
exaggerated use of universal q'uantifiers and references 111 the ex-trem
ities of scales: eg: I'm com
pletely broke; There's absolutely nothing on the telly this evening. A
lmost all these exam
ples make
reference to an absurdly extreme position on a scale; eg: H
er eyes nearly popped out of her head refers to the highest conceivable point on. a scale of surprise, and It m
ade my blood boil refers to
the highest conceivable point on a scale of anger. A
conversational principle which seem
s to underlie such cases is the principle w
hich enjoins us to 'Say w
hat is unpredictable, and hence interesting.' A
t the risk of proliferating too many pragm
atic principles, I shall tentatively propose, then, an Interest Principle, by w
hich con-versation w
hich is interesting, in the sense of having unpredict-ability or new
s value, is preferred to conversation which is boring
and predictable. One com
mon w
ay in which this principle m
an-
HY
PERBOLE A
ND
LITOTES
147
ifests itself in our everyday linguistic experience is the temptation
we feel, w
hen retelling a personal anecdote, to embroider on the
anecdote various kinds of elaboration and exaggeration. Another
sign of this principle is the way in w
hich hyperbolic expressions becom
e weakened through a process of dim
inishing returns (in this respect they resem
ble euphemism
s). If overstatements are
used frequently, an addressee inevitably adjusts his interpretation so that they lose their interest value and becom
e predictable. There is thus a perpetual tug-of-w
ar, in human conversation, be-
tween the M
axim of Q
uality and the Interest Principle. 10
If hyperbole is, in this sense, a natural tendency of human
speech, it is difficult to understand why the opposite tendency of
litotes is so often observed. Part of the explanation is that there is a dialectic betw
een hyperbole and litotes somew
hat parallel to that betw
een politeness and irony. Just as irony is a 'second-order principle' w
hich sacrifices politeness for the sake of the CP, so
litotes is a 'second-order principle' which sacrifices the interest-
ingness of overstatement for the S?ke of the honesty of. under-
statement. If hyperbole suffers from
diminishing returns because
of incredulity, it is a salutary tactic to move in the opposite direc-
tion, and to restore credibility by using descriptions which so
obviously fall short of what could be truthfully asserted that they
cannot be supposed exaggerated. Litotes therefore regains the credit w
hich goes with strict observance of the M
axim of Q
uality, and w
hich is sacrificed by hyperbole. To elucidate the m
otivation for litotes further, I shall call upon yet another principle: one that has been acknow
ledjed by psychologists under the title of the 'Pollyanna H
ypothesis' .. 1 This states that people w
ill prefer to look on the bright side rather than on the gloom
y side of life, thus resembling the optim
istic heroine of Eleanor H
. Porter's novel Pollyanna (1913). Interpret-ing it in a com
municative fram
ework as a 'Pollyanna Principle'
means postulating that participants in a conversation w
ill prefer pleasant topics of conversation to unpleasant oM
's. The negative aspect of this principle is, of course,
EUPH
EMISM
: one can dis-guise unpleasant subjects by referring to them
by means of
apparently inoffensive expressions (eg workers are 'm
ade redun-dant' instead of being 'dism
issed'). But another aspect is the
tendency to understate the degree to which things are bad. Thus
the 'minim
izing' adverbials. of degree a bit, a little, and a little bit are specialized tow
ards negatively evaluated terms:
[39] The paint was a bit dirty.
[40] tThe paint w
as a bit dean.
A SU
RVEY
OF TH
E INTER
PERSO
NA
L RHETO
RIC
i\,.nother restriction, which seem
s to have become
ized in gramm
ar, is that a bit and tJ little can occur with the
tively eval'aative adverb too, but not the positiv.eiy evaluative-
i!dverb enough: [,p:] She is ill little too young for the job. [42] *She is a little young enough for the job:
Another adverb w
hich often implicates a negative evaluation is
rather, which again tends to add a do\vntoning effect to the term
it m
odifies:
[43] The employees w
ere rather unenthusiastic about the move.
[44] The employees w
ere rather enthusiastic about the move.
Although both [43] and [44] are perfectly gram
matical, [44] is
pragmatically abnorm
al in comparison w
ith {43]. A third tech-
nique for understating pessimistic evaluation .is the one w
e have already observed -
the use of negation: [45] The m
eeting was not partic-ularly w
ell attended. [46]
tThe meeting w
as not particularly badly attended. A
s a negative proposition is assumed to '(ieny a positive expec-
tation, [45] takes as its norm the corresponding positive proposition
'Tn.e meeting w
as particularly well attended'. H
ence even if the m
eeting had been quite well attended, it w
ould be truthful to assert [45]. In. this w
ay, the understatement disguises a bad report
in a form w
hich on the face of it pe.mits a good interpretation.
T."le unfavourable interpretation is urived at indirectly, by im-
plicature, and is thus weakened. The opposite proposition-[46] is
pragmatically less favoured, because it runs counter to the Pol-
lyanna Prin,dple. It would occur only in an unusual context,
where for som
e reasoq bad attendance was expected.
We can see, then, that litotes is a w
ay of underplaying aspects of m
eaning which are pragm
atically disfavoured. In [45], it is the ' pessim
istic judgement that the m
eeting was badly attended that is
. thus-mitigated. In other exam
ples, it is impoliteness that it is m
iti-gated, as in the follow
ing examples of m
odest self-praise: [47] A
ctually, I'm rather good at crossw
ord puzzles. [48}
rather_proud of our classless profession. .
Hyperbole and, litotes are not single pragm
atic principles, but "father general tendencies w
hich occur whenever som
e pragmatic
principle brings about a distortion of the truth. The influence of the PP, for exam
ple, causes both polite overstatement and polite
understatement. T
he influence. of the Pollyanna Principle causes
CO
NC
LUSIO
N
!49 both
and euphemistic understatem
ent B
ut m
a type of litotes (illustrated by [44}) w
hich ferns to
as an antidote to the opposite tendency o exaggerate m
keepmg w
ith the Principle. of Interest: [49] She's not a bad-looking girL [so] There are som
e rather splendid murals on the N
orth WaH
.
which untypically acts in m
itigat!on of ·
? '. appears to be a guarantee of the honest\' t of the speh aker s opm
ton, reassertipg, against the Principle of In: erest, t e value of the C
P.
6.5 Conclusion
has moved from
relatively fum ground -
the o_
-to a
uncertain area where I have s
cuiated on the .role m
. conversational rhetoric of such traditionfrhetori-devlces ;s
trony, h:yperboie, and litotes. I have suggested how
ese can. e
mtegrated into the G
ricean paradigm of
conversational pnnciples and implicatu:res
thereb h i .
. for indirect relationships betw
een and
wluch supplem
ent the maxim
s of the CP and the pp I
. y present a sum
mary of the principles and m
axims of the
TAB
LE 6.1 Interpers1lnal R
hetoric
First-crder principles
Cooperative
Principle
Politeness Principle
Interest Principle Pollyanna Principle
Higher-order
p ri.ncip les
Irony B
anter
Quantity
Quality
Relation
Manner
{ Tact G
enerosity l A
pprobation tM
odesty ·
Agreem
ent Sym
pathy Phatic?
A SU
RVEY
OF TH
E INTER
PERSO
NA
L RHETO
RIC
sonal Rhetoric as it has been enlarged by the additions of this
chapter (Table 6.1). There is clearly a great deal to be done in elaborating this plan, and in solving som
e of the problems it
raises. One of the problem
s is this: describing irony, hyperbole, and sim
ilar effects in terms of pragm
atic principles emphasizes
the social perspective on language at the expense of the psycho-logical. Thus, for instance, the choice betw
een hyperbole and totes can in part be described in term
s of a goal-oriented model,
but must also take account of differences in the personality, the
attitude, and so on of the language user. This is true of all com-
ponents of the Interpersonal Rhetoric, but is probably less true of
the CP and PP than of the other principles.
Another aspect of the subject w
hich this chapter has neglected is the typological study of cultures and languages in relation to the Interpersonal R
hetoric. So far, our knowledge of intercultural
differences in this sphere is somew
hat anecdotal: there is the observation for exam
ple, that some eastern cultures (eg China
and Japan) tend to value the Modesty M
axim m
uch more highly
than western countries; that English-speaking culture (particu-
larly British?) gives prominence to the M
axim of Tact and the Irony
Principle; that Mediterranean cultures place a higher value of the
Generosity M
axim and a low
er value of the Modesty M
axim.
These observations assume, of course, that such principles, being
the general functional 'imperatives' of hum
an comm
unication, are m
ore or less universal, but that their relative weights w
ill vary from
one cultural, social, or linguistic milieu to another.
Although these m
atters remain unclear in detail, the Interperson-
al Rhetoric provides a fram
ework in w
hich they may be system
-atically investigated.
·
Notes
1 .. The existence of both second-person and third-person politeness is w
ell demonstrated in languages w
here there exist special polite and honorific m
orphemes in relation to both addressee and referent. See
Kuno (1973:18-22) on the honorific system
of Japanese. Com
rie (1976), in discussing languages w
ith rich honorific systems, disting-
uishes between three· separate axes of politeness: the speaker.:..
addressee, speaker-referent, and speaker-bystander axes. 2. I am
indebted to Hideshi Sato for this and the follow
ing Japanese iilu§trations.
3· The pragmatic and sem
antic implications of hedged perform
atives are explored in Fraser (1975).
4· Relevant here is G
offman's (1963, 1967, 1971) w
ork on face and territories o
f self.
NO
TES
5· See Sacks, Schegloff, and Jefferson (1974) on turn-taking con-
ventions. 6. O
n the comm
unicative value of silence, see Verschueren (forth-
coming), C
hapter 3· 7· See note 2, C
hapter 4, p 102. 8. A
ritualized form of banter is the activity of 'sounding' (a cerem
onial exchange of insults) practised in the black com
munity of N
ew Y
ork, as studied by Labov (1972). This language-gam
e depends for its effect on the understanding that the allegations m
ade by each party are recognized as untrue, and therefore on the fact that they cannot be m
istaken for real 9· G
rice (1975:53) cites hyperbole and litotes (or meiosis) as floutings
of the Maxim
of Quality.
10. De B
eaugrande and Dressler (1981: 144, 16o, 213) exam
ine 'interes-tingness' as a desideratum
of a text. On the one hand, they associ-
ate it with unexpectedness and inform
ativeness ( cf the Maxim
of Q
uantity), and on the other hand they see it as in conflict with
'knownness' {cfthe M
axim of Q
uality). II. The Pollyanna H
ypothesis (Boucher and O
sgood 1969} has been used to explain w
hy words of pleasant associations predom
inate over those w
ith unpleasant ones, and also why speakers tend to conceal
the badness of things through negative expressions (see Clark and
Clark 1977=538-9). R
ather than reflecting a human tendency to be
optimistic, it m
ay represent the tendency to associate the normal
with the good, and the abnorm
al with the bad. If so, a fam
iliar case of com
petition between the Pollyanna Principle and the Interest
Principle is found in (uncensored) newspapei:s and new
s broadcasts, w
here interestingness, and hence newsw
orthiness, is strongly asso-ciated w
ith what is unpleasant. ('B
ad news is good new
s.')
Chapter 7
Com
municative G
ramm
ar: an example
JOH
NSO
N: 'M
y dear friend, clear your mind of cant. You m
ay talk as other people do: you m
ay say to a man. "Sir, I am
your most hum
ble servant." You are not his m
ost humble servant. ... You tell a m
an. "I am sorry you had such bad w
eather the last day of your journey, and w
ere so much w
et." You don't care six-pence w
hether he was w
et or dry. You may talk in this m
anner; it is a mode of talking·
in Society: but don't think foolishly.' [Jam
es Boswell, The Life of Johnsnn. 15 M
ay, 1783]
7.1 Com
municative G
ramm
ar and pragmatic force
In the preceding chapters I have tried. to develop in more detail
the approach to pragmatics briefly stated in C
hapters I and 2. My
aim is now
to give an example of how
this approach works out in
practice, by applying it to the pragmatics of negation and inter-
rogation in English. The analysis w
ill necessarily be rather in-form
al, but it will be elaborated in m
ore detail than has been practicable so far.
One of the im
plications of this model is that w
e can analyse any gram
matical category (say negation, m
odality, or interrog-ation) on
distinct levels. The most fam
iliar level to linguists is the syntactic: w
e can describe, for example, how
to form nega-
tive sentences. or clauses in English. The second level is the sem
antic: the level, that is, of sense rather than of force: here we
consider, for example, the negative operator in relation to prop-
ositional logic. And the third level is the pragm
atic: we have
already considered, ·for example, one aspect of the pragm
atics of negation in 4·5· I, w
here the sub-maxim
of negative uninforma-
tiveness was discussed. W
e may describe as coM
MUNICATIVE
GR
AM
MA
R an approach to gram
mar w
hich aims to relate these
three levels of description to each other. In w
hich direction should the analysis be made? Should w
e start w
ith syntax, and work out from
there to semantics and finally
to-pragmatics, or should w
e work in the opposite direction? In
principle, there seems to be' no good reason for preferring one
order to the od1er; but in practice, it is natural to start with w
hat is relatively w
ell known and clear-cut -
the syntax -and to go
from there to w
hat is relatively context-variable and unclear-the pragm
atics. In deciding on this direction, we im
plicitly seem to
CO._M
MU
NICA
TIVE G
KI\M
MA
R A
ND
PRA
GM
ATIC
FOR
CE
I 53
take up the position of the addressee, who (as argued in 2.5.2)
arrives at the force of an utterance by decoding its sense and then, by heuristic problem
-solving, works out its force, or pragm
atic in-terpretation. This interpretative direction is the one taken, for exam
ple, by Grice with his concept of conversational im
plicature, . and by Searle in his discussion of 'indirect speech acts'. M
ethod-ologically, this seem
s to be the soundest way to approach the elu-
cidation of pragmatic force, since if w
e agree (see 2.4) that pragm
atics, like semantics, is a study of publicly conveyed m
ean-ing, rather than of the private thoughts or judgem
ents of this or that person, then it is inevitable that w
e start with the text itself -
what is publicly observable -
and then attempt to reconstruct
from this the m
eaning which is conveyed, given certain assum
p-tions and .certain kinds of know
ledge shared by s and h. It is for t..his reason that I suggested in
I .4(i) that the . best role for the pragm
atic analyst is that of a bystander or observer: the third per-son in the exchange.
· A
nother reason for preferring to work from
syntax . to prag-m
atics, or (if you will) from
form to function, is the difficulty of pin-
ning down w
hat the force of an utterance is, given (as argued in 3.4) that pragm
atic force is essentially non-categorical and inde-finitely variable according to context, and ·that pragm
atic descrip-tion involves indeterm
inacies and continuous values. ln fact,
before we em
bark on the task of exemplifying pragm
atic analysis in m
ore detail, the question must be .considered: w
hat form does the
description of pragmatic force take? To provide a provisional
answer, I shall bring together a num
ber of observations which can
be made about pragm
atic force, on the basis of earlier chapters: L
The force of an utterance U
is defined, for our purposes, by a set of conversational im
plicatures F (I use 'implicature' in the
broad sense of 'a proposition pragmatically inferred from
(i) the sense of U, (ii) the assum
ption that s is ob'§erving, in cer· tain respects and to certain degrees, the principles and m
axims
of the Interpersonal Rhetoric; (iii) contextual kno'>vtedge').
2. The set of implicatures F is ordered in such a w
ay that for each im
plicature it is possible to trace a path whereby it can be in-
ferred, perhaps via intermediate im
plicatures·, to a certain de-gree of probability, from
(i)-(iii) above. Tne length of tr.is path is a m
easure of the indirectness of the implicature.
3· Each implicature has a degree of confidence associated w
ith it; this· degree of confidence m
ight be formalized as a probability
of the implicature's being part of w
hat s meant to convey by
.
4· A subset of the set of implicatures F defines the m
ocuti.cmary '
154 C
OM
MU
NIC
ATIV
E GRA
MM
AR : A
N EX
AM
PLE
force of U. This subset (which in the sim
plest cases has only one m
ember) defines a m
eans-ends analysis of which the
uttering of U is tqe central action, and thereby defines the pre-
sumed illocutionary goal of s in uttering U.
5· Another (not alw
ays disjoint) subset of Fdefines in what re-
· spects and to w
hat extent s is presumed to be observing· the
maxim
s of the Interpersonal Rhetoric. The im
plicatures in this subset define s's presum
ed rhetorical goals, such as observing certain m
axims of the PP.
6. Many
implicatures
are associated
with
some value
on a
pragmatic scale. For exam
ple, an implicature in the form
of a propositional attitude, such as 's believes that (X
]', will be
associated with a subjective value as to the degree of con-
fidence of that belief. If such values are represented by Greek
letters, •... then an im
plicature should be more fully ex-
pressed in a form such as 's believes"' that [X
]'. Similarly, if an
implicature has to do w
ith s's observance of the PP, then some
value should be attached to the relevant attitudinal predicate to indicate how
polite s is being. We have already sufficiently
noted the scalar nature of politeness: one cannot reduce tact, for exam
ple, to a simple yes-no ch0ice betw
een 'sis obeying the Tact M
axim' and's is not obeying the Tact M
axim'. A
lso, the scalar nature of properties such as inform
ativeness and truthfulness has already been noted in, for exam
ple, the trade-off betw
een politeness and sincerity. 7. Som
e implicatures contain m
ore than one 11ttitudinal predi-cate, one proposition being em
bedded in another. For exam-
ple, an implicature m
ay read: 's assumes"' that [h is being
modestP]'. In principle, this em
bedding of one propositional attitude in another can proceed indefinitely.
These general statements about for-ce w
ill do for the present. O
bviously, a stricter formulation of the notion of 'pragm
atic force' is possible, and I shall return briefly to that question in the final chapter. A
t present, it is probably the indeterminacy of that
notion which w
ill strike readers accustomed to m
ore categorical theories, such as Searle's speech-act theory. I can only say that this indeterm
inacy is necessarily a prominent factor in any m
odel w
hich attempts to represent realistically how
linguistic com
-m
unication works.
Yet one further kind of indeterm
inacy must be m
entioned: that w
hich comes betw
een Popper's expressive and signalling functions
of language. In
discussing m
eaning, w
e norm
ally assum
e that what is expressed by s and w
hat is signalled by s are equivalent: that w
hat s plans to comm
unicate, and what is actually
COM
MU
NICA
TIVE G
RAM
MA
R AN
D PRA
GM
ATIC FO
RCE 155
conveyed to h are the same. This assum
ption is, indeed, built into the pragm
atic definition of meaning (2.4). B
ut there are two
respects in which this m
atch between s's m
eaning and h's mean-
ing may fail. First, there m
ay, as I have already argued, be an elem
ent of planned indeterminacy: the force of U
may be to som
e extent left for negotiation betw
een sand h. Second, there may be
an element of unplanned indeterm
inacy: for example, in the dis-
cussion of the Maxim
of Relation (4.3) I suggested that relevance
is a matter of degree, and that to a certain point it is a m
atter of choice w
hether h recognizes s's remarks as relevant to a particu-
lar conversational goal or not. To take one of Grice's original
examples
(1975:54), supposedly
an extract from
a tea-party
conversation: A
: Mrs X
is an old bag. (A
mom
ent of appalled silence) B: The w
eather has been quite delightful this summ
er, hasn't it?
.
In Grice's interpretation, B blatantly violates the M
a.xjm of R
e-latiorf, and hence im
plicates that A has comm
itted a social gaffe. B
ut another reading of it, perhaps less likely in this case, is that B is genuinely em
barrassed by.A's rem
ark, and ineptly changes the subject, w
ithout intending that anyone should notice his embar-
rassment or displeasure. In either case, A
's response may be of
three kinds:
(a) A, coarse, thick-skinned fellow
that he is, fails even to notice B's em
barrassed change of subject. (b) A notices B's inept change of subject, and attributes it to B's
genuine embarrassm
ent at A's rem
ark. (c) A notices B's inept change of subject, and attributes it to B's
intention of drawing the com
pany's attel\tign to the fact that A has com
mitted a social gaffe.
And of course, there are further possible responses, such as B's
noticing the change of subject, but .not regarding it in any way as
a reaction to A's previous rem
ark. One point m
ade by this illus-tration is that w
hat may be part of the intended m
eaning may not
be a part of the conveyed meaning and vice versa. A
nd a secgn<.t point, m
ore important for the present discussion, is that it is im
-possible to determ
ine a boundary between an uttetance w
hereby s im
plicates some proposition p (in this case 'A
h:as -a
social gaffe') and an utterance which m
erely cqro.¢unicates the unintended inform
ation that s believes,p. 4o revert to the sim-
plest type pf example: w
hen is ayawn;rm
ply a sign.of boredom,