Principles of Pragm

atics G

eoffrey N. Leech

LONG

MAN LINGUISTICS UBRARY

Title no 30

PRINCIPLES OF PRAGMATICS

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

QQ

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,