Al-Sharafi - Textual Metonym.pdf

Textual Metonymy

A Semiotic Approach

Abdul Gabbar Mohammed Al-Sharafi

Textual Metonymy

1403_913323_01_preiv.qxd 11/19/03 8:40 PM Page i

This page intentionally left blank

Textual MetonymyA Semiotic Approach

Abdul Gabbar Mohammed Al-Sharafi

1403_913323_01_preiv.qxd 11/19/03 8:40 PM Page iii

© Abdul Gabbar Mohammed Al-Sharafi 2004

All rights reserved. No reproduction, copy or transmission of thispublication may be made without written permission.

No paragraph of this publication may be reproduced, copied or transmittedsave with written permission or in accordance with the provisions of theCopyright, Designs and Patents Act 1988, or under the terms of any licencepermitting limited copying issued by the Copyright Licensing Agency, 90Tottenham Court Road, London W1T 4LP.

Any person who does any unauthorised act in relation to this publicationmay be liable to criminal prosecution and civil claims for damages.

The author has asserted his right to be identified as the author of this work in accordance with the Copyright,Designs and Patents Act 1988.

First published 2004 byPALGRAVE MACMILLANHoundmills, Basingstoke, Hampshire RG21 6XS and175 Fifth Avenue, New York, N.Y. 10010Companies and representatives throughout the world

PALGRAVE MACMILLAN is the global academic imprint of the PalgraveMacmillan division of St. Martin’s Press, LLC and of Palgrave Macmillan Ltd.Macmillan® is a registered trademark in the United States, United Kingdomand other countries. Palgrave is a registered trademark in the EuropeanUnion and other countries.

ISBN 1–4039–1332–3

This book is printed on paper suitable for recycling and made from fullymanaged and sustained forest sources.

A catalogue record for this book is available from the British Library.

Library of Congress Cataloging-in-Publication DataAl-Sharafi, Abdul Gabbar Mohammed, 1967–Textual metonymy: a semiotic approach/Abdul Gabbar Mohammed Al-Sharafi.

p. cm.Includes bibliographical references and index.ISBN 1–4039–1332–3 (cloth)1. Metonyms. 2. Semiotics. 3. Discourse analysis. I. Title

P301.5.M49A39 2003401�.41––dc21 2003054871

10 9 8 7 6 5 4 3 2 113 12 11 10 09 08 07 06 05 04

Printed and bound in Great Britain byAntony Rowe Ltd, Chippenham and Eastbourne

1403_913323_01_preiv.qxd 11/19/03 8:40 PM Page iv

To those who made this book possible by their constant prayers:

To my parentsTo my wife and my children

a sign of gratitude to their company and a gesture of gratitude fortheir unfailing support

1403_913323_02_prevxii.qxd 11/19/03 8:40 PM Page v


vii

Contents

List of Figures ix

Acknowledgements x

System of Transcription xi

Introduction: The Structure of the Book 1

1 Theoretical Framework and Historical Background 51.0 Introduction 51.1 Rhetoric and textlinguistics 51.2 Figurative language and textlinguistics 81.3 Metonymy in western rhetoric 111.4 Metonymy in Arabic rhetoric 20

2 Metonymy in Modern Figurative Theory 352.0 Introduction 352.1 Theories of metonymy 36

3 Metonymy and Semiotics 803.0 Introduction 803.1 Semiotics: a general perspective 803.2 On the notion of signification 853.3 On representational semiotics and metonymy 963.4 Metonymic aspects of the linguistic sign 1013.5 A relational model of metonymic signification 1043.6 A textual model of metonymy 106

4 Metonymy and Text Cohesion 1094.0 Introduction 1094.1 Metonymy and cohesion 110

5 Metonymy and Text Coherence 1365.0 Introduction 1365.1 Metonymy and schema theory 1385.2 Metonymy and schemata 1455.3 Metonymy and scripts 1605.4 Metonymy and scenarios 1685.5 Metonymy and plans 1765.6 Metonymy and goals 181

1403_913323_02_prevxii.qxd 11/19/03 8:40 PM Page vii

viii Contents

Notes 184

Bibliography 193

Author Index 205

Subject Index 209

1403_913323_02_prevxii.qxd 11/19/03 8:40 PM Page viii

ix

List of Figures

1.1 Classification of types of transference in Arabic rhetoric 211.2 The three dimensions of contiguous transference in

Arabic rhetoric 223.1 Metonymy and the Peircean notion of sign 913.2 A relational model of metonymic signification 1053.3 A model of textual metonymy 1074.1 A semiotic model of text 1114.2 Patterns of form for concept in example (46) 1304.3 Patterns of hyponymic relations in example (47) 1314.4 A metonymic model of hyponymic relations in

example (47) 1324.5 Patterns of concept for concept relation in example (52) 1344.6 A model of meronymic relations in example (56) 1355.1 A unified model of knowledge structures 1435.2 The three types of inheritance 1485.3 The office schema model in example (15) 1545.4 The person schema model in example (15) 1545.5 A blend of the two models in example (15) 1545.6 Metonymic referential relations in example (18) 1575.7 Metonymic relations within the town schema in

example (18) 1585.8 Metonymic relations within the characters’ schema 159

1403_913323_02_prevxii.qxd 11/21/03 6:24 PM Page ix

x

Acknowledgements

This book owes a great deal to many people and institutions. First of all, Iwould like to express my indebtedness and gratitude to Dr James Dickins,my teacher and supervisor, who made the experience of doing this researcha great pleasure and who directed my sometimes scatterbrained thinkinginto constructive creativity. I would like to extend my sincere thanks toSheikh Sultan al-Qäsimi who made this research possible through theSharjah Scholarship Grant. My profound gratitude goes to Professor TimNiblock and other committee members of the Sharjah Scholarship Panel forselecting my proposal. I wish to express my thanks to the ORS (OverseasResearch Studentship) committee for contributing to the funding of thisresearch project. I am indebted to Dr Janet Watson for all help and support.Many thanks are also due to A. Ehteshami, for help and encouragement.

Sincere thanks are due to Gilles Fauconnier of the University of California,San Diego, Guenter Radden of the University of Hamburg, Richard Waltereitof the University of Tübingen, Christian Vosshagen of the University ofHamburg, Andrea Blank of Philipps-University Marburg and Dan Fass-Canada,for email correspondence and advice regarding current research on metonymy.Special thanks are due to Mohammed Khattäbi of Ibn Zuhr University,Morocco for his illuminating response to my inquiry regarding metonymy inArabic rhetoric. I am grateful to Klaus-Uwe Panther of the University ofHamburg and Linda Thornburg of the University of Etövös Lorand inBudapest, Anne Pankhurst of the University of Edinburgh, Brigitte Nerlich ofthe University of Nottingham and Richard Waltereit of the TübingenUniversity for sending me some of their papers related to metonymy. My sin-cere thanks go to Randal Holme of the Department of Linguistics, Universityof Durham for his invaluable and illuminating comments on an early draft ofthis book.

My sincere thanks go to Dr Ahmed Al-Imädi, Mr Abdull Aziz Maarafia, Dr Huda Bouslama, Dr Madani Othman, Mr Al-Noor Rahma and Mr AmeenAl-Himaryi, for all their encouragement.

ABDUL GABBAR MOHAMMED AL-SHARAFI

1403_913323_02_prevxii.qxd 11/19/03 8:40 PM Page x

System of Transcription

xi

1403_913323_02_prevxii.qxd 11/19/03 8:40 PM Page xi


1

Metonymy is traditionally defined as a process of substituting one word foranother with which it is associated. This however is a reductionist view of thenature of metonymy; my aim in this book is to take metonymy beyond theseconfines and investigate the textual potential of the trope. I therefore proposea semiotic approach to metonymy to account for the three-dimensional aspectsof the trope, arguing that metonymy is not only a process of substitutionbetween words, as the majority of classical rhetorical treatises perceive it to be.Nor is it just a process of substitution between concepts, as the majority of mod-ern figurative accounts of the trope claim. In fact, I believe that metonymyincludes a further level of signification of things and objects, which I willdemonstrate to be primarily the dimension of context. Out of this semiotictreatment of metonymy I propose a textual model of the trope to account forthe nature and role of metonymy in text. This textual model makes use of sucha semiotic tridimensional model of metonymy and relates it to another tridi-mensional model of text that realises text as language, text as cognition andtext as context. Let me now provide a definition of metonymy as I treat itthroughout this book. Metonymy is a process of REPRESENTATION in which oneword or concept or object stands for another by contiguity or causality.

I explore in this book whether the following statements can be substantiated:

(i) The potential of metonymy has been largely underestimated.(ii) Metonymy is cognitive and semiotic in nature.

(iii) A semiotic approach to metonymy leads to a comprehensive theoryof metonymy as representation that cuts across the ontological, epis-temological and linguistic dimensions of metonymic signification.

(iv) Based on the representational view of metonymy, a textual theory ofmetonymy could be developed.

The structure of the book

Chapter 1 outlines the general theoretical framework and the historicalbackground. In section 1.1, I show that there is a strong link between

Introduction

1403_913323_03_int.qxd 11/19/03 8:40 PM Page 1

rhetoric and textlinguistics. Section 1.2 argues that there is a link betweenfigurative theory and textlinguistics. The assumption underlying the dis-cussion is that figurative language is a fundamental part of our cognitionand not merely a form of deviation from normal modes of thinking. Thisdiscussion draws from cognitive linguistics sources which argue that figura-tive language is not merely a tool for adornment and embellishment or amatter of defamiliarisation as it is commonly considered among formal the-ories of criticism. Rather, it is an important mechanism of our perception.

Section 1.3 goes into a specific discussion of the definitions of metonymyin the western rhetorical traditions, specifically the Greek and the Latin ones.This section tells us that the definitions provided for metonymy in these tra-ditions limited the scope of the trope to either a substitution between wordsor a substitution between concepts. The section concludes with an observa-tion that the history of metonymy in western classical rhetoric suffers fromthis limitation and therefore we need to look for some other source of rhetor-ical scholarship. This takes us on finally to section 1.4, which discusses thetrope in Arabic classical rhetorical scholarship. The Arabs recognised threeforms of transference, which could correspond to what is known in westernrhetoric as metonymy. The section has to propose a reorganisation of the fig-urative spectrum in Arabic rhetoric bringing the three modes of figuration,namely kinäyah ‘implicitness’, majäz ‘aqlï ‘cognitive or rational transference’and majäz mursal ‘loose or non-similarity transference’ to bear on the notionof representation as a fundamental metonymic operation.

The aim here is to construct a semiotic account of the trope in the Arabicrhetorical scholarship in such a way that kinäyah would correspond to treat-ments of metonymy as a representation between words, since the essence ofthis type of transference in Arabic rhetoric is basically linguistic. In addition,the intention is to relate majäz ‘aqlï ‘rational or cognitive transference’ tocognitive accounts of metonymy, because, as its name suggests, this type oftransference in Arabic rhetoric is essentially cognitive or rational. Majäzmursal ‘loose transference’ is also linked to the notion of ontological signi-fication specifically with the representation of objects because the relationsthat underlie this type of transference in Arabic rhetoric are mainly exis-tential relations including part–whole, whole–part, situationality, position-ality, adjacency and so on. I conclude this section with the argument thatin neither tradition was there any clear attempt to bring all the dimensionsof metonymy together in a unified approach to elucidate how metonymyfunctions in binding text and providing grounds for its interpretation. Inshort, the role of metonymy in text generation and text organisation isclearly lacking in both traditions.

Chapter 2 investigates whether any semiotic treatment has been proposedin any modern account of metonymy, the field either of figurative theory orof cognitive semantic studies. The chapter reaches the conclusion that thesame reductionism can still be witnessed also in modern accounts of

2 Textual Metonymy

1403_913323_03_int.qxd 11/19/03 8:40 PM Page 2

metonymy. The only exception to this generalisation is the treatmentproposed by Radden and Kövecses (1999), which is a semiotic analysis parexcellence, albeit that it suffers from a lexical orientation in most of its dis-cussions. Radden and Kövecses’s study does not proceed from this semioticanalysis to discuss the textual powers of metonymy. The treatments inrecent accounts are restricted to the lexical level of signification, and, to thebest of my knowledge, no work has been done on a semiotic model of thetextual level of metonymic signification which is characteristic of my semi-otic approach.

I then provide a more detailed account of the semiotic dimension ofmetonymy, taken up in Chapter 3, where I develop a semiotic theory ofmetonymy that would correspond to the triadic representations of thenotion of sign in semiotics. On the basis of an argument that metonymy isan ‘index’, metonymy is stated to be an indexical sign: it is a ‘signifier’ inthe form of words or expressions and it is a ‘signified’ in the form of eitherreal objects in the world or cognitive concepts in the human brain. Thechapter develops a relational model of metonymy, which recognises repre-sentation as the primary metonymic type of relation, and assigns relations ofcontiguity and causality to this principal relation. The chapter then moveson to propose a textual model of metonymic relations based on the relationof representation. Based on Radden and Kövesces’s (1999) inventory ofmetonymic relations, the model highlights specifically the following ninetypes of relations:

(a) Metonymic representational relation of form for form.(b) Metonymic representational relation of form for concept.(c) Metonymic representational relation of form for thing.(d) Metonymic representational relation of concept for form.(e) Metonymic representational relation of thing for form.(f) Metonymic representational relation of concept for concept.(g) Metonymic representational relation of concept for thing.(h) Metonymic representational relation of thing for concept.(i) Metonymic representational relation of thing for thing.

The model is interactive in nature but it is possible to specify (a) to (e) relationsto account for the formal and semantic aspects of text cohesion while theremaining relations, (f) to (i), account for text coherence. The interactivenature of these relations captures the textual potentialities of metonymy inthat in text there is always dynamic interaction between forms, conceptsand contexts. The higher level of textuality, though, realises cognitive andcontextual relations like those of CONCEPT FOR CONCEPT, CONCEPT FOR THING,THING FOR CONCEPT and THING FOR THING, the forms playing a major role as trig-gers to these cognitive and contextual processes.

Chapter 4 explores the proposed relation between metonymy and textcohesion and provides examples of how metonymy actually contributes to

Introduction 3

1403_913323_03_int.qxd 11/19/03 8:40 PM Page 3

the cohesion of text. The chapter integrates the metonymic relations (a) to(e) above into a model of cohesive patterning in text. The chapter takesHalliday and Hasan’s (1976) model of cohesion in English as a framework.Chapter 5 integrates the four metonymic relations numbered (f) to (g) aboveinto a model of coherence in text. The chapter takes as a framework the workdone in the conceptual representation of knowledge structures, notably thatof Schank and Abelson (1977), Minsky (1975) and Rumelhart (1977). I pro-pose that knowledge structures such as schemata, frames, scripts, scenarios,plans and goals are essentially mental structures that are based onmetonymic relations of PART FOR WHOLE or WHOLE FOR PART, CAUSE FOR EFFECT

and EFFECT FOR CAUSE relations and as such the movement between these con-ceptual structures in text-processing is facilitated by the representationalrelations of CONCEPT FOR CONCEPT, CONCEPT FOR THING, THING FOR CONCEPT, andTHING FOR THING. Perceived as such these knowledge structures are funda-mentally metonymic in nature and once triggered by linguistic forms in thetext establish relations of metonymic reasoning between parts of text, eitheras being parts forming wholes or vice versa, or as causes pertaining to effectsor vice versa. The chapter presents a reorganisation of these knowledgestructures into a unified model based on whether the particular knowledgestructure is descriptive or procedural in nature. Within the proceduralknowledge structures, I distinguish between those knowledge structures thatare conventional and those which are arbitrary and this is related to the twomajor principles of metonymic reasoning of contiguity and causality.

4 Textual Metonymy

1403_913323_03_int.qxd 11/19/03 8:40 PM Page 4

5

The impulse to speak and think with metonymy is a significant partof our everyday experience. Traditionally viewed as just one ofmany tropes, and clearly subservient in most scholars’ minds tothe master trope of metaphor, metonymy shapes the way we thinkand speak of ordinary events and is the basis for many symboliccomparisons in art and literature.

Gibbs (1999:61)

1.0 Introduction

This chapter provides a theoretical and historical context for the book.Section 1.1 deals with the relationship between rhetoric and text linguistics.Section 1.2 considers the relationship between figurative theory and text lin-guistics. Section 1.3 deals with the history of metonymy in western rhetoricaltraditions. Section 1.4 moves on to discuss metonymy in the Arabic rhetori-cal tradition. The aim of these two latter sections is to show that the treatmentof metonymy, as one of the main figures discussed in both rhetorical traditions, suf-fers from two major problems:

(i) Theoretical reductionism, by reducing the nature of the trope to a mere substi-tution of words and neglecting its cognitive and pragmatic dimensions.

(ii) Practical reductionism, by reducing the role of the trope to the level of lexicalsubstitution and neglecting its potential power at the level of text.

1.1 Rhetoric and textlinguistics

Before outlining the relationship between rhetoric and textlinguistics it is nec-essary to start with the questions ‘What is rhetoric?’ and ‘What is text linguis-tics?’ Let me begin by a definition of textlinguistics. Crystal points out that

in recent years, the study of texts has become a defining feature of a branch of linguistics referred to (especially in Europe) as textlinguistics,

1Theoretical Framework and Historical Background

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 5

and ‘text’ here has central theoretical status. Texts are seen as languageunits which have a definable communicative function, characterized bysuch principles as COHESION, COHERENCE, and informativeness whichcan be used to provide a FORMAL definition of what constitutes theiridentifying textuality or texture. (Crystal 1991:350)

So textlinguistics is the study of human textual interaction or their use oflanguage in real-life situations. In this sense the scope of the discipline cov-ers both written as well as spoken texts and in fact this is the sense sustainedthroughout my discussion.

Now let us move to define rhetoric. Wales defines rhetoric as follows:‘from Greek techne rhetorike “art of speech”, originally a discipline concernedwith the skills of public speaking as a means of persuasion’ (Wales1989:405–6). In his discussion of the relationship between rhetoric, stylis-tics and textlinguistics Enkvist argues that ‘the expansion of linguistics tocover text and discourse increasingly motivates a new reading of rhetoricand stylistics’ (Enkvist 1985:11). He then defines rhetoric as ‘that branch oflanguage study which is teleologically oriented towards effective communi-cation. Everything that fits into this loose and spacious envelope thenbecomes, actually or potentially, part of rhetoric’ (ibid. p. 16). Plett alsohighlights the aspect of ‘persuasion’ as a fundamental feature of rhetoric. Heargues that ‘rhetoric originally was a technique or art of producing persua-sive texts’ (Plett 1985:61). Plett asserts that rhetoric as a technique ‘wasalways characterised by a set of rules. These did not coexist in a randommanner, but regularly strove towards logical consistency and structuralcoherence.’ He maintains that these rules constituted a system which

retained an amazing solidity which lasted for almost two thousand andfive hundred years. It was handed down from generation to generationand formed the basis for the production of public and private texts suchas political speeches, sermons, letters, declamatory exercises, advertise-ments, and, most significant of all, literature. Thus rhetoric can truly betermed a science of discourse. (Plett 1985:61)

These spacious definitions of rhetoric allow for the discipline to account forhuman communication in general. Textual interaction, being one majortype of human communication, is then regarded under this definition as a branch of rhetoric. Enkvist (1985) proposes four text theories whichaccording to him provide a convenient means of comparing rhetoric andtextlinguistics and ‘reflect the history of textlinguistics in terms of its expan-sion from the study of intersentential links to the investigation of every-thing that goes into the pragmatics of discourse’ (p. 23). The first is thesentence-based model or theory of text. Enkvist argues that the text linguistsin this tradition ‘tried to describe the cohesive ties that cement sentences into

6 Textual Metonymy

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 6

texts, often by extending traditional grammatical methods’ (p. 23). Enkvistassociates the work in this tradition with the work of Halliday and Hasan(1976).

The second theory of text according to Enkvist is called ‘the predication-based model’. This arose from the shortcomings of the previous modelconcerning the difficulties of manipulating or altering the sentence divisionof the text. But such manipulation, according to Enkvist, ‘was necessary toreveal the relations between texts and to answer the question “could twotexts in fact be regarded as variants (allotexts) of the same underlying set ofelements (texteme)”?’ (p. 23). This model starts from an ultimate basic unitof text which Enkvist calls ‘text atoms, relations between the text atoms, anda text strategy steering the linearisation, grouping, coordination and embed-ding processes that produce surface texts from the underlying set of textatoms’ (p. 23).

The third theory is that which Enkvist calls ‘the cognitive model’, which‘tries to describe some area of human cognition, usually in the form of cog-nitive networks’. This model emerged as a result of the need to know wherethe sentences in the first model or the predications in the second modelcome from. Perceived as such, the cognitive model provides a higher char-acterisation of sentences as cognitive propositions in the mind before theyare expressed as sentences. Enkvist relates this model to the work of Minsky(1975) and Schank and Abelson (1977) in semantic schema or frames.

The fourth theory of text proposed by Enkvist is the interaction-basedmodel, which highlights the multidimensional aspect of human communi-cation. This model stresses the conceptualisation of a sender, a receiver and therelations between them as well as other situational and contextual factorsinfluencing the communication activity. Enkvist relates these models of textto the various parts of classical rhetoric.1 The ‘sentence-based’ model and ‘the predication-based’ model of text, according to Enkvist, correspond to thethird part of rhetoric known as ‘elocution’ or style. The ‘cognitive’ model,Enkvist argues, corresponds to the part of rhetoric known as ‘invention’which is the search for material to be textualised. Usually this material existsin a cognitive form in the mind of the speaker. The ‘interaction-based’ modelis argued to correspond to the nature of rhetoric as emphasising the effec-tiveness of communication. It should be noted here that the link I intend toestablish between rhetoric and textlinguistics is laid down along the lines pro-posed by Enkvist. I shall, therefore, focus my discussion on the nature androle of metonymy in text. I shall, however, limit the discussion to the first andthe third models, that is the sentence-based model and the cognitive model.

The reason for this limitation is that the second model, the predication-based, is too logical to be incorporated into my argument, given the fact thatthe overall orientation of my position is cognitive. This orientation will obvi-ously prevent the adoption of the logical view of language and cognition.Besides, the logical view stipulates that there is a one-to-one correspondence

Theoretical Framework and Historical Background 7

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 7

between concepts and categories and assumes that language is a reflection ofreality. This undermines the approximative nature of communication andoverlooks the inherent nature of fuzziness in concepts and categories.Consequently, this makes the space for figurative treatment within thisframework rather difficult. For the first model I discuss Halliday and Hasan’smodel of cohesion in Chapter 4. Minsky’s and Schank and Abelson’s work onthe cognitive model of text is discussed in Chapter 5. With regard to the lastmodel, that is the interactional model, I argue that the semiotic approachthat I propose for the treatment of metonymy is interactionist because it cap-tures contextual factors as well as cognitive and linguistic factors involved intextual communication and metonymic signification.

1.2 Figurative language and textlinguistics

In textlinguistics there is usually a tendency to consider the distinctionbetween literal and figurative language a matter of degree (cf. Goatly1997:14). The reason for this is that textlinguistics, since its emergence inthe 1960s, has incorporated a large number of ideas from various disciplines –notably cognitive psychology and sociology. It has also benefited a greatdeal from philosophical and anthropological theories of human communi-cation and language use.2 This diversified input has made the nature of thediscipline largely interdisciplinary and essentially semiotic. According tothis view, there is no clear line between figurative and literal languagebecause a great deal of human communication is approximative,3 andlinguistic concepts are generally fuzzy (cf. Goatly 1997:14).

Textual practice is then a reflection of this fuzziness in linguistic conceptsand the approximative nature of communication. Therefore, it is difficult toclaim that the text is a reflection of either the state of affairs in the real worldor the conceptual representation of this state of affairs. There is a great dealof interaction and negotiation between the world of text, the world ofconcepts and the world of things. This interaction between these three dif-ferent worlds gives a good chance for imprecise representation and leavesthe process of text generation, organisation and interpretation only imper-fect. This means that we employ figurative language for the purpose ofapproximation and compromising exactness in almost all types of languageand not only poetic language. Consider the following examples:4

(1) I can hear you.(2) Zena went shopping.(3) Tom hit John.(4) Adam likes apples.

Although these examples are individual utterances and they look very normaland typical of everyday language, their interpretation involves a great deal

8 Textual Metonymy

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 8

of help from cognitive processes that are typical of what might be called ‘fig-urative understanding’. Example (1) involves a double level of non-literalunderstanding. When we hear people we only hear their voices and notthem as a whole, but because the voice of someone is very typical of theirpersonality and their self, the voice is used to represent the person. A sec-ond level of non-literality is evident in the fact that when we say ‘I can hearyou,’ the interpretation of the utterance is not understood as ‘ability’ as itsform suggests. Rather, the interpretation is that ‘I am hearing you’ in actu-ality. Example (2) shows another form of figurative understanding of theutterance. We know that the act of shopping involves many sub-actions. Inthe example we have no problem imagining them, although they are notmentioned in the text. Because this act is ritualised in our life, the wholestands for the parts of this whole and we come to activate this ritualisedknowledge and to make sense of the text.

Example (3) looks very literal but in fact is not, because it is not possiblethat the whole body of Tom hit the whole body of John. In actuality, it isonly one part of Tom that hit one part of John; perhaps the former’s handand the latter’s nose were involved in this action. Again our understandingof this text is made successful only through non-literal understanding of it.Because the hand and nose are parts of the bodies of both Tom and John,the whole of their bodies are used to signify the parts. In example (4) thereis an element missing but we never normally realise it. We know that Adamdoes not like apples ‘for their own sake’ and that when we say this we meanAdam likes eating apples. We understand all these examples, and in fact wealways miss the fact that they are examples of figurative language becausethe cognitive structure of our brains is, as it were, structured along the linesof similar principles.

The production stage of communication starts with the state of affairs tobe communicated or communicated about. This, according to Goatly(1997), is the physical stage because it exists in the world of things andobjects or entities. However, it could be argued that this level is not onlyphysical but also involves some psychological and abstract phenomenawhich pertain to abstract concepts. Through a mental process which can beparalleled with ‘stringing of concepts into thought’ this state of affairsbecomes thought which in turn by another mental process becomes a proposition which is realised by linguistic data as text. So the productionof a text is a result of an interactive process between reality, mind and lan-guage. The reception stage goes along similar lines but in a reverse order,starting this time with the text and then building up propositions as to whatthis text may mean. Then from the proposition there is a process of guess-ing and inferencing to arrive at the thought which the speaker might havehad in mind about a certain state of affairs. The development of communi-cation through these different worlds is bound to show some sort of imper-fect correspondence between one world and another. We know for sure now


1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 9

that there is not a perfect correspondence between objects and linguisticcategories because the relationship is not that clear-cut. Culture plays amajor role in forming our linguistic categories of objects we see in reality. A prototypical house for a European is not the prototypical house for aBedouin Arab. Categories shift from language to language, culture to cultureand from time to time within the same culture and the same language.

The understanding of the following text (5) is dependent on activatingpieces of knowledge that are not mentioned in the text. In other words, theinterpretation of this text goes beyond its literal expression to a figurativedimension of text interpretation. How is this possible?

(5) Mary went to the kitchen. She found the fridge empty. She openedthe freezer and took out some ice-cream.

It is easy to interpret this text, because it activates a schema of a kitchen andthis is what makes our understanding of the text possible and easy. In fact,the items referred to using the definite article are regarded as given andknown. The text is made interpretable because it utilises a routinised scriptwhich helps us to expect the rest of the text since we know what someonegoing into the kitchen would do next. The text is also possible to interpretbecause it makes use of a general pattern of causal connectivity. We knowthat people go into kitchens because they want to eat. To achieve this goalwe can easily interpret the opening of the fridge and then the freezer. In fact,we could also predict that if Mary did not find the ice-cream in the freezerthe text would continue making other searches until the goal of gettingsomething to eat is fulfilled. We can conclude that Mary ate the ice-cream,although this is not mentioned in the text at all, but from our knowledge ofsimilar situations we can expect someone who opens the freezer and takesout some ice-cream to eat it.

This series of prediction, expectation and supplying of information nototherwise expressed in the text is the link I want to establish between figu-rative language and textlinguistics and this is the treatment I propose formetonymic reasoning in text. To conclude, a textual approach does notrecognise these figures as forms of deviation but rather as fundamental partsof our cognitive structure that help provide coherence to reality. Thisapproach views figures in general not only as ways of conveying meaning atthe lexical level but also at a much wider level of textual communication.Sometimes the whole text is employed to realise one single figure of speech,such as metaphor or allegory. Orwell’s Animal Farm is a good illustration ofsuch texts, which realise one particular figure of speech. The whole novelfunctions as an allegorical account of the Russian communist revolution.Jakobson’s approach to metonymy as characterising realist prose5 is justanother attempt to map the workings of figures onto textual models. In fact,the textual model I am proposing for metonymy is not just this general

10 Textual Metonymy

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 10

attempt. It goes beyond this to provide an investigation of a more detaileddescription and explanation of how metonymy contributes to text creation,organisation and interpretation.

1.3 Metonymy in western rhetoric

According to Stanford,

Herodotus and Thucydides, and later Plato, used the word µε�o��́�ε��

to mean to call by a new name, to change something’s name, and in thepassive voice to receive a new name. But in none of its contexts doesµε�o��́�ε�� suggest any of the significance of the term µε��µí� as after-wards applied to metonymy, a particular kind of metaphor. (Stanford1936:3)

This quotation opens our eyes to two vital issues related to the discussion ofmetonymy in western classical rhetoric. The first issue is that because rhetori-cians and philosophers were interested in the poetic use of language theyregarded metaphor as primary to the figurative domain and neglectedmetonymy because it does not involve symbolism and double-unit signifi-cation as there is no transfer on the semantic plane.6 This observation appliesto the western as well as the Arabic rhetorical traditions because both tradi-tions were preoccupied with the study of poetic language. This is usually con-veyed through metaphorical expressions which reveal the excellence of thecreativity of the poet in bringing together different worlds by means ofputting them in a similarity relation through metaphor. The second issue isthat even when there is a discussion of metonymy, the treatment appears tobe very fragmented and rather narrow.

It seems that Greek rhetorical scholarship at one time became entirelypoetic scholarship. Poets and rhetoricians were equally interested in how toattract the attention of their audience by the use of metaphorical language.This involves a transfer on the semantic plane, and the poetic power of boththe speaker and the listener reflects the way analogies are produced by thespeaker and reduced by the listener. Literary scholarship in Greece meantthat both participants had to put in some effort towards the understandingof the poem or the oration. This is why they undermined practical andpurely referential discourse because it was seen as banal and not containinganything new, strange or shocking. This is not surprising to us today as thisview conforms to modern figurative and stylistic theory also, especially asexpounded by Jakobson, who maintains that metaphor is a predominantfeature of symbolic and poetic language, whereas metonymy is the pre-dominant feature of realist prose ( Jakobson 1971:96).

Stanford also argues that the word µε��øo�́, which means ‘metaphor’,‘first appears in Isocrates’ Euagoras (1936:3). Isocrates was trying to draw


1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 11

a distinction between poetic language and non-poetic language. The claimis that prose writers do not have the same linguistic privilege, or what wecall today the ‘poetic licence’, to use whatever linguistic structure is possi-ble to draw the attention of the recipient of the communicative message tothe message itself. Isocrates claims that prose writers are handicapped in thisregard because their discourse has to conform to the forms and terms usedby the citizens and to those arguments which are precise and relevant to thesubject-matter. In other words, Isocrates proposes here that metaphor is a distinctive feature of poetic language because it conveys the experience ofthe world afresh and provides a kind of defamiliarisation in the way the cit-izens perceive the world. Again it is quite clear that the reference is madehere to metaphor and not to metonymy.

If we turn now to Plato to see to what extent his accounts of the rhetori-cal figures can help us in our quest for developing an understanding of clas-sical metonymy, we will be disappointed because Plato’s idealistic philosophyprevented him from dealing with rhetoric except to condemn it and criticiseit as the enemy of truth (see Stanford 1936:4; Harris and Taylor 1997:19). ThePlatonic dialogues reflect the uncomfortable feeling that Plato had towardsrhetoric. Plato’s ethical philosophy launched an attack against the ‘sophists’who were the trainers of rhetoric at that time. However, one should notignore the Platonic debates regarding whether or not there is some intrinsicrelationship between words and their meanings. Householder (1995:92)argues that the question ‘ “is the sound–meaning relation of all or someGreek words inevitable and natural?” is the main topic of discussion inPlato’s Cratylus’. Householder maintains that

Democritus (as quoted in Proclus’ commentary on the Cratylus 16) offeredfour arguments (with four specially coined names) in favour of arbitrariness:(a) ‘homonymy’ or ‘polysemy’, that is the same sequence of phonemes maybe associated with two or more unrelated meanings; (b) ‘polyonymy’ or‘isorrophy’, that is the existence of synonyms; (c) ‘metonymy’, that is thefact that words and meanings change; (d) ‘nonymy’, that is the nonexis-tence of single words for simple or familiar ideas. (1995:93)

We can see that metonymy was a theory of linguistic signification and a process of meaning creation in ancient Greek philosophy. This corrobo-rates the view expressed above, that by being a trope or a semantic operationthat involves movement within one and only one semantic domain, meto-nymy attracted the attention of philosophers interested in meaning-creation more than rhetoricians and orators who were mainly interested inpoetic language.

Aristotle’s treatment of rhetoric and the poetic devices seems to be moreexplicit and perhaps more fruitful. From his treatment of metaphor in hisPoetics and Rhetoric we can derive an understanding of metonymy, since

12 Textual Metonymy

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 12

Aristotle deals with some metonymic relations under metaphor but withoutexplicitly using the word ‘metonymy’. In the Poetics, Aristotle argues that

Metaphor is the application of a strange term either transferred from thegenus and applied to the species or from the species to the genus, or fromone species to another or else by analogy. (1927:ch. 21, 7–11)

In his account, Aristotle discusses four types of metaphor. He regards whatis known today as metonymy as one of these types. Of the four types ofmetaphor Aristotle discusses, the first two apply to what we know today assynecdoche. The third type is what we know today as metonymy and it isonly the fourth type that corresponds to what we know today as metaphor.In addition to the third type of Aristotle’s metaphor, there are some exam-ples which are based on metonymic relations in his fourth type of metaphor.Those Aristotle obviously did not discuss as metonymic.

We can see clearly the metonymic basis of metaphor in Aristotle’s example:

(6) A cup is to Dionysus what a shield is to Ares.

The basis of the figurative relation in this example is metonymic in the firstplace. This metonymic basis starts by providing a contiguous relation inwhich the words ‘cup’ and ‘wine’ are realised as metonymic by virtue of thefirst being the container and the second the content. So when we refer toDionysus’ cup we mean the drink in the cup, that is the wine. Similarly,Dionysus is the maker of this wine, so the relationship here is that of inven-tor for invention. So Dionysus is used to mean the thing he makes, that isthe wine. Metonymic relations also underlie the words ‘shield’ and ‘Ares’ bymeans of the mediating notion ‘war’. Ares is the maker of shields so it is aninventor-for-invention metonymic relationship that is at work here. Alsothere is a contained-for-container metonymic relationship evident in theexample in which the shield is a salient feature of war, thus ‘shield’ is con-tained within ‘war’. So the whole process as we can see is essentiallymetonymic. The metaphor comes only after all these metonymic processeshave been cognitively resolved. The metaphor is realised by taking Dionysus’cup for Ares’ shield and vice versa.

Lodge argues that Aristotle’s definition of metaphor has been very influen-tial throughout the history of rhetoric in the west. Lodge goes on to claim thatbecause Aristotle treated metonymy as a subclass of metaphor, metonymycontinued to be discussed as such till the late 1950s, when Jakobson proposedthat metaphor and metonymy are in fact two distinct figures based on twoopposite principles (Lodge 1977:75–6). However, it should be noted that notonly was Aristotle aware of the effect of metonymy in language as a phe-nomenon, although he did not give it this name, but also he was aware of itin cognition. It is however strange how the connection between these two


1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 13

fundamental aspects of metonymy was missed throughout the history of thetrope. Koch argues that Aristotle sets up three relations to account for theprocess of remembering:

[S]o we track down the sequence [of our ideas] by starting from the presentmoment or from something else and from something similar or oppositeor close [to it]. That is the way remembering comes about. (Koch 1999:141)

The treatment of the notion of ‘contiguity’ by Aristotle as an aid to memoryis suggestive of a cognitive view of rhetoric at the time. However, whenrhetoric was reduced to the critical analysis of written texts, the part ofrhetoric known as ‘memory’ dropped out (see Enkvist 1985:17; Dixon1971:46). This is in fact how the gap began to appear between rhetorical the-ory and the theory of remembering or rather what is nowadays referred to as‘cognitive science’ in general. As a result, rhetoricians after Aristotle did notbenefit from his cognitive treatments of the fourth part of rhetoric. It shouldbe stressed that the cognitive dimension is one of the vital issues that thisbook intends to bring up and relate to other dimensions of metonymy,namely the linguistic and contextual or pragmatic dimensions.

I now move on to discuss Demetrius’ account of metonymy in his treatiseOn Style. The treatise addresses metaphor in five successive sections, whichis an indication of the importance given to the figure in the discussion of style.The work seems to be devoted to the treatment of the features of prose style and the difference between this and poetic style. One gets this impressionwhen Demetrius states:

In the first place, then, metaphors must be used; for they impart a specialcharm and grandeur to prose style. They should not, however, be crowdedtogether, or we shall find ourselves writing dithyrambic poetry in place ofprose. Nor yet should they be far-fetched, but natural and based on a trueanalogy. (Demetrius 1927:II, 78)

There is no mention of metonymy in Demetrius’ work. This supports theview that metaphor dominated the scene of western classical rhetoricbecause it was able to provide this charm and grandeur.

Like Demetrius, Longinus’ On the Sublime also includes no mention ofmetonymy. To conclude this section, it is useful to cite Eco about the dom-inance of the whole rhetorical domain by metaphor. Eco holds that

the term metaphor for many authors – and this is true for Aristotle andEmanuele Tesauro – has served to indicate every rhetorical figure in gen-eral; the metaphor, as the Venerable Bede put it, is ‘a genus of which allthe other tropes are species’. To speak of metaphor, therefore, means tospeak of rhetorical activity in all its complexity. (Eco 1984:87)

14 Textual Metonymy

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 14

Moreover, Groupe mu express scepticism about the incapability of ancientrhetoric to formulate a satisfactory definition for metonymy in the statement

La rhétorique ancienné a été incapable de formuler une définition satis-faisante de la métonymie. (Groupe mu 1970:117)

This observation is quite supportive of the hypothesis I have been discussingthroughout this section, which is mainly that the classical rhetoricalaccounts are not helpful as far as metonymy is concerned. However, theLatin account of metonymy seems to be more elaborate and it is to this tra-dition that I turn now.

If we move to the Latin account of metonymy we find that it begins withthe Auctor’s treatise Rhetorica ad Herennium (86–82 BC) and continues untilLaborintus (1250) by the German Eberhard. Rosiene divides this quitelong tradition into four phases. The first is the classical period which coversthe Auctor, Cicero (106–43 BC) and Quintilian (first century AD). The secondis late antique grammar: this covers Sacerdos (third century), Donatus (mid-fourth century), Charisius (mid-fourth century), Diomedes (late fourthcentury), Pompeius (late fifth century) and Sergius (fifth or sixth century).The third period is the medieval period: this includes Cassiodorus(490–575), Isidore of Seville (570–636), Julian of Toledo (642–690), IsidoreJunior (653–704) and Bede (673–735). The fourth period is late medievalpoetics, which covers Matthew of Vendôme (1175), Geoffrey of Vinsauf (?),Gervase of Melkley (?), John of Garland (1195–1272) and Laborintus (1250).The Latin tradition continues till the Middle Ages and perhaps to theRenaissance.

I shall start my discussion with the definition provided by the Auctor inhis rhetorical treatise known as Rhetorica ad Herennium. The Auctor definesmetonymy as

the figure which draws from an object closely akin or associated anexpression suggesting the object meant, but not called by its own name.(Cicero [The Auctor] 1954:iv, 43)

This definition contains a number of elements that need to be discussed.The first element in the definition above is that of ‘drawing’, which is a cog-nitive process very much similar to ‘abstracting’, that is, abstracting from anobject – or let us say an ‘entity’ because object is rather physical and thisabstraction can occur even within abstract entities. The second element isthat of ‘closeness’ or ‘association’, which gives the sense of ‘neighbouring’both physically and cognitively. This is a fundamental characteristic ofmetonymy and in fact a distinctive feature of metonymic figuration thatdistinguishes the trope from other figurative tropes.


1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 15

That the relation that underlies metonymy is of a neighbouring nature iswhat makes it different from other tropes in general and from metaphor inparticular, the underlying relation of which is that of analogy, similarity andresemblance. The third element is the use of the word ‘expression’ in thedefinition. Rosiene argues that ‘expression’ in both translations is not quitea successful translation of the Latin word ‘oratio’. Rosiene further argues thatwhat the Auctor meant by the word ‘oratio’ is not ‘expression’ as Caplan andAchard translated it. Rather, he meant a Latin translation of the Greek logos,(see Rosiene 1992:178–80). Rosiene then engages in a discussion of theGreek notion of ‘logos’ to conclude that the meaning of ‘expression’ if seenunder the Stoic linguistic philosophy in fact translates among other thingsas ‘relation’ (p. 181). Rosiene (1992:183) argues that the Auctor’s definitionof metonymy and his term ‘denominatio’ which is translated as ‘metonymy’has four distinct processes which are fundamental to the understanding ofthe nature of metonymy. These are:

(i) it denominates things(ii) and it abstracts a relation from these things

(iii) in such a way that a thing may be understood(iv) that may not be not called by its common name.

So in any metonymic signification we have four separate processes. Thesestart by denominating, that is signifying or picking up a referent. Then thereis a process of abstracting a relation between referents. This abstracting isactually the result of perceiving of relations of contiguity between phenom-ena and things. Understanding is the result of the perception of these rela-tions. The metonymic process is then concluded by the process of callingthat thing by a name that is not its common name but one with which it isassociated. Perceived as such then metonymy will be a figure of speech inwhich there is a process of abstracting a relation of proximity between twowords to the extent that one will be used in place of another.

If we move to Cicero proper we find that he begins his discussion ofmetonymy by emphasising its importance and value in the stylistic embell-ishment of the oration. He states:

This [metonymy] is a valuable stylistic ornament; but care must be takento avoid obscurity – and in fact it is usually the way in which what arecalled riddles are constructed; but this mode does not turn on a singleword but consists in the general style, that is, in a series of words. (Cicero1967:III, 167)

It is interesting to note that Cicero’s definition of metonymy swings betweenconsidering metonymy as a trope pertaining to ‘general style’, as Caplantranslates the Latin term ‘oratio’, or to ‘verbal succession and rhetoricalperiod’, which is Rosiene’s translation of the term, and the consideration of

16 Textual Metonymy

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 16

metonymy as a trope pertaining to the word. The first part of Cicero’sdefinition conveys the impression that for him metonymy is a periodicaltrope, as he relates its processes to the series of words rather than to a singleword. However, as one goes on in his definition one arrives at the conclusionthat Cicero reduces the scope and the tropical power of the trope to the levelof the word. In fact, this last reductionist view is what Cicero adheres to informulating his last conception of the trope, when he maintains thatmetonymy occurs as ‘for the sake of ornament one proper name is substitutedfor another’ (p. 167). The definition of metonymy in Cicero is both lexicallybased and ornamentally motivated. This shows clearly how metonymy wasseverely reduced to merely a substitution of names for aesthetic purposes.

The fact that Cicero’s definition lacks the concept of ‘closeness’ whichappears in the Auctor’s definition is an indication of a remarkable change inthe rhetorical tradition at this early stage of the classical Latin period. Thisdoes not suggest that rhetorical theory in this period was not progressing inan evolutionary manner. Rather, the development was individual or per-sonal. I think the Auctor’s concept of ‘closely akin or associated’ is a signif-icant development of a conception of metonymy away from the influenceand bias of metaphor. However, there is a notable regression to the Greekconception of metaphor being the trope of tropes when the notion of ‘close-ness’ is cancelled from the definition of both Cicero and Quintilian and thetrope is reduced to a mere substitution. This reduction makes the troperather vague and difficult to distinguish from other tropes, particularly frommetaphor which also involves substitution.

Now I turn to the definition of metonymy provided by Quintilian, whodefines metonymy as a trope which ‘consists in the substitution of one namefor another, and as Cicero tells us, is called hypallage by the rhetoricians’(Quintilian 1979:VIII, vi, 23). This definition is nothing but another versionof Cicero’s definition we discussed above except that the word ‘proper’ ismissing in Quintilian’s definition. Perhaps this is a significant late antiquechange in the way tropes are viewed in isolation rather than in relation toone other. That is to say that the tendency of classical Latin rhetoricians to statethat in metonymy we encounter a movement from a proper name to anotherwas mainly an attempt to define metonymy in relation to metaphor, whichconstituted a movement from a proper name to an improper one. It shouldfurther be noted that the notion of ‘proximity’ is missing from Quintilian’sdefinition as it is also from Cicero’s.

When we move towards the late antique period we find that the numberof orators and rhetoricians who treated rhetorical tropes in a practical man-ner was increasing because rhetoric was increasingly becoming a subject inthe school curriculum. This meant that the rhetorical treatises of these peri-ods were in large measure practical accounts of how to improve writing skilland how to sharpen critical ability. In the late antique period we first meetSacerdos of the third century, whose definition of metonymy is suggestive


1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 17

of a remarkable departure from previous definitions. This departure isembodied in the fact that the definition does not include the traditionalterms like ‘words’ or ‘name’. Instead we find the term ‘signification’ used.This is a very interesting point because it stresses the issue of the relation-ship between figurative theory and signification, which is the issue to beraised in Chapter 3 of this book.

Sacerdos defines metonymy as ‘[s]peech descending from a proper signifi-cation to a proper [one] through an interpretation of proximity’ (see Rosiene1992:210). In Sacerdos’ definition we notice a return to the notion of ‘prox-imity’, which I think is both a sign of maturation in tropical understandingand also an indication of a continuation from classical traditions. Theaccount is illuminating in the sense that the two prime elements of the cog-nitive processes involved in metonymic processing are included in this defin-ition. These are ‘signification’ and ‘proximity’. In essence metonymic processesare semiotic because there is a use of a linguistic sign which has a meaningother than itself. The notion of proximity is also important because this iswhat makes this signification resolvable and makes it cognitively accessibleand easily retrievable.

After Sacerdos we begin a tradition of naming rather than defining thetrope. In other words, rhetoricians became interested in how to name thetrope or what to call it. This was a real problem for them because the termsused oscillated between Greek metonymia of the grammarians and hypallageof the rhetoricians as Cicero tells us, on the one hand, and between the Latindenominatio, transnominatio and transmutatio, on the other. Donatus, forexample, defines the trope as ‘diction one might call “transnomination” asit were’ (p. 239). So it is the ‘naming’ that is sought here and the hesitationis indicative of a rather confused account of the trope.

For Charisius, metonymy is ‘diction transferred from some significationto another proximity’ (p. 219). There is hope in this definition of a progres-sive account of metonymy. The definition seems compatible with classicaldefinitions and accounts for both the process of signification and the rela-tion of proximity. This is reminiscent of Sacerdos’ account of the trope.However, Diomedes’ definition brings us back to the concept of propriety inthe relation between elements in a tropical domain while at the same timeit lacks the notion of proximity. He defines metonymy as ‘diction translatedfrom some proper signification to another proper [one]’ (p. 226).

Towards the end of late antiquity we witness a return to the reductionistview of metonymy in which the nature of the trope is not given real impor-tance and is only given a name. Metonymy for Pompeius, for example, is‘what one might call denominatio’ (p. 240). Without the hesitationwitnessed in Donatus’ ‘as it were’, Pompeius can be seen as a starting of atradition in which rhetoricians will try to simply give the name ‘denomi-natio’ to the trope without any hesitation. Cassiodorus describes metonymyas the process in which ‘we indicate the meaning of a thing in diverse ways

18 Textual Metonymy

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 18

by means of alien translated words’ (p. 218). Cassiodorus maintains that ‘thetrope is called metonymia in Greek, transnominatio in Latin’ (p. 241). Onething significant about Cassiodorus’ definition is that it shifts the descrip-tion from names or words to ‘things’. This is important for the semiotictreatment of the trope since it tells us that the substitution can actually takeplace between things also, not only names or words.

Isidore defines metonymy as ‘metonymia, transnominatio, the translation[of a name?] from some signification to another proximity’ (p. 226). Julian’smetonymy is ‘what one might call “transnomination” as it were’ (p. 239). Bedegives the same definition for metonymy as ‘what one might call “transnomi-nation” as it were’. Isidore Junior defines the trope as ‘metonymia is transnom-inatio as it were’ (p. 244).

Gervase of Melkley describes the trope as follows: ‘metonymia is the sameas transnominatio. For meta [is] trans, onoma nomen’ (p. 242). Matthew ofVendôme states that ‘metonymia is transmutatio’ (p. 244). Geoffrey ofVinsauf argues that ‘among the figures is a certain figure called metonymiawhen container is posed for contained’ (p. 248).

John of Garland uses the name ‘denominatio’ with description to denotea process ‘when one draws a relation from neighbourly things, by means ofwhich a thing may be able to be understood that was not called by its com-mon name’ (p. 250).

To conclude this section it should be pointed out that my hypothesis out-lined in the introduction to this chapter, that metonymy in western rhetor-ical scholarship suffers from theoretical reductionism, prima facie, is notvalid. This is because the investigation shows that western rhetoriciansacknowledged that metonymy is not only a substitution of words, but also aprocess of substituting things for things. The treatment of metonymy as a process of substituting a word for a word is evident in Cicero’s treatmentof the trope, in Quintilian’s treatment, in Isidore of Seville and in Diomedes.The treatment of metonymy as a process of substitution which involvesthings is evident in the Auctor, Cassiodorus and John of Garland. The treat-ment of metonymy as a process of signification is evident in Sacerdos,Charisius, Diomedes and Isidore of Seville. All other definitions of the tropeby the scholars discussed above are naming definitions.

However, on a deeper level the brief history provided above shows thatalthough these definitions take metonymy beyond its linguistic boundariesand include other dimensions like ontology, these treatments neglect thecognitive basis of metonymy and are not integrated into a coherent model.Rather, they are scattered and random attempts, which therefore appear atone time and disappear at another. Some scholars treat metonymy as sub-stitution between words, others as between things and others as involvingsignification. These treatments were never integrated into a coherent modeland the history is not progressive in the sense that once a step is achievedthe next generation takes the discussion from there and develops it further.


1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 19

Rather, we have several examples of some progress, which was then followedby a total fallback onto a narrow view. In short, the history of metonymycontinued to oscillate between reduced and abrupt definitions and return-ing back to classical definition on the one hand, or resorting to a method ofgiving names rather than definitions, as we have seen towards the latemedieval period on the other. At other times the treatments of metonymyinvolve nothing but giving examples of metonymy.

Besides, the treatments were all carried out under the assumption thatmetonymy is a poetic device which conveys some sort of deviation fromnormal modes of language structure to provide some charm and grandeurto the style. Furthermore, the treatments were all confined to the lexicallevel and no treatment attempted to discuss the role of the trope in pro-viding connectivity to text. It should be illuminating to provide a briefhistory of metonymy in the Arabic rhetorical tradition to see how it wastreated here and whether there were similarities or differences between theArab and western accounts of the trope; more importantly, to see whethermy hypothesis will still hold as far as the Arabic treatment of metonymyis concerned. It is to this aspect that I turn now in the last section of thischapter.

1.4 Metonymy in Arabic rhetoric

The title of this section is somewhat strange because metonymy is a westernconcept and there is some sort of imposition of this notion on Arabic rhetori-cal scholarship. However, in Arabic–English dictionaries there is a constanttendency to translate the word kinäyah as metonymy. This section aims toshow that metonymy as a substitution of one word for another with whichit is related is wider than this dictionary rendering and involves more thanwhat is conventionally regarded as kinäyah. I provide Figure 1.1 to illustratethe classification of figuration in Arabic rhetoric which shows the place ofkinäyah in relation to the overall figurative space in Arabic rhetoric.

Figure 1.1 shows that in Arabic rhetorical theory there are two major typesof transference. The first is linguistic transference and the second is cognitivetransference. Within linguistic transference we have borrowing, covering andother loose relations. I am not interested in transference based on mushäbahah‘similarity’ relations, represented in the diagram by the Arabic term isti‘ärah,that is borrowing. I am interested in three types of transference which I believe are related or at least can be related to build up a model of non-similarity transference. The first of these is the transference based on cogni-tive relations. In Figure 1.1 this is termed majäz ‘aqlï ‘cognitive transference’.The second is the type of linguistic transference which is based onmuläzamah ‘contiguity’ relations, which in Figure 1.1 is termed kinäyah. Thethird type of majäz I am interested in is the type of transference which isbased on loose relations and in Figure 1.1 is termed majäz mursal.

20 Textual Metonymy

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 20

The division of transference into two major categories is a remarkablysignificant attempt by Arab rhetoricians to view transference not only as a linguistic phenomenon but also as a cognitive phenomenon. This takesthe discussions of Arab rhetoricians of transference far beyond formal sub-stitution and considers the whole process in its wider cognitive as well aslinguistic context. I therefore argue that the term ‘metonymy’ involves thefollowing three types of transference discussed in Arabic rhetoric:7

(i) Majäz ‘aqlï, that is cognitive or rational transference. An example of thistype of transference would be ‘The king built the Great Palace’. The kindof transference involved here is not a transfer of word meaning but thetransference is in the attribution of the action of building the palace tosomeone who is not the real doer of the action and this is why it iscalled cognitive transference or figurative predication in which theaction is predicated to some entity that is not the real doer of the actionbut is contiguously related to it.

(ii) Majäz mursal, that is transference based on part–whole, whole–part rela-tions and other relations of adjacency and existentiality. An example ofthis type of transference would be ‘Don’t sit on that chair, it is broken.’Here the whole chair stands for one part of it. In actuality it is not thewhole chair that is broken but part of it is broken, most commonly oneor more of its legs.

(iii) Kinäyah, that is transference based on substitution of names on thebasis of a relation between them. Here the social motive of euphemismis paramount and it is the directive force behind the use of this type oftransference. An example of this would be the Qur’anic euphemistic


majäz (transference)

majäz lughawi majäz ‘aqlï

linguistic transference cognitive transference

majäz al-mushäbahah majäz al-muläzamah majäz al-‘alä’iq al-’ukhrä

isti ‘ärah kinäyah majäz mursal

(borrowing) (covering) (loose transference)

Similarity Contiguity Other relations

Figure 1.1 Classification of types of transference in Arabic rhetoric

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 21

expression ‘if you touch women’ to mean what is more than touching,namely sexual intercourse.

The logic behind this classification is that while the first and third of thesepertain to the worlds of concepts and language respectively, in view of thefact that they are known in Arabic rhetoric as ‘cognitive’ and ‘linguistic’respectively, the second pertains to the world of things and objects, in viewof the fact that the relations underlying it are essentially existential relationssuch as adjacency, positionality and contingency. With this I conceptualisemetonymy in Arabic rhetoric as having three dimensions: the cognitivedimension represented by majäz ‘aqlï, the linguistic dimension representedby kinäyah and the ontological dimension represented by majäz mursal.8

This is the semiotic approach under which I intend to deal with the tropeand develop a textual model in which the three levels of representationsinteract to provide unity and connectedness to text in its three realms.Figure 1.2 shows the three-dimensional transference I am calling contiguoustransference in Arabic rhetoric.

In the following paragraphs, I shall present the definitions provided forthese three types of contiguous transference by Arab rhetoricians.

1.4.1 Kinäyah

We find the first reference to kinäyah by Abü ‘Ubaydah (d. 833/210) in hisbook Majäz al-Qur’an. He refers to it as a linguistic phenomenon and usesthe grammarians’ conception of it when he considers it the omission of thenoun and its substitution with a pronoun. In other words, Abü ‘Ubaydahtreats kinäyah as ‘implicitness’ in linguistic reference where only a pronounis mentioned without any nominal antecedent before it or any resolvablereference after it. The examples he uses include the following verses fromthe Qur’an:

(7) Kullu man ‘alayhä fän. (55:26)‘All that is on it will perish.’

(8) Hattä tawärat bi al-hijäb. (38:32)‘Until it was hidden in the veil of night.’

22 Textual Metonymy

majäz ’irtibätï (contiguous transference)

kinäyah majäz ‘aqlï majäz mursal

implicitness rational transference loose transference

linguistic transference cognitive transference ontological transference

Figure 1.2 The three dimensions of contiguous transference in Arabic rhetoric

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 22

(9) Falawlä ’idhä balaghati l-hulqüm. (56:83)‘Yea when it reaches to the collar-bone.’

Then Abu ‘Ubaydah comments that the pronoun highlighted in (7) means‘the earth’ and in (8) means ‘the sun’ and in (9) means ‘the soul’. This isa subtle explanation of a linguistic phenomenon known nowadays asexophoric reference, where there is no noun in the text and yet a pronounis used to be interpreted by means of context because the reference is out-side the text. The referent might be construed in the immediate context ofsituation or it might be a salient feature of the universe of discourse. Whatis interesting in Abü ‘Ubaydah’s treatment is that he considers kinäyah anexample of implicitness or covering a linguistic item. This treatment,although primitive, constitutes the first attempt to account for this linguis-tic phenomenon by giving it a rhetorical explanation.

Al-Jähiz also alludes to kinäyah in his book al-Bayän wa al-Tabyïn when hereviews the definitions of rhetoric in other cultures. He presents the defini-tion of rhetoric in the Indian tradition as follows: ‘rhetoric is insight in proofand the awareness of when to exploit chances of argument’. Al-Jähiz thencomments that the indication of the awareness of these two principles inrhetoric is to avoid explicitness and resort to implicitness if explicitnesswould be more difficult for the audience (Al-Jähiz 1948:I, 88). According toal-Jähiz, the sign of rhetoric is the use of implicitness9 where it is appropri-ate and the use of explicitness where it is fitting. This, of course, correspondsto the definition of rhetoric in the Arabic culture as the appropriateness ofspeech to the context of speech. If the context requires a grand style then itis mandatory to use grand style, and if it requires a middle or a low stylethen it is considered foolishness to use grand style in such a context.

Al-Mubarrad (d. 898/285) in his book al-Kämil, instead of defining kinäyah,identifies a number of functions of kinäyah. It should be noted however thatthe treatment of kinäyah by al-Mubarrad is entirely linguistic; this is not sur-prising, because of his linguistic and particularly grammatical backgroundand orientation. In fact, most rhetorical remarks were first identified bygrammarians trying to explain the various styles of Arabic sentence con-struction. It is important to note here that al-Mubarrad and other grammar-ians did not do rhetorical study any significant benefit and it was only withthe work of rhetoricians like al-Jurjäni, al-‘Askari, Ibn al-Athïr, al-Sakkäki andothers that rhetoric began to emerge as an art separate from grammar,although it continued to exhibit a strong grammatical affiliation. In additionto the discussion of the linguistic aspect of kinäyah, Al-Mubarrad discussesthe social aspect of kinäyah, which is not relevant to the course of this book.10

I shall turn now to a well-known book in Arabic rhetoric. This is Kitäb al-Badï ‘ by Ibn al-Mu’tazz, who attempted to search for the origins of badï’ inclassical sources of the Arabic language, that is in the classical poetry of the pre-Islamic period and the very early Islamic poetry as well as in the Qur’an and


1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 23

in the Traditions of the Prophet. His aim was to prove that the innovative styleswere not actually new and that they had their origins in classical Arabic poetryand in the Qur’an. Ibn al-Mu‘tazz seems to regard kinäyah and ta‘rïdh ‘allusion’as the same phenomenon. Other Arab rhetoricians, on the other hand, regardthe former as an example of indirect meaning interpreted by means of anattribute or a word semantically related to the omitted word, while the lattercorresponds to a phenomenon whereby the indirect meaning is interpreted bymeans of context.11

The next important author is Qudämah ibn Ja‘far (d. 945/337), one of theleading literary critics of the tenth century. He discussed kinäyah under theheading ’irdäf, which could roughly be translated into ‘contiguity or adja-cency’. He defined it as ‘the process in which the poet intends a meaning butdoes not give a word conveying that meaning. Rather he gives a word that con-veys a meaning contiguous or adjacent to that meaning’ (Qudämah 1958:113).

We can see from this definition that Qudämah considers kinäyah and ’irdäfas the same phenomenon. The treatment of kinäyah and ’irdäf as one andthe same trope has been used by rhetoricians and critics other thanQudämah. For example Khafäji holds the same view and gives al-Buhturi’sline in which he describes the way he killed a wolf as an example (see for a further discussion Kanazi 1989:156). The line runs:

(10) Fa’atba‘tuhä ’ukhrä fa’adhlaltu naslahä bihaythu yakünu al-lubbu wa al-ru‘bu wa al-hiqdu.‘I stabbed it again and made the spearhead disappear where reason,fear and malice reside.’

Kanazi argues that according to Khafäji there is a use of kinäyah in the linein the words ‘where reason, fear and malice reside’. The poet intended to saythe heart but instead of this word he used words for feelings that usuallyoccupy the heart so he used a container–content relationship in thiskinäyah. Then Qudämah gives a line of poetry that describes a woman:

(11) Ba‘ïdatu mahwa al-qurti ’immä linawfalin ‘abühä wa’imma ‘abdu shamsin wa häshimi.‘Her earring is long, her father is either of the Nawfal family, the AbdShams family or the Hashim family.’

Qudämah maintains that the earring being described as long is a kinäyahbecause the intended meaning is that the woman’s neck is long and thisaccording to the Arabs is a sign of beauty in a woman. The semantic adjacencyor the semantic entailment of the expression stems from the fact that if thewoman’s earring ornament is long then this entails that she has a long neck.

Abü Hiläl al-‘Askari, who was both a rhetorician and a literary critic, is saidto have transformed baläghah ‘rhetoric’ into criticism (see Matlüb 1978:140).

24 Textual Metonymy

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 24

He is a proponent of the scholastic school of al-Jähiz and Bishr ibn al-Mu‘tamir and others. This can be understood from his pursuit of the sig-nificance of form in the rhetorical organisation of a piece of speech, be ita poem or a piece of prose. In this sense, he was a supporter of al-Jähiz’s for-mal school (for details see Tabbänah 1981, who discusses al-‘Askari’s theoryof form in a whole chapter (127–50). Therefore, we find that al-‘Askari’s dis-cussion of kinäyah is included under the part of rhetoric known as badï’‘embellishment’, which is basically the adornment of form. In his discussionof kinäyah al-‘Askari treats kinäyah and ta‘rïdh as the same phenomenon. Hediscusses them both under one heading entitled al-kinäyah wa al-ta‘rïdh‘metonymy and allusion’. The definition he provides for kinäyah and ta‘rïdhis ‘the manner in which meanings are expressed in a suggestive way ratherthan a declarative way’ (Al-‘Askari 1952:360). He gives two examples illus-trating kinäyah and ta‘rïdh. Both examples are Qur’anic verses:

(12) ’aw jä’a ’ahadun minkum min al-ghä’iti ’aw lämastumu al-nisä’.(4:43)‘If you come from the privy, or you have touched women.’

(13) wa furushin marfü‘ah. (56:34)‘And on couches raised high.’

In (12), there are two examples of kinäyah where the word ‘privy’ is used togive the meaning of ‘opening the bowels’ and the word ‘touch’ is used todenote an extended meaning beyond touching, namely ‘sexual intercourse’.Both metonymic examples are actually euphemistic examples of metonymyto express taboo meanings. According to al-‘Askari, the word ‘couches’ inexample (13) is used to suggest women, as this type of interpretation iswarranted by following Qur’anic discourse.

Among the literary critics who discussed kinäyah in their critical works isIbn Rashïq al-Qayrawäni (d. 1063/456), who wrote a book entitled al-‘Umdahfi Sinä‘at al-Shi‘r wa Naqdihi. Al-Qayrawäni included a chapter on the rhetori-cal aspect he called tatbï ‘. Looking at the examples he provided, one reachesthe conclusion that he meant by tatbï ‘ the concept of kinäyah. Moreover, al-Qayrawäni’s tatbï‘ corresponds to Qudämah’s ’irdäf. Al-Qayrawäni definestatbï ‘ as a type of signalling and he points out that some people call it trans-ference in which the poet intends to mention something but omits that thingitself and mentions something strongly associated with it (Al-Qayrawäni1907:215). The first example al-Qayrawäni gives to illustrate tatbï ‘ is ’Imru’ual-Qays’ line:

(14) Yabïtu futätu al-miski fawqa firäshihäna’ümu al-dhuhä lam tantatiq min tafadhdhuli.

‘The bits of musk remain in her bed till the morning. A forenoonsleeper who doesn’t wear work clothes.’


1403_913323_04_cha01.qxd 11/21/03 6:34 PM Page 25

In this example al-Qayrawäni argues that kinäyah is in the phrase ‘forenoonsleeper’ by which the poet intended to convey a meaning that the womanis a prosperous one who does not get up early in the morning to bring thenecessary things like wood for cooking because she has servants doing suchthings for her. Another level of reasoning comes into action once we knowthat she is a prosperous woman and that she has servants doing all thehousework for her. This is that she is a soft woman because she does not hurther body with work. Of course, softness is regarded as a sign of beauty andthis is why the poet has described her as such.

In his book Dalä’il al-’i‘jäz, al-Jurjäni defines kinäyah as

the situation in which the speaker wants to convey a meaning which he doesnot convey via the conventional word in the language. Rather he goes to a meaning which is adjacent and contiguous to that which he wants to con-vey; so with this he indicates the conventional meaning. (al-Jurjäni 1946:52)

The examples al-Jurjäni gives to illustrate his definition include

(15) Tawïl al-nijäd.‘Having a long sword.’

(16) Na’üm al-dhuhä.‘A forenoon sleeper.’

(17) Kathïru ramädi al-qidr.‘Having a lot of ash under his cauldron.’

In example (15) we have the expression ‘long sword’, giving the meaningthat the man is tall. Because if the sword is long this entails that the mancarrying that sword is tall. This example seems to exploit a cause–effectmetonymic relationship in which the fact that the sword the man wears islong is because the man himself is tall. This example is similar to the womanhaving a long earring. The fact that the earring is long means that the neckof the woman is long. In example (16) of the woman who sleeps in theforenoon, the meaning intended is that she is a prosperous woman becauseaccording to the life patterns of the Arabs, women get up early in the morn-ing and spend the whole morning working to earn their living. This womansleeps in the forenoon because she has servants doing the job for her. Thismeans that ’Imru’u al-Qays’ lover is beautiful and soft because she does notdo any work. In example (17) the more ash found under one’s pot the moregenerous one is, simply because this means that one cooks a lot, which inturn means that one feeds a lot of guests. Therefore, kathïr al-ramäd ‘he hasa lot of ash’ is a sign that is associated with generosity.

In the three examples given by al-Jurjäni we find a common feature sharedby all of them which is a distinctive characteristic of kinäyah. This is the strongassociation between the term substituted and the term substituted for. Thedefinitions above provide a ground to claim that a principal feature of kinäyah

26 Textual Metonymy

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 26

is the meaning associations in which one meaning calls or conjures otherrelated meanings either from the immediate context of situation as example(15) probably shows or from the broader context of culture like that depictedin examples (16) and (17).

Al-Sakkäki gives a quite different definition for kinäyah in which he statesthat kinäyah is ‘to leave pronouncement of a term and resort to mentioningwhat that term entails to transfer the interpretation from the mentioneditem to the left item’ (Al-Sakkäki 1937:189). What is interesting in the treat-ment of al-Sakkäki is his concentration on the notion of entailment: läzim‘the entailing’ and malzüm ‘the entailed’ in the definition of kinäyah. Thisreminds us of the logical background of the man as one of the leaders of thescholastic school of rhetoric. This school focused on the application of logicto the study of rhetoric and their contribution was in the classification andcategorisation of figurative notions and rhetorical concepts. Al-Sakkäki thengives a number of frequently cited examples of kinäyah like that of thewoman who sleeps in the forenoon, of the man who has a long sword andof the man who has a lot of ash under his pot.

Al-Qazwïni in his ’ïdhäh defines kinäyah as ‘the term used to entail some-thing semantically concomitant with it, with the possibility of intendingthe literal meaning of this term’ (n.d.:183).

The fact that the use of the expression tawïlu al-nijäd ‘long sword’ to meanthe tallness of the body of the carrier of this sword is a type of kinäyah is truein view of the logical inference that is motivated by the expression ‘longswords’ which suggests that the people wearing these long swords must betall. This is the metonymic use because we do not mean the literal meaningof the words ‘long sword’ but we go beyond it to mean its semanticallyrelated concept of tallness. However, the speaker may also intend the literalmeaning of the words ‘long sword’, and this according to al-Qazwïni is thedifference between kinäyah and majäz ‘metaphor’. In metaphor only thetransferred meaning is intended because when you say that Zaid is a lionyou cannot mean that Zaid is the animal in reality. In kinäyah, by contrast,you can mean the literal meaning of the term also. So you can use theexpression ‘He is the man who has a long sword’ to mean that he has a longsword and nothing more.

Ibn al-Athïr (d. 1237/637) defines kinäyah as ‘the word which refers to a meaning that can be taken on both planes, that is the literal and the fig-urative by means of a general term relating the two types of meaning’ (seeAtïq (n.d.:44)).

Ibn al-Athïr gives the example of the word na‘jatan or ‘ewe’ which is usedin the Qur’anic verse given here as (18)

(18) ’inna hädha ’akhï lahu tis‘un wa tis’ün na‘jatan wa liya na‘jatunwähidatun. (38:23)‘This man, who is my brother, has ninety nine ewes and I have onlyone ewe.’


1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 27

Ibn al-Athïr argues that the word na‘jatan ‘a ewe’ in this verse is a metonymicexpression for woman because both ‘woman’ and na‘jah ‘ewe’ agree in onesemantic feature which is that they are both ‘� feminine’. In this case IbnAl-Athïr substantiates his theory of kinäyah as pertaining to the two levels ofsignification; the literal as well as the figurative because the word na‘jah ‘ewe’could be interpreted as ‘woman’ and could also be interpreted as na‘jah ‘ewe’and nothing else.12

Al-‘Alawi in his treatise At-Tiräz gives a three-level definition of kinäyah.Two of these are literal and the third is figurative. The first linguistic meaningof kinäyah is the root kanaya or kanawa,13 which means ‘titling’. The Arabs areused to addressing one another with titles as a sign of respect. For examplethey would call someone Abu Ali (the father of Ali) if he has a son called Ali.The second linguistic meaning is derived from the meaning of ‘covering’. Thisexplains why the third meaning, which is the rhetorical meaning, is also, tosome extent, based on this second definition. In this case, however, thecovering is figurative rather than physical.

Next, al-‘Alawi reviews a set of definitions provided by rhetoricians beforehim, such as al-Jurjäni, al-Mäliki and Ibn al-Athïr. Al-‘Alawi seems to rejectthe definitions provided by these rhetoricians. He attempts a refutation ofthese definitions, which he sees as containing some faulty expressions ofkinäyah. After he refutes the definitions of kinäyah provided by these schol-ars, he provides his own definition in which he states that kinäyah is ‘theterm referring to two different meanings, literal and figurative, without a medium and not in a declarative way’ (Al-‘Alawi 1914:373). Having posedthis definition, al-‘Alawi then explains the elements of his definition. Eachelement of the definition is explained and narrowed down to eliminate allother semantic possibilities of the element. This is a logical method in whichprecision in definition is sought. The use of the word ‘referring’ in the defi-nition is significant because this distinguishes kinäyah from ta‘rïdh(‘allusion’) which is understood by contextual interpretation rather than bysemantic reference. The reason why he uses ‘two meanings’, al-‘Alawi argues,is to differentiate kinäyah from other words that have one meaning, like‘man’ and ‘horse’. The use of the term ‘different’ in the definition is to dis-tinguish kinäyah from synonymy. The use of the terms ‘literal and figurative’is to distinguish kinäyah from simple polysemy, where words have more thanone meaning but all of them are literal. The term ‘without a medium’ is todistinguish kinäyah from simile, which is expressed via a linguistic medium,that is the particle ‘like’ or ‘as’. The term ‘not in a declarative way’ tells usthat kinäyah is indirect and that it is metaphor, which is expressed via directdeclaration of the term.

1.4.2 Majäz ‘aqlï

This is what has been translated earlier as ‘cognitive transference’. It is thetype of figuration in which the verb or the action is predicated to an agent

28 Textual Metonymy

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 28

other than the real doer. In western rhetorical traditions (discussed above insection 1.3) this type of figuration is discussed under the general heading ofmetonymy. In this section I shall review the set of definitions provided byArab rhetoricians for this type of figuration. In the discussion of this type offigurative use attention will be paid to the insights we can extract from thesedefinitions to incorporate into a combined approach to the way metonymyis viewed and how it functions in texts.

In his book al-Sinä‘atayn, Abü Hiläl al-‘Askari is mainly interested in the discussion of majäz in its general sense. That is to say he concentrateson the discussion of ‘isti‘ärah ‘metaphor’ and provides numerous examplesfor this type of majäz in the Qur’an and in the speech of the Arabs whetherpoetry or prose. It seems that at that early stage of Arabic rhetoric the con-cept of majäz was not yet developed into a fully fledged theory asexpounded in the writings of al-Jurjäni or al-Sakkäki, who came later,towards the fifth century and onwards. Therefore, the discussion of majäz‘aqlï is lacking in earlier treatises.

In his book Asrär al-baläghah al-Jurjäni discusses majäz ‘aqlï, including thedistinction between majäz ‘aqlï and majäz lughawi. The discussion is detailedand sophisticated. It strongly suggests the importance of the distinctionbetween haqïqah and majäz because of the religious background this dis-tinction came to represent. In fact, the differences that cropped up betweenthe various Muslim groups involved in the scholastic debates known as ‘ilmal-kaläm was basically the issue of interpreting the verses which exhibit fig-urative use of the language. The main thrust of al-Jurjäni’s discussion ofmajäz is the distinction between three elements in the sentential semanticstructure. The first element is that of musnad ’ilayh ‘subject’, the second isthat of musnad ‘predicate’ and the third element is the relationship of ’isnäd‘predication’. On the basis of this distinction al-Jurjäni proposes his theoryof majäz. He maintains (Al-Jurjäni 1954:345–60) that if the transfer is in theword that is functioning as predicate then the type of figuration is lughawï‘linguistic’ because it is going to be a play on the meaning of the word – aquality that is provided by the language as a conventional system. However,if the transfer is in the predication then the type of figuration is ‘aqlï ‘cog-nitive’ because the action will be attributed to some entity other than thereal doer – a phenomenon that can be interpreted only by means of acognitive judgement that there is some kind of relationship between the realdoer of the action and the subject used. This relationship has to be arelationship other than similarity or else the figuration will be that ofmetaphor.

One very interesting point al-Jurjäni puts forward in his discussion is thatof ‘intentionality’ in majäz ‘aqlï. He states that the use of the deviated formin the construction of majäz ‘aqlï has to be intentional. This is to say thatthe user has to opt for this deviated form to achieve aesthetic effect, and thathe should not believe that the subject used is the real doer of the action. He


1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 29

illustrates this from the Qur’an, including the verse in which God speaks onbehalf of the nonbelievers who say:

(19) wa mä yuhlikunä ’illa al-dahr. (45:24)‘And nothing but time destroys us.’

Here we cannot say that these nonbelievers were opting to use majäz ‘aqlïwhen they attributed the action of destroying to time instead of God. Thereason for this is that they really think that it is time that destroys them andthey said this without the intention of using language figuratively in thisparticular example. The definition that al-Jurjäni provides for majäz ‘aqlï is‘any sentence whose semantic judgement is transferred from that which itnormally conveys with the purpose of figuration’ (1954:356).14

In his book Dalä’il al-’i’jäz al-Jurjäni discusses majäz, which he thinks isvital to the understanding of how language is used in real-life situations tocreate certain aesthetic effects. Al-Jurjäni defines majäz ‘aqlï as that type offiguration which does not involve a transfer in the meaning of a word buta transfer in the predication or rather in the attribution of actions to enti-ties that do not normally do them. Al-Jurjäni gives as an example the versein which God says:

(20) ’ulä’ika alladhïna ’ishtaraw al-dhalälata bil- hudä famä rabihattijäratuhum. (2:16)‘They are those who bought error for guidance so their trade did notprofit.’

Al-Jurjäni dwells on the concept of ‘profit’ being a verb predicated to ‘trade’which is obviously not capable of doing any action. This predication, al-Jurjäni (1946:227) maintains, is a good example of majäz ‘aqlï because thetransfer is not in the nature of the meaning of the verb ‘profit’ itself, but inthe fact that it is being predicated to ‘trade’ – a subject that is not animateand hence not capable of performing an action of profiting. Al-Jurjäni’sargument centres on the fact that the meaning of the word ‘profit’ is by nomeans transferred because the original meaning of it is still intact. However,it is the predication of this verb to ‘trade’ that makes the desired aestheticeffect by means of a cognitive transfer.15

Al-Sakkäki defines majäz ‘aqlï as ‘speech that is transferred from its con-ventional meaning for the purpose of creating some kind of interpretation’(Al-Sakkäki 1937:185).

Among the examples Al-Sakkäki gives are the following:

(21) yunbitu al-rabï‘u al-baql.‘The spring grows radish.’

(22) yashfï al-tabïbu al-marïdh.‘The doctor cures the patient.’

30 Textual Metonymy

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 30

(23) kasä al-khalïfatu al-ka‘bata.‘The Caliph clothed the Ka‘bah.’

(24) banä al-wazïru al-qasr.‘The minister built the palace.’

Al-Sakkäki’s discussion of majäz ‘aqlï does not differ very much from al-Jurjäni’s discussion. It seems, however, that al-Sakkäki’s discussion is moreof a logical nature, and that the rhetorical significance of majäz ‘aqlï is lostin his detailed logical discussion of the issue.

Ibn al-Athïr includes majäz ‘aqlï in his discussion of the notion of majäz.He introduces more or less the same arguments provided by rhetoriciansbefore him. However, he concentrates on majäz lughawi because it is, accord-ing to him, the core of figurative use of language since there is a transfer inthe conventional meaning of the word whereas majäz ‘aqlï involves a trans-fer in the construction or in the predication of actions to subjects other thanthe real doers of those actions. This is more or less a similar treatment tothat of al-Jurjäni.

Al-Qazwïni in his book ’ïdhäh defines majäz ‘aqlï as ‘the predication of theverb or its meaning to an agent that it is not its agent in the conventionalsense for the purpose of interpretation’ (Al-Qazwïni n.d.:21).

Al-Qazwïni engages in a detailed discussion of the notion of majäz ‘aqlï. Hediscusses the various forms of this type of figuration which he attributes tothe relationship between the verb on the one hand and the subject, object,infinitive, time, place and cause on the other. It is very interesting to notethat al-Qazwïni discusses majäz ‘aqlï as an aspect of ma‘äni ‘notions’. In thisrespect, his analysis differs from that of those who preceded him whoregarded this type of figuration as an essential aspect of bayän ‘eloquence’. Infact, al-Qazwïni draws our attention to this point at the end of his discussionof this majäz because it discusses the states of predication which is essentiallyan aspect of ma‘äni – the branch that deals with the way the construction ofsentences reflects the appropriateness of the speech to the context.

Al-‘Alawi’s treatise of At-Tiräz does not account for majäz ‘aqlï in the wayother contemporary rhetoricians do.16 He provides a detailed discussion offifteen types of majäz mufrad ‘transference in the individual words’ and heincludes some examples of majäz ‘aqlï among the examples he provides forthese types. Also he discusses this sort of majäz but under a different name,that is under what he calls majäz murakkab ‘constructional transference’. Thisis precisely what al-Jurjäni calls majäz ’isnädï ‘predication transfer’ and whatal-Sakkäki calls majäz ‘aqlï ‘cognitive transfer’, and it is this latter name thatcame to designate this type of majäz in later treatises.

1.4.3 Majäz mursal

Majäz mursal, as the name suggests, is a type of transference. Arab rhetori-cians call this type of figuration majäz mursal for two reasons. First, the


1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 31

relations underlying it are open and not confined to a specific type of rela-tion. Second, the relationships underlying it are not those of similarity (Atïqn.d.:143). There is a great deal of similarity between majäz mursal and majäz‘aqlï because they both refer to a phenomenon whereby the use of a wordor a construction is intended to convey a meaning other than its conven-tional one with the aid of a strong connection between the meaningintended and the conventional meaning of the word or the construction.The first difference between the two however is that while the first occursin individual words the second occurs in the construction. Furthermore,another vital difference between these two types of majäz is that while thefirst one is an example of majäz lughawi the second is an example of majäz‘aqlï. This means that the former utilises the linguistic possibilities of wordmeanings and builds the transfer on a kind of deviation from the norms ofthe word meaning provided by the language.

By contrast, the latter capitalises on a deviation from cognitive processesthat are the outcome of transfer of predication. While the words themselvesare used in their conventional sense, it is the construction or the predica-tion that deviates from normal language construction in that verbs oractions are attributed to subjects other than their real doers. This might beseen as ‘personification’ in the western sense, which according to Wales is‘a figure of speech or trope in which an inanimate object, animate non-human, or abstract quality is given human attributes’ (Wales 1989:349).Wales gives as an example of personification the following lines from Gray(the personification is in bold):

(25) Here rests the head upon the lap of EarthA youth to fortune and to fame unknown.Fair Science frowned not on his humble birth,And Melancholy marked him for her own.

(Elegy Written in a Country Churchyard)

In the example above, the verb ‘frowned’ is attributed to an inanimateabstract noun ‘science’ which is unable to do any sort of frowning. This isregarded as a major feature of poetic language which attributes a sense ofanimation and sometimes a sense of humanity to non-animate or non-human objects.

I think the difference between majäz ‘aqlï and personification is that majäz‘aqlï is a much broader concept than personification because it not onlyincludes verbs or actions but also extends to include predication, which inArabic includes all forms of noun phrases. Besides, this transfer is based onlogical connectedness and cognitive association between the subject usedand the subject omitted.

Al-Jurjäni alludes to majäz mursal although he does not give it this name.In his book Asrär al-baläghah, al-Jurjäni goes into detail about the various

32 Textual Metonymy

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 32

forms of majäz and suggests that majäz is the area where there had beendispute before him because the issue is very much related to the disputesbetween Muslim religious groups regarding the names and attributes of God.Therefore, al-Jurjäni attempts to outline a standard to which Muslims canrefer to in their debates. Al-Jurjäni begins his discussion of this figurativeaspect by introducing a definition of majäz in general. His definition runs:‘majäz is every word that is used for a meaning other than that for which itwas originally invented by virtue of a connection between the first and thesecond’ (Al-Jurjäni 1954:325).This ‘connection’ is what came to be knownin later rhetorical circles as the ‘relation’ that ties the two meanings, that isthe conventional meaning and the figurative meaning.

Al-Sakkäki discusses majäz mursal under the heading of ‘linguistic figura-tion attributed to the meaning that is devoid of any exaggeration of like-ness’. He then defines it as ‘the transfer of the meaning of the word from itsoriginal meaning with the aid of a context to another meaning for the pur-pose of establishing a form of connectedness between the two respectivemeanings’ (Al-Sakkäki 1937:200).

It is clear from al-Sakkäki’s definition that context is needed to determinethe interpretation of this figurative use. The emphasis on the establishmentof a form of connectedness between the two meanings of the two words inquestion is a characteristic of this type of majäz, and the set of relationsunderlying it illustrate this aspect in more detail.

Al-Qazwïni is considered the first Arab rhetorician to use the word mursalto describe this type of majäz lughawi (Haddärah 1989:59). In his book’ïdhäh, al-Qazwïni divides majäz according to the type of relation underly-ing each type into two types. The first he calls mursal and the second’isti‘ärah. Then, he defines majäz mursal as ‘the majäz in which the relationbetween the meaning used and the original meaning is not that of similar-ity, like the word “hand” when it means “benevolence” ’ (Al-Qazwïnin.d.:154). Rather, it is a relationship of semantic connectedness in which thehand is the instrument by which the blessing is bestowed. Therefore, onecan call this an ‘instrumental’ relationship.

Although al-‘Alawi is a contemporary of al-Qazwïni we find that al-Qazwïniin his ’ïdhäh is more systematic in his division and classification of the typesof majäz. Al-‘Alawi divides majäz into fifteen types. Some of these types fallunder majäz ‘aqlï like the examples he gives in type number four (Al-‘Alawi1914:I, 70). In this example, al-‘Alawi treats the predication in

(27) Säla Al-Wädï.‘The valley flowed.’

as a type of majäz mufrad ‘figuration in individual words’. However, this typeof predication is in fact a type of majäz ‘aqlï because the transfer involved inthis example is not in the meaning of individual words but in the predication.


1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 33

That is to say that the words ‘valley’ and ‘flowed’ as individual words areused in their literal meaning. The transfer nonetheless is in predicating theaction of flowing to the valley while the real subject is the water and not thevalley. This type of majäz is based on a spatial relationship, that is becausewater runs through the valley we can figuratively say that the valley flowsor in fact a container-for-content relationship is at work here.

To conclude this section it is essential to make the following observation.The treatment of metonymy in Arabic rhetoric is also severely reduced andrather unsystematic. Although the treatment in this section seems system-atic, it is because the model which has been imposed on the data makes the treatment generally systematic. The triadic scheme of signification that is used in this section has considered metonymy in Arabic rhetoric from theperspective of the three domains of metonymic signification. These are the domain of words, the domain of concepts and the domain of objects.The section argues that kinäyah could be the type of signification involvingwords, majäz ‘aqlï the type of signification involving concepts and majäzmursal the type of signification involving objects. The connection betweenthe first two is evident while the connection with regard to the third is plau-sible, especially if we take into consideration that the relations underlyingmajäz mursal are actually relations which conform to the realm of ontology.The relations are those of contiguity, adjacency, totality, partiality, localityand positionality.

The Arabic rhetorical tradition, it could be safely argued, therefore was nobetter than the western tradition as far as the treatment of metonymy is con-cerned. The trope in both traditions suffers from epistemological limita-tions, and from confinement to the lexical level in addition to its treatmentas a phenomenon pertaining to the language of poetry or highly elevatedstyles of diction rather than to everyday language and normal modes ofthinking.

34 Textual Metonymy

1403_913323_04_cha01.qxd 11/19/03 8:41 PM Page 34

35

Metonymy may in fact be more common than metaphor, andindeed is held by some to be the distinguishing mark of realistprose, yet it is seldom subjected to the detailed and lengthy inves-tigations that metaphor undergoes. And not only is it not widelystudied but most accounts of it are unsatisfactory.

Bredin (1984:45)

2.0 Introduction

In the previous chapter we have seen that the majority of studies ofmetonymy in classical rhetoric treated the trope as a process of substitutionmainly involving words and hence reducing the power of the trope to lin-guistic substitution. In addition, metonymy has been treated as a process ofsubstitution at the lexical level only. I have shown that this is a narrow viewin both respects: as will be demonstrated in the course of this book,metonymy is not only a process of word substitution or a process limitedonly to the lexical level. This chapter investigates the nature and function ofmetonymy in modern figurative theory. The aim is to see whether and towhat extent modern accounts have been able to capture these two importantaspects of metonymy. The word ‘modern’ in the title of this chapter refersparticularly to the work carried out in the area of metonymy from the early1950s till the present day. It was in the early 1950s that attempts began toapply structural techniques to the study of literature and figurative language.It is generally accepted that the structural analysis of figurative language hasbenefited the field of figuration in terms of systematisation and paved theway for the creation of modern rhetoric. This is by no means to discredit theworks of I.A. Richards (1936) and K. Burke (1945), which to some extent rev-olutionised the study of rhetoric and gave it fresh insight. Rather, the inter-est is precisely centred on the attempts to carry out structural analysis ofrhetorical devices as principles underlying and regulating human cognitionand communication.

2Metonymy in Modern Figurative Theory

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 35

This chapter draws from three main sources to discuss the concept ofmetonymy. The first source is the rhetorical and figurative studies carriedout recently on metonymy. The second source is those semantic studies thathave investigated metonymy as a semantic phenomenon dealing with shiftsof meanings and referents. The third source is the cognitive approach, whichenriches our understanding of metonymy as a conceptual structure and asa mode of cognition.

2.1 Theories of metonymy

Metonymy is the figure of speech which involves the replacement of oneword with another with which it is associated. This definition has beenalluded to on several occasions in the previous chapter. This section dealswith the various theories of metonymy that have begun to emerge since the1950s. It aims to provide a survey of views regarding the concept in order toassess the status of metonymy in modern figurative theory.

2.1.1 Linguistic theories of metonymy

This section reviews the theories of metonymy that have been proposed bylinguists or people working in the field of rhetorical studies. I consider bothfields – linguistics and rhetoric – to share more or less similar concernsthroughout history. Rhetoric was one of the principal disciplines for thestudy of language in use in classical times and it continued to be so throughthe Middle Ages and the Renaissance. In the post-Renaissance period rhetoricas the study of the language of argumentation declined and was reduced tothe study of a handful of figures. The term ‘rhetoric’ acquired its pejorativesense in this period as mere nonsense and an act of adorning language toconceal factual defects. However, rhetoric and linguistics nowadays workhand in hand to explore language in use and to study all forms of effectivecommunication. This is to say that there should not be any confusionbetween the two disciplines as far as this section is concerned. Moreover, itshould be stressed here that the most natural home for rhetoric in modernlinguistics would be textlinguistics or discourse analysis. This is what I thinkrhetoric is and it is what I advocate throughout the course of this book.

2.1.1.1 Jakobson

In his paper ‘Two Aspects of Language and Two Types of Aphasic Distur-bances’ Roman Jakobson (1971) attempts to bring literature under thescrutiny of linguistics to give the study of literature more precision andobjectivity. Jakobson outlined what was a very influential and controversialtheory at the same time, about the application of linguistic techniques com-bined with fundamental aspects of figurative theory to the study of litera-ture and writing in general and in fact to various other aspects of cultural

36 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 36

phenomena. His paper initiated a considerable amount of interest and drewthe attention of linguistic and literary circles to the roles of the two poles ofmetaphor and metonymy in language in the development of discourse.

Jakobson’s theory of metonymy and metaphor comes in a highly technicaltreatment of the subject of language disorders. From this specialised discus-sion which seems initially irrelevant to both linguistics and critical theory,1

Jakobson builds his theoretical constructs of ‘selection’ and ‘combination’ byrelating these two concepts to the Saussurean ones of paradigmatic and syn-tagmatic arrangements of the linguistic sign. Jakobson argues that ‘speechimplies a selection of certain linguistic entities and their combination into lin-guistic units of a higher degree of complexity’ (Jakobson 1971:72). From thishe establishes his semiotic theory of the linguistic sign involving metaphorand metonymy as two fundamental modes of signification.

The next step came naturally: to apply these two concepts to shed newlight on the traditional typology of tropical figures. Among those who havediscussed rhetorical figures Jakobson has been the most reductionist,2 reduc-ing the figures to two main ones, metaphor and metonymy. The Renaissanceperiod reduced the classical tropes into four: metaphor, metonymy, synec-doche and irony. Jakobson reduced this fourfold classification into two.Jakobson regards synecdoche as belonging to the metonymic realm, becauseboth express contiguous relations of genus to species and vice versa.

I am concerned with outlining Jakobson’s theory of metonymy in anattempt to show how his theory has benefited the subsequent study ofmetonymy. Jakobson argues that metonymy is based on contiguity whereasmetaphor is based on similarity. The question that I want to raise, and thathas been raised by a number of scholars before me (see Lodge 1977, 1988;Osterwalder 1978; Bansloben 1996), is: what does Jakobson mean by theterm ‘contiguity’? The term ‘contiguity’ seems to have caused remarkableconfusion among the linguists, cognitivists and literary critics who haveattempted to study Jakobson’s theory of tropes. Perhaps the term was alsoconfusing for Jakobson himself, since he stated that ‘metonymy, based on adifferent principle, easily defies interpretation’ (Jakobson 1971:95). Thisobservation comes after Jakobson tries to give a justification of why the lit-erature on metaphor is far larger than that on metonymy. His justificationis that metonymy being based on contiguity is more difficult to interpret.As I understand him, Jakobson wants to say that the relation underlyingmetonymy is heterogeneous rather than homogenous. In other words, theprinciple underlying metonymy does not provide the researcher with ahomogenous means to deal with it. The most plausible interpretation of thisis that ‘contiguity’ is a heterogeneous term or an umbrella term coveringmore than one principle of interpretation.

From a linguistic perspective, ‘contiguity’ can be interpreted as linguisticcontiguity, or more precisely concatenative contiguity, the most obvious

Metonymy in Modern Figurative Theory 37

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 37

aspect of which is linguistic deletion. The evidence that theorists adoptingthis approach provide is that examples like

(1) The keel ploughed the deep.

actually realise an underlying deep structure of the type in (2) below:

(2) The keel of the ship ploughed the deep sea.

The word ‘keel’ stands in contiguous relation to the word ‘ship’ and theword ‘deep’ stands in contiguous relation to the word ‘sea’ because ‘deep’ isa metonym referring to ‘sea’ through one of its attributes. In this case thetwo forms coexist in a virtual code, but because of problems related to‘redundancy’ the surface structure realises only the informatively essentialelements. From a cognitive perspective, ‘contiguity’ can be interpreted as‘cognitive contiguity’. In this case, example (1) above would be interpretedas the concept ‘keel’ is contiguous to the concept ‘ship’ and the concept‘deep’ is contiguous to the concept ‘sea’ by means of cognitive adjacency,because they coexist in some form of adjacency in a conceptual schema.

The idea that ‘contiguity’ is heterogeneous is also evident in anotherrespect. The principle of ‘similarity’ underlying metaphor is always inter-preted as a mapping between two entities pertaining to two differentdomains by means of a similarity established between a ‘tenor’ (T) and a‘vehicle’ (V)3 on the basis of a particular ‘ground’ (G). In metonymy, how-ever, it is difficult to establish such a coherent metalanguage for the simplereason that these metalinguistic terms are actually non-existent in theprocess of metonymic signification. Although it is true that we can accountfor metonymic relations by means of terminology like ‘trigger’ and ‘target’,the feature that characterises metonymy is that the connector that links thetwo terms is not based on one relation as is the case with metaphor. Rather,it is based on a multiplicity of relations. Therefore, it is essential to accountfor metonymy by means of a typology of metonymic relations or meto-nymic concepts, rather than formulas as in the case of metaphor. The list ofmetonymic relations provided by linguists and cognitivists, which is calcu-lated at 46 by Radden and Kövecses (1999), is a good example of how diverseand heterogeneous the concept of contiguity is.

One can still argue however that even if contiguity has diverse manifesta-tions, this does not necessarily mean that the interpretation of metonymy isproblematic, as Jakobson argues. The problem, probably, lies in the linguis-tic or more precisely the formal approach Jakobson is imposing on the inter-pretation of tropes. I think that had Jakobson approached the issue with acognitive orientation, his conclusion would have been different. The reasonis that cognitively speaking metonymy is in fact much easier to interpretthan metaphor because the signification of metonymy, unlike metaphor,

38 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 38

does not involve a transfer between different domains, which normallyinvolves more effort in message comprehension and processing. Rather, thesubstitution takes place between different aspects or elements of the samedomain. This feature should make interpretation easy rather than difficult. Itshould be easier to understand metonymy because one does not need tomake a conceptual shift between domains but shifts within the samedomain.

As the French linguist Gaston Esnault argues, metonymy ‘does not open upnew paths to follow as metaphorical intuition does; instead it hurries over thestages in paths that are too well-known and shortens the distances so as tofacilitate the rapid intuition of things that we already know’ (cited in Nerlichet al. 1999:362). Nerlich et al. comment on this quotation and argue thatmetonymy therefore ‘enables us to say things quicker, to shorten conceptualdistances’ (p. 363). They further assert that metonymy is ‘a universal strategyof cost-effective communication’. The plethora of studies on metonymy thatwere produced after Jakobson’s paper (see Bohn 1984:534–50) shows thatJakobson’s thesis about the polarity of metaphor and metonymy in theprogression of discourse and indeed in areas other than linguistics is in factfundamental and intellectually significant. The only problem with Jakobson’streatment is his linguistic or rather his formal approach, which led to thecategorisation of metonymy as problematic.

The argument I am putting forward here is that a linguistic approach onits own is not sufficient to account for metonymy because it will wronglyassume that metonymy is difficult to deal with by means of the traditionallyestablished terms of tenor, vehicle and ground. From this wrong assumptiona wrong conclusion would be drawn that it is difficult to interpret metonymysimply on the ground that metonymy does not offer a homogeneous meansfor the study of its various manifestations. My argument is, therefore, thatdiversity is not at all a problem for interpretation. In fact, it seems that wedo not understand things only because they are similar to one another butalso because they stand next to each other. This means that the principle ofproximity is just as powerful as that of similarity in aiding recognition.

Metonymy is realistic, it is referential and it is contextually relevant, andthis is what makes it a concept of powerful interpretive force. As soon as weapply a cognitive methodology to approach the issue we shall find thatmetonymy presents us with such remarkable interpretative power that itmust be seen as an aid to cognition and by implication to interpretation. Thefact that metonymy is inherently referential makes it more geared to the lit-eral side of the scale if there is one. In this case it is more easily interpretable.The concern of metonymy with details that belong to what semanticists call‘semantic fields’ make metonymy an inherently semantic and cognitive phe-nomenon that is not exotic to our normal conceptualisation processes.

It should be noted that Jakobson’s placement of metonymy on the axis ofcombination of linguistic signs is the major impulse behind my proposed


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 39

textual theory of the trope. This is because metonymy, being combinational,should explain the syntagmatic relations at the text level, an aspect of textgrammar that has not been resolved.

2.1.1.2 Groupe mu

‘Groupe mu’ is the nickname of a group of French linguists comprisingJacques Dubois, Francis Edeline, Jean-Marie Klinkenberg, Philippe Minguet,François Pire and Hadelin Trinton, who were interested in reviving rhetoric.They saw in it a subject that could help the study of language and commu-nication. This was a view that had gone into oblivion as a result of themedieval reductionist approaches to rhetoric which reduced the scope of thisdiscipline to the study of how to ornament and adorn the style of the mes-sage rather than how to compose, persuade and convince by means of argu-mentation. The Groupe’s theory of metonymy is developed within a generaltreatment of the fundamentals of rhetoric. After considering both metaphorand synecdoche, the Groupe discuss metonymy. The first reference theymake is to Jakobson’s theory of metonymy as opposed to metaphor.

The Groupe criticise traditional rhetoric as a pre-scientific whole (Rosiene1992:24) and attempt to apply the principles of the structuralist project toprovide some coherence and systematisation to rhetoric. The main aim of thestructuralist project as Rosiene sees it is ‘to place the “essentially taxonomic”character of classical rhetoric upon firm scientific ground’ (p. 33). TheGroupe accuse classical rhetoric of offering an unsatisfactory definition ofmetonymy on the ground that the definitions provided by classical rhetori-cians merely enumerated either the types or the examples of the trope ratherthan defining the nature or the function of the trope. Rosiene (pp. 173–99)opposes this view and argues that the Groupe’s claim is not quite rightbecause in the definitions of the classical rhetoricians one finds really usefulaccounts of metonymy. He gives as an example the definition of metonymyprovided by the Auctor4 whose definition of metonymy has been discussedin Chapter 1, section 1.3, and is reproduced here for convenience: [T]he fig-ure of speech which draws from an object closely akin or associated anexpression suggesting the object meant, but not called by its name (Cicero[the Auctor] 1954: IV, xxxii). Rosiene argues that this is a satisfactory defin-ition of metonymy because it involves four fundamental operations thatcharacterise the process of metonymic understanding, namely: (i) denomi-nation (ii) abstraction; (iii) intellection; (iv) appellation.

The Groupe take metonymy to mean the substitution of one verbalexpression for another, whenever the expressions are related to one anotherwithin a web of connotative associations. They present their theory ofmetonymy under two main headings, each of which pertains to a funda-mental principle underlying the process of metonymy. They begin theirdiscussion by borrowing the concept of ‘contiguity’ from Stephen Ullmann,who used it in his definition of metonymy as a transfer of meaning ‘based

40 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 40

on sense-contiguity’ (1951:223), and then attempt to assess the relevance ofthis concept to metonymisation. The Groupe argue that the notion of ‘con-tiguity’ might not be sufficiently helpful and as a result might not be ‘abeginning of a satisfactory theory of metonymy’ (1981:120). However, itcould be a useful notion to provide a specification of metonymy. In this casemetonymy could be easily identified and distinguished from other closelyrelated tropes, notably metaphor. It does not seem at all surprising that theGroupe are actually concerning themselves with the particularities of anddifferences between tropes, because their main thrust is to provide a com-plete system of all the tropes. However, despite their vigilance, they fall intothe trap of reductionism. In fact, they actually reduce the tropes or at leastpropose a reductionist approach to the interpretation of tropes. They useAristotle’s definition of metaphor in which he talked about species–genus,genus–species as a basis to postulate that in fact the trope of tropes is synec-doche (see for further detail Bredin 1984:46).

With reference to metaphor, the Groupe maintain that metaphor

is not a substitution of meaning, but a modification of the semanticcontent of a term. This modification is the result of the conjunctionof two basic operations: addition and suppression of semes.5 In otherwords, metaphor is the product of two synecdoches. (Groupe mu,1981:107).6

With regard to metonymy the Groupe’s view centres around the argumentthat if ‘contiguity’ is spatial, temporal and causal then there would be littledifference between metonymy and synecdoche, for in both cases there is arelation of some sort of contiguity. Leech confirms this when he commentson Webster’s Third New International Dictionary’s definition of metonymy as‘a figure of speech that consists in using the name of one thing for that ofsomething else with which it is associated’. Leech comments that this defi-nition ‘covers all rules of transference, including that of metaphor, sincesimilarity is a form of association’ (Leech 1969:152).

It seems that the Groupe are unhappy with linking the concept ofmetonymy, or perhaps any rhetorical figure, to the ontological domain sincethis makes it difficult to abstract relations and establish systems from a diversevariety of phenomena. The Groupe are also not content with the idea ofassigning the workings of metonymy to the realm of language alone. Theyhave expressed unhappiness with the term ‘thing’ which keeps recurring inmany definitions of metonymy as a transfer of reference of ‘things’. They con-sider the problem of metonymy to lie in this particular aspect of pragmaticunderstanding of the concept of metonymy. Reference to things implies arealistic and contextual account of the trope which tends to highlightthe types, species and examples of metonymy rather than derive a universalprinciple to account for the phenomenon. In other words, ontological


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 41

approaches to metonymy were, for the Groupe, reminiscent of previous unsci-entific accounts of rhetorical tropes as taxonomies of unrelated phenomena.The Groupe emphasised a twofold theory of metonymy in which both ‘thesemantic plane’ and ‘the material plane’ are accounted for. In our terminol-ogy their theory is at the same time epistemological and ontological.

The Groupe aim to place metonymy on well-defined borders from othertropes. Part of their unrest with traditional rhetoric is the confusion betweenmetaphor, metonymy and synecdoche especially after the general treatmentof Aristotle’s metaphor which in fact subsumed metonymy and synecdoche.The Groupe, therefore, cite Du Marsais’ account of the distinction betweenmetonymy and synecdoche in which he argues that in the first figure the

relation between the objects is such that the object whose name is bor-rowed remains independent of the thing whose idea it awakens and doesnot form a whole with it … while the union found between objects insynecdoche assumes that the objects form a whole, such as the whole anda part’ (Groupe mu 1981:120)

The Groupe argue that ‘within the vagueness of this formulation, we find,believe it or not, metonymy’s specific character.’ It seems that the Groupefound what they seek in terms of separating metonymy and synecdoche. Sometonymy is a relation which does not involve union between the objects,and synecdoche is a relation that does. Although this seems a promising cri-terion in terms of which metonymy and synecdoche could be distinguished,the criterion itself is certainly derived from a linguistic conception of thenature of both relations rather than a cognitive conception. Studies insemantic domains and semantic fields support the view that it is extremelydifficult to distinguish between metonymy and synecdoche. In fact, theconcept of ‘contiguity’, the relevance of which to metonymy the Groupeattempt to discuss, makes the possibility of drawing a line betweenmetonymy and synecdoche extremely difficult. The argument of the Groupeseems futile because they tend to impose a metaphorical model on the inter-pretation of metonymy when they talk about the relations between‘objects’, although we know and the Groupe know also that the relationsunderlying both metonymy and synecdoche do not involve more than oneobject or one semantic domain. When the Groupe try to distinguishmetonymy from synecdoche they argue that the relation that underliesmetonymy remains independent and does not involve union with theobjects. However, when they compare metaphor and metonymy, also to sep-arate metonymy from metaphor, they identify metonymy with ‘co-inclu-sion of semes in a whole’ on the semantic plane, and identify it with‘membership in a material whole’ on the material plane (pp. 120–1).

I find the Groupe’s position confusing and quite contradictory. Theirargument that ‘in the metonymic operation, the passage from the first

42 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 42

starting term (S) to the resulting term (R) is made via an intermediary term (I)that includes S and R in either mode �7 or mode �’8 is ample evidence thatthere is a sense of totality in the metonymic operation and that the criterionof union or totality is not a useful tool to distinguish the two commonly con-fused tropes. In fact, when the Groupe offer an example to illustrate theirpoint, they resort explicitly to the notions of unity and totality in their expla-nation of what is going on in the interpretation of the example

(3) Take up your Caesar.

spoken by a teacher to tell his students that they were to continue theirstudy of De Bello Gallico. When they come to explain the metonymic oper-ation involved in the example they point out that

the intermediary term will be the spatio-temporal totality, including thelife of the famous consul, his loves, his literary works, his wars, his times,his city. In this totality of type �, Caesar and his book are contiguous.

It is clear from this discussion that the theory of metonymy offered by theGroupe does not provide criteria which are sufficiently well defined to sep-arate this trope from synecdoche. The Groupe confuse the two tropical oper-ations of metonymy and synecdoche and instead of distinguishing the twotropes they in fact blur the view of the two tropes altogether. I think this isthe natural result of any attempt to separate the two tropes – because theyare simply inseparable, especially if looked at from a ‘levelling’9 point ofview for the purpose of developing a coherent theory of how ‘contiguity’functions in discourse. This is the theme of this book and I advocate theview that metonymy and synecdoche are inseparable, especially when theyare taken as the manifestations of contiguous relations in both cognitionand language.

The Groupe then move to discuss the notion of ‘connotation’ and its rolein the metonymic operation. Two important points can be made about theGroupe’s position in this respect. The first concerns their view that meta-phor is denotative whereas metonymy is connotative. This model fallssharply in contradiction with the model Osterwalder (1978) provides andwhich views metonymy to be denotative and metaphor connotative (p. 23).Osterwalder argues, and perhaps many structuralists would agree, thatmetaphor, being a trope based on a mapping from one level of conceptual-isation to another on the basis of a similarity established between the twolevels, is in fact connotative or brings connotative experience into play, asthe degree of similarity and the distance between the tenor and the vehiclewill be determined by the cultural values of the society in question.

Recalling Esnault’s statement that metaphor opens new paths ofmetaphorical intuition, it is plausible to see how connotative metaphor is.


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 43

It makes sense also to see how the gap that metaphor tries to create andbridge at the same time is by no means understood by means of the deno-tation of the conceptual sense of the word. In other words, metaphor pro-duces an infinite multiplicity of meanings. Metonymy, however, can be seenas a reductive operation of understanding in which the multiplicity of mean-ings is actually reduced to fit a particular context. This context may be eitherlinguistic, by means of syntagmatic contiguity, or cognitive, by means of spa-tio-temporal contiguity. Given this position, let us see why the Groupe holdthe view that metaphor is actually denotative and metonymy is connotative.

The Groupe state that

metaphor brings out denotative semes, nuclear semes, included in thedefinition of terms. Metonymy, on the other hand, brings out the con-notative semes, that is, those contiguous to the array of a larger groupingand combining to define this grouping. (1981:121)

This argument is simply built on a wrong assumption, that metaphoricalinterpretation does not call for general meanings. It is based on a myth thatwhat is involved in the similarity between the two things is in fact the crit-ical features that define each thing, or what is known in semantic studies asthe intentional features of that thing. This is obviously not true: the essenceof metaphor is to allow imagination an open space to draw fictional worldsto try to reconcile the tension created by the violation emanating from themetaphorical operation. My position, which is opposed to that of theGroupe and Osterwalder, is that there is a great deal of connotation involvedin both tropes. I agree with Osterwalder that metaphor involves connota-tion and disagree with him that metonymy is only denotative. In the sameway I agree with the Groupe that metonymy involves a great deal of con-notation and disagree with them that metaphor is only denotative. In anyact of figuration there is a connotative background and what might some-times seem denotative is nothing but an act of lexicalisation through theprocess of demetaphorisation or demetonymisation whereby the signifier–signified relationship gets naturalised to the extent that people think it isreal and mandatory while in reality it is a connotation.

The second point that needs to be commented upon is the model ofmetasememe interpretation provided by the Groupe. This model underesti-mates the figurative nature of both metaphor and synecdoche because itclaims that the interpretation of these two processes takes place in a normal,straightforward manner. Only when this straightforward and default modeof interpretation fails is a metonymic interpretation sought. It placesmetonymic understanding as the last resort, to be sought only when allother options are exhausted. The Groupe maintain that

the reader has recourse first to analytic procedures to reduce metasememes.He will first see whether the figure is synecdoche, metaphor or antiphrasis.

44 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 44

It is only when the consideration of the semic datum (nuclear and codedand therefore certain) has failed that he will seek connotative extrapola-tions that will allow him to identify metonymy. (1981:122)

The most readily attainable interpretation of the Groupe’s statement is thatmetonymic interpretation is cognitively marked and farfetched, whilemetaphoric and synecdochic understanding is cognitively unmarked andsalient. This is a strange position and runs against positions supported bysolid evidence from empirical cognitive investigation (see Lakoff 1987;Gibbs 1994; Fauconnier and Turner 2002).

2.1.1.3 Ullmann

Stephen Ullmann’s theory of metonymy is outlined in two of his semanticbooks, namely Principles of Semantics (1951) and Semantics: An Introduction tothe Science of Meaning (1962). I have alluded to Ullmann’s account of meto-nymy when I discussed the Groupe’s notion of ‘contiguity’ in the previoussection. There I mentioned that the Groupe discuss the notion ‘contiguity’which they found in Ullmann’s account of metonymy. Ullmann’s theory ofmetonymy comes within his discussion of semantic change, which heascribes to associations of similarity and associations of contiguity. Like theassociations of similarity, the associations of contiguity, Ullmann argues, canbe either between the senses or between the names. For the former type ofassociation he talks about examples like the ‘ “tongue” as the principal organof speech-activity is associated with the latter. (Resultant semantic change(metonymy): “the English tongue”)’ (Ullmann 1951:80). For the second typeof contiguity association, that is that between names, Ullmann gives theexample of the contiguous link established between the words ‘private’ and‘soldier’ to the extent that this connectedness led to a semantic change man-ifested as ‘deletion’ in occurrences like ‘a private’.

Ullmann returns to a more or less similar argument regarding the moti-vation of semantic change a few pages later where he discusses examples like‘the French “bureau” office’. In this example the motivation of the seman-tic change is based on ‘the most peculiar trait of changes of meaning … thecoexistence of old sense and new within the same synchronous system’(p. 89). In the example above the semantic change is motivated by a conti-guity association established between the office and one of the salient piecesof furniture normally available in an office. This is the ‘bureau’, which is thewriting desk. In fact, the word ‘bureau’ has undergone a double change fromburel, an old French word meaning a type of cloth used for covering desks.Then the word came to stand for the whole writing desk and finally for thewhole office, where there will usually be a desk.10

Ullmann argues that metonymy is ‘less interesting than metaphor since itdoes not discover new relations but arises between words already related toeach other’ (1962:218). However, after supporting his argument by a quota-tion from Esnault regarding the difference between both metaphor and


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 45

metonymy, Ullmann asserts that ‘if metonymy is of limited interest to thestudent of style, it is an important factor in semantic change’. This showsclearly that Ullmann’s theory of metonymy develops out of an interest insemantic change and how this change is semantically motivated. I thinkUllmann’s argument that ‘metonymy is of less interest to the student ofstyle’ is outdated and falls short of accounting for the power of metonymyin both composition and stylistic analysis. In this respect, Jakobson’s paperthat I have discussed in section 2.1.1.1 above shows how metonymy is infact a way of conceptualising and of understanding. Metonymy, Jakobsonargues, is one of the two ways in which a ‘discourse may take place’(Jakobson, 1971:90). Also, metonymy for Jakobson is one of the two modesin which ‘an individual exhibits his personal style, his verbal predilectionsand preferences’ (p. 91).

Ullmann tries to provide a classification of various metonymies based onthe various types of contiguous relations underlying each of the metonymicoccurrences. He talks of metonymies based on spatial relations. The exam-ple he gives for this type of metonymy is the shift from the Latin coxa ‘hip’to French cuisse ‘thigh’ as these two organs are two contiguous parts of ourbody and therefore the semantic shift was contiguously motivated. I thinkexamples of this semantic shift can be found in many other languages, par-ticularly those languages which exist in two different varieties,11 like Arabic,where one variety is classical (at least in origin) and the other modern. Itshould be easy to trace the meaning change from the classical to the mod-ern varieties and vice versa. My guess is that a large proportion of sensechange that has taken place between the two varieties is metonymicallymotivated.

The second type of contiguity is that of temporal relations and in this typeof relation the stress is on time contiguity. Either preceding or followingactions or events come to stand for an event or a state. So the Arabic wordssabüh and ‘asha meaning ‘breakfast’ and ‘supper’ respectively are both derivedfrom the temporal meanings of the periods in which these meals are usuallyeaten. The word sabüh is derived from subh meaning ‘morning’, ‘asha from‘ishä’ meaning ‘evening’. So the name of the time period came to representthe meal eaten around that period. Ullmann then talks about metonymic rela-tions other than the spatial and temporal contiguities. Here he talks about therelation of ‘the part for the whole’. The example he gives for this relation isthe word ‘redbreast’ for ‘robin’ in which the part of the bird, that is its breast,comes to stand for the whole bird (Ullmann 1962:218–19). This last type ofmetonymy shows that Ullmann does not distinguish between metonymy andsynecdoche as some rhetoricians do. It seems that a semantic theory of tropeswould usually undermine the minute differences in favour of deriving moregeneral rules governing the phenomenon or the results of the phenomenon;in Ullmann’s case it is the semantic change that results from the metonymicextension.

46 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 46

The previous three types of relations are relatively easy to interpret becausethere is some sort of logical link between the substituted element and the ele-ment substituted for, whether that relation is spatial, temporal or part–wholecontiguity. However, there are examples where the interpretation ofmetonymy is difficult to attain by means of intensional judgment from thecanonical and critical features of the item itself but rather has to be made onthe basis of the extensional properties and events that surround the utteranceboth situationally and culturally. The examples Ullmann provides for this cat-egory of metonymic relations are those designating social class, like ‘redcoat’,‘redcap’ and ‘bluestocking’. These are culturally motivated metonymies andone needs familiarity with the culture to be able to interpret them. The samething applies to certain metonymies which apply to the Arabic context butseem to be difficult to interpret in other contexts. The ramäd ‘ash’ examplediscussed in chapter 1, section 1.4, and repeated here as (4) is a case in point:

(4) Kathïru al-ramädi ‘idhä mä shatä.‘He has a lot of ash in the winter.’

The fourth category of semantic change Ullmann discusses is that whichrelates to what is usually called in figurative accounts of metonymy the‘inventor-for-invention’ relation or ‘producer-for-product’ relation. The exam-ple Ullmann gives for this category is the utterance by a physicist, ‘Oneampère is the current that one volt can send through one ohm.’ Ullmannargues that this statement involves three examples of metonymy. All are‘inventor-for-invention’ metonymies in which the surname of each of theinventors – the Frenchman André-Marie Ampère, the Italian AlessandroVolta and the German Georg Ohm – came to stand for these standard unitsof electrical measurement he invented.

One more aspect of Ullmann’s theory of metonymy that needs to be elu-cidated here is his distinction between contiguity of senses and contiguity ofnames. He calls the first category ‘metonymy’ but the second he calls ‘ellip-sis’. I have outlined the definition of ‘contiguity of senses’ above. As regardsthe ‘contiguity of names’, Ullmann defines it as ‘those words which oftenoccur side by side’ (p. 222). The association that exists between these wordsis not due to a semantic relatedness that is inherent to the lexical item byvirtue of sharing a similar semantic field. Rather, it is a conventional one thatexists between words as a result of their co-occurrence in similar linguisticenvironments. Two words keep recurring in similar linguistic environmentsuntil a conventional link is established between them. A further develop-ment takes place when one of these two words is usually omitted because itsmeaning has transferred to the sense of the other word. A good example isfound in the utterance

(5) She is going to the ladies.


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 47

The original form is ‘She is going to the ladies’ toilet,’ but because there hasexisted a link between the two words the word ‘ladies’ is regarded as infor-mative enough and the addition of the word ‘toilet’ could be regarded asredundant.

Ullmann’s distinction seems to be based on a semantic principle that dis-tinguishes ‘system’ and ‘use’. In the ‘contiguity of senses’, to which Ullmanngives the term ‘metonymy’, the ‘system’, that is the semantic properties ofthe lexical items involved, are seen to be inherently related owing to somesort of a contiguity relation. However, in the ‘use’ case the two items cometo have this connectedness because of rules of use. In other words Ullmann’sview throws some light on the notion of ‘motivatedness’. He seems to arguethat there are two types of this. The first is cognitive motivatedness in whichthe meaning is cognitively motivated because it appears in domain relation-ships. The second is the situation in which the meaning is conventionallymotivated by rules of use. In the view of metonymy as I am developing it,there is no such distinction between conventional motivatedness and cog-nitive motivatedness.

2.1.1.4 Bredin

Bredin (1984) begins his paper entitled ‘Metonymy’ by criticising the gen-eral tendency to augment metaphor at the expense of metonymy, which hethinks could be more common than metaphor. He then moves on to assessthe modern status of the trope, especially with reference to its being ‘thedistinguishing mark of realist prose’ (p. 45). Here he thinks that the effortof Jakobson, which was further augmented by the study of Lodge, gavemetonymy a real boost and called attention to the potential of this tropeas a mode of cognition. Bredin attacks the ‘enumeration’ accounts of meto-nymy which he thinks are prevalent even in the most contemporary refer-ence works. The problem does not stop, according to Bredin, at trying toprovide exhaustive lists of metonymic instances and species. In fact, itextends to the general disagreement among ‘various authorities’ as to whatthese instances are. The most scientifically alarming problem is that thenumber of examples that have been suggested by scholars in the field havevaried considerably and there seems to be no attempt to develop an organ-ising principle that would systematise the relation between objects or namesunderlying metonymy.

Bredin holds that in the history of metonymy there has been a great dealof confusion and obscurity as to both the nature of the trope and its rela-tionship to other closely related tropes like metaphor and synecdoche. Thehistory of metonymy has swung between taxonomic enumeration of speciesand types on one side and reductionism of the system of tropes on the other.The former was actually the central feature of the classical rhetoric projectof which Aristotle, the Auctor and Cicero are chief examples. The latter isthe central feature of the structuralist project of which Jakobson, Lodge and

48 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 48

the Groupe mu are good examples. Whenever there is an attempt to deriverules and principles there tends to be some sort of reductionism in the treat-ment of the data. When there is an attempt to stress the types and speciesthere is usually a tendency to enumerate examples rather than explain thenature of the trope. Bredin is not content with either approach. He is keento explicate the nature of metonymy and at the same time to place a well-defined border between metonymy and other closely related tropes, that ismetaphor and synecdoche.

The ‘transfer of the name of a thing to something else’, which is the stan-dard definition given to metonymy in many rhetorical treatises, is seen byBredin to be too general to the extent that it is in fact a definition of tropein general, since any trope involves a transfer. He thinks that this is a begin-ning of a definition of any trope because it outlines the genus to which eachtrope belongs. Bredin then emphasises the need to give what he calls the‘differentia’, which distinguish metonymy from other tropes. It is here thatdifficulty arises, Bredin argues (p. 46). For Bredin the difference between dif-ferent tropes lies in the difference in the nature of the relation which holdsbetween the two elements between which the transfer takes place.Therefore, characterising the nature of the relation that holds between whatis sometimes called the ‘trigger and target’12 in metonymic interpretation isthe beginning of the solution to the confusion and obscurity in metonymicunderstanding.

Of course Bredin cannot claim that he is the first to think of this, becausebefore him Jakobson and even Ullmann thought about the nature of themetonymic relation and tried to characterise it in an attempt to distinguishthis type of relation from other figurative relations especially from that ofmetaphor. Bredin expresses his discontent with the accounts of this relationthat were offered by classical rhetoricians like Quintilian, Bede and Doyle,who according to Bredin favoured a very general definition of metonymythat suits the definition of trope in general and completely ignored thenature of the relation underlying the trope. He also does not agree withthe formulations provided by Cicero, Abrams and Preminger, who charac-terise the relation between the objects involved in the transfer in generalterms as ‘closely associated’ or ‘closely related’. Nor does Bredin feel happywith Fraunce’s ‘agreement’ or with Jakobson’s ‘contiguity’, on the ground thatthese accounts are still too general to give the specific character of metonymicrelations.

Thus, Bredin rejects all previous accounts of metonymy, and places him-self in a position to offer a new account of the nature of the trope andthe relation underlying it. Bredin begins this endeavor by a provocativeclaim regarding the essence of the trope. He argues that the essence ofmetonymy is not a transfer in the ‘sense’ of the name of a thing to some-thing else but rather a transfer of the ‘reference’ of a thing to the referenceof something else. He alludes to Frege’s distinction between ‘sense’ and


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 49

‘reference’ and uses this distinction to serve his goal of situating metonymyon the ontological level of representation as an aspect of reference. It is worthmentioning that the notion of transfer in metonymy has been contested andthe majority of researchers in metonymy assert that there is no actual trans-fer in metonymy, as there is only one domain involved in the operation.What is actually involved is ‘substitution’? The word ‘transfer’ seems to bebiased towards the implication of two domains, while the term ‘substitution’reveals an internal movement between parts and wholes and other singledomain spaces.

The methodology Bredin uses to achieve his goal, which is to provide anunambiguous definition of metonymy, is to compile a list of metonymies inan attempt to derive a general principle from the mass of examples andspecies. He gives a list of 11 metonymic relations. Each of these can bereversed, giving a total of 22 examples. I am not particularly concerned withthe typology offered by Bredin, just as I am not interested in various othertypologies proposed in the literature. This is because typologies are notimportant to the course of this book, which tries to transcend the enumer-ation of types, species and examples in order to probe into the nature of thetrope and derive general rules as to how it actually operates in real-life lan-guage use. Therefore, I am here concerned with the treatment of Bredin’sattempt to present metonymy as independent from other related tropes.Bredin’s adoption of the view that metonymy actually operates on objectsrather on senses leads him to criticise the Groupe’s conception of the webof connotative associations. This is simply because the connotations pertainto the verbal expression rather than to the objects themselves. Bredin rejectsthe Groupe’s theory of metonymy as insufficient to reveal the specific char-acter of metonymy in an unambiguous manner.

Bredin thinks the Groupe’s use of the notion ‘connotative association’blurred the picture regarding which of these connotative relations is actu-ally metonymic and which is not. There are examples of connotative rela-tions that seem related, but they cannot be examples of metonymy, arguesBredin. The words ‘inflation’ and ‘economy’ are related, but ‘replacing oneexpression by the other would be neither figurative nor intelligible’ (p. 50).Bredin argues that one may agree with the Groupe that there are two modesof connotative association, one being conceptual and the other material, a view the second half of which would suit Bredin’s view of metonymy.Even in this case, however, Bredin argues that the problem is that accordingto the Groupe’s theory of metonymy it ‘is not enough simply to classifyconnotations as semantic or referential. For what we want to know is whichconnotative relations are, in addition to being connotative, metonymical aswell’ (p. 50).

Bredin then moves to discuss yet another theory of metonymy, proposedby Albert Henry. According to Bredin, Henry draws a distinction between twofundamental modes of meaning. One is the intensional mode and the other

50 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 50

is the extensional mode. Metonymy is then related to the mode of intension.This mode means that ‘the intension of a word is the set of constituent con-cepts which, taken together, constitute the “concept entity” designated bythe word’ (p. 51).

Henry’s argument is that metonymy arises when one constituent conceptis used to designate the whole ‘concept entity’. On the same grounds thatBredin criticises the Groupe’s notion of the web of connotative associations,Bredin also attacks Henry’s notion of ‘concept entity’. Bredin gives an exam-ple to show that Henry’s notion of ‘concept entity’ does not necessarily cre-ate a metonymy. The semic field of the expression ‘publishing company’‘includes such concepts as ‘book’, ‘distribution’ and ‘profit’; yet it is impos-sible to employ the name of any of these as a metonymical replacement for“publishing company” ’ (p. 52). Bredin concludes that the accounts ofmetonymy since ancient rhetoric have been rather general, lacking an analy-sis of what Lakoff and Johnson (1980:37) call the ‘systematic’ character ofmetonymy. Bredin states that ‘metonymical connections are not random,but are specific types of connection’. Bredin further asserts that ‘it is thisspecificity which is lacking in Groupe mu and in Henry, despite their bestefforts’ (p. 52).

I would argue that the most legitimate proposal to provide ‘specificity’ canbe found only within the confines of a cognitive theory of meaning in whichthe notions of ‘semic fields’ and ‘concept entity’ could be cognitively stud-ied to see what actually forms a concept totality or a semic field in the cog-nition of language speakers. This is with regard to the solution to the‘specificity’ issue in metonymic operation. A more important issue here isthat in this book I am not particularly interested in pursuing the debatesregarding the differences between metonymy and other related tropesbecause it is believed that some overlap is inevitable since we are dealing withmodes of thinking rather than with formal systems of logic or language.Cognitive theories of metonymy will be the topic of section 2.1.2 below. Letme now turn back to Bredin’s theory of metonymy.

Cognition, Bredin argues, ‘is a connected series of encounters betweensubject and object in which we analyse things into their constituents, syn-thetically combine objects in larger or more abstract wholes, such as con-cepts, classes and kinds.’ After outlining the distinction between analyticand synthetic dimensions of cognition Bredin then relates the first to hisconcept of structural relations, which he thinks are essentially intrinsicbecause they are relations within things. To the synthetic dimension ofcognition Bredin relates the concept of ‘synthetic’ relations, because theseare extrinsic relations which involve a great number of synthesisingprocesses to group things among things, not within things.

Bredin associates synecdoche with the analytic structural relation, as it isbelieved that the trope involves relations within the same object: part forwhole or whole for part. This is reminiscent of the Groupe’s account of


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 51

metonymy when they refer to Du Marsias as ‘one of the rare classical rhetori-cians to have asked himself summarily about the difference betweenmetonymy and synecdoche’ (Groupe mu 1981:120). Du Marsais assigns tometonymy a feature of independence from both objects of signification,whereas in synecdoche there is union between these objects. It is worthpointing out here that Bredin’s classification of relations into structural andextrinsic is not theoretically justified, because it might be argued that allrelations are in fact structural, whether they are intrinsic or extrinsic; oth-erwise they cannot be called relations.

The extrinsic type of relation breaks down into two types of relation, onesimple and the other dependent. Bredin relates metonymy to the simple rela-tion because metonymy, according to him, ‘neither states nor implies the con-nection between the objects involved in it…it relies wholly upon thoserelations between objects that are habitually and conventionally known andaccepted’ (p. 57). Bredin relates metaphor and other tropes to the dependenttype of the extrinsic relation and asserts that these are dependent because theyinvolve a twofold synthesis: ‘a conceptual synthesis based upon a sharedproperty of the relata. (A and B have length) and the synthesis of the relationitself (A is longer than B)’ (p. 56). Bredin argues that the difference betweenmetonymy and metaphor lies in the fact that ‘metaphor creates the relationbetween its objects, while metonymy presupposes that relation’ (p. 57).

Bredin holds that metonymy can never articulate a newly discoveredinsight and thus lacks the creative depth of metaphor. He concludes thatmetonymy is ‘irresistibly and necessarily conventional’. Because metonymyis based on a simple extrinsic relation, and because it presupposes the rela-tion between the objects involved, it is conventional, in the sense that itinvolves relations that are known already to exist between the objectsinvolved in the metonymic operation. We know that a ‘keel’ is part of a‘ship’ in example (6):


The problem of metonymic interpretation is its arbitrariness, which makesit subject to change from one culture to another and from one generationto another within the same culture. Towards the end of his paper Bredinargues that ‘it is no doubt this arbitrariness which has caused the remark-able and general failure to discern its true nature’.

Two points should be raised in connection with Bredin’s treatment ofmetonymy. First, metonymy is not a hackneyed operation where the infor-mation conveyed is somewhat banal and already known. Were this the casethen the trope would not be regarded as a trope in the first place. It is truethat metonymy involves the transfer of the name or the object on the basisof well-known relations of contiguity and causality. However, it is not truethat metonymy does not add new insight simply because the motives

52 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 52

behind the transfer cannot be regarded as commonplace. Rather, they areideologically loaded and convey a great deal to the overall meaning via themetonymic interpretation. The choice of one part to represent the whole orthe whole to represent a particular part may not be seen as an innocent one.Always there is extra meaning resulting from the simple choice of things tostand for other things.

In example (6) above, there is strong evidence that metonymy doesconvey new information. It is not a matter of conventionality that the word‘keel’ comes to stand for the word ‘ship’; why not any other part of the ship?Is this random? Or arbitrary? It is certainly not. Metonymic relations aremotivated and purposeful. In fact, in metonymic operations there is a greatdeal of suppression of certain aspects of cognition for the sake of fore-grounding other aspects. Indeed, metonymy is perhaps the most prominentexample of motivatedness in language. Therefore, assigning the features of‘conventionality’ and ‘arbitrariness’ to the operations of metonymy willnot do justice to the trope in any objective way. The reason is that thenotion of ‘conventionality’ boils down to the concept of ‘symbolicity’ inlinguistic signification. I shall discuss these aspects of linguistic significationin the next chapter, section 3.4.3. Here I want to stress the fact thatmetonymy resides in the indexical mode of signification which is causal orcontiguous.13

2.1.1.5 Lakoff

Lakoff’s theory of metonymy is first encountered in his Metaphors We LiveBy, which he co-authored with Mark Johnson (1980). Lakoff and Johnson(1980) begin their discussion of metonymy by giving a general definition ofthe trope. They hold that metonymy is ‘using one entity to refer to anotherthat is related to it’ (p. 35). They attempt to distinguish metonymy from per-sonification, which is commonly confused with metonymy. The reason forthis confusion is that personification involves assigning animate or humanproperties to non-animate or non-human subjects. This is a common aspectof the metonymic operation also. So in the example

(7) The Times has not arrived at the press conference yet.

which Lakoff gives as an example of metonymy, the verb ‘arrived’ is attrib-uted to a non-animate and non-human subject. The difference that Lakoffand Johnson see between metonymy and personification is clarified in theexample

(8) Inflation robbed me of my savings.

They think this example is personification because there is no reference toa person. Rather, the term ‘inflation’ is given the qualities of an animate and


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 53

human subject as a matter of metaphorical personification. Inflation refersto inflation. However, in the example

(9) The ham sandwich is waiting for his check.

as well as example (7) above, there is an actual entity referred to in the exam-ple and the words ‘the ham sandwich’ and ‘The Times’ are used to refer tosomething other than themselves. The reference is to a person who is relatedin some way to these entities.

This start makes Lakoff and Johnson (1980) maintain the view that

metaphor is principally a way of conceiving of one thing in terms ofanother, and its primary function is ‘understanding’. Metonymy, on theother hand, has primarily a referential function; that is, it allows us to useone entity to stand for another. (p. 36)

It looks as if Lakoff and Johnson are attempting a reductionist view ofmetonymy as nothing but a referential trope. This undermines its rolein interpretation and understanding. If metonymy is merely referential thenit is of marginal importance to cognitive processes because it does notcontribute much to the conceptual structuring of our experience. Its ‘stand-ing for’ relation is characteristic of an ontological account of experienceonly and does not participate in the epistemological dimension of ourexperience. It should be noted that Lakoff and Johnson later admit thatthe role of metonymy is not and cannot be merely referential. Metonymy,they argue, contributes to understanding in that the item used highlightsparticular aspects that are deemed particularly important to the communi-cation.

Lakoff and Johnson (1980) refer to the metonymic example

(10) We need some good heads.

in which ‘heads’ stands for ‘people’ because of the part–whole relation thatstands between heads and people. However, the function of metonymy doesnot stop at the level of referentiality alone. Rather, it extends to the level ofcognitive informativity because the use of ‘heads’ to stand for people in thisparticular example is not random or innocent. In fact, it is purposeful, it isinformative and it serves the function that we need ‘intelligent’ people notjust any people. In this case ‘intelligence’, ‘heads’ and ‘people’ stand incontiguous relations and they enter into multiple signification. ‘Heads’stands for ‘intelligence’ because ‘intelligence’ is thought to be a property ofthe human brain, which is situated in the head, and the head is part or

54 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 54

perhaps the most important part of the body. By the same token we canthink of the example

(11) I have been reading Shakespeare.

in which the use of the term Shakespeare to refer to his work is not a resultof a random or purely innocent choice, but of a purposeful choice to signifythat the whole oeuvre of this author, his comedies, his tragedies and his son-nets and so forth, were all in my mind even though I was reading only oneparticular work. Therefore, there is an act of implied cognitive hyperbolewhich augments our conceptual space.

Lakoff and Johnson (1980) extend the argument of the cognitive powerof metonymy to include even seemingly pure referential examples ofmetonymy. Example (12) below:

(12) The ham sandwich wants his check.

is used as an act of ‘dehumanising’ in which case the aspect of the referenthighlighted is that of being a customer nothing more and nothing less. Onecould also add that in example (13) below

(13) Table 10 is waiting for the order.

the metonymy is not purely referential but serves also to highlight a physi-cal aspect of the restaurant’s arrangement which is all that is important inthe conceptual structuring of the experience of the waiter or waitress. It isof particular relevance to the course of this book to include the statementwith which Lakoff and Johnson conclude their chapter on metonymy. Theyassert that in fact, the grounding of metonymic concepts is in general moreobvious than is the case with metaphoric concepts, since it usually involvesdirect physical or causal associations’ (p. 39).

Lakoff and Turner (1989) discuss further the hypothesis which Lakoff andJohnson make in (1980) regarding the referential nature of metonymy. Thisrelates metonymy to the pragmatic phenomenon of deixis, which is definedby Crystal as

the term used in linguistic theory to subsume those features of languagewhich refer directly to the personal, temporal or locational characteristicsof situation within which an utterance takes place, whose meaning isthus relative to that situation; for example now/then, here/there, I/you,this/that are deictic expressions. (Crystal, 1991:96)


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 55

In example (14)

(14) We are ahead of time.

Lakoff and Turner (1989) claim that there are two metonymies involved inthis example. They argue that

we stands for the point that we are at on the path from past to futureevents, and time stands for the point on that path at which we werescheduled to be at the present time. (p. 46)

Whether the example is readily interpretable as metonymic or not is irrele-vant. The interesting part of the argument is the linking of metonymy tothe process of deictic reference. This point will be elaborated further whenI discuss Stallard’s and Nunberg’s theories of metonymy in this chapter, insections 2.1.2.4 and 2.1.2.5 respectively. The point will also be raised in thenext chapter when I discuss metonymy as an index in section 3.4.2.

Lakoff and Turner (1989) introduce the notion of ‘schema’ in relation to theset of conceptual structures that metonymic expressions activate. I think thisnotion is very useful in the explanation of the nature of metonymy. Seeingmetonymy as being primarily a cognitive process involving schematic map-ping between two parts or aspects of the same conceptual domain will makeclear how basic the notion of schema is to the operation and interpretationof metonymy.14 Lakoff and Turner (1989) provide as an example of howmetonymy activates schemas the lines from Yeats’s ‘The Second Coming’:

(15) … but now I knowThat twenty centuries of stony sleepWere vexed to nightmare by a rocking cradle … (Lakoff and Turner1989:101).

In this example the ‘rocking cradle’ stands for the baby inside the cradlewhich is Jesus Christ and the sense of rocking is suggestive of the cyclicreturn of Jesus. These images are activated in our schema once we read‘twenty centuries’, which is a reference to Christ’s first coming and evokes asense that if there is a first coming there will surely be a second. This senseis particularly activated by the word ‘rocking’, which evokes images of con-tinuity and repetition.

Lakoff’s (1987) theory of idealised cognitive models (ICMs) is significantlyrelevant to the course of this book and it merits discussion here. GeorgeLakoff (1987) begins his theory by emphasising the importance of categori-sation. He argues that

There is nothing more basic than categorisation to our thought, percep-tion, action, and speech. Every time we see something as a kind of thing,

56 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 56

for example a tree, we are categorising. Whenever we reason about kindsof things – chairs, nations, illnesses, emotions, any kind of thing atall – we are employing categories. (Lakoff 1987:5–6)

From this quotation we can draw a conclusion that categorisation is essen-tially a metonymic process because according to Lakoff whenever we seesomething as a kind of thing, for example a tree, we are categorising. Thiskinship or associative relation between elements within the same domain isthe essence of metonymic signification and has been identified by rhetori-cal scholars since ancient times.

This quotation also shows the importance and prevalence of metonymyin our cognition and language and hence in our textual interaction, beinga reflection of these two aspects. Lakoff moves on to outline his theory ofidealised cognitive models (ICMs) by defining an ICM as

a complex structured whole, a gestalt, which uses four kinds of structur-ing principles:

(i) prepositional structure, as in Fillmore’s frames;(ii) image-schematic structure, as in Langacker’s cognitive grammar;

(iii) metaphoric mappings, as described by Lakoff and Johnson;(iv) metonymic mappings, as described by Lakoff and Johnson.

Each ICM, as used, structures a mental space as described by Fauconnier.(Lakoff, 1987:68)

The most relevant of these structuring principles to our discussion now isthat which relates to ‘metonymic mappings’, which essentially means thereferential function between parts and wholes of entities. In fact, the exam-ple Lakoff provides to illustrate his notion of ICMs is a fundamentallymetonymic example. He states:

Let us begin with Fillmore’s concept of frame. Take the English wordTuesday. Tuesday can be defined only relative to an idealised model thatincludes the natural cycle defined by the movement of the sun, the stan-dard means of characterising the end of one day and the beginning of thenext, and a larger seven-day calendric cycle – the week. In the idealisedmodel, the week is a whole with seven parts organised in a linearsequence; each part is called a day and the third is called Tuesday.

The ICM of the concept ‘week’ is the conceptualisation of the genus, speciesand examples of this model as a holistic entity – a gestalt. Lakoff explains theconcept of ICM by means of a metonymic model in which institutions standfor places where they are located in examples like ‘The White House isn’t say-ing anything,’ in which the element B ‘the place’ is used to stand for the ele-ment A ‘the institution’ within the same ICM. Lakoff asserts that ‘a cognitive


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 57

model may function to allow a salient example to stand metonymically fora whole category’ (p. 90).

It is this saliency principle that is highlighted in the textual theory ofmetonymy proposed in this book, in which one element in the metonymicmodel can actually stand for and thereby signify the whole model or anyother element within the same model, and hence provide texture, cognitivemapping and coherence to conceptual experience when processing text. Itis the saliency of an example and its relationship to other peripheral exam-ples within a category that is the central issue of Lakoff’s notion of ‘radialcategory’, in which there is a central sense, which could be seen as the cen-tre of a wheel, and extended senses that go out from the centre, makingthem radial. Radial category construction is the situation where a categoryis structured in this way. This conception is characteristically relevant to thetextual theory of metonymy, especially the pragmatic dimension of this the-ory, where metonymy is brought to bear on the coherence relations withintext. This is an aspect which will be employed in Chapter 5 of this book.

Lakoff and Johnson (1999) maintain that ‘our concept of time is cognitivelyconstructed by two processes, one metonymic (based on correlations withevents) and one metaphoric (based on motion and resources)’ (p. 167). Theygive a specific example of the event-for-time metonymy, which is ‘Harry hada heart attack during the rock concert’, in which ‘the rock concert standsmetonymically for the length of time during which the rock concert occurred’(p. 154). The authors argue that there are also metonymic expressions in lan-guage and particularly in written language where the punctuation markstands metonymically to the meaning of that punctuation mark (p. 254). Thisis particularly relevant to the textual model I am proposing for metonymy, inwhich a particular form stands for a particular meaning in a text because ofan associative relation between them. Lakoff and Johnson (1999) also discussthe product-for-process type of metonymic relationship and they provide anexample like ‘The book is moving right along’ in which the book stands forthe process of writing the book (p. 203). A conceptualisation of this naturewarrants a more comprehensive scope of the nature of metonymic significa-tion in text, which is the main proposal of this book.

2.1.2 Cognitive theories of metonymy

In this section I intend to complement the discussion in the previous sec-tion, which focused primarily on the linguistic theories of metonymy by anaccount of the increasingly growing interest in the trope in cognitive stud-ies. The discussion will include work done in cognitive semantics and in arti-ficial intelligence.

2.1.2.1 Fass

Fass discusses metonymy as one of the contiguity-based tropes along withsynecdoche, merism, antonomasia, periphrasis, toponymy, euphemism anddysphemism. He chooses for the trope a definition proposed by Perrine

58 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 58

(1992): ‘the substitution for the thing meant of something closely associatedwith it’ (Fass 1997:32). Fass holds that the distinction between metonymyand synecdoche is disappearing15 and argues that the difference in naturebetween metonymy and metaphor leads to a difference in function. Whilemetaphor maps things or names among domains by virtue of its compari-son and similarity power, metonymy maps things or names within domainsby means of its connectivity and contiguity power. Fass further asserts thatthe function of both metonymy and metaphor can be further distinguishedon the lines that

metaphor is a brief and often creative way of illustrating certain features ofa word by drawing from another domain, whereas metonymy is a conciseway of allowing one thing to stand for another within one domain. (p. 48)

Fass points out to a phenomenon he calls ‘twice-true metonymy’, whichhighlights the process of meaning transfer involved in metonymic interpre-tation. He selects an example from Fauconnier (1985:24):

(16) Ari painted a tanker.

Fass mentions that this example has two interpretations, one literal and theother metonymic. In the metonymic interpretation the ‘tanker’ means ‘animage’ of a tanker. As for the literal meaning, ‘Ari painted a tanker’ means‘Ari covered the tanker with paint.’ I would argue that this phenomenoninforms us about the issue of factuality inherent in the nature of metonymy.This is to say that in the example, whether in its literal or its metonymicinterpretation, there is an actual tanker involved. Perhaps the multiple inter-pretation is caused by an inherent ambiguity in the verb ‘paint’ itself. Itmight have acquired two independent senses. One meaning is ‘to cover withpaint’ and the other meaning is ‘to draw with paint’. In any case, the dis-tinctive feature of metonymy and synecdoche is that there is a referentialfunction involving actual referents as opposed to metaphor which involvesimaginative concepts. In the example

(17) John is a fox.

there is no actual reference to a fox but rather to a feature that the foxhas which is in turn shared by John. The view of metonymy as a form ofindirect reference has been proposed by Nunberg (1977) and is discussed insection 2.1.2.5.

In his account of metonymy Fass mentions also the process in which thereis a chain of metonymic interpretation; he selects an example from Reddy(1979:309):

(18) You’ll find better ideas than that in the library.


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 59

In this example a chain of metonymies is evident. According to Reddy, citedin Fass, ‘ideas are expressed in words, words are printed in pages, pages are in books, books are found in libraries’ (Fass 1997:73). Although thischain of metonymies is based on a part–whole metonymy, there are chainsof metonymy which are based on other relations also. The causal relationunderlying the chain of metonymy in the ‘ash’ example mentioned inChapter 1, section 1.4, in relation to metonymy in Arabic rhetoric is a casein point. The ash stands for a lot of cooking. A lot of cooking stands for alot of food. A lot of food stands for a lot of guests and a lot of guests standsfor generosity, which is the ultimate metonymy intended when we saythat someone has a lot of ash under his cauldron. Perhaps this example isculture-specific and is not generalisable. Nevertheless, the point to be madeis that in chain metonymies there is a considerable space for cultural varia-tion owing to the connotative nature of this type of multiple signification.

Fass’s cognitive account of metonymy and his attempt to find a solutionto the problems created by the arbitrariness of the trope is very interesting.If metonymy is arbitrary, the argument forwarded by Bredin as we have seenearlier in section 2.1.1.4, then metonymic relations are in fact open-ended.This makes the trope a murky area where there are no clear boundaries,which may in some sense support Jakobson’s statement that metonymydefies interpretation. Fass’s proposal to constrain ‘what can and cannot be ametonymy’ on the basis of viewing metonymy as relationships between caseroles should be regarded as a genuine attempt to address the specific char-acter of metonymy. However, it is still an account of species and types ratheran account of principles underlying the nature and process of metonymicoperations.

Fass seems to be interested in a grammatical analysis of metonymy, andhis proposal of case role relations is indicative of this inclination. Howeverhe goes beyond this to consider grammatical variation in the realisation ofmetonymy. He refers to metonymies in noun sense extensions such as

(19) A watched pot never boils.

in which the noun ‘pot’ is extended to mean the contents of the pot and notthe pot itself. These metonymies have been called CONTAINER-FOR-CONTENT

metonymies. It is also worth mentioning here that the traditional view ofmetonymy is that it is a trope that occurs only in nouns, hence the ‘nymy’root from the Greek word for noun. However, this is not actually true becauseverbs, for example, can function metonymically as well. In the example

(20) I went to court.

the verb ‘went’ is used metonymically to stand for the whole scenario thattakes place when people go to courts, because ‘going to court’ is one act of

60 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 60

the whole scenario of going there and standing in the dock and arguing yourcase, and so on and so forth. Similarly, in the example

(21) I went shopping.

the verb ‘went’ signifies the whole scenario of shopping which includesgoing from one shop to another, scanning the products, deciding what tobuy and what not to buy, thinking about the prices and perhaps bargainingabout these prices. Fass’s account of metonymy falls within this view ofthinking that metonymy is actually prevalent in all language.

The discussion of ‘metonymic objects’ which Fass cites from Curme(1964:134) makes us think of how metonymy could function as a semanticinterpretation of certain syntactic processes. The example cited is

(22) He wiped the table.

in which the table is not the real object of the verb because the wiping off isnot done to the table but to the dust on the table. In this example there isclear reference to the fact that metonymies occur with objects in the sameway they do with subjects. Owing to the contiguous relationship between thetable and the dust on it, it becomes conceptually possible to metonymicallysubstitute one for the other. Fass goes beyond the noun sense extensions andmetonymic objects to discuss other metonymic lexical combinations. Hetalks about preposition–noun metonymies as in examples like

(23) After the alarm.

which is a preposition–noun metonymic combination, as the prepositionand noun both cooperate to create this metonymy. ‘After the alarm’ meansafter the sounding of the alarm, because after ‘expects an event as its object’(p. 92).

Fass also discusses adjective–noun metonymies, citing examples fromQuillian (1969:469) like

(24) Young client.

which ‘involves inferring that the client’s age is being judged young, whichis not explicit’. Other examples of adjective–noun metonymies include

(25) Jealous letter.(26) Angry report.

Fass argues that a possible analysis of these adjective–noun combinations isin terms of a metonymic chain consisting of a PROPERTY FOR WHOLE metonymy,


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 61

where ‘jealous’ is the PROPERTY and ‘person’ is the WHOLE, followed by an ARTIST

FOR ARTFORM metonymy in which ARTIST is ‘person’ and ARTFORM is ‘letter’. Afteroutlining the various types of grammatical variation in metonymic opera-tions, Fass moves on to review some of the accounts related to the interpre-tation of metonymy. Fass tells us that not much literature on theinterpretation of metonymy is available and that the ‘best work is in the com-putational literature’ (p. 93).

Fass is interested not only in exploring the nature of metonymy but alsoin investigating how metonymy is recognised and interpreted. I am moreinterested in discussing his account of metonymy recognition than that ofmetonymy interpretation. I believe that metonymy recognition necessarilyinvolves metonymy interpretation, because metonymy cannot be recognisedwithout it being interpreted. In other words the distinction between meto-nymy interpretation and recognition is baseless. Fass distinguishes betweentwo approaches in the interpretation of metonymies. One is conventional orconventionalised, the other novel. The distinction is based on the assump-tion that the former is the approach which considers metonymies as ‘knowl-edge-specific relationships and that metonymic interpretation is theapplication of those relationships’ (p. 93). The latter regards metonymies as‘essentially arbitrary relationships’ (p. 93).

Fass relates the conventional view to the work of Lakoff and Johnson andthe works inspired by them. The novel view is related to the work of Stallard(1993), who seeks a characterisation of the argument structure of meto-nymic utterances, and of Nunberg (1995), who looks at the predicative andreferential aspects of metonymies. In fact, the typologies proposed formetonymic relations fall under the conventional view because they assumethat the trope is or has been conventionalised into certain concepts or types.Within the conventional view, there exists a notion similar to the conceptof metonymy. This is the concept of ‘coercion’, which according to Grosz‘occurs whenever some property of an object is used to refer indirectly to theobject’ (Grosz et al. 1987:213, cited in Fass 1997:94).

With regard to metonymic recognition, Fass distinguishes three views thathave been offered in the literature. The first concerns the syntactic clue of‘proper nouns with indefinite determiners’. The words ‘Ford’ and ‘Boeing’are typical examples of proper names used for products they make. A use ofany indefinite article before these names renders the expression metonymicand signals the non-literal usage. However, in names like ampère, ohm andvolt, which were discussed above, the matter is not as straightforwardbecause the substitution has been very much conventionalised and as aresult has been naturalised to the extent that one finds it difficult to recog-nise the metonymy in such instances. Therefore, there are perhaps differentlevels of recognition within this view depending on the extent of how live,active or dead the metonymy is.

The second view of how metonymy can be recognised is through viola-tions of selection restrictions. This is, according to Fass, ‘the most common

62 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 62

approach to metonymy recognition’ (p. 97). It is the most common becausethe overall view of figures is that they are semantic deviations, and henceviolate the normal categorisation that verbs subcategorise in their argumentstructure. A point has to be raised here though. This concerns the notion ofviolation of selection restrictions and how it relates to tropes in general andto metonymy in particular. Metaphor is recognised by violations of selectionrestrictions, and so is personification. So what makes the violations in thecase of metonymy distinct? Consider the following examples:

(27) Table 10 is getting angry.(28) The sea is getting furious.(29) Inflation robbed me of my savings.

Examples (27), (28) and (29) all exhibit violations of selection restrictions.‘Tables’ cannot get angry whether they are 10 or 11 because ‘angry’ is anattribute peculiar to animate entities. The ‘sea’ cannot be ‘furious’ because‘furious’ is an adjective that pertains only to animate objects. Also ‘inflation’cannot rob; only human beings can rob. It seems as if violation of selectionrestrictions is a working strategy in the recognition of metonymy as well asmetaphor and personification. It also seems as if the process of recognitiongoes in the following manner. First, violation of selection restrictions givesrise to the recognition of a figurative use in general. There follows anotherprocess to recognise the type of transfer involved. I argue that it is at thissecond stage that a distinction is drawn between the various types of viola-tions leading to various types of figuration. In the case of metonymy therecognition of the trope is characterised by a recognition of a discrepancybetween a surface subject and a surface predicate. The violation of the sur-face structure of the utterance conceals a harmony in the underlying deepstructure in which the underlying real subject is in agreement with the pred-icate in terms of semantic features.

Fass offers a third view of metonymic recognition based on Stallard’sapplication of Grice’s maxim of quantity. This maxim is in fact one of fourdicta that make up what Grice calls the ‘Cooperative Principle’, which is‘make your conversational contribution such as is required, at the stage atwhich it occurs, by the accepted purpose or direction of the talk exchangein which you are engaged’ (Grice 1975:45).

Fass tells us then that Stallard argues that the maxim of quantity preventsa referential interpretation and supports a predicative interpretation ofexample (30):

(30) Which airlines fly from Boston to Denver?

The reason is that a predicative interpretation conforms to the maxim ofquantity and thus does not give superfluous information while a referentialinterpretation does give redundant information. This is because if the


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 63

intended meaning is the set of flights and not the set of airlines then thereis redundancy because flights are necessarily on airlines.

2.1.2.2 Gibbs

This section deals with Gibbs’s account of the nature and role of metonymyin language and thought. The immediately striking thing about Gibbs’schapter entitled ‘metonymy’ is the statement he makes at the outset of thechapter, which runs: ‘[T]his chapter explores the role of metonymy inthought and language’ (Gibbs 1994:320). This is quite remarkable, as itshows clearly that Gibbs is not interested in the nature of the trope as it hasbeen debated for more than two thousand years. Instead, he is proposing anew way of looking at metonymy. This dwells on the role or function ofmetonymy rather than the nature and the principles underlying it. It isworth pointing out here that this is generally the view I am putting forwardin this book. I argue that the most useful way of understanding the work-ings of metonymy is by looking at both its nature and its role in discourseand its function in understanding.

Gibbs does not provide a definition of metonymy apart from his generalcomment on our understanding of a poem, which he mentions at the begin-ning of the chapter. In this regard he states that this understanding ‘dependson our ability to think metonymically at the mention of parts of some eventand infer something about an entire situation’ (p. 319). Gibbs asserts ‘thatmetonymy is a fundamental part of our conceptual system.’ Like Lakoff andJohnson (1980), Lakoff (1987) and Lakoff and Turner (1989), Gibbs talks aboutmetonymic models and thinks that ‘speakers’ frequent use of metonymicexpressions and listeners’ understanding of these utterances are motivated bymetonymic models that form a significant part of our everyday conceptualsystem’ (pp. 320–1). The first issue Gibbs deals with in his treatment of therole of metonymy in language and thought is how to set metonymy out asdistinct from metaphor and synecdoche. This has been a concern of manyrhetorical theorists, linguists, philosophers and cognitive psychologists. I havealready referred to such debates in this chapter (section 2.1.2.1). Here I ammore interested in Gibbs’ discussion of metonymy as a cohesive device.

Gibbs’s interest in the cohesive power of metonymy reflects the basicthrust of this book and his assertion that ‘many conversational inferencesabout what speakers mean by what they say require metonymic reasoning.Metonymy serves in many such instances as an important cohesive devicein text and discourse understanding’ (pp. 323–4) constitutes the basichypothesis I aim to validate. The textual function of metonymy Gibbs refersto here is not only the global function identified by scholars like Jakobsonin which the literary production of poets, novelists and playwrights couldbe classified globally according to whether it is metonymically or metaphor-ically dominated. Rather, it goes beyond this generality and becomes morespecific when we also look at local patterns of connectedness.

64 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 64

Gibbs discusses the role of metonymy in thought, drawing evidence fromexperimental cognitive psychology about the ways people categorise andclassify objects in reality. Of particular relevance to this issue is the notionof prototypes which involves judging ‘certain members of categories asbeing more representative of those categories than other members’ (p. 325).Gibbs agrees with Lakoff (1987) on the view that one source of prototypeeffects is metonymic. The following example shows how people usemetonymic models in their everyday life:

(31) A: How did you get to the airport?B: I waved down a taxi.

Gibbs maintains that this example shows how ‘conversational implicature’helps us make inferences about the speaker’s intended meaning. Gibbs thenquestions the epistemological basis of these conversational inferences. Heasks ‘How does a listener infer that B actually found a taxi to take him to theairport?’ (p. 327). The answer is that this is possible through metonymic util-isation of idealised models in which one part of the model evokes the entiremodel. This is what happens in the interpretation of B’s utterance in whichthe listener will take ‘waving down a taxi’ as representative of the wholescenario of getting to the airport.

Similar to this is his example of anaphoric reference:

(32) They were told to expect the prime minister at twelve the next day.Punctually at noon the car drove up in front of the State Department.

Here the noun phrase ‘the car’ in the second sentence anaphorically refersto the phrase ‘prime minister’ in the first sentence. Gibbs highlights theissue of conceptual anaphors and even argues that they are sometimes moreeasily interpretable than grammatical anaphors because people applymetonymic processes when they attempt to interpret discourse. Gibbsprovides this example

(33) A: I need to call the garage.B: They said they’ll have it ready by five o’clock.

and argues that ‘a series of experimental studies demonstrate that people rateas more natural and read more quickly sentences with conceptual anaphors…than they do sentences with appropriate singular pronouns’ (p. 329).

As the example above shows, the pronoun ‘they’ is grammatically incor-rect because it does not agree with its antecedent in number. The garage issingular and the pronoun is plural. However, the interpretation of this utter-ance is made easier with the plural pronoun because of our pervasive abilityto establish a metonymic relation of PLACE FOR PEOPLE WORKING IN THE PLACE.The garage stands for the people who work in it. It is conventionally known


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 65

that garages are usually run by more than one person. Therefore, the use ofthe pronoun ‘they’ is more natural than ‘it’.

Gibbs argues also that metonymy is a main source of our ability to under-stand implicit cause–effect relations. These relations are activated by a set ofcausal associations between certain lexical items. These associations areessentially metonymic in nature. The example Gibbs provides to illustratethis point is the tale:

(34) He wanted to be king.He was tired of waiting.He thought arsenic would work well.

There are causal and procedural connections between the parts of this talewhich make our understanding of the tale depend on general knowledgerather than on the linguistic structure of the tale. It can be argued thatbetween the word ‘king’ and the word ‘arsenic’ there is a conceptual associ-ation of some kind. It can be asserted that within the ‘semic field’ or the‘concept entity’ of the word ‘arsenic’ there is a sense of assassination of apolitical rival. This sense gets activated by the word ‘king’ which is a politi-cal term. This lexical signalling of causal relations underlying the narrativestructure of the tale facilitates our interpretation of the tale if explicit infor-mation about the plan of the tale is missing.

Understanding metonymic expression is an important aspect of Gibbs’stheory of metonymy. In this regard Gibbs talks about two main principlesunderlying the understanding of metonymic expressions. These are ‘sensecreation’ and ‘sense selection’. Gibbs argues that in metonymic under-standing the listener or reader not only performs an act of selection of theappropriate sense of the word used from the set of potential senses availableto him in his mental lexicon. In fact, they also have to apply the principleof ‘sense creation’ because in many cases the sense intended by the utter-ance will not be specified within the mental lexicon. As soon as the listeneror reader realises the lack of a contextually appropriate sense in their men-tal stock, and given that the utterance is anomalous if interpreted literally,they will try to create a sense that suits the context and facilitates theirunderstanding of the utterance. This is a temporarily created sense which isfacilitated by means of other clues within the utterance, mainly lexical clues.

The example Gibbs gives to illustrate his argument of sense selection andsense creation is

(35) John fired the tuxedo because he kept dropping the tray.

Gibbs argues that readers do not find a difficulty in understanding the utter-ance despite the apparent anomaly due to the use of the word ‘tuxedo’. Howthey find it easy to interpret the utterance, according to Gibbs, is by first

66 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 66

recognising that there is a sort of discrepancy between the literal meaningof the word ‘tuxedo’ and the overall meaning of the whole utterance.Second, listeners and readers resort to the process of sense creation since ‘thecontextually appropriate meaning of ‘tuxedo’ cannot be selected from ashort list of potential meanings in the lexicon’ (p. 337). I think that themetonymic interpretation of the utterance is guaranteed by virtue of therelation of association between the word ‘tuxedo’, which means a dinnerjacket, with the person wearing it. Nevertheless, the problem now is how lis-teners and readers interpret the word ‘tuxedo’ as a butler. In principle theword ‘tuxedo’ could stand for many types of person. Essential here is therole of lexical signalling within text to establish relations of associationsbetween lexical items. The verb ‘fired’ establishes a relation between ‘he’ andthe ‘tuxedo’ as that between employer and employee. Then the word ‘tray’further specifies the schema in which ‘an employee with a dinner jacketholding a tray’ is activated, thus creating the sense of ‘tuxedo’ as butler.

Gibbs also considers the role of metonymic reasoning in the interpretationof colloquial tautologies. These are expressions used to give some sort ofemphasis to the utterance or to convey certain attitudes, whether positive ornegative, towards the nouns mentioned in these tautological constructions.There are two main types of tautologies; one has the construction N (singu-lar) is N (singular) such as

(36) War is war.(37) Politics is politics.

The other has the structure N (plural) will be N (plural) such as

(38) Murderers will be murderers.(39) Rapists will be rapists.

Gibbs argues that these expressions exhibit a metonymic reasoning in that amention of a whole stands for one particular quality of these wholes.According to the context of interpretation, ‘the whole boy’ in the expression

(40) Boys will be boys.

will come to stand for one property of boys or the boy as a token will standfor the whole class of boys. Boys will be boys means boys will be unruly orsweet and lovable.16

Gibbs also addresses the role of metonymic reasoning in the interpreta-tion of indirect speech acts. Gibbs argues that the use of metonymic rea-soning in the form of indirectness in making requests and orders helps tomaintain face among the participants in the conversation. The issue of facehas been raised in sociological and pragmatic studies of language in use and


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 67

pertains to the psychology of the addresser when he expresses the directiveor regulative function of language through which he is able to modify anddirect the behaviour of the addressee. In most cases of indirect speech actsthere is a metonymic relation of the type PART FOR WHOLE or WHOLE FOR PART.So in the request:

(41) Can you lend me ten dollars?

metonymic reasoning is involved in the interpretation of the question as arequest to give the person the amount of money specified and not about theability to lend. Our understanding of such examples is facilitated by ourmetonymic reasoning which stipulates that people infer a whole series ofsteps from only part of this whole scenario.17

The obstacle hypothesis, which Gibbs discusses in relation to the metonymicinterpretation of indirect speech acts, suggests that the ‘apparent conven-tionality of an indirect request depends largely on the extent to which anutterance specifies an addressee’s projected obstacles in complying with thespeaker’s request’ (p. 354).

This means that the particular act of the whole exchange which thespeaker feels might constitute an obstacle to the listener comes to stand forthe whole exchange. For example, having the time might be an obstacle toanswering the question ‘What is the time?’ So the request of a passer-byabout the time is appropriately carried out with the part that is believed toform the obstacle. That particular part of the exchange comes to representthe whole speech act. The question then appears in the form, as Gibbsargues,

(42) Do you have the time?

I find Gibbs’s theory of metonymy compatible with his overall aim of argu-ing for the poetic nature of mind and supporting his argument with empir-ical evidence. This is why his theory of metonymy is very distinct from otheraccounts. He concerns himself centrally with the role of metonymy in lan-guage and thought, providing examples of how metonymy is interpreted innatural language data. Unlike other researchers he does not occupy himselfwith the investigation of the classification of the trope. For example, he isnot particularly interested in enumerating the species or types of metonymy.This is by no means to say that these are not important. Rather, Gibbs’s the-ory of metonymy is functional18 and this is what makes it particularly rele-vant to the theory I am developing throughout this book. One effect of thisorientation on Gibbs’s account of the four principal tropes is that he treatsmetonymy and synecdoche as the same phenomenon. He is not very inter-ested in ‘minute’ differences, especially given that his goal is to outline thefunction of these closely related tropes in discourse. It should be noted that

68 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 68

the same approach is adopted in this book in that the two tropes ofmetonymy and synecdoche are not sharply distinguished because the ori-entation is functional and in their function metonymy and synecdochetend to overlap.

2.1.2.3 Radden and Kövecses

Radden and Kövecses state at the outset of their paper that their aim is topropose a coherent conceptual framework for metonymy. They also contestthe traditional view of metonymy which treats the trope purely as a matterof language. They argue that their theory goes beyond the traditional defi-nition of metonymy ‘as a figure of speech that consists in using the nameof one thing for that of something else with which it is associated’ (1999:17).

Radden and Kövecses identify four fundamental shortcomings of the tra-ditional view of metonymy. The first of these is the treatment of metonymyas an aspect of language. This treatment falls short of current findings incognitive studies which assert that metonymy is essentially conceptual innature. The second point Radden and Kövecses raise pertains to the natureof the relation underlying metonymy. While they admit that the notion ofcontiguous connection between entities is fairly uncontroversial, they ques-tion whether the term is clearly enough defined to identify the specific char-acter of metonymy. They refer to the debate surrounding the nature ofcontiguity between ontological and cognitive studies. Each of these looks atthe term in accordance with its own theoretical framework. So ontologicaltheories look at metonymy as a set of relations in the ‘world of reality’,whereas cognitive theories consider metonymy as the set of relations in theconceptual world.

The authors express their satisfaction with the notion of ICMs (idealisedcognitive models) proposed by Lakoff (1987:68). An ICM is, as Gibbs states,‘a prototypical “folk” theory or cultural model that people create to organ-ise their knowledge’ (Gibbs 1994:58). Gibbs further argues that ‘ICMs makesome sense, given that ICMs are idealised and don’t fit actual situations ina one-to-one correspondence but relate many concepts that are inferentiallyconnected to one another in a single conceptual structure that is experien-tially meaningful as a whole’ (p. 58).

Radden and Kövecses argue that these ICMs ‘may capture metonymicprocesses best’ (1999:20). This is because they do not account for the imme-diate conceptual components of a particular domain but rather characterisethe cultural models of which the domains are parts. The notion of ICMs,Radden and Kövecses argue, is ‘not restricted to either the world of reality,the world of conceptualisation, or the world of language but … may cutacross these ontological realms.’ In this way, the authors think, the notionof ICMs will solve the problem of scattered interest in metonymy amongcognitivists, philosophers, linguists and rhetoricians, and bring them allunder one unified theory.


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 69

The third issue Radden and Kövecses discuss with reference to theshortcomings of the traditional view of metonymy is the notion thatmetonymy ‘has referring function (only)’. The authors attribute this short-coming to the erroneous assumption that metonymy is a substitution ofthings. Metonymy, they argue, also occurs ‘at the purely conceptual level(categorisation, linguistic reasoning), at different levels of language (lexis,morphology, syntax, discourse), and as a linkage interrelating differentontological realms (concepts, forms and things/events)’ (p. 21).

The fourth point deals with the way the traditional view defines metonymyas ‘substitution’. Radden and Kövecses argue that ‘metonymy does not sim-ply substitute one entity for another entity, but interrelates them to form anew, complex meaning’ (p. 19). In their attempt to solve this problem theyrefer to Langacker’s cognitive explanation of metonymy in which he talksabout metonymy as a ‘reference-point phenomenon’ in which ‘the entity thatis normally designated by a metonymic expression serves as a reference pointaffording mental access to the desired target’ (Langacker 1993:30).

The discussion of these four shortcomings of the traditional view ofmetonymy leads Radden and Kövecses (1999) to formulate their own defin-ition of metonymy, which is entirely cognitive. Their definition runs as fol-lows: ‘[M]etonymy is a cognitive process in which one conceptual entity, thevehicle, provides mental access to another conceptual entity, the target,within the same idealized cognitive model’ (p. 21). Radden and Kövecsessuggest that the development of a theoretical framework for metonymyinvolves answering four fundamental questions related to ICMs andmetonymy. These questions revolve around the following four issues:

(i) the ontological realms in which metonymy occurs;(ii) the conceptual relationships which may lead to metonymy;

(iii) the cognitive principles governing the selection of the preferred vehicle;(iv) the factors overriding the preferred default routes and yielding ‘non-

default’ cases of metonymy.

In their discussion of the first issue Radden and Kövecses argue thatmetonymy occurs within and cuts across three realms: the world of concepts,the world of things and events, and the world of forms. Examples ofmetonymies that result from the interaction between concepts within onerealm in which a concept stands for another concept include:

(form) concept (A) for concept (B)(mother) ‘mother’ for ‘housewife-mother’

The interaction between forms and concepts, in which the metonymic oper-ation cuts across two different realms, is exemplified by:

form (A) for concept (A)dollar for ‘money’

70 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 70

The interaction between the realm of concepts and the realm of things isexemplified by: Concept (A) for thing (A)

‘cow’ for thing cow

The interaction between the realm of form, concepts and things is exempli-fied by:

form–concept (A) for thing (A)cow–‘cow’ for thing cow

The interaction between two forms and one concept is exemplified by formsof clipping and acronyms:

form (A)–concept (A) for form (B) concept (A)UN for United Nations

As for the second issue, ‘conceptual relations which may lead to metonymy’,Radden and Kövecses present a lengthy discussion of various types ofmetonymic relations. The main idea of the classification is that

the types of metonymy-producing relationships may be subsumed undertwo general conceptual configurations:

(i) Whole ICM and its part(s)(ii) Parts of an ICM (p. 30).

They argue that the first possibility realises whole–part relations and theirconverses. The second shows part–part metonymies. The authors provide adetailed explanation and enumeration of the various types of metonymicexpression that result from certain conceptual relations. The discussion iscarried out in two broad sections. The first section deals with configurationswithin the whole-for-part conceptual system and the reverse of this system.The second section deals with the configurations within the part-for-partconceptual system.

Radden and Kövecses seem to have probed deeply not only into the natureof metonymy but also into the motivation behind the metonymic relationand the selection of the particular vehicle. They maintain that metonymycan be motivated by certain expressive goals that the speaker has in mindbecause of social needs or for rhetorical purposes. In the first case themetonymic expression violates the communicative principle ‘Clear overObscure’. So the speaker will resort to obscure constructions in order toachieve this expressive need. In the second situation it is claimed that theviolation involves one or more cognitive principles. This leads to the cre-ation of a non-default metonymy that exhibits striking aspects of figuration.Owing to the underlying intentionality, the resulting metonymic expressionshows personal style and personal preference of metonymic vehicles.

Radden and Kövecses relate euphemism-based metonymies to the violationof the communicative principle in that there is tendency to favour obscure


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 71

over clear expressions, which is against the cooperative principle ofcommunication. However, for sociolinguistic reasons pertaining to keepingface and solidarity, the violation is licensed. The example given is the use ofthe word ‘redundancies’ instead of ‘dismissals’. The argument is that ‘dis-missals’ is a more direct word than ‘redundancies’ and that the former mightconstitute a threat and hence a negative face, whereas the latter is a muchmore mitigated expression. The lack of clarity in the expression

(43) Where can I powder my nose?

is justified on the grounds that it is a euphemistic expression for the moreliteral and relevant expression

(44) Where can I urinate?

Thus the speaker overrides the cognitive principle of relevant over irrelevantby selecting ‘powdering the nose’ which is totally irrelevant to the act ofurinating as the vehicle of the metonymy.

With regard to the violation of cognitive principles, Radden and Kövecsesassert that metonymy is first and foremost a figurative mode of thought.Like other figurative modes of thought, it is usually employed to achieverhetorical effects as in humour, jargon, literature, persuasion, slang, poetryand the like. The result of the violation of cognitive principles results natu-rally in the creation of unmotivated original and live metonymies that hadattracted the attention of rhetoricians, philosophers and linguists sinceancient times. This feature has also recently attracted the attention of cog-nitive psychologists who have extended their analyses to original and livemetonymies to look into the nature of more conventional and more cogni-tively motivated metonymies, as these are believed to shed more light onthe conceptual structure of our cognition.

2.1.2.4 Stallard

David Stallard (1993) distinguishes between two kinds of metonymy – refer-ential and predicative. He begins by asserting that ‘the phenomenon ofsemantic coercion is quite a common one in natural language’ (p. 87). Whatis special in Stallard’s treatment of the trope is that he challenges the univer-sal view of metonymy as a process of indirect reference in discourse. He con-tends that this view is not adequate to account for data in which the referenceis actually to the direct surface noun phrase and the expression still exhibitsa metonymic operation. Stallard’s approach is semantically oriented and hedevelops the distinction between the two types of metonymy on the basis oftwo main examples, the interpretation of which, he thinks, differs radically:

(45) The ham sandwich is waiting for his check.(46) Which airlines fly from Boston to Denver?

72 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 72

Stallard argues that in example (45) there is an act of indirect reference. Theproof of this is the intra-sentential anaphora in which we find the pronom-inal ‘his’ referring to the subject noun phrase. This shows that the agree-ment is actually with the deep or indirect antecedent noun phrase ‘theperson ordering the ham sandwich’ and not the surface or direct nounphrase ‘the ham sandwich’. The indirect reference is necessitated by virtueof the violation of the selection restrictions. This renders the literal inter-pretation problematic, since ham sandwiches do not normally wait, onlyhumans do. By contrast, in example (46), although there is a violation ofselection restrictions in the predication of the verb ‘fly’ to ‘airlines’ whichdo not normally fly, literal interpretation is actually necessary. The dilemmaarises when we realise that a non-literal interpretation would yield an absurdanswer to this question – for example, if the listener were to answer thisquestion by giving the set of flights going from Boston to Denver.

Stallard indeed discovered a deep problem in previous attempts ofmetonymic interpretation. But this means that the majority of metonymicexamples are in fact predicative, because the agreement is usually with thesurface and direct literal antecedent. It seems that in the case where there isa shift in rank between the literal and direct referent and the non-literal andindirect one, there tends to be a referential function of metonymic relations.In fact, this is what Stallard asserts when he says that

as for the referential type of metonymy, we have found only a few casesof it in this corpus. We hypothesize that the reason for this is that refer-ential metonymy, involving as it does an encoding of a reference in termsof a categorically different thing, is a more marked and unusual event inpsychological terms. Predicative metonymy on the other hand involvesno such operation, merely the convenient making-way of a predicate fora non-standard but related argument. For this reason, our work preferspredicative metonymy as the default choice in processing when no otherevidence is present. (p. 93)

Stallard defines ‘referential metonymy’ as the process in which ‘themetonymic noun phrase does indeed have an intended referent related tobut different from its literal meaning’ (p. 88). Predicative metonymy, how-ever, Stallard continues, is the process in which ‘the actual and intendedreferent of the noun phrase is just the literal one, and it is more accurate tosay that the predicate is coerced’ (p. 88). He suggests two criteria by whichwe can generate the correct reading. The first one is ‘the external bind-ing agreement condition on applying the referential metonymy schema’ (p. 92). Stallard continues explaining this first criterion: ‘[I]f an NP’s externalsemantic context agrees with its literal referent, but not its referen-tially coerced version, then referential metonymy is ruled out for that NP.’The second criterion is the intra-sentential anaphora which figures out in


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 73

examples like:

(47) The ham sandwich is waiting for HIS check.(48) Which airline flies to ITS headquarters?

Stallard argues that in example (47) ‘the pronoun “his” cannot agree with theliteral referent, but can agree with the metonymically interpolated PERSON,and so provides evidence for the referential reading’. In example (48), Stallardcontinues, ‘the pronoun “its” cannot agree in number with the interpolatedset of flights, but can agree with the singular “airline” and so provides evidencefor the predicative reading’. So in both examples we have linguistic constraintsdetermining the preferred reading of the metonymic type involved.

One could perhaps further demonstrate the distinction by adding anappropriate adjectival phrase to the relevant metonymic noun. Considerexamples 49–52 below:

(49) *The ham sandwich, which was carefully prepared in our ownkitchen, is waiting for his cheque. (the asterick signals unaccept-ability)

(50) The ham sandwich, who’s obviously in a hurry, is waiting for hischeque.

(51) This airline, which was founded in 1983, flies from London toNewcastle.

(52) *This airline, which was built in Japan, flies from London toNewcastle.

In the example

(53) Which airlines fly from Boston to Denver?

the constraint that determines the reading of this utterance as predicative isnot a linguistic constraint however. It is a pragmatic constraint related toGrice’s maxim of quantity. This stipulates that any superfluous and redun-dant information is rejected by virtue of the principle that communicatorsare generally cooperative and do not provide more information than isneeded (Grice 1975). The pragmatic constraint tells us that the intended ref-erent in the example above is in fact the surface NP because if it were theinterpolated NP it would lead to some sort of superfluous informationbecause all flights are actually on airlines.

Fass comments on these proposals and argues that Stallard’s analysis‘though impressive appears to be incomplete’ (Fass 1997:86). Fass thinksthat Stallard did not explain why examples like

(54) Nixon bombed Hanoi. * They sang all the way back to Saigon,

fail the agreement test.19 He asks what type of metonymy the one in thisutterance is. Is it referential or predicative? According to the rules Stallard

74 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 74

provides it is predicative, because the pronominal reference ‘they’ is ruledout since the intended referent is actually the surface noun phrase which issingular. I think the referential/predicative metonymic distinction is alsouseful in other respects. For example, it has been rather difficult to integrateexamples of personification where it is felt that there is some metonymicrelation underlying such expressions.

(55) Inflation robbed me of all my savings.

The traditional account of example (55) is that it is ‘personification’. However,Stallard’s predicative metonymy extended the scope to include transfer onverbs as well as nouns. Here the noun phrase intended is actually the surfacenoun phrase ‘inflation’. However, there is still violation of selection restric-tions in that ‘inflation’ does not actually rob. So the solution is that there isa predicative metonymy involved in which the verb ‘rob’ actually changes itsargument features.

2.1.2.5 Nunberg

I move on now to discuss Geoffrey Nunberg’s treatment of metonymy in hispaper ‘Transfers of Meaning’ (1995). Nunberg distinguishes two types ofmetonymy. The first is ‘predicate transfer’ metonymy and the second is‘deferred reference’ metonymy. Nunberg states: ‘By ‘transfers of meaning’ Imean the ensemble of productive linguistic processes that enable us to usethe same expression to refer to what are intuitively distinct sorts of cate-gories of things’ (1995:109). Nunberg tries to distinguish between transfersand rhetorical figures on the ground that the former category is essentiallya linguistic phenomenon whereas the latter is a conceptual matter. Heasserts that

granted that there is a salient correspondence between monarchs andcrowns, for example, it still has to be explained why the word crown canbe used to refer to monarchs – or for that matter why this fact shouldhave any linguistic consequences at all. (pp. 109–10)

Nunberg (1995:109–10) introduces the distinction between the two types ofmetonymic reference in his discussion of the mechanism of transfer. He pro-vides two examples uttered in the context of

a customer hands his keys to an attendant at a parking lot and says either:

(56) This is parked out back.

or

(57) I am parked out back.


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 75

Nunberg argues that both examples involve metonymies in the sense that bothsubjects in (56) and (57) can be thought as related in some way to the car andthe location where it is parked. Whereas the reference in (56) is actually to theintended deep referent, that is the car, and not to the surface demonstratum,in (57) the reference is to the surface noun phrase. The evidence of this char-acterisation of the nature of referentiality of the NP in the two utterances isbased on some linguistic tests which provide evidence for the distinction. Thefirst is the gender test in which example (56), ‘This is parked out back’, is trans-lated into Italian (an inflecting language showing gender differences). Thetranslated version shows that the referent rather than the demonstratum deter-mines agreement in the Italian example. Similar results will be obtained whenexamples like (56) are translated into Arabic, an inflecting language for genderand number and person. The inflection for gender and number would clearlyshow that the intended reference in example (56) above is deferred and thatthe real reference is made to the car and not to the key. This demonstrates thatthe above example is in fact a case of deferred reference.

The second test is that of conjunction in which we conjoin additionalinformation about the subject in the first clause. The conjunction will showus whether the agreement in the second clause depends on the surface NPor the deep NP. If the first case holds then it is a matter of deferred reference.If it is the second case which applies then it is an example of predicate trans-fer. The example Nunberg gives for this is:

(58) This is parked out back and may not start.

The conjunction in (58) above supports the view that the intended referenceis not the surface demonstratum but the implied referent ‘the car’, becausethe additional information conjoined is essentially a description of a car andnot of a key. Regarding example (57), both the gender and the conjunctiontests support the view that the intended referent is the surface NP. A tagquestion to ascertain the example above is:

(59) I am parked out back, aren’t I?

A conjunction test will reveal that it is only possible to conjoin informationreferring to the person having the car rather than to the car itself. So theconjunction examples Nunberg gives are:

(60) I am parked out back and have been waiting for 15 minutes.(61) *I am parked out back and may not start.

For this type of meaning transfer, Nunberg suggests the terms ‘deferredostension or deferred indexical reference’. This is

a process that allows a demonstrative or indexical to refer to an objectthat corresponds in a certain way to the contextual element picked

76 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 76

out by a demonstration or by the semantic character of the expression.(p. 111)

Example (61) exemplifies another kind of transfer, Nunberg argues:

The principle here is that the name of a property that applies to some-thing in one domain can sometimes be used as the name of a propertythat applies to things in another domain, provided the two propertiescorrespond in a certain way.

Nunberg calls this type of metonymic transfer ‘predicate transfer’, indicat-ing that the transfer that is evident in utterances like (61), for example, isnot a transfer of the noun phrase but a transfer in the semantic propertiesthat the predicate projects on its arguments. This treatment is similar toStallard’s treatment of predicative metonymy in the ‘airline’ example above.

2.1.2.6 Fauconnier

I perceive Fauconnier’s theory of ‘mental spaces’ to be essentially a theoryof metonymic representation. Fauconnier (1985) begins his theory of‘mental spaces’ by stressing his view that metaphor, metonymy or other so-called rhetorical devices are ‘not deviations or parasitic on core semanticsand literal meaning’ (p. 1). He maintains that ‘they are taken to be central,and much more widespread than is usually assumed’. Fauconnier accountsfor metonymic reasoning in terms of an application of a pragmatic functionor a connector to link between two objects a and b, so that a description ofa may be used to stand for b. With this reasoning Fauconnier provides acoherent metalanguage to deal with metonymic representation. He calls theelement a the reference ‘trigger’, element b the reference ‘target’ and thepragmatic function the ‘connector’. Connectors are seen by Fauconnier tobe part of idealised cognitive models (ICMs) in the sense of Lakoff (1987)(Fauconnier 1985:10). He then defines mental spaces as ‘constructs distinctfrom linguistic structures but built up in any discourse according to guide-lines provided by the linguistic expressions’ (p. 16).

Fauconnier in this definition alludes to the cognitive nature of mentalspaces because they are actually ‘domains which we set up as we talk orlisten, and that we structure with elements, roles, strategies and relations’(p. 143). Fauconnier further asserts the ‘process’ nature of mental spacesby arguing that ‘it will be convenient to speak of mental spaces as being builtup during ongoing discourse, rather than to refer to the correspond-ing sequence of sets’ (p. 144). Fauconnier’s conception of mental spacesas an ongoing process of meaning creation corresponds closely to theconception of metonymy sustained throughout this book as a meaning-creating process and not as a product. Fauconnier’s identification principle(IP) may be seen as a formally explicit characterisation of the representa-tional process of metonymic signification. On the basis of a pragmatic


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 77

function, more than one element within one space can be linked and assuch can be signalled.

In fact, Fauconnier argues that a pragmatic function can also link elementsfrom different spaces. This conceptulisation is conducive to the theory ofmetonymy as a text binding process because the pragmatic interpretation ofdiscourse provides numerous links between elements within one space andalso links between elements in different spaces. For the former case,Fauconnier provides examples like

(62) Plato is the red book; Homer is the black book. (p. 143)

In this example, the connector is between writers and books which are bothin one domain or mental space. For the latter case, Fauconnier provides thefollowing example:

(63) In that movie, Cleopatra is Liz Taylor. (p. 144)

Here the connector is linking between characters and actors which are indifferent spaces. Given our discussion earlier of the nature of metonymy asa process of substitution between elements within the same domain, whichhas been taken to be the critical distinction between metonymy andmetaphor (which involves substitution between different domains), shall westill consider the link between different spaces a case of metonymy or a caseof metaphor? In fact, it may be argued that the process is still essentially aprocess of metonymic signification because the ‘space builders’, such as thephrase ‘In that movie’, which Fauconnier defines as ‘expressions that mayestablish a new space or refer back to one already introduced in the dis-course’ (p. 17), are actually responsible for providing a hyperdomain or ahyperspace which binds the different spaces.

This binding is referred to by Turner and Fauconnier as ‘mental binding’,which is defined as ‘a basic mental operation whose uniform structural anddynamic properties apply over many areas of thought and action, includingmetaphor and metonymy’ (2000:133). Turner and Fauconnier (2000) assertthat

the work on conceptual blending [another term for mental binding] hasshown that in addition to such mapping, there are dynamic integrationprocesses which build up new ‘blended mental spaces’. Such spacesdevelop emergent structure which is elaborated in the online constructionof meaning and serves as an important locus of cognitive activity. (p. 133)

They argue that the blended space, which is the result of mapping of two ormore spaces, contains meaning structures that do not come from the sourceor the target. This of course does not rule out the fact that much of the

78 Textual Metonymy

1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 78

semantic structure of the emerging blend comes from both the source andthe target, but the theory of conceptual integration emphasises an impor-tant point, namely the existence of principles of online construction ofmeaning in human communication. These principles create novel elementsin the blend not originally existing in the source or target. Fauconnier andTurner’s theory of conceptual blending warrants a conception of metonymicsignification as the originator of such conceptual blends and the referencepoints within these blends, and this in turn accounts for coherence in tex-tual interaction. Fauconnier and Turner (1999) show the pervasiveness andimportance of metonymy in conceptual integration by arguing that‘metonymy plays an important role in constructing conceptual integrationnetworks’ (p. 77). They propose a set of optimality principles for blending.These principles revolve around the idea of maintaining tightly integratedblends that if manipulated remain as a whole, keeping the original connec-tions to the input space intact. The blend alone is responsible for recon-structing the input and there is a need to maintain a match between therelations of the elements in the blend and the relations of their counterpartsin the input.

Finally, there has to be a functional motive for an element to exist in theblend (Turner and Fauconnier 2000:139). The authors propose an optimal-ity constraint, which they call the ‘metonymy projection constraint’. Herethey argue that there are cases where two elements can be projected from aninput because of a metonymic association between them. In such cases, themetonymic distance between the elements in the blend is shortened andthis, according to Turner and Fauconnier, is the tightening of metonymiesunder projection.

In their recent account of conceptual blending Fauconnier and Turner(2002) argue that there is a complete system of interacting principles behindconceptual blending. These principles include ‘change, identity, time, space,cause–effect, part–whole, representation, similarity, category, intentionalityand uniqueness’ (pp. 93–102). It should be noted here that many of thesevital principles are actually metonymic operations in which there is map-ping of source onto a target. Cause–effect and part–whole are of particularrelevance to metonymic mapping because the former involves causality rela-tions of the type ‘X is the cause of Y’ in which case we understand one bymeans of another owing to the conceptual binding between the two elementsin the conceptual networks (pp. 75–88). Part–whole mappings are essentiallymetonymic because they operate within a contiguous space (pp. 97, 315).


1403_913323_05_cha02.qxd 11/19/03 8:41 PM Page 79

80

‘When I use a word,’ Humpty Dumpty said in a rather scornful tone,‘it means just what I choose it to mean – neither more nor less.’

Lewis Carroll, quoted in Aitchison (1995:91)

3.0 Introduction

The previous chapter discussed metonymy in modern figurative theory. Theaim of this chapter is twofold. The first is to develop a semiotic approach tometonymy. In this connection I argue for a theory of motivatedness in lin-guistic signification based on the conceptualisation of metonymy as a modeof contiguous and causal signification. Metonymy is viewed essentially as away of abstracting a relation between concepts, words and objects. This rela-tion is fundamentally a relation of representation. As such metonymy resemblesto a great extent the notion of a sign which is perceived as a three-dimensionalentity covering the three modes of knowledge, i.e. of words, of concepts andof objects or things. This chapter aims to develop a semiotic approach tometonymy based on this rudimentary assumption. The second aim is todevelop a textual model of metonymy based on this semiotic approach.

In this chapter I shall give a general overview of semiotics and discuss meto-nymy as representational signification. The aim is to link semiotics, significa-tion, communication and metonymy together and propose a textual model ofmetonymic signification. I hope to bring to the understanding of the reader theview that metonymy is not only a semiotic operation on lexical items; morethan this, it is a process of representational signification that stretches beyondthe individual lexical item to the text level. In short, metonymy with its signi-fying power is seen to be a major factor in text cohesion and coherence.

3.1 Semiotics: a general perspective

It is therefore possible to conceive of a science which studies the role of signsas part of social life. It would form part of social psychology, and hence of

3Metonymy and Semiotics

1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 80

general psychology. We shall call it semiology (from the Greek se-meîon,‘sign’). It would investigate the nature of signs and the laws governingthem. Since it does not yet exist, one cannot say for certain that it willexist. But it has a right to exist, a place ready for it in advance. Linguisticsis only one branch of this general science. The laws which semiology willdiscover will be laws applicable in linguistics, and linguistics will thus beassigned to a clearly defined place in the field of human knowledge.(Saussure 1983:15–16)

With these words the Swiss linguist Ferdinand de Saussure was actually pro-claiming the birth of a new discipline or at least predicting its birth. But inany case, it was a discipline that had the right to exist because it is there inour everyday conscious or unconscious life or, as Saussure put it, ‘it has … aplace ready for it in advance’. The scholar’s task with regard to this disciplineis to identify the rules and the principles that underlie and govern theprocess of signification. As far as the subject-matter of this discipline is con-cerned it was already there even before Saussure because it is the essence ofcognition and communication. In fact, the concept of signification is notnew or a breakthrough of the nineteenth or the twentieth century. Rather,it is a concept that was debated in ancient Athenian society. What is usuallyreferred to in books dealing with semiotics as ‘Plato’s puzzle’1 is a good illus-tration that the notion of signification has attracted the attention ofphilosophers from ancient times.

The main concern of semiotics is the study of the system of signs. There isnot a big difference between semiology and semiotics. The former has its ori-gins in the theory propounded by Saussure around the early years of the twen-tieth century and the latter goes back to the work on signification undertakenby the American philosopher Charles Sanders Peirce, who gave it the name‘semiotics’. In the course of this section I shall need to refer to the differencesbetween both theories and the implications of such differences on the devel-opment of the discipline in Europe and America. Stam et al. define semioticsas ‘the study of signs, signification and signifying systems’ (1992:1). If semi-otics is the study of how signs are created, transmitted and interpreted, thensemiotics is actually a theory of meaning.2 This is in fact the premise on whichI have decided to discuss the semiotic dimension of metonymy. Semiotics, Ibelieve, will be of immense help in the explication of the meaning of the con-cept of metonymy or rather in broadening our understanding of metonymy.Moreover, taking metonymy as a form of signification promotes the conceptto the level of communication, instead of keeping it restricted to the level oflexical meaning.

Semiotics is also defined as the science of signs, especially by those whodraw from Peirce, its founder. These quote his famous statement in whichhe asserts that logic can be seen as ‘the science of the general necessary laws of signs’ (Peirce 1955:vol. 2, para. 227). However, the correctness of the

Metonymy and Semiotics 81

1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 81

attribution of the quality of ‘science’ to this discipline is still far from estab-lished. Semiotics has ‘thick’ descriptive power, by which I mean that itallows for a deep treatment of the data and looks at the data from a multi-dimensional perspective. Yet it suffers from a lack of agreement as to its‘scientificity’. Perhaps the reason is that although it has thick descriptivepower it does not, however, have predictive power, which is the basic prin-ciple of scientific inquiry. Semiotics cannot predict the future of any signi-fying system or what could be signified in a particular situation because thisdepends entirely on various factors that are difficult or perhaps impossibleto bring under control. All semiotics claims to do is to provide causal expla-nation by relating objects and actions to their underlying norms of socialand cultural systems; semiotic explanations do not have predictive power.Linguistic semiotics for example does not aim to predict, although it ‘showswhy the sequence has the form and meaning it does by relating it to thesystem of language’ (Culler 1986:73).

Semiotics is largely qualitative and favours thick description of the phe-nomenon in which a grounding theory emerges from the data, rather thanimposing on the data the preconceived categories and patterns of an a priorisystem. In fact, the vastness of the phenomena that semiotics purports to dealwith3 makes it difficult to consider it a scientific discipline because almost allsocial sciences have something to do with semiotics in one way or another.This obviously reduces the definiteness and discreteness of semiotics.

According to Ullmann (1962:15), C. Morris divided semiotics into threeareas:

(i) Semantics: the meaning of signs (the relationship of signs to what theystand for).

(ii) Syntactics: the relations between signs.(iii) Pragmatics: the ways in which signs are used and interpreted.

This classification is very much influenced by linguistics, and the levels rep-resented here are used widely in linguistic studies. The significant differenceis that while the classification of linguistic data begins with the material orthe substance and then moves on to the form, in semiotics we have onlyform because the substance as far as semiotics is concerned is presumably notimportant, since what is important is the systematisation of this substance.4

The relationship between semiotics and linguistics is quite special,because among all systems of communication, or rather of signification,language is the most systematised and the most conventionalised one. Infact, no other forms of signification qualify as fully fledged systems of com-munication because these systems have not yet been fully convention-alised into proper systems of communication.5 In discussing the conceptof arbitrariness Saussure refers to the relationship between linguistics and

82 Textual Metonymy

1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 82

semiology as follows:

We may therefore say that signs which are entirely arbitrary convey betterthan others the ideal semiological process. That is why the most complexand the most widespread of all systems of expression, which is the onewe find in human languages, is also the most characteristic of all. In thissense, linguistics serves as a model for the whole of semiology, eventhough languages represent only one type of semiological systems.(Saussure 1983:68)6

Dickins argues in this connection that ‘this extraordinary communicativeability of natural languages is linked to (and can be regarded as a conse-quence of) an extremely high degree of complexity at the abstract analyticallevel’ (Dickins 1998:17).

Dickins further asserts, quoting Michael Lamb, that

natural languages seem to be of such a higher degree of complexity ascompared with other semiotic systems that one might reasonably expectany analytical situation exhibited in other semiotic systems to be exhib-ited in natural language also. One would not, however, expect analyticalsituations exhibited in natural language necessarily to be exhibited inother semiotic systems.

Eco treats the issue of verbal and non-verbal signification as problematic asfar as semiotics is concerned. He asks: what is the proper object of study ofsemiotics? Is it verbal language, which is language proper, or is it all otherlanguages? Eco concludes that

the problem could be solved by saying that the theory of signification andcommunication has only one primary object, i.e. verbal language, all otherlanguages being imperfect approximations to its capacities and thereforeconstituting peripheral and impure instances of semiotic devices…so thatlinguistics is not the most important branch of semiotics but the model forevery semiotic activity. (Eco 1976:172)

Hawkes holds that the main thrust of semiotics is to study communication andin this it overlaps with structuralism because, as he argues, the ‘interests of thetwo spheres are not fundamentally separate’ (Hawkes 1977:124). Hawkesdefines structuralism as ‘a way of thinking about the world which is predom-inantly concerned with the perception and description of structures’ (p. 17).In fact, Hawkes predicts that the future will witness a unification of semioticsand structuralism under one heading called ‘communication’ (p. 124).However, this comprehensive view is by no means the only way semiotics has


1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 83

been discussed. Semiotics has also been reduced to a mere methodology ofstructural analysis. In this connection, Lye argues that

structuralism enables both the reading of texts and the reading ofcultures: through semiotics, structuralism leads us to see everything astextual, that is composed of signs, governed by conventions of meaning,ordered according to a pattern of relationships. (1996:2)7

Sapir also conceives of communication as a fundamental characteristic ofhuman beings. He maintains that ‘every cultural pattern and every singleact of social behaviour involves communication either in an explicit orimplicit sense’ (Sapir 1949:104).

It should also be stressed that semiotics has been used in textlinguistics.Semiotic-oriented approaches to text analysis8 have provided text linguisticswith really deep-rooted insights into the nature of textual communicationand how meaning is created through communication, in contradiction tothe view that communication is the result of meaning. In fact, Deacon et al.maintain that semiotics and what they call critical linguistics are the two keyapproaches to the analysis of written texts. They argue that ‘critical linguis-tics regards the linguistic patterning of texts and utterances as indeliblysocial in character’ (Deacon et al. 1999:148). Deacon et al. seem to call for a bond between semiotic and cognitive approaches to the study of text.Critical linguistics is the type of linguistics that guarantees such a bondbecause it is broader in scope and methodology.

Through semiotics we see everything as textual as far as it is composed ofsigns, governed by meaning conventions and ordered in patterns by meansof various relationships. Semiotic-conscious text analysis grants the textanalyst a grounding theory that emerges from the text itself and makes thedata alive. Semiotics makes text analysis an intellectually demanding andrewarding practice rather than a mathematically dry endeavour. This is tosay that traditional text analysis was mainly concerned with the quantifica-tion of the number of times each linguistic item is mentioned in the text.This quantification, although apparently scientific, deprives the treatmentof the text as a living organism which creates meanings all the way throughits generation and interpretation. Quantification is based on the assumptionof fixed meaning and fixed interpretation. However, semiotics is based onthe assumption of multiple interpretation and this assumption is part andparcel of an illuminating textual analysis.

To conclude this section I think it is essential to cite an important definitionof semiotics provided by Umberto Eco, who writes ‘semiotics is in principle thediscipline studying everything which can be used in order to lie’ (Eco 1976:7).According to Eco ‘anything that cannot be used to tell a lie cannot be used totell the truth and in fact cannot be used to tell [anything] at all’. The assump-tion underlying this definition is presumably the paramount importance Eco

84 Textual Metonymy

1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 84

places on the idea of intentionality and conventionality as two importantaspects of signification. It looks as if these two concepts lie at the heart of Eco’sconception of what could qualify as a sign. However, this property would leadultimately to the limitation of Eco’s semiotics to the conventional systems ofsigns that are used for communication, notably human language, becauseother semiotic systems seem to lack the property of lying.9 Lying in languagecan occur in various ways, for example literally by flouting the maxim of qual-ity according to Grice’s maxims or figuratively by using a figure of speech likeirony, hyperbole or metaphor. In fact, both of these ways manifest themselvesin figurative language. Cook gives an example where the use of a figure ofspeech actually leads to the flouting of the maxim of quality. In, for example,

(1) I’ve got millions of beer bottles in my cellar. (Cook 1989:31)

there is an exaggeration involved, and this figurative speech is mainlyachieved via the flouting of the maxim of quality. In other words there is anact of lying in the example. In fact, any literal interpretation of the examplewould lead to absurd responses like ‘You English people must have very bighouses!’

According to speech act theory, correspondence between sentence mean-ing and utterance meaning in speech results in literal speech acts (Searle1975:60). So declarative sentences are seen to be assertions, interrogative sen-tences questions and imperative sentences orders. But figurative languagealways plays on the lack of correspondence to create some sort of tensionbetween literal and figurative interpretation. This aspect is also discussed bySearle when he discusses metaphorical expressions and the issue of truth con-ditions (see for details Searle 1979:113). A woman in classical Arabic cultureupon seeing the Caliph said the following:

(2) ’ashkü ’ilayka qillata al-jirthän fï baytï.‘I complain to you of the lack of mice in my house.’

Taking this sentence as a declarative sentence realising a complaint wouldtotally miss the point. The utterance requires a figurative interpretationbased on a causal inferencing of the type ‘mice have left my house becausethere is no food there’. An indirect speech act, which is a request for finan-cial help, would therefore result from this interpretation.

3.2 On the notion of signification

In this section I shall discuss the notion of signification. This will require a discussion of the concept of sign. Most of the work consulted here will beby Saussure and Peirce. The section aims to provide a general overview ofthe nature of signs and the various definitions that have been provided for


1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 85

the process of signification. Also of importance will be the discussion of thevarious types of signs and more particularly the various modes of significa-tion. The ultimate aim of this section is to develop a semiotic approach tometonymic signification.

3.2.1 The nature of signs

In his Encyclopedic Dictionary of Language and Languages, David Crystaldefines the word ‘sign’ as

a feature of language or behaviour which conveys meaning, especiallywhen used conventionally within a system (such as speech, writing, ges-ture, dance); also called a symbol (but many writers make a distinctionbetween these terms). (Crystal 1992:353)

Wales provides the following definition: ‘From Latin signum “mark, token”,sign is sometimes used interchangeably with symbol to denote “something”which stands for, or refers to something else, in a meaningful way’ (Wales1989:419).

The principle of interpretation says ‘a sign is something by knowingwhich we know something more’ (Peirce cited in Eco 1984:2). According toPeirce, a sign is ‘something which stands to somebody for something insome respect or capacity’ (Peirce 1960:vol. 2, 228). Morris asserts that

something is a sign only because it is interpreted as a sign of somethingby some interpreter. Semiotics, then, is not concerned with the study ofa particular kind of phenomenon, but with ordinary phenomena insofar(and only insofar) as they participate in semiosis. (Morris 1938:20)

In essence signs are mediators between messages and signals or betweenthought and expression. Signs are meaningful units. These units includewords, images, sounds, gestures or objects. When we invest these units withmeaning, then they become signs. So there is a process of semiosis that theword, image, gesture or any form of signification has to undergo for it to beregarded as a sign laden with meaning. To this end Turner argues that forsomething to qualify as a sign ‘it must have a physical form, it must refer tosomething other than itself, and it must be recognised as doing this by otherusers of the sign system’ (Turner 1992:17). In this concise definition of thenotion of sign, Turner provides us with three criteria for anything to qual-ify as a sign or rather as a signifier. It should have a physical form and thiscould be a sound, an image, an object or a gesture. In other words this sig-nifier should be something that can be seen, heard, touched, smelt or tasted.It must signify something other than its own entity and it must be conven-tionalised in the sense that it must be recognised by the users of the code asdoing so.

86 Textual Metonymy

1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 86

It is worth mentioning in this connection that Saussure’s view of both thesignifier and the signified is a mentalistic one because he excludes referenceto any object in the world. He asserts that ‘the linguistic sign is then a two-sided psychological entity’ (Saussure 1983:66). Moreover, Saussure pointsout that ‘a sign is the combination of a concept and a sound pattern’.However, he then alters this terminology by proposing ‘to keep the term signto designate the whole, but to replace concept and sound pattern respectivelyby signification and signal’ (p. 67). For Saussure both signification and signalare concepts that reside in the mind of the speaker or perhaps in the col-lective mind of the speech community.10 Just like langue, which is the sys-tem of rules of a particular language being stored in the collective mind ofthe speech community, signification could be a mental concept that isstored in the collective mind of the speech community in some schematicshape. In this case every individual in this particular speech communitywould have some sort of manifestation of this system. Saussure’s semioticthought, especially the concept of ‘signification’, could be perceived alongthe lines of his linguistic thought.

Perhaps the most important thing to point out about Saussure’s theory ofsemiology is the way he deals with the concept of ‘sign’ and how this affectshis semiological theory in general. What Saussure calls ‘sign’ is the arbitrar-ily established link between a particular signifier and a particular signified.This link is conventionally established and has to be learned before one canactually use it. But what does Saussure mean by the arbitrariness of this asso-ciation between the form and content? For Saussure a link is arbitrary if andonly if it is established by convention alone. Any natural connectionbetween the signifier and the signified would render the outcome a symbolrather than a sign.11 In this sense, words like ‘house’, ‘white’, ‘see’ etc. areessentially concepts pertaining to psychology and they ‘become linguisticentities only by association with sound patterns’ (Saussure 1983:101). Thusthey would naturally qualify as signs because there is no natural connectionbetween these forms and the contents they denote and this is why they aredifferent in different languages. It is only the conventions of English thathave established such a link between the word ‘cat’ and the small, furry, four-legged animal usually kept in houses as a pet.

But in languages there are words that are not totally arbitrary, i.e. theyexpress some kind of intrinsic natural relation between the word and themeaning it denotes. The clearest example is in onomatopoeia, where theword echoes the thing it refers to. This category and the category of excla-mations are treated as ‘symbolic’ and thus the term ‘symbol’ is used insteadof sign as far as the Saussurean theory is concerned (see the discussion of ono-matopoeia in Saussure 1983:69). Therefore, the term ‘sign’ is reserved inSaussure’s use of it to the total arbitrary relation between the signifier and thesignified. This implies that for Saussure all motivated associations are not sig-nification systems because they are marginal, as it were, to signification. It


1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 87

should be stressed here that it is the notion of ‘arbitrariness’ in Saussure’stheory of semiology that is going to be contested in the textual approach ofthis book because the assumption is that propositional and textual significa-tion is largely motivated. In fact, a semiotic theory of metonymy leads to theconception that linguistic signification is highly motivated because our selec-tion from the paradigmatic axis and our combination on the syntagmaticaxis is loaded with intentions, ideologies and biases.

Peirce’s theory of semiotics exhibits a different conception of the nature ofsign, signification and interpretation from that of Saussure.12 These issueswill be dealt with in other sections of this chapter. What I intend to elucidatehere are the key issues peculiar to Peirce’s semiotics. It should be noted thatwhile many examples of the differences between Peirce’s semiotics andSaussure’s semiology are treated in various sections of this chapter, the inten-tion is to look with particular focus at the issue of the scope of the disciplineas perceived by Saussure and Peirce. The scope of semiotics portrayed bySaussure is quite narrow compared with that proposed by Peirce. Saussure’snarrow conception of semiology stems from the fact that he perceives thediscipline to be the science of signs as he defines them. Saussure treats signsas arbitrary and conventional. Other ‘signs’, which are produced as a resultof natural motivation or generally ‘signs’ that are motivated, such as smokeas a sign of fire are not regarded as signs according to Saussurean semiology.Now this conception of sign if compared with Peirce’s semiotics is only onesubcategory of the vast conception of the Peircean sign. In fact, whatSaussure conceives as a sign proper corresponds to what Peirce calls a ‘sym-bol’, which is arbitrarily signified.

Peirce’s semiotics goes beyond the concept of sign as symbol to includesign as index and sign as icon. It looks at the concept of sign from a philo-sophical perspective as a unit of understanding, knowledge and interpreta-tion. He is interested in deriving a rule as to how signification operates ingeneral, whether the source of this signification is a human being or is nat-ural. For Peirce a sign like smoke coming out of a house is as important a signas the utterance ‘There is smoke coming out of that house,’ although the for-mer is a natural sign that is produced by a non-human sender. Linguisticsigns for Peirce are only a part of the system of signs that can be seen innature. While Saussure stresses the importance of the sign being composedof a linguistic expression in order to qualify as a sign, Peirce regards as a signany element in nature that yields knowledge and leads to interpretation.

Semioticians generally regard the concept of sign as a unitary concept andthe division of the elements comprising it as a matter of pedagogy ratherthan a real division. They see sign as a coin having two faces which cannotbe separated from each other. Saussure compares the sign to a human being,comprising body and soul. Better still, according to Saussure, is comparisonto a chemical compound such as water which he argues ‘is a combination ofhydrogen and oxygen; but taken separately neither element has any of

88 Textual Metonymy

1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 88

the properties of water’ (Saussure 1983:102). This type of approach hasencouraged such integrated conceptions of the notion of sign. Lewis main-tains that ‘the sign incorporates both signifier and signified: it is the mater-ial entity made meaningful’ (1991:27). In Lewis’s quotation we have a clearreference to the current trend of materialising the concept of sign. WhileLewis agrees with Saussure regarding the integrated structure of the notionof sign, we find a departure from Saussure’s mental conception of significa-tion. It seems that there is a movement to bring the concept of semiotics intoconcrete manifestation in socio-cultural contexts. I believe this is likely to bea fruitful development, as communication cannot be thought of only as amental phenomenon. In fact, it is more naturally a real-life phenomenonthat takes place within a material socio-cultural context. Perceived as such,communication is seen as the process of meaning creation. The process ofmeaning-creation in communication is far from an arbitrary process. Rather,it is a process of motivated selection of concepts and motivated combinationof these into thought.

Some semioticians (see Wells 1977:3; Anderson and Trudgill 1992:75)think of the concepts of signifier and signified as corresponding to the lin-guistic concepts of form and content. The signifier is paralleled with ‘form’and the signified with ‘content’. This view looks rather naive. The way sig-nifiers are looked at as ‘form’ that is used to carry ‘content’, i.e. meaning,leads to some problematic understanding regarding the role of the signifier.First, this view looks at the signifier merely as a carrier of meaning, but notas an element that is itself meaningful. Second, it assumes a one-to-onecorrespondence between content and meaning. This might not be asstraightforward as it is being presented by those who attempt to link semi-otic concepts to linguistic concepts, especially if we take figurative languageinto consideration (see Gibbs 1994:ch. 2). In fact, meaning is the result of aprocess of interpretation and a result of a long series of inferences, and it isnot merely a matter of relating a particular signifier to its conventionallyestablished content or meaning. The problem becomes more particularlyrecognisable when we have one signifier representing more than one signi-fied, as is the case in polysemy, or have one signified being represented bymore than one signifier, for example in synonymy. Regardless of the differ-ences regarding the specific features of the signifier and the signified, andwhether it is legitimate to relate these semiotic concepts to their linguisticcounterparts of form and content, there is a general tendency among semi-oticians to agree that the concept of sign involves all the three elements ofsignification. These are the sign itself, the signifier and the signified.

Given that Saussure’s concept of langue is influenced by Durkheim’s notionof ‘collective mind’,13 it is equally reasonable to argue that his concept of sig-nification is also influenced by the same view. In this case Saussure could beinterpreted to mean that the reference of the sign is not necessarily the objectin the world but the mental concept the language users construct of that


1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 89

object. Each of the members of the speech community is expected to havehis own mental concept of that thing in the world, but these mental con-cepts will not be very different from each other because they are united witha stereotype of that thing. This perfect and holistic representation is whatresides in the collective mind of the speech community. The link I intend toestablish from this discussion is that the concept of sign in the Saussureanconception, although mentalistic in nature, is itself metonymic in the sensethat the signification or the concept and the sound pattern or the signalstand in a metonymic relation of representation between language and cog-nition. Reality is virtually ignored in Saussurean semiology but nonethelessthe model is still characteristically metonymic.

For Eco a sign is ‘always an element of an expression plane convention-ally correlated to one (or several) elements of a content plane’ (Eco 1976:48).In fact, this definition of sign is reminiscent of Saussure’s definition of signas the correspondence between a signifier and a signified. In a sense it is therelation of correspondence that makes the sign and hence there is nothingphysical called ‘sign’. It is an abstract relation that is drawn between a sig-nifier and a signified by means of conventionalised norms. Hjelmslev sug-gests that ‘it appears more appropriate to use the word sign as the name forthe unit consisting of content-form and the expression-form and establishedby the solidarity that we have called the sign-function’ (Hjelmslev 1961:58).Hjelmslev thus perceives ‘sign’ as a unit that is established by virtue of a cor-relation between an expression and content. This correlation becomes a sortof a code that as Eco argues ‘provides a rule to generate signs as concreteoccurrences in communicative intercourse’ (Eco 1976:49). This does notmean, however, that Hjelmslev is concretising the concept of sign by look-ing at it as a unit. This unit is a theoretical construct and the focus is mainlyon the set of relations that are established in the structure of the sign func-tion. This is because the sign function is a type of a language and languagefor Hjelmslev is a ‘structure, not to be discovered by the linguist but to beestablished by him’ (see Mulder and Hervey 1972:7).

Peirce’s model of sign, on the other hand, looks quite different because itperceives the concept of sign as comprising three facets, rather than two asin the Saussurean tradition. This is important to the course of this book,which aims to establish a three-dimensional model of metonymic significa-tion. Saussure’s notion of ‘signifier’ is called ‘representamen’ in Peirce’smodel and this is the form the sign takes. This can be both material andabstract. In other words, an abstract concept can function as a signifier andcan trigger some sort of interpretation of its own. The second aspect of signis that of the interpretant, which according to Peirce is not an interpreterbut rather the sense the representamen signifies. Now according to theabstract view of the concept of representamen, the interpretant could as wellqualify as a sign. So we have a kind of cyclic signification in which one signtriggers more than one interpretation in a successive way. Peirce’s third

90 Textual Metonymy

1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 90

concept is the object, and here the referent is the object in the world.While Saussure’s signified or reference is abstract, Peirce’s is material: thething in the real world. However, this does not mean that the Peirceanconcept of sign is material or that the sign is only physical. Rather, the signmight be both physical and abstract. It is the referent in the Peircean sensethat is physical. Figure 3.1 illustrates Peirce’s model and relates it tometonymy.

Figure 3.1 shows a great deal of similarity between the notion of ‘sign’ andthe notion of ‘metonymy’. This similarity is warranted by the view of bothnotions as representational. A sign mediates between a representamen, aninterpretant and an object. In the same way, metonymy mediates betweena word, a concept and an object. The notions of sign and metonymy facili-tate ‘stand-for’ relations between forms, concepts and objects. This is therepresentational view of metonymy that is sustained throughout this bookand will be discussed in detail in section 3.3 of this chapter.

Saussure’s concept of sign is structural in the sense that he was seeking anexplanation of the internal structure of the sign and the processes of signifi-cation and interpretation. In other words his concern was to build a modelwith emphasis on internal structure within the sign system. This led to histheory of linguistic signification, an insightful hypothesis that helped tofound the study of linguistics on semiological grounds. By means of its inter-nal structure, language creates its reality without the need to refer to the pat-terns of reality. The word ‘cat’ in the English language exists in the languagewithout need for reference to the four-legged, long-tailed animal. How doeslanguage do this? The answer is simply by means of internal structure throughwhich the word ‘cat’ behaves according to the laws of its status as a noun, i.e.an element in the linguistic structure and not according to its physical refer-ent as an animal. While Saussure’s semiology is instrumental, Peirce’s theoryof sign is representational. From an epistemological perspective Peirce


Interpretant�

� �

Representamen Object

Sign

Concept�

� �

Word Object

Metonymy

Figure 3.1 Metonymy and the Peircean notion of sign

1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 91

approaches the issue of signs and signification in an attempt to know howrepresentamina ‘stand for’ their objects.

The Saussurean emphasis on the internal structure of the sign systemessentially means that language creates its own categories by reference to itsown structural system. Deacon et al. (1999) in this respect maintain that ‘themeaning of a word exists only within a language system, only in differenti-ated relations to other words in that system’ (p. 136). Hawkes (1977) holdsthat ‘language … does not construct its formation of words by reference tothe patterns of “reality”, but on the basis of its own internal and self-sufficientrules’ (p. 17). This emphasis also justifies an early structural interest in look-ing at language as the determinant of our worldview.14 Because language cre-ates its own world, it imposes this world on reality, so that:

Human beings do not live in the objective world alone, nor alone in theworld of social activity, but are very much at the mercy of the particularlanguage which has become the medium of expression of their society. Itis quite an illusion to imagine that one adjusts to reality essentially with-out the use of language and that language is merely an incidental meansof solving particular problems of communication or reflection. The factof the matter is that the real world is to a large extent built upon the lan-guage habits of the group. No two languages are ever sufficiently similarto be considered as representing the same social reality. The worlds inwhich different societies live are distinct worlds, not merely the sameworld with different labels attached.15 (Sapir 1949:162)

Or as Whorf put it later on:

And every language is a vast pattern-system, different from others, inwhich are culturally ordained the forms and categories by which the per-sonality not only communicates, but also analyzes nature, notices orneglects types of relationship and phenomena, channels his reasoning,and builds the house of his consciousness. (Whorf 1956:252)

Bickerton writes ‘the categories into which we divide nature are not innature, they emerge solely through interaction between nature and our-selves’ (Bickerton 1990:53). This structuralist perspective regarding thedialectic relationship between language and thought has allowed for a hugebody of cultural studies, which sprang from anthropological linguistics andsociolinguistics. The essence of signification is the interaction between cog-nition, language and reality. The structuralist quest has always been toexplore the relationship between these three fundamentals.

Saussure’s theory of sign and signification tells us that language with itsinternal structure does not merely reflect reality, rather it creates its ownreality.16 Saussure’s concentration on the internal structure of the linguistic

92 Textual Metonymy

1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 92

sign and his emphasis on the notion of conventionality inform us that hisview of language is that of a medium as well as a topic. In other words, real-ity is what the language creates. This view is quite important for the theoryof motivatedness in linguistic signification that I am concerned with in thischapter. This is because if the process of meaning creation is purely con-ventional, i.e. culturally produced, then there is cultural motivatedness inthe creation of these meanings, as we will encode our ideologies and per-ceptions into the process of semiosis. It is true that this aspect is not as clearif we conceive of signification as taking place at the level of individual lex-ical items. However, once the signification process is seen at the level of textwhere the interaction between language, mind and reality is presumably inits most perfect form, the relationship between the signifier and the signi-fied is not at all innocent and is largely motivated by all sorts of cultural andeven ontological factors. The reason why motivatedness is much clearer atthe level of text is that once we select from the paradigmatic axis and formstrings of language to serve a particular communicative function, theprocesses of selection and combination are never innocent.

An important conception in this regard is Saussure’s concept of the arbi-trariness of the linguistic sign. As we have seen, according to Saussure, thereis no intrinsic relationship between the signifier and the signified (cf. Saussure1983:67). That is, the relationship is a matter of convention and the solesource of this is the speech community. One result of this is the tremendousnumber of languages spoken around the world. This means that the signifieris fundamentally a cultural product and is bound to change from one lan-guage to another unless the word in one language is borrowed from anotherto refer to the same meaning. In this case the act of borrowing is itself anintentional communicative act. Again we need to interpret this notion of arbi-trariness in a special way because an arbitrary theory of language significationmeans in essence that there is no physical correspondence between the sig-nifier and the signified. This means that nature does not impose its categorieson language; rather language creates categories to describe and explain nature.If we conceive of language as a social reality present in the unconscious of thespeech community, then this unconscious will actively operate when mean-ing is created to reflect all sorts of rational motivatedness in linguistic signifi-cation, especially at the textual level. The arrangement and organisation ofsigns is heavily influenced by ideological purposes and personal intentions aswell as cultural patterns of cognition.

In this connection Fairclough argues that

[i]deological struggle pre-eminently takes place in language. We can thinkof such struggle as not only in language in the obvious sense that it takesplace in discourse and is evidenced in language texts, but also over lan-guage. It is over language in the sense that language itself is a stake insocial struggle as well as a site of social struggle. (Fairclough 1989:88)


1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 93

Similarly, Kress and Hodge advocate a theory of motivatedness in linguisticsigns when they maintain that ‘linguistic signs are always motivated con-juncts of form and meaning’ (Kress and Hodge 1993:205). This would meanthat any use we make of language will be influenced by our ideologicalstance, the values we hold and the biases we adopt. Language is not an inno-cent reflection of reality; rather, it is a means by which we construct reality.What can be concluded from the above discussion is that our lexical andgrammatical choices are not innocent choices. There is always a struggle toattain desirable things and a struggle to avoid undesirable ones. Hence text-linguistics should account for this important aspect of language as a socialfact and as a product of social processes.

Hatim and Mason argue that the notion of motivatedness underlies theessence of language because ‘it is this notion which provides the essentiallink between textual occurrences and the context in which they are embed-ded’ (1997:24). Birch argues, on the same lines, that aspects of language ‘sig-nal different points of view, these in turn signal different realities, and theserealities determine and are determined by different (multiple) ideologies’(Birch 1993:43).

He argues further that ‘there is in any use of language a struggle for dom-inance; a struggle to bring about change’. Ghadessy puts forward a more spe-cific argument about motivatedness in linguistic signification when heargues that ‘our lexical and grammatical selections are not innocent choices.There is always a struggle for power which results in ideologically conflict-ing systems of classifying and controlling the world’ (Ghadessy 1993:3).

The view of motivatedness in linguistic signification seems to be some-thing that may not be disputed and if there is any chance to perceive lan-guage as innocent it is simply because we share the myths and biasesexpressed in our native languages. Cameron argues in this connection that‘there is always a point of view in language but we are apt to notice it onlywhen it is not the one we share’ (1995:74). Fairclough asserts that

the operation of ideology can be seen in terms of ways of constructingtexts which constantly and cumulatively impose assumptions upon textinterpreters and text producers, typically without either being aware of it.(1989:83)

The above discussion aims to argue against the notion of arbitrariness in orderto establish a theory of metonymy as a theory of motivated signification.

3.2.2 Types and modes of sign

There are various types of signs and there are various classifications ofsigns.17 Perhaps the classification which is most useful and relevant to ourdiscussion here is Peirce’s classification, which is based on the relationshipbetween the sign and its object. In a sense this can be seen as a classification

94 Textual Metonymy

1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 94

based on the mode of signification. According to this conception, signs havethree modes of signification. If the relation between the signifier and the sig-nified or rather between the sign and its object is that of similarity then thisis an iconic signification. The type of sign thus designated is an ‘icon’. If therelationship between the signifier and the signified, or rather between thesign and its object, is that of symbolism in that the signifier symbolicallysignifies the object, then this mode is symbolic and the sign thus designatedis a ‘symbol’. The third mode of signification is perceived when the rela-tionship between the signifier and the signified or rather the sign and itsobject is that of causality in the sense that the signifier causes the signifiedor the object of the sign. In this case the former is called the ‘cause’, the latteris called the ‘effect’ and the sign thus designated is an ‘index’.18

Another type of classification proposed by Peirce is based on the nature ofthe interpretant which each sign represents. This classification includes‘rheme’, ‘proposition’ and ‘argument’. A rheme can be understood as a refer-ring expression and representing a particular kind of object. Therefore,proper names would fall into this category because they merely name anentity in the real world. A proposition on the other hand involves two enti-ties. This is because propositions generally express an assertive point of viewregarding the state of affairs in the real world. In order to understand propo-sitions one has to go beyond the proposition itself which is the sign accord-ing to this conception to examine the conditions of the world of theutterance to judge the sign as true or false. An argument is a sign thatinvolves mediation between two entities: the premise and the conclusion.

An example can be provided to illustrate these three divisions of signaccording to the interpretant or the sense of the sign. If I point to a chairand say ‘chair’ then this is a referring expression and thus is understood asa way of naming or indicating that particular object. So the rheme is under-stood as a one-way process in which I do not need to draw a relationshipbetween two or more interpretants because there is no proposition about thechair yet. Now if I point to this chair and say ‘this chair is comfortable’, I am making a proposition and to confirm or falsify this assertion I need toexamine the conditions of the real world, i.e. to see whether the chair isactually comfortable or not. In this sense the interpretation of the interpre-tant requires a kind of a two-level process in which the sense of the signifi-cation is resolved by means of taking recourse to the world outside thesign-vehicle but which is part of the sign-object. The third example is thatwhen I proceed further to make an argument about the state of the worldand say ‘Because this chair is comfortable I always like to sit on it.’ Here thesign is mediating between a premise and a conclusion regarding the senseof the argument. This last type or model of signification is very relevant tothe intent of this book in two respects. First, it is related to the semiotic con-ception of metonymy as causal linking between significations and proposi-tions. Second, it is related to a textual theory of metonymy because it allows


1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 95

for the notion of signification to go beyond the sentence level to accountfor textual relations underlying discourse.

3.3 On representational semiotics of metonymy

This section aims to link the representational theory of semiotics and therepresentational nature of metonymy. I shall discuss the representationaltheory of semiotics by examining two theories that treat the process of sig-nification as a process of representation. The first of these theories is thatproposed by Aristotle in his book De Interpretatione. Aristotle’s statement ofhis sign theory runs as follows:

Now spoken sounds [phonai] are symbols [symbola] of affections in thesoul [pathemata], and written marks symbols of spoken sounds. And justas written marks are not the same for all men, neither are spoken sounds.But what these are in the first place signs of – affections of the soul – arethe same for all; and what these affections are likenesses of – actual things[ pragmata] – are also the same … A name is a spoken sound significant byconvention, without time, none of whose parts is significant in separa-tion. For in ‘Whitfield’ the ‘field’ does not signify anything in its ownright, as it does in the phrase ‘white field’. I say ‘by convention’ becauseno name is a name naturally but only when it has become a symbol. Eveninarticulate noises (of beasts, for instance) do indeed reveal something,yet none of them is a name. (Aristotle 1984:25; cited in Keller 1998:25)

Underlying Aristotle’s representational theory of signs is the assumptionthat the world of concepts is universal, and the world of actual things is uni-versal too. It is only the world of names that changes because of the arbi-trary nature of this name. So according to this conception signs ‘represent’.That is, by way of conventionality, they symbolise affections of the soul,which naturally resemble the actual things in the world. This is a cleveraccount designed to solve the dilemma of nature vs. convention whichGreek philosophers long debated about.19 It is clever because it combines theconventional aspect of name and at the same time it accounts for the nat-ural resemblance of affections of the soul to the actual things in the world.The question that arises is, how is it possible to verify that the affections ofthe soul actually resemble the actual things in the world? And, in fact, moreimportantly what is the individual role of cognition? According to thisconception all humans think in the same way but express these thoughtsdifferently. I think this is quite a faulty postulation because although a considerable part of human thinking is objective, i.e. shared by the groupor at least standing to some extent independent of the individual, the major-ity of the cognitive processes involved in experiencing reality result frompersonal interaction between the individual and the world. This is what

96 Textual Metonymy

1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 96

is meant by the term ‘idea’ that Frege talks about as being different from‘sense’ in this particular personal-vs-communal or subjective-vs-objectivedichotomy.

From the above quotation it is possible to distinguish three levels of rep-resentation that concern the concept of sign and signification. These arewhat Aristotle calls ‘spoken sound’, ‘affections of the soul’, and ‘actual things’.These correspond respectively to the linguistic, epistemological and onto-logical dimensions of phenomena. Aristotle’s presumption that sounds sym-bolise affections of the soul which are like actual things in the real worldgives rise to a crucial question as how these sounds or sound-forms symbol-ise these affections. A possible answer is that sounds symbolise their con-tents which are the affections of the soul here by means of standing for themor representing them. But this is not enough an answer because the samequestion comes to mind again: on what basis do sounds represent or standfor affections? It seems as if we are in a vicious circle in which no matterhow precise we are in forming our answers to such questions the outcomewill always lead to another question.

In a sense language as perceived by Aristotle is a representation of a repre-sentation, or in other words language is a representation of an imitation.Aristotle suggests that language is conventional but cognition is imitativeand universal. This conception is strongly challenged by cultural and anthro-pological studies. Most notably, the criticism of this theory comes from workdone on some American Indian languages like Hopi and Maya in theAmericas, by linguists like Edward Sapir, Franz Boas and, subsequently,Benjamin Lee Whorf, Berlin and Kay, and Rosch. The final formulation of therelativistic hypothesis was what is now known as the Sapir–Whorf hypothe-sis, the main argument of which is that humans are at the mercy of theirnative languages and dissect nature along lines laid down by those languages.The main issue of this hypothesis is that language is not innocent: it doesnot merely represent reality, rather it creates and constructs reality.

The relationship between the representational theory of signs and theconcept of metonymy follows from the fact that metonymy is largely rep-resentational. Metonymy is essentially a cognitive process in which certainparts of things ‘stand for’ these things in general. Hence metonymy providesa cognitive reference point for representation and as such it is an importantaid to categorisation which, according to Lakoff (1987), is the most basicaspect of our thought, perception, action and speech. This brings the processof metonymy very close to the phenomenon of linguistic representation inwhich a particular expression is conventionally associated with a particularcontent. Similarly, metonymy is a cognitive phenomenon where one thingbecomes associated with another thing until it conventionally stands for iton many occasions. So in the example

(3) Sana’a denied the accusations.


1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 97

because ‘Sana’a’ is the place where the Yemeni government is based, thiskind of contiguous association allows the former concept to stand for thelatter, i.e. it allows Sana’a to stand for the Yemeni government.

Let us examine now a theory of signs as representation expounded by themathematician and logician Frege. Frege (1966), in his paper entitled ‘OnSense and Reference’, provides a theory of sense and meaning in which healso conceptualises the three levels of phenomena discussed earlier, i.e. thelinguistic, the epistemological and the ontological. He states that ‘a propername (word, sign, sign combination, expression) expresses its sense, standsfor or designates its reference. By means of a sign we express its sense and des-ignate its reference’ (Frege 1966:61). Frege talks of linguistic expressions likenames, predicates and sentences. He also talks about ‘sense’, which could betaken to correspond to the epistemological level of representation discussedabove. However, Frege distinguishes between two epistemological levels ofexperience, which might be called the communal and the personal. Hisanalogy of observation of the ‘moon’ is indicative of this distinction. Forhim if someone observes the moon through a telescope, then the moonitself will be the reference. The term ‘reference’ here means the object of theobservation, i.e. what can be called the ontological level of representation,albeit in a different sense from the way I am using ‘ontological’ in this book.That is to say, ‘reference’ is not always the object of observation. Frege meansby ‘reference’ also the truth-value of a designated predicate, i.e. a sentence.The sentence, according to Frege, results from the designation of a particu-lar proper name to a specific predicate. When this sentence is created its ref-erence becomes its truth-value.

Again on Frege’s analogy, the moon is the reference and, according to himalso, the image on the lens of the telescope is called the ‘real’ image. Fregecalls this the ‘sense’. Then Frege talks about the retinal image and to this hegives the term ‘idea’ or ‘experience’. Now what is the difference between‘sense’ and ‘idea’? Frege answers this question by arguing that while the for-mer is the ‘objective’ meaning, the latter is the ‘subjective’ meaning. Hestates that

the optical image in the telescope is indeed one-sided and dependentupon the standpoint of observation; but it is still objective, inasmuch asit can be used by several observers. At any rate it could be arranged forseveral to use it simultaneously. But each one will have his own retinalimage. (p. 60)

So ‘sense’ is what everybody gets from the sign as its direct literal meaning,and ‘idea’ is the personal input into this general interpretation.

Frege perceives of signification to be triadic in the sense that the sign,which he regards as the linguistic level of representation or more preciselythe linguistic expression, is related to other two corresponding levels of

98 Textual Metonymy

1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 98

representation. These are the ‘sense’, which could be the ideational andcognitive level, and the ‘meaning’, by which Frege means the ‘reference’.20

This is the actual thing in the world. But the question that arises now is whatis precisely the relation underlying this type of correspondence? The answerto this question demands that we refer to his statement mentioned earlierin which he explicitly maintains that a proper name stands for its reference.Frege, as I mentioned earlier, is a mathematician and a logician. His generaltheoretical framework is the rules of logic and his main interest is to achievea theory of meaning that is essentially an attempt to explain the two majorconcepts in his article’s title, i.e. ‘sense’ and ‘reference’. It would seem as ifFrege is taking language for granted, and it is this treatment that makes histheory more of a theory of interpretation than one of signification. Senseand reference are the mental idea and the actual thing respectively.However, language is in fact underlying the objective interpretation of thesign which Frege calls ‘sense’.

The fact that Frege’s theory of meaning is perhaps more formal thancommunicative does not mean that it is totally devoid of interest for a the-ory of meaning as signification and more importantly a theory of meaningas representation. Although Aristotle’s and Frege’s theories of meaning areessentially representational, they exhibit certain differences. Perhaps themost notable of these is their account of the epistemological level. WhereasAristotle thinks it is one for all human beings because it is the affections ofthe soul which resemble the actual things in reality, Frege conceives of twolevels of epistemology: one universal or at least collective and the other per-sonal and individual. Yet both theories regard the relations underlying sig-nification as representational in which the sign ‘stands for’ the referent.

I shall now move on to discuss the relevance of this theory to the theoryof metonymy as I am proposing it. The distinction Frege proposes between‘sense’ and ‘idea’ is very relevant to the theory of metonymy for two rea-sons. The first reason concerns the way the Fregean theory of meaningallows for both a personal and subjective meaning, and a general and objec-tive meaning. Now the whole realm of rhetorical figures and figurative lan-guage can be comfortably located on the personal subjective side of themeaning creation process, whereas the literal meaning can be seen to resideon the general and objective side of the continuum. In fact to this end Fregehimself points out that even this notion of ‘idea’ – the personal side ofmeaning – has also the potential to achieve a consensus. According to Fregethis is the way art is created because people do not depend on sense alonewhen they interpret art. Rather, they make a great deal of use of humanideas, the absence of which ‘would make art impossible’ as Frege puts it(Frege 1966:61). So Frege distinguishes sharply between the sense of theword, the reference of the word and the idea that the word arouses in thehearer. This intermediate level of representation, which Frege calls ‘idea’, iswhere he places the interpretation of arts and it is the same level where I


1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 99

place figurative language as opposed to the literal language that may be saidto correspond to the Fregean notion of ‘sense’.

To drive home a point, Frege conceives of three terms, namely, ‘sign’,‘sense’ and ‘reference’. The fact that he recognises the sign to stand for thesense, which in turn stands for the reference, shows that he is totallyinvolved with literal objective signification here. Perceived as such, Frege’stheory of meaning might not be as relevant as it is hoped to be for the the-ory of metonymy as a process of signification. However, Frege does not stopat this level of explanation; rather he goes further to outline his theory of‘idea’, i.e. his view of certain aspects of personalised meanings, or, more pre-cisely, those ideational aspects that are aroused by the words in the hearersof these words. In a sense this is the level of the multiple interpretation ofthe sign. This does not mean that a particular sign has more than one inter-pretation. Rather, it means that the sign undergoes a cyclical process of sig-nification and interpretation. The sign signifies its referent and this referentitself becomes a sign on its own, arousing certain ideational meanings thatmight be peculiar to the hearer of the utterance. It is at this level of ideationalrepresentation that a potential for a figurative theory becomes available,because it is the level at which the initiator of the signification encodes hisown ideas, myths and beliefs and tries to lead the receiver to accept thesemyths and beliefs. It is also the same level whereby the receiver adds his ownideas, myths and beliefs to the sense of the word or the utterance.21

A simple example to illustrate the point might be the artificial sign of a fire alarm. This sign itself has been a target for mechanical significationthat took place at a previous stage, namely the detection of fire or smokewhich gave the signal to the fire alarm to make it ring. This fire alarm, per-ceived as a signifier, yields various further types of signification, and this Ibelieve is the part where there might be a correspondence with the notionof ‘idea’ according to Frege and where figurative language use, as I argue,resides. People’s responses to this sign differ widely. Some might run away.Others will call the fire brigade. Others, according to the sort of ideas thissign arouses in them, might prefer to remain seated, especially those whoare familiar with false alarms. This latter group might leave the place justlike others but they will not have got the same meaning from the sign; thatis, the sign will not be a sign of a real fire for them.22

I have tried to show that the Fregean notion of ‘idea’ is a useful tool for map-ping figurative uses of language as being subjective and to a large extent aes-thetic. This is true because literary language involves a great deal of personalpreference for certain styles, personal use of imagery, and so on. However, notonly is Frege’s representational theory of meaning useful to metonymy in thisgeneral sense, but also it is quite promising for the theory of metonymy in a more specific way. That is to say, because Frege’s theory of meaning accountsfor all aspects of meaning starting with the linguistic, the epistemological andthe ontological, it is helpful in treating metonymy as a process of signification

100 Textual Metonymy

1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 100

encompassing all these levels. Metonymy is a signifying process whereby onelinguistic item is substituted for another with which it is associated. Now thecriterion on which the substitution is based might not be shared by all recipi-ents of the message, and might not be a normal way of perceiving signifyingrelations. In other words these relations might be examples of ‘ideas’ in theFregean terminology simply because after all they are personal preferences forestablishing a relation between the literal and the figurative meaning. Yet thisdoes not mean they are left to random whims for their creation and interpre-tation. The reason is that the relations that underlie such a substitution arequite objective ones, and these make the sign creation and interpretationremarkably objective.

Soskice provides useful insights into the nature of representation in themetonymic operation. These, in fact, complement those of Frege’s andAristotle’s representational view on signification. Soskice argues that metaphorand metonymy seem to be similar on the surface but in fact are different asfar as their underlying representational relations are concerned. Metonymytends to be referential and this referentiality carries with it some factuality.She maintains that

Metonymy and synecdoche seem superficially similar to metaphor, butthey are functionally (that is, semantically) different. In metonymy andsynecdoche, one word or phrase stands in for a more straightforward ref-erence and this standing in is of a different nature from that which char-acterizes metaphor. The plugging in of an adjunct for the whole, or a morecomprehensive term for a less, is essentially an oblique and less prosaicway of making a direct reference. Instances of metonymy and synecdochepoint one directly to the absent item. It would be a failure in compre-hension if, on hearing the phrase ‘the White House said today’ one won-dered if shutters and doors opened like mouths. (Soskice 1985:57)

Soskice concludes that ‘with metonymy and synecdoche, meaning is largelysubsumed by the reference it makes’ (p. 58). The relation of referencebelieved to underlie the process of metonymy is quite important to the the-ory of metonymy as I am trying to develop it, particularly with reference tomy interest in developing a textual model of metonymic relations in text.

3.4 Metonymic aspects of the linguistic sign

This section is directly concerned with the explication of the metonymicbasis of the linguistic sign. The main aim is to argue that metonymy under-lies a great deal of language use. The fact that we do not often recognise thisdoes not mean that this influence does not exist. This is simply becausemetonymy is such an integral part of our cognition that most metonymiesactually go unnoticed. Here, I am concerned with the semiotic dimension


1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 101

of metonymy and more particularly of metonymy as a feature of linguisticsigns.

3.4.1 Metonymy as signification

If signification, as I pointed out earlier in this chapter, is a process of namingtwo entities and then abstracting a relation of ‘standing for’ between them,then metonymy is this and something more. This is to say that metonymyinvolves the process in which one thing comes to give a meaning other thanitself by means of various relations, and hence involves a process of signifi-cation or rather is a process of signification. Metonymy also adds to this a process of thinking on which the relation underlying the signification isfounded. This could be a relation of causality or contiguity in its variousfacets. This in turn means that the processes of metonymy are not arbitrarybecause they involve a considerable amount of thinking and rationality.

In fact, if we trace the development of the theory of metonymy over timewe find that signification is a common notion used in relation to the dis-cussion of metonymy regardless of whether the rhetorical tradition is thewestern or the Arabic. Among the western classical rhetorical treatises thattreat metonymy as signification or at least make explicit use of this notionare those of Sacerdos, Charisius, Isidore of Seville and Diomedes.23 In theArabic rhetorical tradition we find allusions to the notion of signification byal-Jurjäni who in his definition24 of the trope uses the word ‘yümi’u’, whichtranslates as ‘indicates’ or even as ‘signifies’, as an allusion to the process ofsignification by indication. This in fact makes al-Jurjäni’s definition evenmore relevant to the theory of metonymy than other classical proposals.

3.4.2 Metonymy as an index

In the discussion of the various modes of signification I mentioned thatPeirce perceived signification as having three modes. I am concerned herewith the third mode of signification suggested by Peirce. This is the modehe calls ‘index’ and he means by this the mode of signification in which thesignifier and the signified are related to each other in some indexical rela-tion. Peirce maintains in this connection that

the demonstrative pronouns, ‘this’ and ‘that’ are indices. For they callupon the hearer to use his powers of observation, and so establish a realconnection between his mind and the object; and if the demonstrativepronoun does that – without which its meaning is not understood – itgoes to establish such a connection, and so is an index. (Peirce 1955:110)

But Peirce’s index is not only deictic. It pertains also to any kind of con-tiguous connection between an interpretant and an object with or withoutthe mediation of the representamen. Nunberg defines the notion ‘indexi-cals’ as ‘expressions whose interpretation requires the identification of some


1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 102

element of the utterance context, as stipulated by their lexical meanings’.25

When a sign like smoke signifies fire this process of signification works onthe cognitive relation of causality in which smoke is perceived to be causedby fire and as such smoke is an index to fire. Now this cognitive linking isnot purely conventional because there is an intrinsic relationship betweenthe sign and the signifier. However, it is difficult to argue that there is noelement of conventionality involved in that people start associating smokewith fire in a conventional sense. In other words if we imagine that there isa community which never saw fire or smoke, it will be difficult for them toconceptualise the causal relation. This brings us back to the overlap betweendenotation and connotation as two inseparable aspects of signs.

Now I want to extend this notion further to include causality within thesymbolicity of the linguistic sign. In other words, language items althoughgenerally believed to be symbolic still exhibit certain aspects of motivated-ness or causality such that one meaning is seen to be the result of another.In language there are certain meanings that are seen to cause the existenceof other meanings. In more technical terms, there tends to be a relation ofcausality between specific meanings. The best example of this causality isthe notion of saliency, which is the major cause of metonymic expressions.So the fact that the customer ordering a ham sandwich is called ‘the hamsandwich’ is because this property is the salient feature of this customer inthis particular context. Thus it is quite plausible to ask, why is the customercalled ‘the ham sandwich’? That is asking for the motivation behind thenaming process. Similarly when we say

(4) The player headed the ball and scored the goal.

we can ask the question why the word ‘headed’ is used as a verb and we soonget the answer that the word ‘headed’ actually stands in a representationalmetonymic relation of part for whole with the whole action of ‘the playerhitting the ball with his head’. Again it is the saliency of the most importantelement in the action that comes to stand for the whole action. I want toexploit this to argue that metonymy is one of the fundamental operationsthat accounts for the phenomenon of motivatedness in language. Metonymyis fundamentally indexical. This indexicality is broader than the concept ofcausality because it goes beyond it to cover relations of contiguity, part–wholerelations and figure–ground relations and of course deictic expressions whichinclude both contextual and textual deixis. As such, metonymy is seen as atheory of pointing in speech action and referencing in text and this gives thetrope an empirically significant role in the creation and organisation of text.Metonymy as a signifying index will broaden the scope of the figure toinclude various facets of cognitive understanding. This indexical interpreta-tion of metonymy is significant for the study of text. I believe that thisconceptualisation will constitute a breakthrough in the study of texts in


1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 103

terms of accounting for these aspects in text and how they provide explicitcohesion and underlying coherence, which is the central aim of this book.

3.4.3 Indexicality vs. symbolicity in linguistic signification

The linguistic sign is generally symbolic in terms of the nature of its signifi-cation. This means that in the majority of cases there is no link whatsoeverbetween the signifier and the signified. This feature has been regarded as oneof the main features of human language and is generally known as ‘arbi-trariness’. In fact, this feature makes the diversity of the languages of theworld a comprehensible phenomenon. If there are numerous languages inthe world today it is because each and every language names things in real-ity according to the conventions of that speech community. The point to bestressed here is that linguistic signs are not only symbolic, as Peirce argues,but they can be treated as indexicals also. The theory of metonymy as index-ical signification proposes that the term ‘index’ be extended from its appli-cation outside language as a natural sign, as is the case with fire alarms, to beapplied with the symbolic nature of the linguistic sign to account for moti-vatedness in linguistic signification. This is the point I have been raising tosupport a view of metonymy as motivated signification. The sign enters intovarious relations with its signified. The sign ‘blade’, in the example

(5) The blade will solve the issue.

enters into a part–whole relation with its signified. This is to say that theblade is a part of the sword, which is intended in the sentence. The mean-ing intended is ‘The sword will solve the issue.’ Yet the sign ‘sword’ whichis the signified of ‘blade’ is itself a signifier for another signified. The sign‘sword’ signifies ‘war’ and this signification is fulfilled through the relationof instrumentality in which the sword is the main instrument of war. Itmight be thought that one needs to specify and say of ‘pre-modern’ war, butthis thought is to be rejected on the ground that although the sword is notused any more in wars of modern times, the word is still used on many occa-sions when there is reference to war. In recent years there has been a ten-dency to use the word ‘gun’ in exactly the same way; this also involvesmetonymic signification.

3.5 A relational model of metonymic signification

I am concerned here with outlining a relational model of metonymy whichwill be a practical tool to develop a textual theory of metonymy. The aimhere is to choose a prime relation that could be said to characterise theprocess of metonymic signification. One might choose the term ‘contiguity’suggested by Jakobson, taking into consideration that it is in fact a generalterm covering all those spatial, temporal and part–whole relations. This


1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 104

seems a good term and it looks comprehensive, but unfortunately it is notcomprehensive enough. Although it takes us far, it does not take us farenough to group all the relations that underlie metonymy under one generalterm. That is to say, the notion of ‘contiguity’ alone misses out the relationof causality, which is a fundamental metonymic relation.

I believe that the crucial relation that underlies the process of metonymyis that of ‘representation’. So metonymy is then a process of signification inwhich the signifier represents a signified which is associated with it in someway. A representational theory of metonymy will help us to extend thenotion of metonymy from its lexical confinement to account for textualanalysis. A representational theory of metonymy will help us look at thefunction of metonymy as that of representing one mode of knowledge byanother. The representational process of metonymy is in fact cyclical, unit-ing all the three knowledge domains.26

So if we have a text we can see this text as representing the cognitiveprocesses underlying it. If by means of cognitive reasoning we interpret some-thing from the text without it being explicitly stated in the formal text, thismeans that the formal text stands in metonymic representational relation ofpart for whole to the cognitive text; obviously the cognitive text is the wholeand the formal text is just a part of it. The same may be said with regard tothe ontological domain. Any formal text will be just one way of encodingsocial experience. So the social potential stands in a metonymic representa-tional relation to the formal actual. This relation of representation is realisedin various ways. Figure 3.2 illustrates the two fundamental metonymic prin-ciples that I take to be constitutive of the textual model of metonymy. Theillustration should read as: metonymy is a process of signification whoseprime nature is that of representation. Various relations underlying metonymyare perceived to emanate from this general relation.

The representational nature of the trope is realised via two main relations.These are the contiguity principle and the causality principle. The contiguityprinciple realises, among other relations, the part-for-whole and whole-for-part


Metonymy

Representation

Contiguity Causality

Part for whole Whole for part Cause for effect Effect for cause

Figure 3.2 A relational model of metonymic signification

1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 105

relations. These are taken to be the most relevant contiguity relations to thedevelopment of the textual model of metonymy. The causality principlerealises the cause for effect and effect for cause relations. These four relationsare the substantial material for the more abstract relations which account forthe interrelations between form, cognition and context. These more abstractrelations will be outlined below when I propose the textual model ofmetonymy.

The relational model of metonymy is in line with the general understand-ing of the patterns of knowledge structure.27 I argue throughout this bookand particularly in Chapter 5 below that knowledge is of two types. One isdescriptive and the other is procedural.28 The first pertains to the knowledgeof the ‘what’ and the second is concerned with the knowledge of the ‘how’.For the first type of knowledge, it is argued, we need contiguity metonymicrelations to describe and express relations between items in the text. In thesecond type of knowledge, it is argued, we need causality relations to realisethe connections between pieces of text and in this we are invited to accountfor intentions and plans and goals of the communicative act as a whole.

3.6 A textual model of metonymy

So far in this chapter I have established the strong relation between the con-cept of metonymy and the concept of sign, on the premise that both ofthem have the power to signify and represent across domains of reality, lan-guage and mind and provide coherence to our experience and communica-tion. In this section, I intend to propose a textual model of metonymy basedon a semiotic interpretation of the processes of the trope. This will be theultimate result of this chapter and will provide the basis for further discus-sion to be carried out in the next two chapters, which will apply this textualmodel to texts to identify their formal connectedness (cohesion) and theircognitive connectedness (coherence). The model I propose (see Figure 3.3)is fundamentally based on the conception of metonymy as representation,taking the metonymic relation ‘stand for’ to be the main and most relevantrelation to the model in hand.

The textual model of metonymy, which I propose in this chapter, is basedon two main sources. The first source is the triadic model of metonymic sig-nification proposed by Radden and Kövecses (1999) which argues thatmetonymic signification cuts across domains. That is the domain of words,the domain of concepts and the domain of objects. This triadic semioticinteraction between these three worlds makes the role of metonymy in bind-ing text very significant in accounting for the interactive nature of textualcommunication between these worlds. The second source is the theory ofidealised cognitive models (ICMs) proposed by Lakoff (1987). The theory ofICMs better characterises metonymy as a process of representation, becausewithin an ideaised cognitive model we can think of metonymic representation


1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 106

to provide the architectural links between genus and species, wholes andparts and even wholes and other wholes or parts and other parts associatedwith the entity at hand.

The model is divided into two main parts. The first is the world of forms andthe second is the world of concepts. I treat the first level as the level of formalsignification in text which involves five metonymic relations. These are:

(a) CONCEPT FOR FORM;(b) FORM FOR FORM;(c) FORM FOR CONCEPT;(d) FORM FOR THING;(e) THING FOR FORM.

I term this level of metonymic signification the level of cohesion. At thislevel the three realms of language, mind and reality interact but the inter-action is led by language and thus form has primacy. The second is the levelof cognition. This combines both the realm of concepts as well as the realmof objects because in textual interaction objects in reality do have indepen-dent existence outside cognition although they are seen as salient featuresof context in its immediate sense, i.e. situation, and in its wider sense, i.e.culture. This level is represented by the following relations:

(f) CONCEPT FOR CONCEPT;(g) CONCEPT FOR THING;(h) THING FOR THING;(i) THING FOR CONCEPT.

Two points need to be raised here about the model. The first is that themodel is interactionist in nature and as such the semiotic level builds on theformal level. The second concerns the fact that the model also incorporates


Representation by contiguity and causality

FO

RM

AL LE

VE

LS

EM

IOT

IC LE

VE

L

Level of inventionLevel of generationLevel of coherence

Level of arrangement and elocutionLevel of organisationLevel of cohesion

CONCEPT

FOR

FORM

CONCEPT

FOR

CONCEPT

CONCEPT

FOR

THING

THING

FOR

CONCEPT

THING

FOR

THING

FORM

FOR

FORM

FORM

FOR

CONCEPT

FORM

FOR

THING

THING

FOR

FORM

Figure 3.3 A model of textual metonymy

1403_913323_06_cha03.qxd 11/21/03 6:36 PM Page 107

ideas from rhetoric. So the formal level in the model below corresponds tothe parts of classical rhetoric known as ‘arrangement’ and ‘elocution’, whichcan be both termed ‘the level of text organisation’. This I have termed specif-ically ‘the level of text cohesion’. The semiotic level corresponds to the clas-sical term ‘invention’ and gives an idea about where the ideas and the formsin the text come from. This can be termed ‘the level of text generation’.I have termed this level ‘the level of text coherence’.


1403_913323_06_cha03.qxd 11/19/03 8:42 PM Page 108

109

The development of a discourse may take place along two differentsemantic lines: one topic may lead to another either through theirsimilarity or through their contiguity. The metaphoric way would bethe most appropriate term for the first case and the metonymic wayfor the second, since they find their most condensed expression inmetaphor and metonymy respectively.

Jakobson (1971:90)

4.0 Introduction

The previous chapter proposed a semiotic approach to metonymy anddeveloped a textual model for the trope. This chapter applies the model oftextual metonymy to account for text cohesion. The following hypothesesunderlie the chapter:

(i) If metonymy is essentially referential (see Lakoff and Johnson 1980;Lakoff 1987; Stallard 1993; Gibbs 1994; Nunberg 1995; Stirling 1996;Warren 1999) then it should help us understand more about referenc-ing in text. In fact, it is this referential basis, which is also the indexi-cal1 basis of metonymy developed in Chapter 3 of this book, thatenables metonymy to account for such a wide range of semantic rela-tions as the ones to be discussed in this chapter.

(ii) If metonymy is the manifestation of the syntagmatic and combina-tional dimensions of signification (see Jakobson 1971) then a theoryof metonymy is in fact a theory that should help us understand howcombinations in text are actually licensed.

(iii) If metonymy is based on relations of causality and contiguity (andalmost all researchers on metonymy agree on this), then metonymyis capable of providing a complete theory of text cohesion, given thefact that text connectedness is essentially based on these two basicrelations.

4Metonymy and Text Cohesion

1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 109

The chapter specifically applies the following metonymic relations:

A: 1. Form for form2. Form for concept to account for ‘reference’ in text.3. Form for thing

B: 1. Form for form2. Form for concept to realise ‘substitution’ in text.

C: Concept for form to realise ‘ellipsis’ in text.

D: 1. Form for concept2. Form for thing to deal with patterns of lexical3. Form for form cohesion in text.

Metonymic processes and relations have as their primary function the cohe-sive link between ideas, between concepts and between discourses. Theabove hypotheses have been developed and nurtured since the early part ofthis book, albeit in an implicit form. They awaited a relatively comprehen-sive account of the nature of metonymy to allow them all to be assimilatedinto the development of the textual theory of metonymy. In Chapter 1,sections 1.1 and 1.2, I outlined briefly how textlinguistics originated fromclassical rhetoric and benefited from various modern disciplines such as styl-istics, anthropology, sociology and cognitive psychology. In this chapter,I am concerned mainly with considering a theoretical framework formetonymy as a textual cohesive device in relation to the theory of cohesionas outlined by Halliday and Hasan (1976).

4.1 Metonymy and cohesion

In this section, I intend to outline the relationship between metonymy andcohesion. Although my discussion of metonymy as a textual cohesive deviceis based on Halliday and Hasan’s (1976) framework, I am actually develop-ing a new insight and approaching the topic from a new perspective. Forexample, I do not discuss cohesion as a set of surface text ties only, but fromthe point of view of its creation by metonymic relations in text. I have a strong intuition that the cohesive devices that Halliday and Hasan proposecan actually be seen as metonymic relations contributing to the texture ofdiscourse and making it stand as a unified whole.

A fundamental feature of my treatment of this semantic aspect of text isthat it is semiotic. This springs from the general semiotic approach of thewhole book. The semiotic approach to the interpretation of cohesion is evi-dent in the treatment of the concept in Halliday and Hasan’s framework. Thisframework treats the concept of cohesion as a semantic concept in whichone element signifies the meaning of another within the text. However, itoverlooks extralinguistic signification as exemplified by exophoric reference


}}

}

1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 110

and cognitive processing of text. Occurrences have a meaning not by them-selves but by being linked to other occurrences in the form of ties. Hallidayand Hasan’s theory of cohesion is entirely a theory of meaning in its actual-ity. It is a formal theory of meaning in text, in that it is the study of meaningas created by a set of formal features in the text.

It should be noted that Halliday and Matthiessen’s more recent work (1999)has also started to address the issue of the relationship between functional lin-guistics and cognitive processing. They place cohesion within the textualmetafunction which they associate with language as information as opposedto the notions of language as reflection and language as action, which are asso-ciated with the ideational metafunction and the interpersonal metafunctionrespectively. Halliday and Matthiessen argue that the textual metafunction isnot confined to information structure represented by theme – rheme organi-sation. Rather, this metafunction extends this aspect to account for the waylanguage points to itself as form and meaning (p. 528). They conceptualisetext as wording in which lexicogrammatical selection expressions are needed.The authors also conceptualise text as meaning in which semantic selectionexpressions are realised (p. 384).

‘Text’, for Halliday and Hasan, is ‘the word which is used in linguistics torefer to any passage, spoken or written, of whatever length, that does form a unified whole’ (1976:1). For Halliday and Hasan, however, a text is a seman-tic unit and not a unit of form. They conclude from this conceptualisationthat ‘[text] is related to a clause or sentence not by size but by REALIZATION, thecoding of one symbolic system in another. A text does not consist of sen-tences; it is REALIZED BY, or encoded in sentences’ (p. 2). In Chapter 3, section3.2.1 above I related the notion of sign in its three-dimensional nature to theconcept of metonymy which I established on a similar basis. I thus estab-lished a semiotic approach to account for the three fundamental aspects ofsignification. The first is the three-dimensional model of metonymy. The sec-ond is the three-dimensional model of sign. Here I present the third aspect,which is the three-dimensional aspect of text. Figure 4.1 is an illustration ofthe triadic representation of text.

Metonymy and Text Cohesion 111

Epistemology (cognition)

�

� �Language (form) Ontology (context)

Figure 4.1 A semiotic model of text

1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 111

Contextual and cognitive dimensions interact with the linguistic formspresented in the piece of language to make it ‘text’. Text is not even thecharacteristic outcome of ‘texture’, which is the actual presence of linguistic‘ties’ in text, as Halliday and Hasan argue. Text is essentially the process of‘meaning creation’ using explicit and implicit resources in and outside thetext. There is a continuous process of interaction, negotiation, expectationand prediction that a formal definition of text misses. In text production andtext interpretation sentences are not value-free because our understanding ofthese sentences necessarily requires a context. So it is better to view ‘text’ asrelated to utterances rather than to sentences.2 Each utterance calls for a con-text for its interpretation, and each utterance provides a context for the inter-pretation of those that follow.3

Halliday and Hasan discuss an important aspect related to textness. Thisis the concept of ‘texture’ which they view as a distinctive property of text.Although Halliday and Hasan’s notion of text is fundamentally semantic,their notion of texture, however, is formal,4 because it is the ‘linguistic fea-tures present in that passage which can be identified as contributing to itstotal unity and giving it texture’ (p. 195). Thus, Halliday and Hasan seem todisregard cognitive and contextual features that also give texture to text. Iargue that as soon as we incorporate metonymic processes as textual cohe-sive devices we will be able to view cognitive principles and contextual fac-tors as contributing effectively to the texture of text.

One of the principal devices of cohesion is that of reference. As we haveseen earlier in Chapter 2, sections 2.1.2.4 and 2.1.2.5, Stallard (1993) andNunberg (1995) have proposed a theory of metonymy as a referential phe-nomenon. They suggest that metonymy accounts for the phenomena ofindirect reference or deferred reference. Examples of this indirect referenceinclude:

(1) A: The Cabinet decided in favour of an increase in petrol prices yesterday.

B: They anticipated a lot of protest from the people, though.

In (1) above we notice an indirect anaphoric reference in the use of the pro-noun ‘they’ to refer to a singular inanimate antecedent. This is a violationof a fundamental property of the pronoun ‘they’. This pronoun normallyrefers to a plural antecedent. The solution to this ‘textual tension’ is pro-vided by a metonymic interpretation of the whole exchange in which‘the Cabinet’ is actually a surface noun phrase (NP) standing for an under-lying deep NP, ‘the ministers in the Cabinet’, with which the pronoun ‘they’naturally agrees.

I would, in fact, go even further than this quite limited treatment ofmetonymy as only a phenomenon of indirect referentiality in text to arguethat, in reality, all types of reference are metonymic, in the sense that there


1403_913323_07_cha04.qxd 11/21/03 6:35 PM Page 112

is a process of substitution in referencing and there is a process of significationin substitution. This view is even more plausible if we take into considera-tion that ‘cognitive reference’ is primary while ‘textual reference’ is sec-ondary, because for a thing to be present in text is a special case of it beingpresent in cognition at large. So in a sense whatever we find in a text standsin a part – whole relationship to what is in cognition. This is a propermetonymic relation. When we use a ‘pronoun’ we use it to stand for a par-ticular ‘noun’. Both of these terms, i.e. noun and pronoun, are actuallyforms of language. So there is a stand-for relation at stake. This is a funda-mental process of metonymic interpretation in which one form stands foranother form. Moreover, as we have seen in Chapter 2, section 2.1.2.3, whenI outlined Radden and Kövecses’ theory of metonymy, which in particularprovides a wider application of metonymy and claims that a relation of FORM

FOR FORM is an important metonymic relation, we could easily argue that alltypes of reference in text are actually metonymic of this type. All other typesof cohesive devices can be dealt with in the same way as we have done withreference. (This is not the place to elaborate on this; it will be discussedfurther in various sections of this chapter.)

A semantic theory of text connectedness should not limit itself to thenumeration of cohesive elements and cohesive ties, because this quantifica-tional approach has proved to be rather shallow. If text is a semantic unitand not a unit of form it follows then that a textual theory must be basedon a semantic interpretation of the nature of the relation that unites bothelements in the cohesive tie. To ensure semantic interpretation of text weneed to account not for the relation that connects ‘John’ to ‘he’ in the fol-lowing example:

(2) John is a student. He studies at Durham.

but also we need to account for the identity of reference and the seemingdiscrepancy between the elements of a cohesive tie. Consider the followingexample:

(3) The lobster did not leave any tip because he was not happy with theservice.

A semantic theory of text connectedness is limited to the statement that thenoun phrase ‘the lobster’ is connected to the pronoun ‘he’, which is not suf-ficient. A metonymic theory of text with its pragmatic dimension explainsthis discrepancy by resorting to the context of situation and hence identi-fying the phrase ‘the lobster’ as the person ordering lobster.

Substitution as a cohesive device is fundamentally metonymic, asmetonymy itself is basically a process of substitution. Now we can think of‘substituting’ one element for another in text as a relational operation which


1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 113

provides a text with formal connectedness.5 By the same token we canconceptualise metonymic substitution as playing a central role in providinglogical and cognitive texture and connectivity. This is to say that metonymyprovides text with a kind of schematic coherence in that both the virtualmeaning and the actual meaning are brought up and activated in the mindof the reader to give the reading of the text a more encyclopaedic interpre-tation. This is not at all to say that metonymy as a process of substitutiondoes not provide formal cohesion. On the contrary, substituting one ele-ment for another is a fundamental part of the FORM-FOR-FORM metonymicrelation. It is only to say that metonymic substitution in text provides bothformal connectivity in text, that is cohesion and cognitive connectivity, andrelevance in text, that is coherence.

If we recall Ullmann’s theory of metonymy as semantic change and hisdesignation of a change based on ‘contiguity of names’ as ‘deletion’ wewould appreciate highly the kind of logical link to be established nowbetween metonymy and ‘ellipsis’. Metonymy as a representational relationof CONCEPT FOR FORM, where an empty slot stands for a previously mentionedform, is essentially based on spatio-temporal contiguity. This feature occursin linguistic structuring of utterances and in text organisation in general. Sowhen we say


we actually conceive of two types of contiguity. One is cognitive and is basedon the PART-FOR-WHOLE relation in which ‘the keel’ stands for the whole ship.The other is a spatio-temporal contiguity created by the spatio-temporalsequencing of the principle of combination of the syntagmatic axis of thelinguistic structure. We could also say

(5) The keel of the ship ploughed the deep sea.

Perhaps this was the original expression and as it became conventionalised,people started to use the form ‘the keel’ to stand for the whole expression‘the keel of the ship’ and ‘the deep’ to stand for the whole expression ‘thedeep sea’. The point I want to make here is that in metonymy as a processof ellipsis there is not only logical or cognitive ellipsis but also formal ellip-sis, as we can see from this example. Similarly, the example

(6) I want to go to the gents.

is a good illustration of a metonymically based ellipsis in which the originalexpression is ‘the gents’ toilet’, but as this expression became very routinisedthe word ‘toilet’, perhaps for euphemistic reasons, dropped out of theexpression and the form ‘gents’ came to represent the whole original


1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 114

phrase.6 As we shall see, ‘ellipsis’ plays a major role in text cohesion and textcoherence since people tend to assume that their partners in the discourseare actually active inferencers, so they leave out many details on the assump-tion that these will be supplied by the general knowledge of the universe ofdiscourse. This phenomenon of ellipsis can be accounted for from ametonymic perspective as well. Perceived as such, metonymy ensures econ-omy and compactness in text and thus shortens distances of interpretation.

Lexical cohesion is also a resourceful area in which metonymy can befruitfully applied. I shall reserve my detailed discussion of the role ofmetonymy in relation to each of these cohesive devices, with exampletexts, to the appropriate section. All I am concerned about now is provid-ing a general introduction to the relevance of the theory of metonymy as a textual tool for the theory of cohesion as propounded by Halliday andHasan, who discuss cohesion under two main headings, grammatical cohe-sion and lexical cohesion. These will be considered here under the sameheadings.

4.1.1 Metonymy and grammatical cohesion

Grammatical cohesion is the type of cohesion that includes reference, sub-stitution and ellipsis. Halliday and Hasan maintain that these three aspectsof grammatical cohesion are ‘clearly grammatical [in] that they involveclosed systems: simple options of presence or absence and systems suchas those of person, number and proximity and degree of comparison’(1976:303).

4.1.1.1 Metonymy and reference

Reference is treated in this book as a semiotic phenomenon realising signi-fication between language, cognition and context. So the view expressedhere of reference is that of a phenomenon that cuts the domains of forms,concepts and objects, and not merely of textual reference, to which it isreduced in Halliday and Hasan’s (1976) model, which treats ‘reference’ asthat linguistic phenomenon in which certain items in language make refer-ence to something else for their interpretation instead of being interpretedsemantically in their own right (1976:30). The authors call such ‘referenceitems’ ‘directives’, as they ‘indicate that information is to be retrieved fromelsewhere’. This is not very different, it should be noted, from the treatmentof the phenomenon called PRO-FORMS by de Beaugrande and Dressler (1981).Pro-forms are forms that ‘can stand in the surface text in place of more deter-minate, context-activating expressions’ (1981:60).

In a more recent account of the issue, Halliday and Matthiessen (1999)consider reference as ‘the way of referring to things that are already semiot-ically accessible: either actually, in the text, or potentially, in the context ofsituation’ (p. 530). It can be noted that there is a growing interest in contextin Halliday’s theory of text. Here it is the context of situation. Halliday and


1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 115

Matthiessen raise the issue of actual reference and potential reference andthis suggests that they view text as an actualisation of one of the manypotential choices the language user has.7

Both of these views, that is reference and pro-forms, are similar to thenotion of ‘sign’, which in addition to the definitions mentioned in Chapter 3,section 3.2.1, as we have seen in Chapter 3 can also be defined as ‘an ele-ment that conveys a meaning other than itself’. I relate to these two per-spectives a third one which concerns the conceptualisation of metonymy asa pro-form also. If the reference item points at another item then that issome sort of signification. This signification can be symbolic or iconic orindexical. With regard to ‘reference’ it is always indexical because there is anact of pointing each time there is reference. In the example

FORM FOR THING

FORM FOR CONCEPT

FORM FOR FORM

(7) John went to the market. He bought some apples.

the reference item ‘he’ does not mean anything by itself apart from ‘its ownconceptual signification’, that is a third-person-singular pronoun. Its func-tion in the organisation and interpretation of text is dependent on perceiv-ing it as a component in a signifying system or an ‘idealized cognitivemodel’ in Lakoff’s terminology or ‘space’ in Fauconnier’s account. The reso-lution of such a holistic frame requires a cognitive movement to theantecedent.

Whether we perceive the relation between ‘John’ and ‘He’ as forms or asconcepts or even as entities and objects we cannot escape a metonymicinterpretation of this relation. By means of the metonymic representationalrelations of FORM FOR FORM, FORM FOR CONCEPT, CONCEPT FOR FROM, FORM FOR

THING, THING FOR FORM it is possible to propose a universal model of referencein text. The first two relations account for the phenomenon of reference inlanguages like English and Arabic which have the tendency to use the propernoun or the common noun at the beginning of the text and rely on theuse of pronouns to refer back to the proper or common noun throughoutthe discourse. The latter relations account for the phenomenon of referencein languages like Hebrew, Japanese and Chinese. According to Callow,‘Hebrew, unlike English, prefers to use proper names to trace participantsthrough a discourse’ (Callow 1974, cited in Baker 1992:183). Baker alsopoints out that

in some languages like Japanese and Chinese, a totally different patternseems to be in operation. Pronouns are hardly ever used and, once a par-ticipant is introduced, continuity of reference is signalled by omitting thesubjects of following clauses. (p. 185)


} }

1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 116

In this case reference is, in fact, some kind of ellipsis. It is, therefore, theconcept of the identity of the participant that is retrieved each time anaction or an event is attributed to that subject. As such it is the metonymicrelation of CONCEPT FOR FORM that is in operation here. The remaining rela-tions account for the pragmatic dimension of reference and particularly thereference involved in the use of deictic expressions.

Halliday and Hasan classify ‘reference’ into two broad categories accord-ing to whether this reference is situational, that is referring to somethingoutside the text, or textual, that is referring to something within the text.Whether it is exophoric or endophoric, based on its function, reference isclassified into three types:

(i) personal reference;(ii) demonstrative reference;(iii) comparative reference.

I shall take an example of personal reference to show how this is directlyrelated to metonymic reasoning. All first- and second-personal pronounsshow a tendency to express exophoric reference, although they do notalways do so. On the other hand, third-person pronouns tend to expressendophoric reference. However, in all these cases we are better off if weinterpret this reference as metonymic, whether it is direct or indirect. Thisis to say that ‘reference’ in its reduced form as textual reference is able toexplain phoric relations of the type:

FORM FOR FORM, FORM FOR CONCEPT, FORM FOR THING

(8) Tim is a professor of politics. He lectures at Durham.

FORM FOR FORM, FORM FOR CONCEPT, FORM FOR THING

(9) Janet is a lecturer in Arabic. She speaks very good Arabic.

The relationship between ‘Tim’ and ‘He’ in example (8) and ‘Janet’ and ‘She’in example (9) is cohesive because both elements form a tie and the iden-tity of reference is maintained. But what are we going to say about texts likethe following?

FORM FOR FORM, FORM FOR CONCEPT

(10) Halliday is on the top shelf. You will find it in paperback.


(11) Table 4 is getting impatient. He has been waiting for a long time.


1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 117


(12) The ladies is getting dirty. It needs to be cleaned as soon as possible.


(13) The diabetes is leaving tomorrow. He is better now.

A metonymic interpretation of the relation between the pronouns in theexamples above and their antecedents is obligatory because in each instancethere is a case of deferred reference and an apparent discrepancy in terms ofsurface agreement. The identity of the reference is not maintained. Meto-nymic processes capture reference relations better because they operatewithin idealised cognitive models that include the three dimensions of sig-nification. They include the forms, the concepts and the things. Themetonymic relations of form for form explain referencing in text as a formalphenomenon, but these relations do not stop at that level. Rather, they alsoexplain cognitive relations between the elements in the text as concepts oras things.

4.1.1.2 Metonymy and substitution

This section applies the metonymic representational relations of FORM FOR

FORM and FORM FOR CONCEPT to explain patterns of substitution in text. Thebasis of my argument here regarding the close relationship betweenmetonymy and substitution is the fact that metonymy itself is a form of sub-stitution. Metonymy is a process of substituting one item for another withwhich it is associated. This is precisely the case with Halliday and Hasan’snotion of ‘substitution’, which they define as ‘the replacement of one itemby another’ (1976:88). Halliday and Hasan maintain that ‘substitution, onthe other hand, is a relation within the text. A substitute is a sort of counterwhich is used in place of the repetition of a particular item’ (p. 89). The exam-ples the authors give to illustrate this particular point are the following:

FORM FOR FORM

(14) My axe is too blunt. I must get a sharper one.

FORM FOR FORM

(15) You think Joan already knows? – I think everybody does.

‘One’ in example (14) substitutes for the word ‘axe’ and ‘does’ in example(15) substitutes for the word ‘knows’. Halliday and Hasan divide ‘substitu-tion’ into three main types according to the grammatical function of the


1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 118

substitute item. These three types are:

(i) NOMINAL SUBSTITUTION: ONE, ONES, SAME

The following are some examples of nominal substitution taken fromHalliday and Hasan:

FORM FOR FORM

(16) These biscuits are stale. – Get some fresh ones.

FORM FOR FORM

(17) This bread’s stale. – Get some fresh.

FORM FOR FORM

(18) Mummy will you buy me a bus? I want that red one.

(ii) VERBAL SUBSTITUTION: DO

Here are some examples provided by Halliday and Hasan for verbal substitution:

FORM FOR FORM

(19) The words did not come the same as they used to do.

FORM FOR FORM

(20) Does Granny look after you every day? – She can’t do at weekends,because she has to go to her own house.

(iii) CLAUSAL SUBSTITUTION: SO, NOT

I take the metonymic relation of FORM FOR CONCEPT responsible for clausalsubstitution relations because although the connector ‘so’ in examples (21)and (22) below refers to the clauses ‘You agree to have a battle,’ and ‘Is theregoing to be an earthquake?’ respectively, it clearly does not refer to theactual form of these grammatical units but to the proposition expressed bythem. Let us consider the following examples of clausal substitution pro-vided by Halliday and Hasan:

FORM FOR CONCEPT

(21) ‘Of course you agree to have a battle?’ Tweedledum said in a calmertone.‘I suppose so’, the other sulkily replied, as he crawled out of theumbrella.


1403_913323_07_cha04.qxd 11/21/03 6:36 PM Page 119

FORM FOR CONCEPT

(22) Is there going to be an earthquake?– it says so.

Halliday and Hasan’s notion of substitution relates to metonymy in variousways. Substitution is largely anaphoric in nature in that it usually refers backto elements mentioned earlier. The nature of this reference is nothing butsome sort of internal signification, that is textual signification in which oneform signifies another. Halliday and Hasan maintain that the differencebetween reference and substitution is that while the former is semantic thelatter is essentially grammatical.

Substitution involves categories that not all languages have, and thisexplains why Halliday and Hasan’s model of cohesion is a model of cohe-sion in English. For example, Arabic seems to lack ‘auxiliary verbs’ whichtake the place of main verbs to show connectivity via substituted forms. Ifthe following example (from Baker 1992:187) is to be translated into Arabic:

FORM FOR FORM

(23) You think Joan already knows? – I think everybody does.

where ‘does’ stands in FORM FOR FORM metonymic relationship with ‘knows’,there will be difficulty in retaining the same substitution pattern in theArabic rendering of the expression. Arabic would repeat the same verb, asthe example in (24) shows:

FORM FOR FORM (THE SAME FORM REPEATED)

(24) hal ta ‘taqid ’anna jawan ta‘rif? ’a‘taqid ’anna al-kull ya‘rif.

However, Arabic seems to accept nominal substitution in some markedstructures like (26) below:8

FORM FOR FORM

(25) My axe is too blunt. I must get a sharper one.

FORM FOR FORM

(26) Fa’sï asbaha sähinan. Yajibu ’an ’äkhudha wähidan ’äkhar.‘My axe became blunt. I must get another one.’

where wähidan ‘one’ stands for fa’s ‘axe’. There is also substitution of thetype FORM FOR CONCEPT and is attested in both Classical as well as modern


1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 120

varieties of Arabic. Let me begin with an example in classical Arabic, takenfrom the Qur’an. Usually this substitution is motivated by factors related toeuphemism and is usually expressed via the verb fa‘ala ‘to do’. In the Qur’anfor example, the Pharaoh argues with Moses about the latter’s childhood inthe former’s palace. The text goes on as follows:

(27) qäla ’alam nurabbïka fïnä walïdan wa labithta fïnä min ‘umurikasinïna, wa fa‘alta fa‘lataka allatï fa‘alta wa’anta min al-käfirïn, qälafa‘altuhä ’idhan wa’ana min al-dhällïn. (26:18–20)

Translation:(Pharaoh) said: ‘Did we not cherish you as a child among us, and did younot stay among us for many years of your life? And you did your deedwhich you did while you were ungrateful. Moses said: “I did it then whenI was in ignorance.” ’

The reference in the use of the verb fa‘ala ‘did’ and its derivatives inexample (27) above is to the action of Moses’ killing of an Egyptian. It seemsas if the euphemistic function of the substitution involved in the versesabove is dominating. Because killing and death are considered and also wereconsidered at that time ‘taboo’, the reference to them was conveyed via ageneral lexical item which is the verb fa‘ala ‘did’. Similarly, this lexicalitem is still used in almost all dialectical varieties of Arabic today to conveythe same euphemistic function. In a criminal report one would normallyencounter the following:

(28) Ba‘d ’an fa‘ala al-mujrimün fi‘latahum harabü.‘After the criminals did their deed they ran away’.

Now it could be argued that the two examples above are actually of themetonymic type FORM FOR CONCEPT, rather than examples of substitution parexcellence, which is conveyed via the metonymic relation of FORM FOR FORM,for the simple reason that there are no forms which have been substituted,that is it is a euphemistic style of covering taboo expressions. However, if weconceive of text as an integral world of the socio-cultural practice of a speechcommunity then the notion of intertextuality and the concept of interac-tions between texts will be the norm of social cognition. As such, we willrealise that the form fa‘ala ‘did’ and its derivatives in the Qur’anic verseabove is actually a substitution of the form qatala ‘killed’, which is men-tioned in another Qur’anic verse – in a different chapter within a narrativeof the same story – and hence the substitution is actually a FORM FOR FORM.The same thing applies to the criminal report example. One would easilyassume that the actual form of ‘killing’ or ‘raping’ has been mentionedbefore, and these forms, that is fa‘ala and its derivatives, have as theirreference forms also.


1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 121

4.1.2 Metonymy and ellipsis

The aim of this section is to implement the metonymic representationalrelation of CONCEPT FOR FORM to explain patterns of ellipsis in text. Ellipsis is‘substitution by zero’, as Halliday and Hasan argue (1976 (p. 142)). Theauthors define ellipsis as ‘something left unsaid’. Halliday and Hasan admitthat this definition might be misleading because of its generality, since noteverything left unsaid is an example of ellipsis. The interpretation of anytext involves a great deal of ‘inferencing’ in which we supply much of infor-mation from outside the text. Is this ellipsis? For Halliday and Hasan it isnot, nor is it cohesive, as it does not relate to textual relations. So what thenis ellipsis? Halliday and Hasan point out that when talking about ellipsis ‘weare referring specifically to sentences, clauses and so on whose structure issuch as to presuppose some preceding item, which then serves as the sourceof the missing information’ (p. 143). They elaborate this point further andstate that ‘an elliptical item is one which, as it were, leaves specific structuralslots to be filled from elsewhere’. So the idea that differentiates ellipsis fromany other form of information supplying or inferencing is that it forms a slot in the structure of the text.

Like ‘substitution’, ellipsis is divided into three main parts:

(i) NOMINAL ELLIPSIS Halliday and Hasan define this type of cohesion as ellipsis that takes place within the nominal group. They give the followingexamples as illustrations:

CONCEPT FOR FORM

(29) Which last longer, the curved rods or the straight rods? – Thestraight � are less likely to break.

CONCEPT FOR FORM

(30) Which hat will you wear? – This is the best �.

CONCEPT FOR FORM

(31) The first three buds fell off. We’ll have to watch the next �.

(ii) VERBAL ELLIPSIS Halliday and Hasan define this type of cohesion as ellipsiswithin the verbal group. They distinguish between two types of ellipsis:

(a) Lexical ellipsis. This occurs from the right, that is the lexical element inthe verbal group which normally occurs to the right of the verbal group.


1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 122

CONCEPT FOR FORM

(32) Have you been swimming? – Yes, I have .

(b) Operator ellipsis. This is ellipsis from the left; the initial element orelements from the verbal group are omitted except the lexical verb. This isvery similar to clausal ellipsis because the subject and the auxiliary verb areomitted.

CONCEPT FOR FORM

(33) What have you been doing? – Swimming.

(iii) CLAUSAL ELLIPSIS According to Halliday and Hasan verbal ellipsis involvesother elements in the clause which are not part of the structure of the verbalgroup but elements of the clause structure. This justifies their coining ofclausal ellipsis to account for omissions external to the structure of the ver-bal group. Halliday and Hasan define the notion of clause in English as ‘theexpression of the various speech functions, such as statement, question,response and so on, [which] has a two-part structure consisting of MODAL

ELEMENT plus PROPOSITIONAL ELEMENT’ (p. 197).Halliday and Hasan define these two elements as follows: ‘The MODAL ele-

ment, which embodies the speech function of the clause, consists in turn ofthe Subject plus the finite element in the verbal group.’ With regard to theother element in the structure of the clause, the authors continue, ‘ThePROPOSITIONAL ELEMENT consists of the residue: the remainder of the verbalgroup, and any Complement or Adjunct that may be present’ (p. 197).

Halliday and Hasan conclude that ‘the two types of verbal ellipsis arederivable from these two major divisions of the clause’ (p. 197). This is tosay that in elliptical clauses it is either the modal element that is omitted orthe propositional element. Halliday and Hasan argue that modal ellipsisoccurs in response to WH- questions, that is questions beginning with whwords like what, where, when and so on and provide two examples of thistype of ellipsis as follows:

CONCEPT FOR FORM

(34) What were they doing? – Holding hands.

CONCEPT FOR FORM

(35) What were you doing? – Swimming?

As regards the propositional type of ellipsis, Halliday and Hasan (1976) arguethat it is ‘associated with those instances where the mood and the polarity


1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 123

are the principal components of the message: typically, responses tostatements and yes/no questions, where the subject is presupposed by a ref-erence item’ (p. 198). They provide the following two examples as illustration:

CONCEPT FOR FORM

(36) The plane has landed. – Has it ?

CONCEPT FOR FORM

(37) Has the plane landed? – Yes, it has .

The classification of clausal ellipsis in Halliday and Hasan into two majorgroups as seen above corresponds to some extent to the division of hadhf‘ellipsis’ in the Arabic rhetorical tradition. According to ‘Atïq ‘the musnad[predicate] and musnad ’ilayhi [subject], which are the two fundamental com-ponents of the sentence, are subject to various operations for the sake ofrhetorical purposes’ (1985:122). Among these operations is hadhf which,according to ‘Atïq, is divided into three types. The first is the hadhf of musnad’ilayhi, which ‘Atïq explains in detail, especially with regard to the reasonsleading to this type of ellipsis. Among the examples of this type of ellipsis‘Atïq gives the following from the Qur’an:

CONCEPT FOR FORM

(38) Wa’in tukhälitühum fa � ’ikhwänukum (2:220).‘If you mix their affairs with yours then your brothers.’

In the above verse there is an example of ellipsis in which the subject ‘they’is elided from the second clause. The verse should read ‘if you mix theiraffairs with yours then [they are] your brothers’. Here again we have themetonymic representational relation CONCEPT FOR FORM which provides unityand connectedness to the text. Whenever, this verse is read there is a cog-nitive slot felt to be filled with a subject. According to ‘Atïq, the passive con-structions fall into this category of ellipsis because they involve the omissionof the subject, usually to hide the doer of the action for safety reasons orbecause it is so well known that it need not be mentioned. An example givenfor this type of ellipsis is the following:

(39) Kusirat al-näfidhatu.‘The window was broken.’

We could imagine the context of this utterance to be in a house, for exam-ple, where the mother is reporting the incident of one of the children having


1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 124

broken the window. She opts for the passive, deliberately omitting thesubject so that the father does not beat the child.

The second type of ellipsis in Arabic is related to the omission of the mus-nad ‘predicate’. An example of this type of hadhf is the following:

CONCEPT FOR FORM

(40) Man huwa ahsan shä‘ir ‘Arabi?Al-Mutanabbï .‘A: Who is the best Arab poet?B: Al-Mutanabbï.’

Here we have an ellipsis of the predicate because a complete answer to the question should have been ‘Al-Mutanabbï is the best Arab poet’. Themetonymic representational relation is obviously at work here to providecontinuity for the text by means of the CONCEPT FOR FORM.

The third type of ellipsis in Arabic is the omission of the ‘object’. Amongthe Qura’nic examples of this type of ellipsis are the following:

CONCEPT FOR FORM

(41) Wallähu yad‘ü ’ilä däri al-saläm. (10:25)‘God calls to the Home of Peace.’

CONCEPT FOR FORM

(42) Walaw shä’a Allähu � lahadäkum ‘ajma‘ïn. (6:149)‘Had He willed He would indeed have guided you all.’

In example (41) we have the object ‘all his servants’ omitted to give the utter-ance a generalising force. The verse means ‘God calls [all His servants] to theHome of Peace.’ Example (42) indicates an ellipsis of the object ‘your guidance’to make the utterance more effective. The verse means ‘Had He willed [yourguidance] He would indeed have guided you all.’ In all these examples the typeof ellipsis involved is that of syntactic ellipsis, which is not significant for a textual interpretation of ellipsis. Nonetheless, we can see the metonymic rep-resentational relation of CONCEPT FOR FORM at work even though the concept isactually made obligatory because it is sometimes the object of a transitive verb.

4.1.3 Metonymy and lexical cohesion

This section aims to explain the role of the metonymic representational rela-tions of FORM FOR FORM, FORM FOR CONCEPT and FORM FOR THING in the developmentand organisation of lexical patterning in text. Lexical cohesion is the categoryof cohesion which concerns connectivity between lexical items in the text. It isthus different from the grammatical cohesive devices discussed so far, in that it


1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 125

is not a feature of grammatical dependency and also it does not represent a relation between forms but between meanings and concepts. Lexical cohe-sion is the result of a network of semantic relations underlying the selection oflexical items in the text. The main argument underlying this section is thatmetonymy explains all the types of lexical cohesion and adds to this the factthat it brings all these relations under one coherent system of conceptual rela-tions that apply globally in the text to ensure maximum ease of processing ofthe text as a unified whole. In the following sections this argument will be madeclear with examples showing how metonymy accounts for the relations of lex-ical cohesion in a more satisfactory way than the term ‘lexical cohesion’ itself.

I argue that a stand-for relation is a general relation that satisfactorilydescribes all types of metonymic understanding and at the same time describesall types of cohesive devices in text. Although this relation is rather general, itis always based on certain specified cognitive principles and these actually con-strain the vagueness of the term ‘stand for’.9 I also argue that in the process oftextual analysis we are required to show the set of cognitive models and con-ceptual schemata that actually generate the text and make it stand as a unifiedwhole. These models can be attained only through a characterisation of whatrelations actually underlie the generation and organisation of the text.Therefore, I propose a new way of analysing a text which brings more than onefactor together to the analysis process. First, it is not concerned only with theactual; it also cares about the virtual. Second, it does not concern itself onlywith the organising principles of text but rather it goes deeper and is concernedabout the generating principles of text. Many organising principles are as theyare because of the influence of the generating principles of text on them.

4.1.3.1 Metonymy and reiteration

Halliday and Hasan use ‘reiteration’ as an umbrella term to cover certain lex-ical cohesive devices such as repetition, synonymy, superordinates and generalwords. They define ‘reiteration’ as a general phenomenon in which ‘one lex-ical item refers back to another to which it is related by having a commonreferent’ (p. 279). Reiteration as a lexical cohesive device is quite similar to‘reference’ in that both devices have an anaphoric referential function. Theonly difference between them is that while the former uses pronouns the latteruses various forms of semantically related lexical items having the same refer-ent. So in the following example the words ‘bull’, ‘animal’, ‘creature’, ‘buffalo’,and so on are all words which have the same referent, that is the male cow.

(43) I was on the farm last weekend. I saw a bull there. The creature wasfriendly. animal

bovinebisonbuffalobull


1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 126

This example suggests a general principle of relatedness between the lexicalitem used and the noun with which it is co-referential. This relatedness mightbe synonymy, as in ‘buffalo’ and ‘bison’, or superordinate, as in ‘animal’,‘bovine’ and ‘creature’, or it might be a simple repetition, as in ‘bull’. Perceivedfrom a formal perspective, the metonymic representational relation underlyingthe choice of any of the above lexical items would be that of FORM FOR FORM. Sothe form ‘creature’ stands for the form ‘bull’ in this particular example.However, since Halliday and Hasan define the nature of the relation betweenthe lexical items in reiteration as that they have a common referent, thismetonymic relation will not be sufficient to characterise the semantic linkage,although it is useful in informing us that the two forms are related, thus pro-viding formal connectivity. It seems, therefore, that we need cognitive and con-textual relations to account for the connectivity between the lexical items inthe above example. Two fundamental relations are actually responsible for theconnectivity between the lexical items. These are the FORM FOR CONCEPT and theFORM FOR THING. In this way we account for the relationship between the signi-fier, which is a linguistic form, and the signified, which might be a cognitiveconcept in the Saussurean sense or an ontological object in the Peircean sense.

Let us consider the following example of lexical reiteration from Arabic, a language10 that is argued by Beeston to make more frequent use of reiter-ation than English (Beeston 1970:112).

(44) ’i‘lam ’annahü fi al-zamäni al-sälif dhakarü ’annahu käna rajulun minal-hukamä’ rafïqan bi al-tibbi, dakhala ’ilä madïnatin min al-muduni,fara’ä ‘ämmata ’ahlihä bihim maradhun khafiy lä yash‘urüna bi‘il-latihim, wa lä yuhissüna bi dä’ihim alladhï bihim. Fafakkara dhälikal-hakïmu f ï ‘amrihim kayfa yudäwïhim liyubri’ahum min dä’ihimwa yashfïhim min ’illatihim allatï ’istamarrat bihim, wa ‘alima’annahü ’in ’akhbarahum bimä hum fïhi lä yastimi‘üna qawlahu walä yaqbalüna nasïhatahu, bal rubbamä näsabühu bil-‘adäwati wasta‘jazü ra’yahu, wastanqasü ’ädäbahu, wastardhalü ‘ilmahu.Fahtäla ‘alayhim fï dhälika lishiddati shafaqatihï ‘alä ’abnä’ jinsihi,wa rahmatihï lahum wa tahannunihï ‘alayhim wa hirsihï ‘alämudäwätihim.

Translation ‘You should know that in times gone by, it was said thatthere was a wise man who was well versed in medicine. This man oncewent into a town and saw that the mass of its people were afflicted by ahidden illness whose ill-effects they could not feel and whose sicknessthey could not sense. The wise man thought about them and how to treatthem, to cure them of their disease and to heal them from their sicknesswhich had continued to ail them. He knew, however, that if he told themthat they were sick, they would not listen to his saying, and they wouldnot accept his advice. In fact, they might even take him as an enemy.


1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 127

They would ridicule his opinion, belittle his views and disdain hisknowledge. So he decided to go about his work surreptitiously out of hisdeep kindness towards his fellow human beings, and out of his mercyfor them and his compassion and his caring for treating them’.

The reiterated lexical items are made in bold in the two versions of the text.Let me spell out the patterns of reiteration in the text above. We have eightpatterns of reiteration in the text above, as shown in Table 4.1. Each lexicalitem in each pattern stands in a metonymic representational relation ofFORM FOR CONCEPT in relation to the other lexical items in the same pattern,because all the lexical items in each pattern belong to one general concept.We should not forget the interactive metonymic relations of CONCEPT FOR

CONCEPT which connect the patterns together to unite the text and provideit with cohesion. For example, one cannot ignore the metonymic relationof contiguity between the first three patterns because they include veryclosely associated concepts like ‘disease’, ‘feel’ and ‘treat’. Similarly, it is noteasy to ignore the associative relationship between patterns four and fivebecause they convey a set of closely contiguous concepts like those of‘advice’ and ‘accept’. By the same token the relations of association betweenthe concepts ‘opinion’ and ‘ridicule’ in patterns six and seven is inevitable.The interactive nature of these associative relations ensures that the text isunified as a whole and that it is cohesive.

4.1.3.1.1 Metonymy and synonymySynonymy is the semantic relation which

may be said to occur if the items are close enough in their meaning toallow a choice to be made between them in some contexts, without therebeing any difference for the meaning of the sentence as a whole. (Crystal1991:340)


Table 4.1 Patterns of reiteration in example (44)

Pattern one Pattern two Pattern threemaradh (disease) yash‘urün (feel) yudäwïhim (treat)‘illätihim (illness and ill-effects) yuhissün (sense) yubri’ahum (cure)dä’ihim (sickness) yashfïhim (heal)

Pattern four Pattern five Pattern sixyastami‘üna (listen) qawlahu (saying) asta‘jazü (ridicule)yaqbalün (accept) nasïhatahu (advice) astanqasü (belittle)

astardhalü (disdain)Pattern seven Pattern eightra’yahu (opinion) shafaqatihi (kindness)’ädäbahu (views) rahmatihi (mercy)‘ilmahu (knowledge) tahannunihi (compassion)

hirsihi (caring)

1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 128

Cruse treats synonymy as the case whereby ‘certain pairs or groups of lexicalitems bear a special sort of semantic resemblance to one another’ (1986:265).He asserts that ‘synonyms are, then, lexical items whose senses are identicalin respect of “central” semantic traits, but differ, if at all, only in respect ofwhat we may provisionally describe as “minor” or “peripheral” traits’ (ibid.,p. 267).11 Lyons (1981:50) suggests a three-level classification of synonyms:

(i) Synonyms are fully synonymous if, and only if, all their meanings areidentical.

(ii) Synonyms are totally synonymous if, and only if, they are synonymousin all contexts.

(iii) Synonyms are completely synonymous if, and only if, they are identicalon all (relevant) dimensions of meaning.

Lyons then argues that absolute synonymy is the phenomenon that existsin expressions that are fully, totally and completely synonymous. Obviously,it is impossible that two lexical items could reach this degree of samenessbecause they might be fully synonymous but with a chance of variationowing to certain shades of affective, social or collocational meaning.12

Synonymy is related to metonymy in two different ways, lexical and textual.The lexical relationship springs from the fact that metonymy is a phenome-non in which two different lexical items are brought together to have the samereferent. Then one of these two lexical items (usually the unfamiliar) is usedinstead of the more familiar one. So in the example:

(45) The ham sandwich is getting impatient.

there is a cognitive synonymy created between the person ordering theham sandwich and the sandwich itself. In this case one of the two expres-sions is used instead of the other, yielding the same meaning albeit with a slightly different stylistic effect. This relatedness between metonymy andsynonymy is as I said lexical and is motivated by a definition of synonymyprovided by Cruse as ‘the lexical relation which parallels identity in themembership of two classes’ (1986:88). The other relation is textual and thisconcerns the fact that synonymy or near synonymy has a referential func-tion in text. The use of synonymous or near-synonymous items in text pro-vide cohesion and they can be treated as metonymic examples of the typeFORM FOR FORM or FORM FOR CONCEPT metonymic relations.13 The following isan example of the generating patterns of synonymy in text to create ideo-logical perspectives.14 The following example15 is taken from Hatim andMason (1997:148–9).

(46) The genuine ulema of Islam have never given in to capitalists,money-worshippers and landlords, and they have always preservedthis decency for themselves. It is a vulgar injustice for anyone to say


1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 129

that the hands of the genuine clergy siding with MohammadanIslam are in this same pot and God does not forgive those who makepublicity in this way or who think in this way. The committed clergyare thirsty for the blood of parasitical capitalists. They have neverbeen in a state of conciliation with them – and never will be.

Figure 4.2 shows the patterns of synonymy in text (46).


Capitalists

Money-worshipers

Landlords

Genuine ulema

Genuine clergy

Committed clergy

Figure 4.2 Patterns of form for concept in example (46)

It should be emphasised here that textual synonymy is slightly differentfrom lexical synonymy in various ways. Lexical synonymy is a feature of thesemantic system of language and the synonyms identified are virtual syn-onyms irrespective of context. This is to say that lexical synonymy iscontext-free. However, textual synonymy is an aspect of meaning in con-text. In my search for synonyms in text I am not looking for virtual context-free synonyms only, but also for the context-bound synonyms or ratherwhat the writer treats as synonyms even if they are not actually synonymsin their semantic structure. This is in line with the general approach of moti-vated signification proposed in the previous chapter, section 3.2. In addi-tion, this approach ensures a contextual and a cognitive treatment of thetext to show us the set of ideologies underlying the generation of text whichare realised by creating a set of cognitive synonyms between elements whichare not synonymous otherwise.

4.1.3.1.2 Metonymy and hyponymyHyponymy is one of the relations of inclusion. Like superordinateness andmeronymy, which are going to be discussed shortly in this chapter,hyponymy is a relation which designates an internal relationship of associ-ation between lexical items. This association is that of belonging or mem-bership. This sense is found in Cruse, who defines hyponymy as ‘therelationship corresponding to the inclusion of one class in another’(1986:89). The sense of inclusion figures also in Crystal, who defines theconcept as ‘the relationship which obtains between specific and general lex-ical items, such that the former is “included” in the latter’ (1991:168). Lyonsdiscusses hyponymy within a wider framework of sense relations, which hedivides into two broad categories. These are substitutional sense relationsand combinatorial sense relations. He relates the former to the Saussureannotion of paradigmatic relations and the latter to the Saussurean notion of

1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 130

syntagmatic relations. Hyponymy is dealt with within the former categoryof relations simply because it is essentially a result of a substitutionalprocess between closely related lexical items. Within the latter category Lyonsdeals mainly with collocational relations because they exhibit several com-binatorial features in terms of frequency and acceptability of co-occurrence(1995:124).

The relationship between hyponymy and metonymy is obvious, since oneof the fundamental relations of metonymy is that of signifying inclusionthrough part–whole relations. The following example from Gibbs (1994:11)illustrates the point. It is a passage that describes a person:

(47) fine eyes, in steel-rimmed glasses and a most expressive and sensitivemouth, by turns tremulous, amused, morally reproving or full of scorn.It was the mouth, one felt, of a man defending the right to be sensi-tive. Physically he was awkward, limp and still at the same time. Hewould stand askew, as it were, holding himself together by gripping hisleft hand in his right. By contrast his gestures were most graceful.

Figure 4.3 illustrates the patterns of these hyponymy relations which relateto the description of the man. Some of these relations are hyponymy–hyponymy relations while others are hyponym–superordinate relations. Onthe whole all relations are metonymic, whether those of species to speciesor those of species to genus, because the principle is the same, that is all ofthe relations are of inclusion. The part–whole relations can be representedmore explicitly, as in Figure 4.4.

The relation underlying the diagram on the left is that of physicalhyponymy, which is taken to mean a relation of physical inclusion. The rela-tion underlying the diagram on the right is that of cognitive hyponymy.According to the triadic semiotic model of metonymic signification proposedin this study, the distinction between the cognitive and the physical or onto-logical is realised in such a way that physical hyponymy signifies ontological


Eyes Mouth Glasses

Man

Left hand Right hand

Figure 4.3 Patterns of hyponymic relations in example (47)

1403_913323_07_cha04.qxd 11/21/03 6:37 PM Page 131

inclusion, while cognitive hyponymy signifies epistemological inclusion. Soit could be argued that the diagram on the left shows ontological hyponymicrelations, while that on the right shows epistemological hyponymic relations.However, both of these types of relation are actually metonymic, given thatmetonymy covers both epistemology and ontology.

4.1.3.1.3 Metonymy and superordinatenessSuperordinateness is the converse of hyponymy, and the two notions implyone another. A hyponym entails its superordinate whereas the superordinatedoes not entail its hyponym. This statement can be reformulated differently,drawing from the literature in semiotics as follows: a hyponym denotes itssuperordinate whereas a superordinate connotes16 its hyponym.

(48) A: I saw a lion.B: I saw an animal.17

A entails B because the relation of inclusion is logically necessary that a lionis an animal. In semantic terms we can say that A entails B. However, B doesnot necessarily entail A, although it may presuppose it. The relation throughwhich we move from the genus to the species is a relation outside languagebecause it relates to culture and it relates to the notion of ‘prototypicality’.However, the relation through which we move from the species to the genusis essentially linguistic. In the examples

(49) A: I cooked lamb.B: I cooked meat.

if you cooked ‘lamb’ you necessarily cooked ‘meat’ because ‘lamb’ is part ofthe concept ‘meat’. However, if you cooked ‘meat’ it is not necessary thatyou cooked lamb.

Let us consider the following two examples:

(50) A: I bought a chair.B: I bought some furniture.

If we can think of a culture where the chair is a prototype of furniture thenwe can say that the whole typically signifies the part. So in this culture when


ManGestures

Gripping Holding

Eyes Mouth Glasses Left hand Right [hand]

Figure 4.4 A metonymic model of hyponymic relations in example (47)

1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 132

the category ‘furniture’ is mentioned it typically signifies ‘chair’. This categoryis not only culture-specific but also context-specific. So if somebody says ‘I need to buy furniture for my house’ we activate a schema of sofas and bedsmore than desks. If someone says ‘I need to buy furniture for my office’ weactivate a schema of chairs, an office desk, and so on. In all of these caseswe are able to infer the part from the whole.18 Also, in the examples

(51) A: I hit him on the nose.B: I hit him on the face.

A entails B but B does not entail A because the part does not entail thewhole.

The textual theory of metonymy as I am developing it in this book allowsus to combine aspects of both denotation and connotation in our analysis.We combine both language and context. We combine both the semanticsand the pragmatics of text in the process of text analysis, text description andtext interpretation, and both the conventional and motivated aspects of lan-guage in text analysis. Moreover, the theory of metonymy as a textual tooldoes not necessarily require that the superordinate be mentioned in the text.Metonymy allows us to conceive of a superordinate as soon as the parts arementioned, even if it is not present in the text. The notion of superordinateif interpreted metonymically is a powerful tool for the analysis and interpre-tation of text and relates coherently with the metonymic principle of actualfor virtual, especially when ‘virtual’ is seen to correspond to ‘whole/superor-dinate’ and ‘actual’ to ‘part/hyponym’. Within such an approach phonolog-ical recurrence in text for example has meaning because it signifies a systemof repetition of similar sound species which belong to an abstract entity thatstands as the ideal form of the sound repeated throughout the text. The samealso applies to morphological repetition and syntactic parallelism, as well asto the semantic relations in text.

Let us consider the following example from Hoey (1991:37), which illus-trates lexical patterns in text. I am interested in the patterns of superordinatesand hyponyms because they are of direct relevance to this section:

(52) A drug known to produce violent reactions in humans has beenused for sedating grizzly bears Ursus arctos in Montana, USA, accord-ing to a report in the New York Times. After one bear, known to be a peaceable animal, killed and ate a camper in an unprovokedattack, scientists discovered it had been tranquillised 11 times withphencyclidine, or ‘angel dust’, which causes hallucination andsometimes gives the user an irrational feeling of destructive power.Many wild bears have become ‘garbage junkies’ feeding from dumpsaround human developments. To avoid potentially dangerousclashes between them and humans, scientists are trying to rehabilitatethe animals by drugging them and releasing them in uninhabited


1403_913323_07_cha04.qxd 11/19/03 8:43 PM Page 133

areas. Although some biologists deny that the mind-altering drug wasresponsible for [the] uncharacteristic behaviour of this particularbear, no research has been done into the effects of giving grizzlybears or other mammals repeated doses of phencyclidine.

Figure 4.5 illustrates the patterns of superordinacy–hyponymy in text (52).We can see that the CONCEPT-FOR-CONCEPT metonymic relation is responsiblefor connecting lexical items like ‘humans’, ‘scientists’ and ‘biologists’ becausethey are metonymically related as superordinates and hyponyms. Similarly,the lexical items ‘animals’, ‘mammals’ and ‘bears’ are also metonymicallyrelated via the metonymic representational relation of CONCEPT-FOR-CONCEPT.

4.1.3.1.4 Metonymy and meronymyMeronymy is a semantic relation of inclusion in which the semantic domainof a lexical item is seen to be branching in a lexical hierarchy. Meronymy isthe lexical relation of part–whole. Defining meronymy is not easy, as Crusesuggests (p. 160). Perhaps the best way to define meronymy is to say thatthe lexical relation is perceived in the formula ‘A has X.’ If A has X then Xis a meronym of A. Let me give some examples:

(53) Hands have fingers.(54) Faces have noses.(55) Cars have wheels.

As we have seen in various parts of this book, the part–whole relation is a fun-damental relation of metonymic reasoning. Therefore, the link betweenmetonymy and meronymy is evident. Cruse (p. 157) discusses the nature ofthe relation of meronymy and suggests that there is a distinction between twonotions related to part–whole relations. The first is ‘meronomy’ and the sec-ond is ‘meronymy’. The difference between these two concepts as Cruse per-ceives it is that ‘a meronomy is a lexical hierarchy whose relation of dominanceis the lexical relation of meronymy’ (p. 180). So meronomy is the structure


Humans Animals

Scientists Mammals

Biologists Bears

Figure 4.5 Patterns of CONCEPT FOR CONCEPT relation in example (52)

1403_913323_07_cha04.qxd 11/21/03 6:52 PM Page 134

whereas meronymy is the relation. This interpretation is evident from Cruse’slater clarification that ‘the semantic relation between a lexical item denotinga part and that denoting the corresponding whole will be termed meronymy’(p. 159). As regards meronomy, Cruse argues that ‘there is first of all a truemeronomy, whose structure is determined by purely linguistic criteria’ (p. 160).

Relevant to this book is Cruse’s distinction between ‘meronomy’ and‘labelled part–whole hierarchy’. The former, according to Cruse, is the ‘truemeronomy’ which is based on linguistic criteria. The latter is ‘formally iden-tical to the corresponding extra-linguistic hierarchy’. Cruse makes it clearthat his concern in his book is meronomy because it is the linguisticallybased hierarchy. I argue here that a theory of metonymy accounts for bothlinguistic and extralinguistic inclusion relations and hierarchies. The textualtheory of metonymy is a theory of linguistic and extralinguistic (that is cognitive) stand-for relations, the most notable relation among these beingCONCEPT FOR CONCEPT, which can account for part–whole relations. Thefollowing passage illustrates how metonymy explains both linguistic as wellas extralinguistic networks of part–whole relations:

(56) The city was quiet. The streets were all covered with snow. Hewalked up the street. The shop round the corner was the only placeto find something to eat. In the food section he stretched out hishand to the shelf to take some cheese and a piece of bread. He paidand continued his aimless night journey.

In the above text we can think of a ‘city’ meronomy in which the part–wholerelations are not actually linguistic but extralinguistic, that is ontologicalrelations. This is to say that street does not really relate to city by means ofsemantic inclusion but by means of cognitive inclusion. Figure 4.6 is a modelof the meronymic relations in the above passage:


A piece of bread

City

Street

Shop

Food section

The shelf

Some cheese

Figure 4.6 A model of meronymic relations in example (56)

1403_913323_07_cha04.qxd 11/21/03 7:15 PM Page 135

136

As a conceptual process of locating a reference point, metonymybelongs to the wider set of strategies of finding a point in the com-mon reference space between a speaker and a listener that can serveas a bridge or link to the intended referent.

Dirven (1999:275)

5.0 Introduction

In the previous chapter I have shown that metonymic relations of FORM FOR

FORM, FORM FOR THING, THING FOR FORM, CONCEPT FOR FORM and FORM FOR CONCEPT

can actually be integrated into a textual model of formal and semantic con-nectedness in text. This I have termed ‘the cohesive power of metonymy’.In this chapter I argue that metonymic processes as text-organising princi-ples are not limited to the set of formal or semantic bonds in text. Rather,these processes go beyond this to explain the pragmatic dimension of textgeneration, organisation and interpretation. This chapter builds on the rela-tions of the formal component and goes beyond that to investigate thedynamic interaction between language, cognition and context in text cre-ation and organisation. The metonymic representational relations to be dis-cussed in this chapter are CONCEPT FOR CONCEPT, CONCEPT FOR THING, THING FOR

CONCEPT AND THING FOR THING. This I term ‘the coherence power of metonymy’in which there is continuity of ideas and interaction between concepts andcontexts. This chapter is particularly an attempt to answer the question how,from a metonymic perspective, a text is perceived to form a unified wholeand how it is perceived as meaningful. There are several approaches to thepragmatic aspect of meaning, and one could be tempted to investigate all ofthese, but I shall limit the scope of this chapter to discussing the theoreticalframework of what has been generally referred to as ‘schema theory’.1 Thefollowing claims underlie the chapter:

(i) If metonymy is a process of locating a reference point, as Dirven’s quo-tation above suggests, then it should help us know how communication

5Metonymy and Text Coherence

1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 136

is made possible via such metonymic processes to bring about onlineconstruction of spaces, frames, scripts and categories.

(ii) If metonymy is a fundamental principle of our cognitive structure,2

then it should be able to explain how cognitive continuities are estab-lished and licensed in text.

(iii) If text coherence means basically text prediction and text expecta-tion then metonymy as a principle of conceptual contiguity andcausality is capable of assisting us in predicting what could be com-ing next in text. A driver upon seeing a sign on the road instruct-ing them to ‘slow down’ will expect another sign along the roadinstructing them to ‘resume speed’ because the two signs belong toone frame.

The ultimate goal of this chapter is to establish a link between discourseand cognition through various types of metonymic processes. Hence,metonymy is to be seen as a process that shapes and binds text to provideconceptual coherence to our textual experience. It should be noted that thelink between discourse and cognition that I am proposing in this chapter isnot unprecedented. In fact, such a link has already been established by somelinguistic theories, notably the systemic functional linguistics theory of reg-ister where there is interaction between three variables: field, tenor andmode. These three variables in turn correspond to another set of three vari-ables at a lower level of representation, which is the level of text or the levelof meaning. These three variables are the ideational component, the inter-personal component and the textual component (see Halliday 1978, 1985;Halliday and Hasan 1976; Halliday and Matthiessen 1999). The func-tional systemic linguistics genre theory has also attempted to establish a link between discourse and cognition by defining genre as a purposefulcommunicative event and relating the concept of genre to cognitivenotions like schemata, prototypicality and family resemblances (Swales1990:45–58). The attempt of this school of thought has been within the‘congruent form’ as Halliday and Matthiessen call it (1999:227), while thethrust of this chapter’s proposal is to discuss textual coherence as a productof an online construction of categories, schemata and scripts via metonymicprocesses.

The chapter will address specifically the relevance of the five major knowl-edge structures commonly discussed in the literature on schema theory. Theseare frames and scenarios proposed by Minsky (1975), schemata proposed byBartlett (1932) and Rumelhart (1977), scripts, plans and goals discussed bySchank and Abelson (1977) and prototypes and stereotypes discussed inUngerer and Schmid (1996), Lakoff (1987) and Gibbs (1994). The chapter aimsto integrate all these notions into a unified cognitive framework and relatethem to the concept of metonymy in its cognitive and textual aspect toaccount for continuities and coherence in text.

Metonymy and Text Coherence 137

1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 137

5.1 Metonymy and schema theory

Schema theory is one of the theories3 that look into the nature of howconceptual structures are constructed in the human brain. More generally,schema theory looks at how knowledge structures are organised in thehuman brain and how they are used in the processing of text. This sectiontries to establish the relationship between the various modes of knowledgeconstruction in the human brain and the various processes of metonymy toshow that our cognition is fundamentally metonymic and that textualknowledge is also metonymically constructed and metonymically inter-preted. The chapter will not consider the debates about the validity or oth-erwise of the various proposed mathematical models of machine intelligencebecause this is outside the scope of this book. What this chapter intends todo, rather, is to investigate the issue of text coherence in its human aspectand relate this to the various metonymic processes, to show that coherencein text is achieved through an interaction between shared knowledge whichis metonymically constructed and organised in memory, and inferencingprinciples that are metonymically motivated.

Collins English Dictionary and Thesaurus (1995) provides a three-fold defi-nition of the concept of ‘schema’:

(i) A plan, diagram, or scheme.(ii) (In the philosophy of Kant) a rule or principle that enables the under-

standing to unify experience.(iii) Logic: (a) a syllogistic figure; (b) a representation of the form of inference.

The three definitions, it can be noted, are related and together provide a uni-fied definition of the concept of schema in the sense we are concerned withhere, that is as representational structure. Schema theory is the theory ofmental representation of stereotypical knowledge. It can be argued thatschema theory had its origins in the way of thinking known as structural-ism and more particularly in the theory of structural psychology known asgestalt psychology. The major claim of gestalt psychology is that perceptionis carried out according to ‘gestalt laws of perceptual organisation’. The mostimportant tenets of gestalt psychology as discussed in Koffka (1935) andWertheimer (1958) are the following:

(i) ‘Principle of proximity’: this is essentially the principle of adjacencywhich states that adjacent elements will be perceived as related.

(ii) ‘Principle of similarity’: similar elements tend to be perceived as oneelement.

(iii) ‘Principle of closure’: perception tends to complete uncompleted figures.(iv) ‘Principle of continuation’: continuous and not severely interrupted

elements will be perceived as wholes.

These laws or principles of perception indicate that perception is essentiallymetonymic because it is based on cognitive relations of contiguity and


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 138

association. In fact, under the principle of association, which is the basicmetonymic principle, we could include all the four principles mentionedabove. This justifies the choice of the model of schema theory as a frame-work for this chapter.

De Beaugrande and Dressler maintain that schema theory goes back toBartlett (1932), who according to van Dijk and Kintsch was ‘a psychologistworking within the gestalt tradition’ (1983:3). De Beaugrande and Dressler(1981:90) trace the development of the theory onward in the works ofRumelhart (1975, 1977b) and Kintsch (1977) as well as others. I am not par-ticularly concerned here with the history and development of the theory, asthis has already been discussed in detail in van Dijk and Kintsch (1983),Schank and Abelson (1977) and Cook (1990) as well as elsewhere. What con-cerns me rather is what the theory postulates and how this relates to theprocesses of metonymy and to the overall development of a textual modelof metonymy. Schema theory postulates that our cognition of the world isessentially holistic.4 The basic claim of schema theory, as Cook puts it, is‘that a new experience is understood by comparison with a stereotypical ver-sion of a similar experience held in memory’ (Cook 1990:1). So in a sensethere are two processes involved in this cognition. One is accommodationin which the person accommodates a concept or an image or a sound, andstructures that concept, image or sound into a stereotype or into a wholewith internal coherence. The other process is that of modification in whichthe person modifies previous concepts or experiences in the light of newexperiences. Here they may either reinforce a previous stereotype or modifyit to accommodate new data.5

It should be noted that schema theory is a theory of human understand-ing, and this applies to the processing of linguistic data as well as to othertypes of data like the sensory data of smell, taste, sight, and so on. Becausewe are able to learn about sensory data other than linguistic data onlythrough language, the linguistic manifestation of our cognition is perhaps themost widely studied phenomenon in schema theory, and it is the type thatI will be concerned with in this chapter. Schema theory is a pragmatic theoryof understanding based on shared knowledge which facilitates the processof inferencing and makes communication possible between people. Therelationship between schema theory and the textual theory of metonymy isevident in virtue of the following observations.

Schema theory postulates the existence of mental structures called‘schemas’ or ‘schemata’ in the brain, which correspond to the three domainsof reality, mind and language. According to Cook, ‘if an intelligence is toprocess discourse, it will need language schemata, text schemata and worldschemata’ (1990:7–8). This interdomain nature of schemata is reminiscentof the interdomain nature of metonymy proposed earlier in this book(section 3.2.1). The textual theory of metonymy developed in this bookpostulates that metonymies operate in facilitating understanding by meansof the conceptual relations they establish in text. As argued above in


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 139

Chapters 2 and 3, metonymy does cross these three domains and providesa coherent understanding of language, epistemology and ontology. I arguehere that these schemata as mental structures are activated by metonymicprocesses because they operate in discourse under metonymic operations ofPART FOR WHOLE and WHOLE FOR PART, CAUSE FOR EFFECT and EFFECT FOR CAUSE.

I shall now give an example of how schema theory works in text-processingand show how this is related to metonymic processes in text. Schema the-ory postulates that the mental structures called ‘schemata’ or ‘frames’ helpus in two ways not only in input-processing, which is the interpretationprocess, but also in output-processing, which is the generation process bymaking our contribution to the discourse concise and relevant. The text in(1) below is interpreted as perfectly meaningful and relevant and thus ascoherent because it builds on an assumption of some further knowledge ofdetail to be supplied by our knowledge store:

(1) I got up, washed, had breakfast and left for work.

The four actions mentioned in (1) above can be termed ‘global actions’, notmerely specific actions, because each of these global actions involves severalother sub-actions that we do not need to mention because they are suppliedby our conceptual structure of previous experience. So each of the fouractions mentioned in text (1) above stands for the sub-actions involved in a metonymic relation of CONCEPT FOR CONCEPT via a PART-FOR-WHOLE metonymicrelation.

Now our understanding of the statement depends on our activation ofthis global knowledge of the everyday routine of getting up, and in fact anyformal provision of further detail of the specifics of this knowledge structurewould be regarded as redundant. The same thing applies to the acts of wash-ing, having breakfast and leaving for work. This concision and compactnessin language is a precise operation of metonymic reasoning. This whole–partmetonymic relation is the default case of both generating and interpretingtext. However, there is also the unmarked form of these two processes. Forexample, if the originator of the above is a witness in court to a road acci-dent, the last global action of ‘leaving for work’ would not be sufficientbecause the judge will most likely interrupt the flow of the text and askthe witness to mention the details of that action. They might be asked togive further details about whether they went to work alone or with some-one else and whether they took the bus, taxi, underground and so on.

All these schematic operations of knowledge structures are nothing butmetonymic operations of the part for the whole and the converse of thisprocess, and they function in activating shared knowledge so as to rid com-munication of redundancy, or in fact to give metonymic details if need be. Iffurther in the text we encounter the phrase ‘the bed’ with the definite article‘the’, we would not be surprised simply because this has been activated via


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 140

the ‘getting up’ schema. Similarly, if the phrase ‘the car’ or any other mode oftransport is mentioned in the definite we will not ask ‘which car?’ because thisis supplied by our ‘going to work’ schema. This is essentially a process ofpart–whole metonymic reasoning. This chapter argues that much of thecoherence of text is heavily dependent on the recognition of metonymic rela-tions underlying the surface structure of the text.

The fact that metonymy is essentially representational,6 operatingthrough a ‘stand-for’ relation, provides ample evidence that metonymy isstrongly linked to the notion of schemata. The basic level of a schema is thecognitive reference-point level according to Rosch (1975), Langacker (1993)or the central level according to Lakoff’s (1987) notion of radial categoryconstruction where there is a central sense occupying the centre of a weband there are some extended senses that go out from the centre. Moreover,the claim that ‘the mind, stimulated either by key linguistic items in the text(often referred to as “triggers”), or by the context, activates a schema, anduses it to make sense of the discourse’ (Cook 1990:7) is very supportive ofthe strong relationship between metonymy and schema theory. The word‘pen’, in

(2) The pen is mightier than the sword.

is a trigger of the schema of ‘knowledge’ of which the ‘pen’ is a part orperhaps an instrument with which it is in a stand-for relation by virtue ofstereotypical cultural association. Schema theory proposes that knowledgeis structured around global structures in our brains corresponding to whatdifferent writers term ‘plans’, ‘goals’, ‘frames’, ‘scripts’ or ‘scenarios’. Oncethis global pattern is activated then the details can be left unmentionedbecause they will be supplied by the schematic structure in our cognition.The various components of one version of schema theory which is the SPGU(scripts, plans, goals and understanding) model of Schank and Abelson(1977) testify to the strong linkage I intend to establish between this versionof the theory and metonymic processes in text.

Bartlett’s schema theory is often confused with Minsky’s frame theory andSchank and Abelson’s script theory. It should be pointed out that there isa great deal of terminological confusion and phenomenal overlap betweenthese theories since all of them are set up to deal with more or less the samephenomena, that is to account for knowledge structures and to understandthe processing and production of texts. This overlap in terminology is typi-cal of the rapidly expanding discipline of linguistics which began in the1970s to cover various new areas not trodden before, especially in the fieldof computational linguistics.

There are also of course obvious differences between all of these terms.These differences will be addressed in various sections of this chapter.Nevertheless, it is generally the case that these theories all agree on a general


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 141

premise, which is, as Maida states, that ‘modules of knowledge imposecoherence on experience’ (see Maida 1987:302). Whether we submit to a frame theory, schema theory or a script theory or to them all together, orin fact to any other approach in the field, the outcome will still be the sameas far as metonymy is concerned. This is because the same premise still holdsfor all of these theories, namely that metonymy is essentially knowledgebased and is systematically structured and causally and contiguously trig-gered. Theories of knowledge structures in general seem to operate on thepremise that there is some sort of basic level categorization, which in thewords of Lakoff (1987:13) means

the idea that categories are not merely organized in a hierarchy from themost general to the most specific, but are also organized so that the cat-egories that are cognitively basic are in the middle of a general-to-specifichierarchy. Generalization proceeds ‘upward’ from the basic level andspecialization proceeds ‘downward’.

This basic level is actually the reference point level or the metonymic leveland it is according to Lakoff ‘functionally and epistemologically primary’.With respect to the problem of terminology it should be noted that thischapter adopts the general term ‘schema theory’ to designate the theory ofmental representation of knowledge structures. For the subsystems of men-tal structure the chapter follows the division provided by Schank andAbelson (1977) in the SPGU. The chapter also incorporates terms from de Beaugrande and Dressler (1981), especially those pertaining to schemata.With regard to the notion of frames, this draws mainly from the discussionoutlined by Maida (1987).

As far as the specific terms are concerned there is quite a lot of disagree-ment about terminology in the literature on schema theory. Some scholars,like Schank and Abelson (1977), talk about scripts, plans and goals as cru-cial components of the theory. De Beaugrande and Dressler (1981) discuss a general notion which they call ‘global patterns’ under which they includeframes, schemata, plans, goals and scripts. Other terms like ‘scenarios’,‘stereotypes’ and ‘prototypes’ have also been suggested (see for exampleSowa 1984 and Minsky 1975). Instead of pursuing the debates and argu-ments of the proponents of one term or the other I shall adopt a differentattitude towards the classification and attempt to integrate these terms in a unified treatment of knowledge structure that also reflects my conceptionof metonymy both as an aspect of knowledge and as an aspect of pragmaticapplication of this knowledge in text. Figure 5.1 shows my proposed classi-fication of knowledge in memory and relates this to theoretical aspects ofmetonymy.

Figure 5.1 is inspired, to some extent, by de Beaugrande and Dressler’sclassification of what they call ‘control centres’, which they include in their


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 142

account of text coherence. By ‘control centres’ they mean ‘points fromwhich accessing and processing can be strategically done’ (1981:95). Theyclassify these control centres into primary concepts and secondary concepts.The model is also inspired indirectly by the notion of ‘primary concepts’proposed by de Beaugrande and Dressler, though their discussion of knowl-edge structures like schemata, frames and scripts and so on was also to someextent confused. The primary concepts de Beaugrande and Dressler suggestare the following:

(i) Objects: conceptual entities with a stable identity and constitution.(ii) Situations: configurations of mutually present objects in their current

states.(iii) Events: occurrences which change a situation or a state within a situ-

ation.(iv) Actions: events intentionally brought about by an agent.

I have related the notion of schemata to the notion of ‘objects’ suggested byde Beaugrande and Dressler to denote conceptual entities with stable iden-tity and constitution. Similarly, I have designated the term ‘script’ for thenotion of situations proposed by those authors to refer to the mutually pre-sent objects in their current states, which is the appearance of the objects ina stereotypical sequence of events. With regard to the notion of ‘events’ sug-gested by the authors I have associated it with the notion of scenarios, tak-ing into consideration that these are not standard situations withconventionally predictable roles and objects. Rather, they are arbitrary eventsand sequences of events that derive their interpretation from the knowledge


Knowledge

Descriptive knowledge (what) Procedural knowledge (how)

Frames and schemata Conventional Arbitrary

Scripts Scenarios

Plans and goals

Prototypes Stereotypes

Figure 5.1 A unified model of knowledge structures

1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 143

of the specific causality of the situation. In fact, it is the next step in de Beaugrande and Dressler’s classification that helps to explain the inten-tionality and direction of these events. So in a sense it is the plans and thegoals which explain the scenarios.

Figure 5.1 shows that we can conceptualise knowledge structures in mem-ory as comprising two major modules. One is the ‘module of entities’, or thestorage of conceptual items each with its own specifications. We can also callthese the ‘module of words’. The second is the procedural model of stringingtogether. We can call this also the ‘module of rules of sequencing and combi-nation’.7 Here the conceptual items are actually combined in various modesto make up models and networks of knowledge structures, that is thought.The model refers to the first module as the knowledge of the ‘what’ and givesit the quality of being descriptive, while it refers to the second module as theknowledge of the ‘how’ and deems it procedural. According to this model,frames and schemata are actually the same thing because they both pertainto the description of the states and conceptual categories. These are assumedto exist in memory in architectural structures based on a hierarchy or someother form of contiguity like causality and association. From now on when-ever the word ‘schemata’ is used it will refer to both schemata and frames.

In my discussion of all these components of our schematic knowledge I shall therefore treat schemata and frames as two aspects pertaining to the same phenomenon, which is the knowledge of things or entities. By thesame token I shall also treat scripts and scenarios as pertaining to the realmof experience and sequenced events. However, as the model shows, they willbe treated separately because they pertain to two different types of sequencedevents: one is conventional and the other is arbitrary. With regard to plansand goals, I argue that they explain the arbitrariness of the scenario.Therefore I shall discuss these two aspects separately to show the subtledifferences between them and how they both give explanation to arbitraryscenarios.

Since scenarios are essentially arbitrary they do not show any internalcoherence because of their very nature as creative and inventive. Goals,therefore, give this internal coherence to these scenarios and plans serve torealise specific procedures to achieve specific goals in the scenario. As forprototypes and stereotypes I have placed them at the bottom of the modelin order to show that both of these terms account for a characterisation ofthe specific features of each of the two paradigms in the model. I argue thatframes and schemata are established via the interaction with prototypicalknowledge and that scripts and scenarios are established by means of stereo-typical knowledge.8 More of this discussion will be provided in sections 5.3to 5.6, which treat the relationship between metonymy and these concepts.The double-headed arrows in the model (Figure 5.1) indicate the interactivenature of the model and the dialectical relationship between its compo-nents. None of these components operates in isolation.


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 144

Let me now outline the theoretical aspects that relate this model ofknowledge in general and language and cognition to metonymy. The twomodules of knowledge postulated correspond to the structural notions ‘paradigmatic’ and ‘syntagmatic’. These are believed to underlie all linguisticand cultural behaviour. The link between the model and metonymy is evi-dent when we know that metonymic processes are based on these funda-mental notions. First, the word or concept or thing is selected from aparadigm and then a syntagmatic relation is abstracted between the item sub-stituted and the item substituted for. These syntagmatic relations are basedon the two principles of contiguity and causality. The descriptive module ofknowledge in the model above is actually the paradigmatic level, whereknowledge is stored in the form of entities and objects as a system. The pro-cedural module outlined in the model above corresponds to the syntagmaticlevel, where, after the selection, entities and objects are stringed and com-bined. More importantly, the model above captures the two fundamentalmetonymic principles, that is that of contiguity and causality. I relateschemata and frames to the metonymic principle of contiguity, because I argue that descriptive knowledge is contiguously stored since it makes use ofconventional knowledge of how things are structured, mainly through cate-gory membership relations of part–whole and adjacency relations. I also arguethat arbitrary knowledge is made coherent only through the metonymic prin-ciple of causality, which involves intentions and interests, plans, scenariosand goals. Hence conventional knowledge is made meaningful through con-tiguity relationships and arbitrary knowledge is made coherent throughcausality relationships. These are fundamentally metonymic.

5.2 Metonymy and schemata

In the previous section I discussed the notion ‘schema theory’ but I did notspecifically address the concept of schemata. What are schemata, how dothey operate in text understanding and how are they related to metonymicoperations in text-processing? This section attempts to provide an answer tothese questions. As I mentioned earlier, the coinage of the notion of schemais usually attributed to Bartlett, who maintains that

schema refers to an active organisation of past reactions or of past experi-ences which must always be supposed to be operating in any well-adaptedorganic response. That is, whenever there is any order or regularity ofbehaviour, a particular response is possible only because it is related toother similar responses which have been serially organised, yet whichoperate, not simply as individual members coming one after another, butas a unitary mass. (Bartlett 1932:201)

We notice a great deal of behaviouristic thinking and terminology in thisquotation from Bartlett. The definition revolves around certain behavioural


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 145

and observable concepts like reactions, responses, experiences and mass. Yetthe definition is obviously a detailed account of the concept of schemata,especially when it relates it to the notion of ‘experience’. This shows thatBartlett is using the term in a general sense to mean the general knowledgestructure which pertains to both object organisation and to sequences ofevents and actions. It should be noted however that the notion of schematadeveloped in this chapter differs from Bartlett’s because it is limited to the setof objects and entities activated by a part–whole or whole–part metonymicreasoning.

Rumelhart defines the concept of schema as

an abstract cognitive representation of a generalised concept or situation.A schema contains, as part of its specification, the network of interrela-tions that characterise the major constituents of the situation or conceptin question. (Rumlehart 1977:290)

The first thing we notice in the definition is the cognitive orientation,unlike Bartlett’s definition which is behavioural. This actually explains eachwork within its predominant intellectual climate. Rumelhart’s definition hasthe same generality as Bartlett’s and both definitions differ from the treat-ment of the notion of ‘schema’ developed here in the manner explainedabove. Schema here is treated as the organisation of descriptive knowledge,which includes knowledge about objects, entities and concepts.

Cook defines schemata as ‘mental representations of typical situations andthey are used in discourse-processing to predict the contents of the particu-lar situation which the discourse describes’ (Cook 1989:69). De Beaugrandeand Dressler define schemata as ‘global patterns of events and states inordered sequences linked by time proximity and causality’ (1981:90). I findCook’s definition more satisfactory than de Beaugrande and Dressler’sbecause it is compatible with the model developed above in section 5.1. Cookactually highlights the specific properties of schemata as mental representa-tions of typical situations and as pertaining to the prediction of the contentsrather than the sequence of these contents.

Minsky, the originator of the frame theory, holds a general view of frameswhich I have regarded here in the model above as the same thing asschemata. Minsky defines the notion of ‘frame’ as

a data-structure for representing a stereotyped situation, like being in a certain kind of living room, or going to a child’s birthday party.Attached to each frame are several kinds of information. Some of thisinformation is about how to use the frame. Some is about what one canexpect to happen next. Some is about what to do if these expectations arenot confirmed. (Minsky 1975:212)


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 146

It is clear that Minsky regards ‘frames’ as general knowledge structures to betreated in exactly the same way as de Beaugrande and Dressler treat ‘globalpatterns’. This is evident in the fact that Minsky designates these frames toboth descriptive and procedural knowledge. Moreover, Minsky’s definition of‘frames’ is the same as de Beaugrande and Dressler’s account of ‘schemata’.The definitions of schemata and frames provided by de Beaugrande andDressler on the one hand and by Minsky on the other are unsatisfactory asfar as my proposed model is concerned. This is because they confuse frames,schemata, plans and scripts and do not allow for a systematic understandingof knowledge structures, which I believe is crucial. In view of this, my modelattempts to designate a general term for the higher-level knowledge struc-tures, and to propose differences between the parts of this superordinateentity based on epistemological classification of these knowledge structures.According to a more recent argument, language knowledge for example isessentially knowledge of words and knowledge of rules (see for further detailsof this proposal Pinker 1999).

It should be noted that de Beaugrande and Dressler’s account of thenotion of ‘frames’ is compatible with the model proposed above. Theydefine frames as

global patterns that contain commonsense knowledge about some centralconcept, for example ‘piggy banks’, ‘birthday parties’, and so on. Framesstate what things belong together in principle, but not in what orderthings will be done or mentioned. (de Beaugrande and Dressler 1981:90)

Although this proposal agrees with my model with regard to frames as aninventory of entities having categorical specifications, the new element inthe model I propose is that it also treats schemas as the same phenomenon.This is something with which de Beaugrande and Dressler would not agree,since they treat schemas the same way as I treat scripts, that is as structuresof sequenced events.

Within descriptive knowledge I also discuss a further knowledge structurewhich concerns the notion of ‘prototypes’. This mainly relates to schemata,as this latter deals with concepts, entities and objects. This is briefly dis-cussed by de Beaugrande and Dressler (1981:91) under the notion of ‘inher-itance’,9 which the authors define as ‘the transfer of knowledge among itemsof the same or similar types or subtypes’. The authors suggest three types ofinheritance, which can be represented as in Figure 5.2.

The model of inheritance is perhaps a further specification of the notionof ‘frame’ or ‘schema’. This is because it is a natural or a general semanticprinciple which specifies class membership in the real world or perhaps inthe semantic cognitive world. Frames, on the other hand, are more generalin this regard in that they account for class membership in a pragmatic sense


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 147

in which not only are the semantic relations of inclusion and exclusioninvolved in class identification, but also some other pragmatic factors whichbring together textual worlds that are otherwise unrelated and which bearon the process of text generation, organisation and interpretation.

So in this section we are actually introduced to a specific type of mentalstructure. This describes entities and specifies their category membership orother contiguous associations between these entities or concepts. This isexpressed in metonymic terms as specifying the concepts’ part–whole rela-tionships and other causal and associative relations between these concepts.We are not however in the realm of time- and context-bound and sequencedevents to which I designate the term ‘scripts’. The notion of scripts is anissue which will be dealt with below in section 5.3. A good example of a schema would be a ‘city’ schema which involves several small parts con-tributing to this global pattern; once this higher-level knowledge structureis triggered, the whole schema is activated.

Schemata also help us localise our perception and understanding. In fact,this is a common technique used by the cinema industry. Some films start a particular scene by zooming in on the visual space and hence the cogni-tion space from the whole to the part in a consecutive series of whole–partmovements. In these films the scene begins with a global shot of the wholecity, then it zooms down to one block, further to one street, further to onehouse and further to one room. This metonymic movement is exploited tofacilitate the audience involvement and emphasise the gestalt or holistic per-ception. This corresponds to the notion of ‘universe of discourse’ in discourseanalysis and it aims to tell the receiver that every thing said or seen in thisparticular episode should be interpreted within the global shot the scenestarted with.

Schemata are also useful in a more practical sense. They help us in keep-ing to the maxims of cooperation in conversation. Of paramount impor-tance among these maxims are the maxims of quantity and relevance. If wegive more information than expected we might be interpreted as being face-tious and demeaning the intelligence of our audience. It is for this reasonthat Arab and western rhetoricians equally define eloquence as brevity.


Class Superclass

Instance Subclass

Analogy

Analogised

Figure 5.2 The three types of inheritance

1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 148

Schemata help us to maintain a friendly atmosphere with the audiencebecause we are not being monotonous and boring.

Schemata help us in recovering ambiguities of reference in text and inresolving problems of definiteness. Consider the following example:

(3) The time was running fast and the train was due in half an hour. Theweather was also very bad. We had to call a taxi. The driver was anold man who could not help very much with the luggage.

In example (3) we find that the noun phrase ‘the driver’ and the nounphrase ‘the luggage’ are definite. This is not compatible with pedagogicalgrammatical rules, which dictate that a noun phrase cannot be definiteunless it is one of its kind or it has been mentioned previously in the textand thus made known to the receiver. However, a textual model ofmetonymy resolves the problem via the schematic knowledge of part–wholerelationships. Metonymy as a schematic representation of knowledge statesthat the schema of ‘taxi’ consists of a driver as an essential part of it. So it isplausible to maintain that the definiteness of the noun phrase ‘the driver’ iscognitively justified and the concept ‘driver’ is made available to the readereven before encountering it in the text because it is salient in the universeof this particular textual world. Similarly, the ‘train’ schema activates a ‘travelschema’, which guarantees that the noun phrase ‘the luggage’ is supplied,thus making it definite.

Classical Arabic grammatical treatises address this issue by postulating a sort of semiotic treatment of it. The phenomenon is known as al al-‘ahdiyyah, which can be roughly translated as ‘the al of knowledge (orfamiliarity)’ because definiteness in Arabic is realised via the definite article al. Arab grammarians were puzzled by this phenomenon of a noun appearingwith al when it has not been mentioned earlier in the text, and is not one ofits kind or a generic noun. They dealt with it by postulating that there areactually three modes of definiteness. In a sense they dealt with the concept ofdefiniteness from the three domains of epistemology, ontology and language.Ibn Hishäm (d. 1360) says: ‘al might also function as a definite article and assuch it is of two types: knowledge-based al and generic al’ (n.d.:49).

Hasan provides a more detailed definition of al al-‘ahdiyyah. He states that‘al al-‘ahdiyyah is the al which attaches to an indefinite noun and adds to ita definite aspect capable of making the referent of this noun definitely iden-tifiable in cases where it would otherwise be general and unspecified’ (Hasann.d.:423).

What concerns us here in this section is the knowledge-based al (the al al-‘ahdiyyah). This is of three types. The first type is that which is termed alal-ma‘hüdatu dhikriyyan,10 that is al which is known by means of it beingmentioned in the text. This can be more idiomatically translated into ‘thetextually known’ al. Among the examples Ibn Hishäm provides for this


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 149

category are the following Qur’anic verses:

(4) Kamä ’arsalnä ilä fir‘awna rasülan fa‘asä fir‘awnu al-rasüla. (73:15–16)‘As we sent down a messenger to the Pharaoh, but the Pharaoh dis-obeyed the messenger.’

(5) Allähu nüru al-samäwäti wa al-’ardh mathalu nürihï kamishkätin fïhämisbähun al-misbähu fï zujäjah. (24:35)‘Allah is the Light of the heaven and earth. The likeness of His Lightis that of a niche within which is a lamp. The lamp is in glass.’

As can be seen in the two verses above, we have the nouns rasül ‘messenger’and ‘misbäh ‘lamp’, each mentioned twice. The first time is without al andthe second time with al. The first is indefinite and the second is definite byvirtue of the fact that it has been mentioned in previous text. This is thereason Arab grammarians termed this type of definiteness or this type of alas al-ma‘hüdatu dhikriyyan, that is ‘textual al’. Among the secular examplesIbn Hishäm gives for this type of al is example (6)

(6) ’ishtaraytu farasan thumma bi‘tu al-faras.‘I bought a horse. Then I sold the horse.’

This type of al is essentially cohesive by the virtue of being textual or for-mal, that is explicitly stated in the text. However, the link is fundamentallysemantic and cognitive because it is a process of cognitive derivation of theaspect of specificity by means of mentioning the noun twice to familiarisethe reader with its reference. There is an interaction of signification betweenthe two worlds of form and things, and this is mediated by the world ofconcepts. So in one type of semiotic domain of signification we can actuallyperceive all the three domains at play.

The second type is that which they call al al-ma‘hüdatu dhihniyyan,11 that isal which is known by means of its being present in the cognition of the inter-locuters or by virtue of the fact that it is well established in the universe ofdiscourse. This could be idiomatically translated as ‘the cognitively known’ al.Here Arab grammarians explicitly refer to the existence of the thing in thecognition of both the speaker and the hearer or the writer and the reader. IbnHishäm gives two Qur’anic verses as examples of al al-dhihniyyah:

(7) ’idh humä fï al-ghäri. (9:40)‘When they were in the cave.’

(8) ’idh yubäyi‘ünaka tahta al-shajarati. (48:18)‘When they gave you their allegiance under the tree.’

The nouns ghär ‘cave’ and shajarah ‘tree’ are definite in both examples, eventhough they are not mentioned earlier in the text. Arab grammarians argue


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 150

that both are deemed definite by virtue of the fact that they are salient inthe cognitive domain of the discourse. In short they are present in the mindsof the receivers.

However, this is not an adequate explanation because the question is, howare they made present in the receivers minds? I think that both occurrencesof ghär ‘cave’ and shajarah ‘tree’ in examples (7) and (8) pertain to the firsttype of al, which is the textual al. Although it is true that both nouns arenot mentioned previously in the text, that is in the Qur’an, there is no waythey can be cognitively constructed from nowhere. One could argue that theschema of travelling which is the topic of example (7), which pertains to themigration of the Prophet Mohammed and his friend Abu Bakr from Makkato Madina, activates the slot ‘cave’ and thus makes ‘cave’ definite. If wethink in this way, however, our argument will be rather weak because a trav-elling schema does not necessarily involve a slot of a cave.

My claim then is that this al is actually textual because both nouns, that is‘cave’ and ‘tree’, are made definite by virtue of the fact that they have beenmentioned in other texts, that is the history of the Prophet which tells us thathe and his friend stayed for three nights in a cave called the Cave of Thawrbetween Makka and Madina. Without this textual information we would finddifficulty in understanding the definiteness of the word ‘cave’. The same thingapplies to the word ‘tree’. If we were not told in another text, that is the his-tory of the Prophet, that the allegiance took place under a tree then we wouldfind it difficult to explain the definiteness of the noun ‘tree’.

Hasan (n.d.:50) provides some examples which I find interesting as far ascognitive al is concerned. The examples include:

(9) ‘indamä yas’alu tälibun zamïlahu: ma al-jadïdu fi al-kulliyyati?‘ “What is new in the college?” Asked by a student of his classmate.’

(10) hal katabta al-muhädharah.‘Did you write the lecture?’

(11) hal ’anta dhähibun’ila al-bayt.‘Are you going to the house?’

The words kulliyyah ‘college’, muhädharah ‘lecture’ and bayt ‘house’ in theabove examples are definite because they are present in the cognition of thetwo participants in the discourse and not because they are previously men-tioned in the text or because they are present in the situation of the dis-course. Nevertheless, one could argue that the definiteness of these nounsin the examples is justified on the grounds of schematic knowledge that thetwo individuals have. This is the schema of ‘studying’ which involves a col-lege, a lecture and much more than this. As regards the third example, it iseasily accounted for by means of a ‘studying script’, which involves thesequenced actions stereotypical of this life situation. We would expect a student finishing college to go to his house.


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 151

The third type of al is what the Arab grammarians call al al-ma‘hüdatu hudhüriyyan,12 that is al which is known by means of being present in thephysical context of the utterance. The item or entity is existent in the situ-ation of discourse. This could be idiomatically translated as ‘the existential’al and conventionally as ‘the ontological al’. It might also be referred to asthe exophorical al, by which is meant that the thing referred to is a saliententity in the context of situation of discourse. Ibn Hishäm gives the follow-ing examples of this type of al:

(12) yä ’ayyuha al-rajul.‘O you the man!’

(13) kharajtu fa’idha al-’asadu.‘I went out, and there was the lion.’

The argument Ibn Hishäm raises concerning these two examples is that thefirst is definite by virtue of it coming after a vocative and the deictic ‘you’.Both the vocative and the deictic are strong indices to the fact that the ref-erent is known by its being present in the physical context. This also appliesto the second example, which has the deictic expression ‘there’. Hasan (n.d.:425) provides another example highlighting the role of contextual factorsin this type of al:

(14) ‘inda ru’yatu kätibin mumsikin bi qalamin fï yadihi taqülu,al-waraqah, taqsidu khudhi al-waraqah al-hädhirah fi maqämi al-hadïth.‘Upon seeing a writer holding a pen in his hand, you say: “the paper”meaning “take the paper” which is present in the context of speech.’

Hasan argues that these utterances are made definite by means of the circum-stantial factors surrounding the speech event. In the first example the situationmakes the paper known because it is existent in the situation. The three modesof definiteness in the Arabic grammatical scholarship again correspond tothree domains of knowledge. The domain of language is represented by the al al-dhikriyyah, the domain of things and objects is represented by the al al-hudhüriyyah, and finally the domain of concepts is represented by the al al-dhihniyyah. This corresponds to my overall argument of metonymy as a representational relation connecting the world of concepts, the world of wordsand the world of things.

As I have mentioned earlier, the terms ‘schema’ and ‘frame’ are both gen-erally used to denote knowledge structures in memory. Thus we encounteruses like schematic representation of knowledge, which roughly means gen-eral conceptual representation of knowledge. This should not concern us,because I have already specified what I mean by schemata and frame withrespect to the scope of this book. Since I have argued above that frames are


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 152

actually schemata, what is said about schemata applies equally to frames. Aswe can talk of a city schema we can also talk about a city frame. We can alsotalk about a room frame, a supermarket frame, a hospital frame, a schoolframe and so on and so forth. All these types of knowledge structuresinclude knowledge about the contents of these domains as referred to inCook’s definition. In fact, if we go back to the originator of frame theory,Marvin Minsky, we find the ultimate legitimacy for our conception ofschemata and frames as descriptive and designating contents and entities orconcepts. The classical example of a frame given by Minsky himself consistsof an imaginary anecdote describing the experience of seeing a room. In hisillustration of this perception and expectation experience Minsky describesan interactionist model of frame activation, assimilation and modification(Minsky 1975:221–2). We are made to believe that the expectations and per-ceptions all pertain to the realm of entities which we expect to see in theroom. This applies equally to the world of language. It is not limited to therealm of vision but applies also to the realm of language-processing, whichis the expression of these visual experiences.

It should be noticed that metonymy operates in the realm of schemata andframes in both directions – that is from the particular to the general, or whatis referred to in metonymic terms as from the part to the whole and vice versa.So if we encounter a text which relies on mentioning wholes or generalitiesand leaving the reader to supply the details from his schematic memory, thenwe perceive this text as coherent simply because the details are taken to begiven and known. Usually metonymic networks operate on both directions inthe same text and this is what gives coherence to this text. For an example ofa text that depends heavily on metonymic reasoning for its coherence andunity in terms of its details, consider the following example of an office schemafrom an introductory part of a crime story after van Dijk (van Dijk 1977:98):

(15) Clare Russell came into the Clarion office on the following morn-ing, feeling tired and depressed. She went straight to her room, tookoff her hat, touched her face with a powder puff and sat down at herdesk. Her mail was spread out neatly, her blotter was snowy and herinkwell was filled. But she didn’t feel like work. She pushed the mailaway and stared out of the window.

In this example we have two types of schema. The first is the office schemawhich is triggered by the word ‘office’ giving naturalness to the parts of thisschema and making them easier to interpret in relation to the general pat-tern. The second is the person schema of Clare Russell herself. For the firsttype we could have the metonymic model of coherence shown in Figure 5.3.The second model is that of the person of Clare Russell herself, which canbe represented by Figure 5.4. Actually the text fuses both schemata in oneto suggest unity between the person and the office and work. The coherently


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 153

blended or integrated schema is represented in the Figure 5.5. This functionof the two schemas is in fact a cognitive process that is similar to conceptualintegration or conceptual blending between Clare Russell and her office (seeFauconnier and Turner 2002:42).

In this example we find the frame or schema of an office activated, thedetails being supplied by our prototypical knowledge of offices, for example


Office

Room

Desk Mail Blotter Inkwell Window

Figure 5.3 The office schema model in example (15)

She

She

She

Her hat Her faceClare Russell

Figure 5.4 The person schema model in example (15)

Clare Russell

She

She

She

Clarion office

Her hat Her face

Desk Mail Blotter Inkwell Window

Figure 5.5 A blend of the two models in example (15)

1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 154

that they contain desks, mail trays, inkwells and so on. So in this examplewe have the global pattern activated and some of the details mentioned asif known. As such the text is perceived as a unified whole describing a schematic representation that we construct as soon as we encounter theword ‘office’. The same thing applies to other types of schemata. For exam-ple in a school schemata we would normally expect teachers, tables, chairs,blackboard and chalk, and depending on when and where this school is con-textualised we could also have some other elements of what constitutesa school like computers, overhead projectors and televisions, and so on.

The above are examples of texts in which the whole is mentioned and someof the details are also mentioned but in which the interpretation of thesedetails is made easy by means of the activation of the global pattern. As I argued above, metonymy functions both ways and in the following we shallsee how parts of a frame or a schema could actually make it easy for us to workout the whole and perceive the context of the text. In the following text forexample the context is a kitchen and by mentioning some details of thisschema we are able to go from the specific to the general even before we actu-ally reach the mention of the word ‘kitchen’, which comes only late in thepassage. The text is from Like Water for Chocolate (1992:9) by Laura Esquivel:

(16) Take care to chop the onion fine. To keep from crying when youchop it (which is annoying), I suggest you place a little bit on yourhead. The trouble with crying over an onion is that once the chop-ping gets you started and the tears begin to well up, the next thingyou know you just can’t stop.

As soon as we encounter the expression ‘chop the onion’ the schema of a ‘kitchen’ is activated. The text actually uses the word ‘kitchen’ a few lineslater only as a qualifier of the word ‘table’ and not as a headword and withfurther parts of this schema. The text describes the birth of a baby called Titaas follows:

(17) Tita was so sensitive to onions, any time they were being chopped,they say she would just cry and cry; when she was still in my greatgrandmother’s belly her sobs were so loud that even Nacha, thecook, who was half-deaf, could hear them easily. Once her wailinggot so violent that it brought on an early labour. And before mygreat-grandmother could let out a word or even a whimper, Titamade her entrance into this world, prematurely, right there on thekitchen table amid the smells of simmering noodle soup, thyme, bayleaves and coriander, steamed milk, garlic and of course onion.

We can see from this text that the parts could easily signify the whole, andthis part–whole signification provides unity and ensures that expectations


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 155

are met. Even the digression to the issue of Tita being in her mother’s bellyand the explanation of her birth does not actually obstruct the functionalcontinuity of our schema because this digression is, as it were, a sub-schemafunctioning to give further explanation of the superschema. This is moreobvious in schema interaction and interplay with other cognitive mecha-nisms like scripts, plans and goals, which will be the subject of sections 5.4–5.6. One particularly interesting aspect that can be raised here is the use ofthe deictic expression ‘there’ in ‘right there on the kitchen table’. This showsthat there is actually strong reason to believe that the ‘kitchen schema’ isnot only supplying all the details but also making them concrete in themind of the receiver. Such an extralinguistic ontological reference can beunderstood because the ‘kitchen schema’ is so pervasive in our cognition astext receivers that the deictic and exophoric reference is easily resolved.

Example (18) below is a narrative text in which I intend to show howmetonymic details establish cognitive structures and create cognitive mod-els of text-processing:

(18) In the town there were two mutes, and they were always together.Early every morning they would come out from the house wherethey lived and walk arm in arm down the street to work. The twofriends were very different. The one who always steered the way wasan obese and dreamy Greek. In the summer he would come outwearing a yellow or green polo shirt stuffed sloppily into his trousersin front and hanging loose behind. When it was colder he wore overthis a shapeless grey sweater. His face was round and oily, with half-closed eyelids and lips that curved in a gentle, stupid smile. Theother mute was tall. His eyes had a quick, intelligent expression. Hewas always immaculate and very soberly dressed. Every morning thetwo friends walked silently together until they reached the mainstreet of the town. Then when they came to a certain fruit and candystore they paused for a moment on the sidewalk outside. The Greek,Spiros Antonapoulos, worked for his cousin, who owned this fruitstore. His job was to make candies and sweets, uncrate the fruits, andto keep the place clean. The thin mute, John Singer, nearly alwaysput his hand on his friend’s arm and looked for a second into hisface before leaving him. Then after this good-bye Singer crossed thestreet and walked on alone to the jewellery store where he workedas a silverware engraver. (McCullers 1961:7)

The first thing that attracts us in this text is the metonymic representation ofthe referential relation of the type CONCEPT FOR CONCEPT to ensure unity andcontinuity of reference in text. Let me show how these referential relations areestablished by means of metonymic structuring, which is largely a schematicmapping of forms onto concepts and things, that is the forms of pronominal


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 156

reference stand in metonymic representational relationship with signifiedconcepts and referents of these referential expressions. Figure 5.6 is an illustra-tion of how these referential relations are actually metonymically structured.

The diagram above shows how we can perceive of referential networks intext as essentially metonymic, building up cognitive models of whole–partand part–whole relations which provide unity and coherence to text. Oneparticularly interesting aspect of text coherence that this text shows is thatit seems that there is a process of textual transformation in terms of theknowledge allocation to particular elements in the text.


The one who The other mute

His

He

The Greek The thin mute

Spiros Antonapoulos John Singer

Two Mutes

They

They

They

They

They

They

The two friends

The two friends

He

His

He

His

His

His

His

Him

His

His

He

Figure 5.6 Metonymic referential relations in example (18)

1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 157

This is a phenomenon that is more especially apparent in texts withcataphoric reference, as is the case in the text above where the descriptionand the use of pronouns and common nouns actually precede proper nouns.Once the receiver of the text reaches the point where proper names are men-tioned, she or he seems to make a transformation in which the name thenreplaces the descriptions by pronouns, nouns and adjectives that precededthe proper name. This textual strategy seems to be vital to assist interpreta-tion and resolve ambiguity in text interpretation.

Another aspect of cognitive connectivity in the above text is found in the organisation of schematic knowledge of the physical setting of the text.The text is anchored in the situational context of a town. Figure 5.7 shows theset of metonymic relations present in the text with regard to the schema ofa town. It is clear from the figure that the schema of the contextual back-ground of the text determines the progression of the text. The figure showsa remarkable balance in the schematic structure of the concept of ‘street’ interms of the details given to each of the two participants’ type of job. In thetop left-hand corner of the figure we have the schema ‘house’ which standsalone in the schematic structure. This indicates that the house is not a sig-nificant element in the generation and organisation of this text. It servesa minor function in the production of the text in that it provides the textwith a point of departure from which to start. This allows the camera eye ofthe text to move away from this point out to describe the outside world, thatis the street and the two people in question. Just as syntactic structures arerepresented in the generative tradition using tree diagrams, one can arguethat textual structures can be similarly represented. Cognitively speaking,the head of this text is the schema ‘street’, which further sub-branches,yielding a right-branching diagram. In syntactic terms, the schema ‘house’is a non-head element and therefore it does not further sub-categorise forany other elements.


The town

The house The street

The main street of the town

Jewelry shop Fruit and candy store

Engraving Silver Fruit Candies and sweets

Figure 5.7 Metonymic relations within the town schema in example (18)

1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 158

One can actually go on and on in the description of several types ofschemata activated in the reading of this text. Where one stops depends basi-cally on the level of delicacy one adopts and the level of detail one wishes theanalysis to account for. Therefore, I shall limit myself here to presenting onemore schematic representation pertaining to the characteristics of the twopeople involved in the text. For the sake of reference I shall present somethingof a surface structure representation of this text, taking into consideration theprocess of transformation I have mentioned in the above model. This processactually takes the element ‘proper names’ and puts it at the beginning of thetext in order to make the interpretation clearer and the processing easier.

It is noticeable from Figure 5.8 that in cognitive domains we do not haveregular patterns like those in linguistic domains, especially if the schemataare employed to realise certain plans to achieve certain goals. For example,in the domain of language we have those languages which are left-branchingand those languages which are right-branching, depending on the place ofthe head in relation to other elements in the structure. In cognitive struc-ture, however, the branching is determined by the intentions of the origi-nator of the message and the requirements of the context. On someoccasions we have right-branching and on other occasions we have left-branching. Figure 5.8 shows left-branching and this suggests that SpirosAntonapoulos could be the main character or element of the narrative,hence a detailed description about his physical appearance and his clothes.It seems as if these features have special bearing on the development of theplot of the narrative. One should, however, be very careful in making sucha generalization, especially in the field of literature where one’s schemata are


The two mutes

Spiros Antonapoulos John Singer

physical featuresclothes

Sweater poloshirt physical features

green stuffed sloppily facegrey shapeless or yellow in front

trousers

hanging looseround oily

behind lips eyelids

tall dress eyes

Figure 5.8 Metonymic relations within the characters’ schema

1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 159

not always satisfied and one’s expectations not always met. This is becauseliterary language relies heavily on defamiliarisation,13 which basicallymeans shocking the receiver by deviating from his established schematicknowledge.

It should be noted that this detailed description is based on metonymicdetails and relies on parts forming wholes or wholes categorising their parts.By contrast, in the description of the other ‘mute’, that is John Singer, we haveonly wholes without mentioning parts. Apart from the concept ‘eye’ there is a remarkable tendency to use general terms in the description of Singer.For example, the concept ‘tall’ which suggests that his body is tall and thisin turn by implication suggests that he has long legs, long arms and longfingers and so on. All these details are not mentioned but our schema of‘someone who is tall’ supplies them. With regard to his clothes again thegeneral approach is adopted in their description. Apart from telling us thathe was always immaculate and very soberly dressed we actually have no ideaabout the details of his clothes. However, the general schema of someonewho is soberly dressed will supply plausible details. In either case, whetherthat of the detailed description or that of the general description, cognitionis aided by metonymic reasoning of the synthetic and analytic structuralrelations between wholes and parts, and is assisted by metonymic under-standing of cognitive movements within this structural network.

5.3 Metonymy and scripts

As the model of knowledge in section 5.1 above shows, the concept of scriptslies at the heart of our stereotypical and conventional knowledge. This meansthat scripts are mental structures of stereotypical knowledge of events. Whatis meant by ‘stereotypical’ is that the knowledge is conventionally availableto the individual by means of cultural saliency. Dyer et al. define the conceptof script as ‘a knowledge structure containing a stereotypic sequence ofactions’ (Dyer et al. 1987:980). They continue: ‘[S]cripts encode culturallyshared knowledge of stereotyped actions that occur in socially ritualizedactivities, such as going to stores, restaurants, and museums; riding trainsand subways; attending plays and banquets; and playing games or drivingcars.’ Two important points can be said about the definition of scripts pro-vided by Dyer et al. The first is that they are essentially related to sequencedevents and actions and not to entities and content. The second is that theyare culturally shared and thus conventional. These two important aspects dis-tinguish scripts from frames and schemata on the one hand and from sce-narios on the other, as captured in my model in section 5.1.

Schank and Abelson provide a similar definition to that of Dyer et al.,emphasising the aspect of a script as a representation of sequenced events.Their definition runs: ‘[A] script is a structure that describes an appropriatesequence of events in a particular context’ (1977:41). They further affirm the


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 160

view of ‘scripts being conventional’, a view also held by Dyer et al. andwhich is strongly emphasised in the model I proposed above in section 5.1.In this regard they maintain that ‘scripts handle stylized everyday situa-tions’. Reisbeck and Schank treat a frame as a fundamentally stable set offacts about the world, whereas a script is more dynamic in that it accountsfor ‘a standard sequence of events that describe a situation’ (Reisbeck andSchank 1978:254). Edwards argues in this connection that ‘Script theoryassumes that events themselves (social events primarily) are more or lessordered and predictable’ (1997:143).

As regards the function of scripts, Schank and Abelson argue that ‘whatscripts do, then, is to provide connectivity’ (Schank and Abelson 1977:40).They provide this connectivity by means of facilitating prediction in text interms of what sequences of events one should expect, in a restaurant forexample. One could imagine the set of actions involved in a restaurant scriptto start with the customer entering the restaurant, looking for a table, sit-ting down, looking at the menu, ordering, eating, paying and then leaving.If one of these acts is not mentioned our scriptal knowledge supplies it.

De Beaugrande and Dressler define scripts as ‘stablised plans called up veryfrequently to specify the roles of participants and their expected actions’(1981:91). Although the treatment conflates the notion of ‘scripts’ with thenotion of ‘plans’, it tells us something important about scripts which weshould take into consideration. This is that scripts are not always global pat-terns of knowledge in the sense that they do not form holistic concepts ofstereotypical situations. They are often reduced to the viewer’s perspectivebut they are still universally standard and culturally shared. This means thatthe script of a restaurant, for example, is not homogeneous. Rather, it variesaccording to the person viewing it. A restaurant script of a customer is dif-ferent from a restaurant script of the cook or the owner of the restaurant. Thisis to say that the set of actions prescribed for each of the above roles will vary.This point further highlights the need for differentiating between schemataand scripts. While all the above participants will have more or less a unifiedview of the schema of a restaurant they would have different representationsof the roles and actions each one should abide by in the restaurant.

In the sentence

(19) I got up, washed and had breakfast.

There is a great deal of detail that is missed out, and to provide this we relyon metonymic inferencing based on script activation. Each of the three actsinvolved in the example actually stands for a whole series of smaller actsneeded to accomplish it. If after this text the following text is encountered:

(20) I was annoyed upon finding out that there was no hot water.

we would not find a problem interpreting it because this sub-act would beavailable thanks to the ‘washing schema’ and to the metonymic relation of


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 161

CONCEPT FOR CONCEPT, which is based on the metonymic relation of PART FOR

WHOLE, which enabled us both to construct this schematic representationand to recover it when the need arose. The script of someone washing acti-vates other detailed scripts in our memory, and one of these is the act ofusing water to wash. Moreover, this subscript on its own becomes a super-script for further minute detail, which this subscript itself activates. Amongthese is the subscript that people get irritated if the water is cold when theyget up in the morning. However, the washing script itself might be said tobe universal because it seems that human beings, irrespective of their dif-ferent cultures, generally wash when they get out of bed. The subscript relat-ing to ‘hot water’ is not universal, because it is quite easy to imagine a worldwhere people do not get irritated if the water is cold in the morning and infact one could imagine a world where people get irritated if the water is hot.Nonetheless, both types of scripts are still culturally shared and thus they donot lose their autonomy, in contrast with those knowledge structures whichare not culturally shared.

I want now to elucidate further the difference I am proposing between aschema and a script. The most important thing about schemata is that theyhelp us predict entities and objects within certain knowledge domains.Scripts, on the other hand, involve more than mere entities. In fact, scriptsinclude these entities in a textured manner. That is, they involve them withthe action sequence of the whole stereotypical act. Therefore, scripts are morecomprehensive than schemata, because while the former group involvesboth entities and actions the latter includes only entities. The treatment Ipropose for metonymy as a script goes beyond the modern figurative accountof metonymy, which can be illustrated as follows:

(21) The ham sandwich is getting impatient.

The lexical view of metonymy deems the treatment of this example as anissue of deferred or shifted reference based on a metonymic process of thetype THE THING ORDERED FOR THE PERSON ORDERING THAT THING. However, thistreatment of metonymy as a figure of speech, that is a lexical phenomenon,fails to capture the power of metonymy as a text-organising principle.

Example (21) is not only metonymic in its lexical or referential aspect butis also metonymic in its textually scriptal aspect. The example activates thescript of a restaurant. This script tells us that the person who ordered theham sandwich had already found a table and sat down and perhaps had readthe menu. The restaurant script also tells us that he must have ordered aham sandwich, although the text does not explicitly tell us so. The scriptsimilarly tells us that he had been waiting for so long that he was gettingimpatient, and that he would perhaps not leave a tip or come again to thisrestaurant. Finally, the script tells us that the person ate a ham sandwich,although nowhere in the text is this mentioned. If we gave this example to


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 162

a reading class and asked them: ‘What did the person eat?’ the answer wouldbe: ‘He ate a ham sandwich.’ This information is supplied by our scriptalknowledge because we know that when people order things in a restaurantthey eat them. In this way, metonymy has a powerful function in terms ofputting the example text in its wider textual context. In fact, this under-standing not only makes the example a part of a wider world but renders itthe creator of this wider context, thanks to our metonymic understanding.The treatment of example (21) as a lexical phenomenon is not sufficientbecause it does not tell us why the person is referred to by the thing heordered. A scriptal metonymic treatment does tell us that the principle ofsaliency is paramount and is a driving force behind text generation. This isto say that the script of a restaurant and the participant roles within thisscript make the metonymic use necessary as a form of dehumanising therelation between the customer and the waitress. This is because all that isimportant for the waitress in this script about Mr or Mrs X is that he or sheordered a ham sandwich.

Schank and Abelson argue that scripts are usually invoked by means ofcertain conceptualisations. These are essentially key elements which activatethe whole script. They maintain that there are four types of what they term‘headers’ (see Schank and Abelson 1977:48–9). The first type of script headeris the ‘precondition header’ (PH) and this ‘triggers the script reference onthe basis of a main script precondition being mentioned in the text’. Theyclaim that, for example,

(22) John was hungry.

is a PH for the restaurant script. This gives a chance for the receiver to pre-dict what is going to be said in the next text from the available data in frontof him. So upon understanding that John was hungry, we could predict thathe is going to try to fulfil a need, which is getting fed. One of the paths inour script for getting fed might be a restaurant. So in this sense the restau-rant is metonymically predicted by means of the metonymic relation CAUSE

FOR EFFECT. The cause of going to the restaurant, which is ‘hunger’, stands forthe effect of this hunger. Hunger activates and facilitates the prediction ofthe entity ‘restaurant’ by means of metonymic causality that can be phrasedas, ‘Because he was hungry he went to a place to eat.’ This universal knowl-edge is followed by the option of cultural knowledge. This is the schema ofa restaurant with the possibility of other paths in the script. For example theperson might go home for a meal or go to a takeway and eat in the park.Nonetheless, all these paths within the script of ‘getting fed’ will be acti-vated and it is the progression of the discourse that will either confirm ordeviate from each of these paths.

The second type of script header is that which Schank and Abelson callthe ‘instrumental header’ (IH). They argue that this comes up in situations


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 163

where there are usually two or more contexts. The idea is that in suchsituations where two frames are activated by the script usually one of themis an instrument for the other main frame. In the example which the authorsprovide, ‘John took the subway to the restaurant,’ there are two frames acti-vated. One of them, the subway schema or frame, is actually an instrumentto achieve the main schema, the restaurant. This conceptualisation of a mainheader and an instrument header in script organisation is useful in manyrespects. It helps in predicting the direction and the progression of the dis-course and explains how branching is possible in some elements in the scriptwhile it might not be in some other elements. It is also useful because it tellsus that the progression of the text will dwell much on the main script, thatis the restaurant in the example above as the target of the text. We expectfurther elaboration on the ‘restaurant script’. The subway script is onlyinstrumental and as such it will not be the main part of the epistemic dimen-sion of the discourse. So, as we can see, certain scripts can be suppressed bythe way we expect the discourse to proceed. We expect no more elaborationon the subway script apart from that which has already been mentioned.However, both scripts, that is the subway and the restaurant script, will bekept open to account for the interpretation of the discourse to come.

The third type of script header is the ‘locale header’ (LH). This explains theparticularly specific types of scripts which as soon as they are mentioned in thetext reinforce the retrieval of stereotypical knowledge of certain sequenced anddetermined actions and events related to that script. Examples of locale head-ers are schools, hospitals, churches, prisons, museums and so on, where boththe entities as well as the set of actions to take place can be easily expected andpredicted. However, there is a problem here with the activation of a particularscript and the various participant roles involved in the script. The best exam-ples of this case are the church, hospital and prison scripts, where each one ofthese would demand a certain specified set of actions to take place.

These actions however will differ according to the role of the participantinvolved. For example we would expect the behaviour of someone going tochurch for the sake of performing the service to be radically different fromthe behaviour of someone visiting the church as a tourist. Similarly, the per-son going to prison as a criminal is not going to behave as someone goingto prison as a visitor. So for the person going to prison as a prisoner wewould use the construction,

(23) He went to prison.

whereas for the person going to prison as a visitor we would use the structure

(24) He went to the prison.

The definite article here has the function of specifying the role of the par-ticipant and thus characterising the type of script to be activated.


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 164

This particular aspect has been formally taken care of by the language andhas become a component of the grammar of English in this example. Arabichowever does not make this formal distinction with regard to the change ofthe role of the individual involved by using the definite article, as is the casein English. However, it seems that Arabic depends heavily on the semanticsof the verb to make such a distinction. So

(25) Huwa dhahaba ’ila al-sijn.‘He went to prison.’

would automatically be interpreted as ‘He was there as a visitor’.

But

(26) ’ukhidha ’ila al-sijn.

would be readily interpreted as

(27) He was taken there as a prisoner.

Arabic tends to use a completely different verb to show the distinction per-taining to the participant roles. By using two different verbs, Arabic makesuse of this potential to provide economy of interpretation and hence to acti-vate one and only one of the two possible scripts. The way the text is con-structed in the passive to show that the person involved was in prison as aprisoner is a subtle aspect of the thematic structure of the semantic systemof the Arabic language.

The fourth type of script header is what Schank and Abelson call the ‘inter-nal conceptualisation header’ (ICH). The case with the previous type ofheader is that two types of scripts are activated, and depending on the role ofthe person involved the interpretation is accessed via one of these twoscripts. The situation in this case is fairly similar; the person involved hastwo roles and the text activates both roles but only one is primary to theinterpretation of the text. The difference between LH and ICH is that whilethe ambiguity in the former is disambiguated by means of formal possibili-ties, that is the definite article, the ambiguity in the latter is disambiguatedby means of resolving the role pattern of participants in the schema. This iscoupled with resorting to the oncoming discourse, which further specifiesthe type of role that should be activated and the type of role that should besuppressed. This is to say it is the progression of the discourse that deter-mines which role is to be activated.

The following example of the ‘waitress’ is given by Schank and Abelson:

(28) John went to visit his friend Mary who was a waitress. While he waswaiting for her, he ordered a hamburger.


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 165

It is characteristic of this situation. The main role of a waitress is to work inrestaurants to give orders to the customers. As soon as we encounter theword ‘waitress’ in a text the metonymic reasoning of PART FOR WHOLE func-tions to help us access the restaurant schema. However, the case here is quitedifferent because the role of a ‘waitress’ here is not called up for its mainunmarked function, that is as a person taking orders in a restaurant, but forits marginal marked function, that is as a friend. The mention of this friendbeing a waitress is a piece of bypassing information and an internal con-ceptualisation, thus the restaurant schema is only marginally activated.

Let us imagine that in a text from a novel in Arabic we encounter the wordma’dhün, which means ‘the authorised marriage contracts conductor’. Theword ma’dhün will automatically activate the script of marriage or wedding.However, depending on the role which this particular element, namely thema’dhün, takes on, the interpretation and the expectation of the progressionof the discourse will change. For example, had we been told in previous dis-course that Ahmad had a friend who is a ma’dhün and that he was comingto visit him in the house, we would expect that the principal role of this ele-ment in the text would be suppressed in favour of the marked role of himas just a friend. However, we need to keep open the path of a possibility ofa marriage contract to be conducted, because we must bear in mind thepotential for script deviation that is characteristic of literary language.

As we have seen throughout the discussion of the notion of script, theprocesses of script activation and text expectation are based on metonymicrelations of PART FOR WHOLE, WHOLE FOR PART, CAUSE FOR EFFECT OR EFFECT FOR

CAUSE. The reason for this is that the actions and sequences in scripts are nor-mally universal or at least culturally shared. So in a sense we have, as it were,inbuilt systems for identifying sequences in such scripts without much dif-ficulty because they are part of our being human or being members of socialcommunities.

Scripts facilitate global understanding but they do not perform this func-tion on their own. In fact, scripts become coherent themselves only whenour cognition supplies the details of a script, or sometimes the logic behindthe specific sequencing of events in the script in the case of the script devi-ating from our shared knowledge. If this logic of sequencing, which is fun-damentally causal based on CAUSE-FOR-EFFECT or EFFECT-FOR-CAUSE metonymicrelations, is missing then, a script will not be able to function properly intext generation, organisation and interpretation. Equally if the part-for-whole reasoning which is essentially metonymic, as we have seen in the var-ious examples in this section, is missing from a script, the whole structureof a script is bound to collapse. This is evident in machine text-processingbecause in case of a machine nothing can be taken for granted and everypiece of knowledge must be computed. This shows what a huge amount ofknowledge and cognitive reasoning the machine needs if it is to be ableto process text. In the case of human processing of text there is usually


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 166

considerable underestimation of the powerful processes of organisation andconceptualisation of thought into coherent discourse. I believe metonymyis one of the fundamental principles of thought organisation which makesit possible for this thought to conform to culturally shared scripts. In short,metonymic reasoning of causality and contiguity is a prerequisite for anyprocess of script activation. More importantly, the conception of metonymyas essentially a relation of representation allows for each mode of knowledgeto stand for another via contiguity or causality. As we have seen so far in theexamples discussed in the previous two sections, metonymy provides men-tal access that helps us move from one domain of knowledge to another inorder to account for the coherence of discourse.

The notion of causality should not however distract us from the fact thatscripts represent standard stereotypical situations that are either universallyor culturally shared. Such distraction could actually cause a confusion whichI want to avoid. If scripts are explainable only through causality then itmight be argued that they are not universal or culturally shared, as we haveto think always about the cause. However, we should not go too far in stress-ing the notion of causality here. Causality here still has a universal aspect orat least a culturally shared aspect. In other words, we can think of globalcausality and local causality. So in the example:

(29) Tom was hungry. He went to a restaurant.

there is a strong causal connection between the two sentences which can beexpressed as, ‘Because Tom was hungry he went to a restaurant.’ The pointI want to make is that this notion of causality should not be confused withthe concept of scenarios which I am going to discuss in the next section.The reason is that this causality is itself universal because as human beingswhen we become hungry we look for food. Even the second part of the textis culturally shared because now in many cultures when people are hungrythey go to restaurants. On the other hand, the causality pertaining to sce-narios is a rather restricted form of causality, and this is explainable only bymeans of other knowledge structures, that is plans and goals.

Consider the following example:

(30) John went to the toilet.

Here the script of a toilet is activated, but hardly anyone would ask whyJohn went to the toilet because the reason is universally known. It is part ofour knowledge that we accumulate throughout our lives and this is preciselywhy we need to use euphemistic expressions in such situations. However,there are many actions in our behaviour that are difficult to explain becausethey are arbitrary, and these require a different version of causality toaccount for them. This is what might be termed ‘personal’ or ‘contextually


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 167

bound’ causality, and it explains certain arbitrary behaviours that do notconform to general stereotypical patterns. An important point that shouldbe made at this stage is that scripts, being stereotypically universal or cul-turally based, are quite limited; it seems that the conception of a script as astereotypical sequence of events leads to the conceptualisation of text-processing as static and involving the stable imposition of ready-madeformulae of standard situations. This is obviously not totally true becausetext coherence and text interpretation are essentially dynamic. This dynamic-ity is evident in many situations where our expectations are not actually met.

Let us now consider the following text:

(31) John was hungry. He was looking for a restaurant. He remembereda good, cheap one he had been to before. But it was quite far away.John thought about taking a taxi but he was not sure he had enoughmoney. He then decided to take the underground. Inside the stationhe went directly to the ticket machine and purchased a single ticketto Baker Street.

In this example we have the general script of ‘someone hungry’ and it is easyto predict what will come next in the text. However, this prediction that theperson in question will have food is only very general. The details of howhe attained this goal are not accounted for in the script and in fact they can-not be specified by our scriptal knowledge because there are many ways, infact infinite ways, to achieve this goal. The details in the above example aregiven to us by the various scenarios which are set to achieve the goal. Theconstruction of these scenarios is arbitrary in the sense that it is not pre-dictable via a global pattern. The topic of arbitrariness in text generation,organisation and interpretation is one I turn to in the following sections.

5.4 Metonymy and scenarios

What are scenarios? How do they differ from other knowledge structures dis-cussed in this chapter? How do they contribute to text coherence and thusfacilitate text interpretation? How do they relate to metonymic reasoning?This section will try to answer these questions. Again the literature on knowl-edge structures conflates the notion of scenarios with those of schemas(Brown and Yule 1983:246) and scripts (Sanford and Garrod 1981:110) andsometimes with that of frames (Minsky 1975:240–6). Scenarios as discussedin this chapter are knowledge structures which resemble scripts but whichdiffer from them in that while scripts are explained by conventionalsequences of actions, scenarios are explained by means of specific causality,which pertains to particular interests and intentions.

As the model in section 5.1 shows, scenarios are treated here as arbitraryknowledge structures. However, a word of caution is necessary regarding the


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 168

notion of arbitrariness. This term has acquired a strong semiotic connotation,especially in connection with Saussurean structuralism, which propoundsthat linguistic signification is arbitrary (Saussure 1983:67) in the sense thatthere is no logical or intrinsic connection between the signifier, that is the lin-guistic expression, and the signified, that is the meaning of that expression.In this case the Saussurean theory of linguistic signification denies ‘motivat-edness’ in language (ibid., p. 69).14 The notion of arbitrariness as presented inthis chapter,15 it should be noted, does not have this terminological load. Thesimple reason for this is that the fundamental principle underlying this bookand indeed many other accounts of the role of metonymy in our languageand thought (see Lakoff and Johnson 1980; Lakoff 1987; Lakoff and Turner1989; Gibbs 1994; Fass 1997; Goatly 1997; Lakoff and Johnson 1999; Pantherand Radden 1999, Barcelona 2000; Fauconnier and Turner 2002) is that thereis a considerable amount of motivatedness in language and cognition andthat metonymic signification is contiguously and causally motivated.

Therefore, the notion of arbitrariness in this chapter should be taken in arestricted sense to denote a situation where the prediction of text relies noton universal or cultural standards but rather on personal intentions andindividual manipulations of context to achieve certain goals. We can, tosome extent, accept the notion of arbitrariness in the Saussurean sense if itis applied to the relation between the signifier and the signified in individ-ual lexical items, as in the examples provided by Saussure himself. Saussurepostulates ‘there is no internal connexion, for example, between the idea “sister” and the French sequence of sounds s–ö–r– which acts as its signal’(Saussure 1983:67). However, it is quite difficult to accept the notion of ‘arbi-trariness’ when we deal with text. The reason is that in the realm of text weare actually dealing with intentions, biases, prejudices and ideologies thatcannot be thought to be arbitrary but which we assume, rather, to bestrongly motivated.

Within one script we might encounter one or more scenarios. If we lookat example (31) above we find that the ‘John hungry’ script allows us topredict that the following text will be an explanation of how this hungryperson will feed himself. We know this because of stereotypical knowledge.So in this case taking a taxi or the underground is a scenario of transferringoneself to one particular place to get fed. Within this scenario of transfer-ring location we can encounter one or more scripts, depending on thechoice of the means of transport. So if the choice was for a taxi, the taxiscript will be activated in our memory and the details will be automaticallysupplied. If, however, the taxi script is activated but then later suppressed infavour of another more convenient choice, as is the case in the exampleabove, the taxi script is then shut and the underground script is activated toaccount for the interpretation of the oncoming text.

The interactive nature of the process of script and scenario activation andin fact between all other knowledge structures in text-processing is crucially


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 169

important,16 because the assumption is that these mechanisms are notintegrated and obviously cannot function in isolation. Rather they collaborateto make sense of text. The schema allows for the activation of a script and thescript enables the activation of a scenario which is explained by the set ofplans and goals governing the whole act of communication. This is, in fact,the rationale behind my integrated model of all these mechanisms. In the lit-erature these are usually discussed separately and with a great deal of overlapand confusion between them. Here I have attempted to systematise them andtreat them as a chain of problem-solving and inferencing mechanisms, suchthat where one fails to solve the problem the other comes in to do so.

One could perhaps extend the argument of conventional vs. arbitraryknowledge to account for general epistemic patterns. For example, onecould argue that descriptive text is generally conventional. It conforms toreal accounts of ontological specifications of properties. It stems from well-established conventions shared by human beings in general. So the sense ofcontiguity is well-rooted in descriptive reasoning and is usually coached viacontiguous associations like part–whole relations, cause–effect relations,maker–made relations and so on. However, if we move to the other end ofthe cline of descriptive vs. argumentative text we have the kind of knowl-edge one might call ‘political’, or perhaps more generally ‘ideologicallyloaded’. This pertains to texts with argumentation in, for example, eitherreligion or politics. This is characteristically unpredictable and dependsheavily on scenarios which are explained only by recourse to the intentionsand interests of the participants. So while descriptive knowledge relies heav-ily on scripts, argumentative knowledge relies heavily on scenarios. The signlooses its referential power in argumentative knowledge, more of the epis-temic ground of the discourse is symbolic and there are many abrupt con-clusions. The referential power of the sign is more or less stable in descriptiveknowledge. Descriptive knowledge can be associated with narrative anddescriptive or expository discourse, while argumentative knowledge can beassociated with argumentative discourse in general. This latter depends onactivation of various scenarios in the process of the interpretation.

Consider the following example from al-Sharq al-’Awsat newspaper:17

(32) kuntu dä’iman dhidd ar-ra’ïs räbïn, kuntu ’u‘äridhuhu wa’uqäwimuhu wa’uqabbihu min tasarrufätihi al-‘anïfati dhidd al-‘arab, waläkinnï hazintu limasra‘ihi, wa sha‘artu bi’annanä khasirnä khasman qawiyyan käna yu‘arqilu al-mufäwadhät, wa känayatashaddadu fi mawäqifihi dhidd al-‘arab waläkinnahu känayuhibbu bilädahu, wa qad härabanä wa ntasara ‘alaynä thummahärabnähu wa ntasarnä ‘alayhi fi 6 ’uktübar. ’istankartu masra‘ahuli’annanä dhidd al- ’ightiyälät sawä’ käna al-majnï ‘alayhi sadïqan’am khasman, wa käna ’ahamma mä fï hädhä al-hädith al-mu’lim’annahu ’athbata ’anna al-‘arab laysü wahdahum maydän al-’ightiyälät, bal ’inna ’isrä’ïl kadhälik tushärikunä hädha l-balä’. wa


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 170

nahmadu alläha ’anna räbïn qutila biyadi yahüdi, wa law käna al-qätilu ‘aräbiyyan lanhälat ‘alayna al-’ittihämät min kulli makän,wa lamä baqiya ‘arabiyun wähidun bighayr ’an yuttaham bi’annahuqätil ra’ïs wuzarä’ ’isrä’ïl.’amma al-’än falqätilu shakhsun wähidunlä maläyïn al-‘arab alladhïna yuttahamüna bikulli shay’in yahduthufi al-‘älam.

Translation I was always against Rabin. I hated his violent actionsagainst the Arabs, and opposed him in whatever ways I could.However, I was sad when he died. I felt that we had lost a powerfuladversary, who, while he may have hampered negotiations andadopted an extreme attitude towards the Arabs, nonetheless lovedhis country. He fought and defeated us in 1967, then in 1973 wefought and defeated him. I felt revulsion at his death, because onemust be against assassination regardless of whether the victim is afriend or an enemy. The most important aspect of this sad incidentis that it showed that the Arabs are not the only people who go infor assassinations. Israel is also afflicted by the same disease. Wethank God that Rabin was killed by a Jew. Had the assassin been anArab, accusations would have been flung at us from all directions.There wouldn’t have been a single Arab who hadn’t been accused ofthe killing of the Israeli prime minister. Now, however, it’s clear thatthe killer was a single person, not the millions of Arabs who findthemselves accused of everything that happens in the world.

The first sentence in the text does not topicalise or set the scene for a globalschema or script, as is usually the case in narrative and descriptive text.Normally, the topicalisation is a syntactic aspect which has a textual role.This role is to activate a conceptual knowledge structure which is theschema to account for the oncoming discourse and integrate it with the acti-vated schema or script. In example (32) above we do not seem to have suchactivation; neither do we have topicalisation, as the sentence begins withthe verb to which the subject is annexed, as is the case in the canonical VSO(verb–subject–object) word order in the unmarked Arabic sentence. So whatdoes this sentence activate and how does it facilitate the interpretation ofoncoming discourse? Actually, the sentence activates a scenario of someoneopposing someone else. We do not know why. The reasons are explainedlater on from the point of view of the originator of the discourse. This pointis crucially important in the scenario interpretation of text because the textis heavily loaded with ideological bias.

The difference between a script and scenario is that while there is somesort of activation of conceptual knowledge involved in both, the activationof the former helps in prediction of further text and establishes a sharedunderstanding between the sender and the receiver of the text. In the latter,on the other hand, there is an activation of some sort of event which is not


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 171

conventionally shared between the sender and the receiver. Rather, thisaction or event is expressed from the point of view of the sender, whichmight not necessarily agree with the receiver’s own perspective. Accordingto our conception of global patterns of knowledge we would expect the textabove to continue with one script in which the writer or sender will con-tinue hurling expressions of hatred and opposition throughout. However,this is not the case. Rather, the text mixes several scenarios, leading to mul-tiple interpretations, which is characteristic of political discourse.

In scenario activation we proceed in the text because we are keen to knowwhy the actions are the way they are presented. In script and schema acti-vation, apart from deviant and defamiliarised schemata and scripts of poeticor literary language, we normally proceed in the text with quite soundexpectations. When we encounter a text beginning with ‘in the hospital…’we activate a schema of a hospital. As a result, concepts like nurses, doctors,beds, X-rays and so on will be supplied by our memory and as we go on inthe text we take these items as given. The same thing applies if we activatethe hospital script. Here in addition to the set of things and objects that aretreated as given we will also be able to supply a great number of sequencedactions and roles for participants in the script. However, with the text aboveit is difficult to imagine a situation that is universal where someone isagainst someone else by convention. This kind of conceptualisation neces-sarily requires that there are specific reasons for this attitude.

In the text above there is no global pattern that is activated and the wholetext is embroiled in a socio-cultural plot which although it depends on widerhistorical events is essentially a sequence of unpredictable actions each withits own causality pattern. Expressions in the text such as ’u‘äridhuhu wa’uqäwimuhu wa ’uqabbihu min tasarrufätihi al-‘anïfah dhidda al-‘arab ‘I wasalways against Rabin. I hated his violent actions against the Arabs’ seem quitepredictable because these are compatible with the sense of ‘someone beingagainst someone’ which has been activated in the first sentence of the text.However, it is strange to find the expression walakinnï hazintu limasra’ihi,‘However I was sad when he died.’ This is because we would not normallyexpect someone who strongly opposes another to feel sorry for his death.However, this is justified if we take into consideration that the text is argu-mentative and it makes use of several scenarios, perhaps the most relevant ofwhich here is that the writer is trying to show sympathy with the deceased.Nonetheless, as will be shown later, this is just one among the scenarios thatare invoked in the text.

As I pointed out earlier, scenario activation is based on specific causality.In the text above the sequence of causality, it should be noted, is not men-tioned in the order which we would normally expect, that is the cause pre-ceding the effect. Rather, we are first introduced to the statement of theexpressions of hatred and opposition, that is the effect even before knowingthe reason. One could argue that the text here is displayed against a wider


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 172

historical and socio-cultural background, which explains the reason behindthis hatred and opposition. However, although this might prove true tosome extent, the text actually expresses the cause but only after the state-ment of the effect, that is tasarrufätihi al-‘anïfah dhidda al-‘arab ‘his violentactions against the Arabs’. The fact that this text is displayed against thebackground of the Arab–Israeli conflict is evident in the use of the word ‘al-mufäwadhät’ ‘the negotiations’ with the definite article although it hasnot been mentioned earlier. This shows that as soon as the text activates theglobal schema of the Arab–Israeli conflict by mentioning the word ‘Rabin’,the concept of ‘negotiations’ comes into mind and is thus treated as known.One could also argue that the word ‘negotiations’ is definite here because ittakes its specified reference from other texts salient in the socio-culturalbackground invoked by the text.

With the word waläkinnï ‘But I…’ a new scenario starts. This is quite differ-ent from the previous scenario which exhibits a rather negative attitudetowards Rabin. In this scenario we are informed that the death of Rabin is caus-ing the writer sadness and that the Arabs had lost a powerful adversary, and soon. On the surface these expressions seem to be praising Rabin but at a deeperlevel the same scenario of negative attitudes is still operating because althoughthe phrase ‘a powerful adversary’, is presented as a way of praising, it is actu-ally loaded with opposition. A powerful adversary is an enemy who is not justan enemy but an extreme enemy. The text goes on to explain that the Arabshad lost this powerful adversary who was hampering the negotiations andadopting an extreme attitude towards the Arabs. Obviously none of these qual-ities is liked by an Arab and the question that comes to mind is how it couldbe possible to claim that the Arabs had really ‘lost’ Rabin? It seems these expres-sions are mentioned here just to mislead the reader and force him to employmultiple scenarios, a feature characteristic of argumentative discourse.

The text is an argumentative text. In this type of text the producer of thetext utilises his power to force certain views and ideas on the receiver.Analysis of the balance of power relations in argumentative texts is animportant technique to reveal the interaction between scenarios and expec-tations, the scenarios being the input from the text producer and the expec-tations the input from the receiver of the text. In this text, however, thereseems to be no pattern of power exerted by the text producer. On the otherhand, the text is torn between apologetic and hostile expressions. There isa general sense of trying to please more than one party rather than impos-ing the producer’s own plans. There is keen interest in compromise betweenconflicts of interest and this is achieved through the activation of multiplescenarios. This activation is warranted by metonymic relations of causalityand contiguity in the sense that each state leads to the existence of anothercausally or contiguously related state.

The sentence waläkinnahu käna yuhibbu bilädahu ‘but he loved his coun-try’ is put in the text misleadingly to show sympathy with Rabin. In fact,


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 173

however, the proposition expressed in this sentence is that he is the enemyof the Arabs if he loves his country. This sentence might be a positiveappraisal if said by an Israeli but it is actually the opposite if said by an Arab.Here it is just another way of allowing for a positive scenario to come in theinterpretation to somewhat mitigate the rather negative general attitude.The text continues by mentioning the military confrontations between thetwo sides, that is the Israelis and the Arabs. The mention of wars is an indi-cation of hostility in which bad memories are recalled for both sides. Thusin view of the scenarios of the text it is easy to see why the writer includedthis point in his article. Now let us consider another part of the text wa käna’ahamma mä fi hädha al-hädith al-mu’lim ’annahu ’athbata ’anna al-‘arab laysüwahdahum maydän al-’ightiyälät ‘An important aspect of this sad incidentwas that it proved that the Arabs are not alone in the field of assassinations.’This point is perhaps the statement of one of the main goals of the wholetext. This is to show the triumph for the fact that not only Arabs assassinatetheir leaders as was the case with Sadat.

Then comes the climax and pivot of the text which expresses the realmotive of the article. This is the part which begins with wa nahmadu alläha’anna räbïn ‘We thank Allah that Rabin …’. This shows clearly that the aimof the whole text is to use the fact that Rabin was assassinated by an Israelito defend the Arabs from potential accusations and to show that they areusually wronged and misrepresented. It is clear from the scenario analysis ofthe text that the whole text is split between the conflict of scenarios. Onescenario is activated and then soon suppressed in view of more globalknowledge related to history and socio-political knowledge. In fact, the con-flict between the scenarios activated and then suppressed is a natural reflec-tion of the conflict between the author as a writer who should adopt anobjective attitude towards the incident and the author as an Arab who hashis own emotions and ideologies. Nonetheless, all these apparently contra-dictory scenarios in fact run coherently to realise the ultimate goal of thetext, which to express the fact the Arabs are usually misrepresented.

The writer seems to be following an objective and universal code of con-duct in mitigating his attitude towards the incident. In fact, it could beargued that he is neither happy nor sad regarding the incident. What he isinterested in is to use the incident to draw analogies of how the incidentreflects bias and prejudice. The author wants to blame the internationalcommunity for applying double standards when it comes to the Arab–Israeliconflict. This is evident from his concluding remarks ‘amma al-’än falqätilshakhs un wähidun lä maläyïn al-‘arab, ‘Now, however, it’s clear that the killerwas a single person, not the millions of Arabs.’

The writer wants to say that had the assassin been an Arab then all the Arabswould have been accused of terrorism but because it is now clear that theassassin was an Israeli the world will treat it as an individual case. The fact thatscenarios are based on localised causality is what makes scenarios basically


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 174

metonymic structures, because metonymy realises causal relations underlyingsense continuity which give the text its unity and coherence. Standard or uni-versal causality is actually part of schematic knowledge because once we readthat ‘John is hungry’ we expect that he will get fed. However, the conflict ofintentions and interests in scenario manipulation is by no means conven-tional. Rather, it is the result of continuous negotiation with the text and itsown epistemic as well as ontological dimensions of knowledge. It is clear fromthe above example that text coherence depends largely on identifying the setof cause–effect relations not only on the linguistic presentation of facts but also by going deeper and consequently wider than that – that is, by tak-ing the formal text as only one resource and searching for other sources ofknowledge that contribute to our understanding of the text as purposeful andthus make it stand as a coherent and meaningful unit.

Minsky raises an interesting point which proved useful to my discussionof the three main knowledge structures, that is schemata, scripts and sce-narios. Minsky elaborates the following point sketched in Winston (1970):

What does it mean to expect a chair? Typically, four legs, some assortmentof rungs, a level seat, an upper back. One expects also certain relationsbetween these ‘parts’. The legs must be below the seat, the back above. Thelegs must be supported by the floor. The seat must be horizontal, the backvertical, and so forth. Now suppose that this description does not match;the vision system finds four legs, a level plane, but no back. The ‘differ-ence’ between what we expect and what we see is ‘too few backs’. This sug-gests not a chair, but a table or bench. (Minsky 1975:252)

On the basis of the above quotation we can draw a conclusion bringingtogether the three concepts I have been discussing in this chapter. Minskytalks about the ‘typical’ features of a chair. I relate this aspect to the notionof schemata which I have argued to contain knowledge of entities. ThenMinsky deals with what he calls ‘relations’. These are obviously the struc-tural relations between these entities. They are structural in the sense of cat-egory membership and thus forming structures. The notion of ‘relations’can also be expanded somewhat to include ‘sequential’ relations, coveringboth ‘spatial and temporal’ dimensions. In this case it is easy to postulatethat this level of ‘relations’ pertains also to the notion of script as I have out-lined it above. Then we come to the notion of ‘deviation’ which is expressedby Minsky using the expression ‘not match[ing]’. I take this level of repre-sentation to belong to the third category where our global or universalexpectations are not met for specific reasons. This is the level of scenarios.

Now theoretically speaking these three levels of representation, that isschemata, scripts and scenarios, correspond to the three levels of metonymicrepresentation. The first is the entity or the selection level where two wordsor things or concepts are made to form a system. This level can also be


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 175

termed the ‘paradigmatic level’ of metonymic representation. It consists ofwords or concepts or things which correspond to the level of typicalityreferred to in Minsky’s discussion and to the notion of schemata in myaccount. The second level of metonymic representation is the level whereone of the two entities is actually put in the representational string. This isthe syntagmatic level of metonymic representation which corresponds tothe level of relations in Minsky’s account and to the notion of ‘script’ in theconceptual structure of our knowledge. The third level is the level of ‘notmatching’ as referred to in Minsky’s argument. This consists of unpre-dictable connections which are explainable only via contiguous means. Thisis the level of metonymic interpretation which involves a great deal of rea-soning and searching for various types of association. The jump from thelexical domain of metonymy outlined here to the textual level of interpret-ing text is actually warranted by the nature of metonymy as an essentiallycombinatorial aspect of language and thought.

5.5 Metonymy and plans

Were coherence in discourse always to be achieved via stereotypical situationswhich describe frequently encountered situations and sequences of eventsthen this would make it easier for us to understand any type of text withoutmuch effort. Or in more technical terms, this would put understanding alwaysin the default mode. In fact, according to this view any model of scriptalknowledge would be sufficient to account for text-processing and therefore wewould not need to look for other types of knowledge structures. A moment’sthought however will reveal that this is not the case because human under-standing is far more complex than this reductionist view of standard knowl-edge structures. There are many types of text that we wrestle with tounderstand. Human generation and interpretation of knowledge is much morecomplex than a model of stereotypicality associated with schemata or scriptscould account for. For this reason Schank and Abelson proposed the notion of‘plans’ to account for situations where ‘people can deal with situations thatthey have never encountered before. They can do this because they have accessto the mechanisms that underlie scripts’ (Schank and Abelson 1977:70).

Schank and Abelson (1977:70) define a plan as a theoretical entity which

is intended to be the repository for general information that will connectevents that cannot be connected by use of an available script or by stan-dard causal chain expansion. A plan is made up of general informationabout how actors achieve goals. A plan explains how a given state orevent was prerequisite for or derivative from another state or event.

They further assert that ‘plans describe the set of choices that a person haswhen he sets out to accomplish a goal’. It is clear from Schank and Abelson’s


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 176

treatment of plans that they regard plans as devices which explain people’sintentions. The nature of plans as used in this chapter is instrumental in thesense that they give connectivity to certain actions that are otherwise dis-connected. The role of plans is explanatory and they give information as towhy a set of sequence is made contiguous, although there seems to be noexplicit thread of continuity between its parts. The example from Gibbs is agood illustration of the role of the metonymic relation of causality inexplaining the role of plans and goals in activating scenarios.

The example tale provided by Gibbs (1994:330) runs as follows:

(33) He wanted to be king.He was tired of waiting.He thought arsenic would work well.

Without activating the metonymic relation of causality, the interpretationof this text is severely hampered. The coherence of this text is based on theactivation of a certain scenario which is established by means of the identi-fication of the plans and purposes underlying the text. The text activates ascenario of someone planning to be king and indirectly alludes to the stateof the affairs that there is already a king and that is why he has to wait fora long time and as a result has become tired of waiting. This state allows forsome action to take place. This action is actually a plan and it aims to getrid of the cause of this tiresome waiting situation, which is the existence ofthe present king in this example. The mention of the word ‘arsenic’ activatesa whole scenario of plotting an act of poisoning the present king to get ridof him. Without metonymic reasoning of causality and without scenarioactivation by identifying the set of plans that underlie the text, the textwould be extremely difficult to understand. This shows how the role ofmetonymic reasoning is fundamental in discourse-processing. Theseprocesses, it should be noted, are not alien to the cognitive structure ofour brain, as it is sometimes believed in certain figurative accounts ofmetonymy. Rather, they are processes that are fundamental to our thoughtand cognition.

De Beaugrande and Dressler define plans as ‘global patterns of events andstates leading up to an intended goal’ (1981) (pp. 90–1). This definitionseems to confuse frames and schemas in the use of the phrase ‘global pat-terns’, especially if we know that the authors regard these latter two as dif-ferent. The definition is rather general also because it uses the term ‘plan’ todesignate events and states which lead to intended goals. Any event oraction, if deliberate, leads to an intended goal, so there is nothing specificabout plans to make them different from scripts for example. Here, I shalldwell on the intentional aspect highlighted by de Beaugrande and Dressler’sdefinition. A plan is a purposeful action in the sense that it serves to explainwhy a particular scenario is presented in the way it is. Let us look at the


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 177

following text from the London newspaper Al-Sharq Al-Awsat (17 April1993). The text shows how metonymic reasoning of causal relations under-lying the text is actually a fundamental aspect of text coherence.

(34) Bada’a nizäm hukm ar-ra’ïs al-‘iräqi saddäm husayn hamlatan mukha-ttatatan litajfïf qitä’ätin wäsi‘atin min ‘ahwär al-‘iräq kawasïlatin lihirmän al-mutamarridïn fi al-janüb min ‘ayi ghitä’, wa li’ijbär sukkänal’ahwär ’nfusahum ‘lä al-khudhü‘ lisaytarati an-nizäm fï baghdäd.

Translation The regime of the Iraqi President Saddam Hussein began aplanned project to drain vast areas of Al-Ahwar in Iraq as a means ofdepriving the rebels in the south of any cover and to force the people ofAl-Ahwar themselves to submit to the dominion of the regime inBaghdad.

The understanding of this text is first of all based on an understanding of ametonymic relation of CONTROLLER FOR CONTROLLED. This is evident in theattribution of the act of draining to Saddam Hussein’s regime, although theact itself is not actually carried out by Saddam Hussein or any member ofhis government. This metonymic reasoning at the beginning of the textreveals and exposes a latent ideology and an implicit blaming attitudeagainst Saddam Hussein by a Saudi-owned newspaper. The attribution of theact of draining Al-Ahwar is a technique to throw the blame for the actionon Saddam Hussein and his regime because they are the controllers of thepeople who actually carried out the draining. The controlled, that is thepeople who performed the action of draining, are basically workers forSaddam Hussein’s regime. Besides they are not known anyway, in which caseit is difficult to blame them. In such cases where the real doer of the actionis not known or not important enough to be known or to be blamed, thenour cognition shifts the reference from the real doer of the action to thecause behind the doing of the action. In this way the responsibility for thisact of environmental catastrophe is thrown onto Saddam Hussein and hisregime.

The coherence of the text also depends on metonymic reasoning in amore specific way. Without identifying the network of plans, goals and sce-narios in the text it is impossible to understand the text. Without establish-ing a relation of cause–effect between the act of draining and the act ofdepriving the rebels of any cover the text will not stand as a meaningfulwhole. Not only that, but also we have to supply information that is notactually mentioned in the text. Our metonymic reasoning of causality willtell us, though the text itself does not help us in this regard, that when anarea is drained of water then the trees dry up and the land is denuded ofcover. Also we have to make yet another metonymic inferencing process by which we use our general knowledge of the world that rebels usually hideunder trees and therefore the removal of the foliage of these trees puts


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 178

the lives of these rebels at risk. Moreover, the fact that the removal of thefoliage of the trees and perhaps subsequently the trees altogether deprivesthe rebels of shelter to hide, forcing them to flee or surrender to the attack-ers, which in this text is the regime in Baghdad. Without these inferencingprocesses text unity and text continuity will be disrupted.

It should also be noted here that the amount of knowledge ourmetonymic reasoning of causality and contiguity supplies us with dependsheavily on the level of detail required and the level of sophistication ourknowledge should have. For example, there are still several metonymicprocesses of cause–effect relations that are not mentioned in the analysisabove, let alone the text. For instance, the causality in the draining processmight be elaborated more in such a way as to show minute details of causalchains, such as the fact that lack of water in the land causes lack of water inthe roots of trees. Lack of water in the roots of trees causes lack of water inthe stem and this leads to lack of water in the leaves. When the leaves donot have water they become yellow, dry up and fall, leaving the tree with-out leaves. In fact, the account would be more detailed than this were we toadopt a zoological or botanical analysis of what actually happens in theroots, stem and leaves when there is no water.

Our metonymic reasoning of CAUSE FOR EFFECT relations underlying the texthelps us interpret the text as a purposeful unit on the basis that each state iseither a result of another state or a state that initiates a further state. Thiscausal connectivity in the text is not the only way we can perceive metonymicroles in text interpretability, however. In fact, the text is also interpretable bymeans of other metonymic relations of structural perception of part–wholeorganisation. The mention of wholes deems the parts of these wholes knownand thus accessible in the memory. This facility ensures that the amount oflanguage we use in our communication is minimised and that language is eco-nomical enough to avoid redundancy. Language as a system of communica-tion stands against a cognitive representation that is far more detailed thanwhat is actually made explicit in speaking or writing, as the text above shows.In this sense the essence of cognition can be seen to represent a metonymicoperation in which the linguistic elements of communication stand for thewhole realm of cognition.

The word ‘arsenic’ in example (33) above helps a great deal in resolvingthe interpretation problems associated with the very elliptical form of thescenario. In the following example we find the word ‘gun’ as a useful toolto activate specific plans which help to realise the goals at which the initia-tor of the text is aiming:

(35) John needed some money.He took the gun and headed towards the local bank.

We have a robbery scenario here, an interpretation warranted by the mentionof the word ‘gun’. We understand that the state of being in need of money


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 179

leads to a state of trying to get money. At this stage there are several plans oractions for the scenario to proceed. John might take one of his valuable thingsand sell it. Within this plan-based scenario, the gun could actually be a valuableantique gun that John has inherited, and because he is now in need of moneyhe has taken the gun and is heading towards the bank to put it in pawn toborrow money from the bank. However, this interpretation is not directlyaccessible because the description of someone needing money and going tothe bank carrying a gun is in its unmarked interpretation highly suggestiveof a robbery scenario.

Another path in the scenario would be that John might phone one of hisrelatives or friends and ask for help. However, the text imposes on us a spe-cific progression. Because the progression of the text through the robberyscenario is supported by default plan interpretation we would normallyexpect that other plans of getting money had failed. If our interpretationwere to be directed towards the ‘valuable antique gun’ interpretation wewould normally expect some more specification to be explicit in the text toguide us to take that particular path of reasoning. For example, we wouldexpect that the text would explicitly state that

(36) John was in need for money. John decided to pawn his antique gunso he carried his gun and headed towards the local bank.

Of course each of these scenarios would activate a particular path of textunderstanding and prediction and each would activate a particular settingand a specific script. We would for example expect the robbery scenario totake place at night when there is no one in the bank, whereas we wouldexpect the gun-pawning scenario to take place during the day and more pre-cisely during work hours.

From the above discussion of the role of plans in text understanding andtext prediction we can see that processes of both understanding and predic-tion in text are interrelated and each depends on the other. It seems that ourunderstanding of a text is based on our prediction of the progression ofcausality with regard to the set of plans underlying the linkage betweenevents and actions. In the same way our prediction of the progression of thetext relies heavily on our understanding of the current state of affairs thatthe text is describing and the mechanism the text is employing to stand asa unified whole. Metonymic reasoning provides text understanding and textprediction with a unique tool that does not limit understanding to one ortwo cognitive principles but rather integrates all the available knowledgestructures to work in an interactive manner to test hypotheses and attest themost probable ones with evidence from the text and from other sources out-side the text. Thus the hypothesis of the chapter is that metonymy as a rep-resentational relation provides mental access that cuts across the threedomains of language, mind and reality.


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 180

One of the examples Schank and Abelson (1977:71) provide for the notionof ‘plan’ is very suggestive of a metonymic relation of CAUSE FOR EFFECT. Theirexample is:

(37) Willa was hungry.She took out the Michelin guide.

In this example there is ellipsis not very much different from the ellipsis in

(38) John went to McDonalds.

In Willa’s story in example (37) above the second sentence strongly invokesan elliptical element, which is ‘to search for a restaurant’. In example (38)there is an elliptical element, which is ‘the restaurant which is calledMcDonalds’. There is not a very big difference between the two types ofellipsis except that in the second example the reference is fairly fixed and isinvoked by well-established cultural semiosis. However, in the first examplethere is only a sense of expectation, and although the causal linkage stronglyactivates a scenario of searching for a restaurant in the guide it is still anexpectation. Example (38) is also metonymic in the causally driven expec-tation sense also. One should expect an elliptical element to continue. Thisis ‘to eat’ because this element is supplied by the schema of a restaurant thatit contains ‘food’ and by our script of going to a restaurant which tells usthat people go to restaurants to eat.

5.6 Metonymy and goals

Goals are more general plans. A plan may describe a specific action in a sce-nario to achieve a general goal. Text coherence is, among other things, adialectic interaction between the fulfilment of the goal, or the declining ofthat goal if it is not fulfilled and resorting to another goal. The followingtext from Naguib Mahfouz’s Za‘baläwi illustrates how text coherence is gov-erned by the overall goal that the text sets out right from the start. The textbegins with the following:

(39) ’iqtana‘tu akhïran bi’anna ‘alayya ’an ’ajid ash-Shaykh za‘baläwi.‘I was finally convinced that I had to find Sheikh Za‘baläwi.’

This start actualises a general goal and text coherence depends heavily onpursuing this goal. In other words the writer is expected to elaborate subse-quently on how he attained this goal. In fact, for a few lines after this state-ment the writer gives the background as to why he needs to find SheikhZa‘baläwi. Mahfouz spends a considerable amount of time giving the readerthe background of this overall goal. He tells us that Za‘baläwi, according to


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 181

the narrator’s father, is a holy man, a saint who was able to solve all sorts ofproblems. The text tells us also that the narrator was in a crisis and that wasthe reason why he was searching for Za‘baläwi. Within the description ofthe actions and events that were carried out to attain the goal of findingSheikh Za‘baläwi, we encounter several other scenarios and plans con-tributing to the achievement of the goal.

For example, the following text tells us that the narrator rememberedwhat his father had told him about the location of Za‘baläwi which is a pre-requisite state for the person to transfer his location to that place to achievethe goal of finding this sheikh.

(40) wa tadhakkartu ’anna ’abï qäla ’innahü ‘arafahu fi bayt ash-Shaykhqamar bi khän ja‘far, wa huwa Shaykh min rijäl ad-dïn al-mushtaghilïn bilmuhämäh ash-shar‘iyyah. Faqasadtu baytahu wa’aradtu at-ta’akud min ’annahu mä zäla yuqïm fïhi fasa’altu bayyä‘fül asfal al-bayt, fanazara ‘ilayya bistighräb wa qäl:– ash-Shaykh qamar! Tarak al-hay min ‘ahdin ba‘ïd, wa yuqäl’innahu yuqïm fi gardin siti wa’anna maktabahu bimaydän al-azhar … wastadlaltu ‘alä ‘unwän maktabihi bidaftar at-tilifün,wdhahabtu ’ilayhi min tawwï fi ‘imärat al-ghurfah at-tijäriyyah.

Translation I remembered that my father had told me that he hadknown him in Sheikh Qamar’s house in Khan Ja’far. Sheikh Qamar was aclergyman whose profession was that of lawyer in Islamic courts. I headedtowards his house. But I wanted to make sure that he was still living there.So I asked the ‘fül’ seller at the basement of the house. The man lookedat me strangely and said: ‘Sheikh Qamar! He left the block long ago. It issaid that he now lives in Garden City and his office is in Al-Azhar Square.’I found out his address from the phone book and went to him immedi-ately in the Building of the Chamber of Commerce.

From the text above we can see that there are several actions that took placeall of which contribute to the realisation of the main goal. Going to see SheikhQamar and asking the man who sells ‘fül’ at the basement and consulting thephone book are all actions to achieve the general goal. In fact, these actionsare predictable or at least easy to process because of the activation of the gen-eral goal of the text. All these actions are causally linked and thus their inter-pretation is processed by means of CAUSE-FOR-EFFECT metonymic relations.Identifying a goal is also a metonymic process in the sense that a text is usu-ally perceived as purposeful, that is as having a general goal. This general goalfunctions like a holistic cognitive domain from which specific actions andplans derive their interpretation and meaningfulness.

The search for Sheikh Qamar which occupies the whole text above is onlyan instrumental secondary goal to serve the general goal of the text, that is to


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 182

find Sheikh Za‘baläwi. The actions within the secondary goal are scenariosrealising the set of particular plans. This chain of interconnectedness ofplans, scenarios and goals is the unique characteristic of metonymic rea-soning of the relations PART FOR WHOLE and CAUSE FOR EFFECT and the converseof these relations. In this case the unity of the text is cognitively maintainedby means of structural integration and cognitive causality. This intercon-nectivity is what provides the text with coherence and meaningfulness andmakes its interpretation possible. In fact, the overall plot of the storyrevolves around this general goal of searching for Sheikh Za‘baläwi. The textgoes on with this goal, which however also includes several digressions toother scenarios and elaborations on various plans in order to achieve it.


1403_913323_08_cha05.qxd 11/19/03 8:43 PM Page 183

184

1 Theoretical Framework and Historical Background

1. See the discussion of these parts in Kennedy (1994) and Dixon (1971).2. To give but four examples: the notion of ‘context’ from anthropology

(Malinowsky 1923); from philosophy and speech act theory (Austin 1962); fromcooperative principle and the maxims of communication (Grice 1975); from soci-ology, politeness and face (Brown and Levinson 1978).

3. Edwards (1997) argues in this connection that ‘In conversation analysis and dis-cursive psychology, the “loose fit” between descriptions and their referents has tobe understood as an essential feature of how language works’ (p. 116).

4. Langacker (1993:29–33) discusses some similar examples regarding metonymyand active zones.

5. See Jakobson’s proposal in his article (1971) and the discussion of this proposalin Chapter 2, section 2.1.1.1.

6. Jakobson observes in this connection that

similarity in meaning connects the symbols of a metalanguage with thesymbols of the language referred to. Similarity connects a metaphorical termwith the term for which it is substituted. Consequently, when constructing ametalanguage to interpret tropes the researcher possesses more homogeneousmeans to handle metaphor, whereas metonymy, based on a different principle,easily defies interpretation. Therefore, nothing comparable to the rich literatureon metaphor can be cited for the theory of metonymy. (1971:95)

7. See Miftäh for a similar attempt to treat both kinäyah and majäz mursal as onephenomenon. In this connection he argues that ‘majäz mursal becomes onespecial case of kinäyah because both involve a relation between two things’(1985:114–15).

8. This type of majäz is based on existential relations like totality, partiality, situa-tionality and positionality. These relations obviously pertain to the world ofthings and objects and not so much to the world of words or concepts.

9. The word used in the Arabic text is ‘kinäyah’ which could be translated asmetonymy but I preferred to use the general meaning which is intended in theoriginal text as contrasted to tasrïh, that is explicitness. It should be noted herethat the term kinäyah in Arabic is always associated with implicitness – a pointthat will be elaborated further in the course of this chapter.

10. Al-Mubarrad’s social functions of kinäyah revolve around the social values desig-nated to covering names by means of title ‘kunayah’. It should be noted here thatsome Arab rhetoricians give examples of what they perceive as kinäyah that arederived totally by means of convention rather than by means of proximity orcontiguity. For example, they would interpret the word na‘jah ‘ewe’ which is men-tioned in the Qur’anic verse inna hädhä ’akhi lahu tis‘un wa tis‘ün na‘jatan ‘thisman, who is my brother, has ninety-nine ewes’ as ‘woman’.

11. Context here refers to the wider notion of context which is the cultural contextas in the case of idioms for example.

Notes

1403_913323_09_not.qxd 11/19/03 8:43 PM Page 184

12. This view is very important in modern treatments of metonymy which consider thetrope as comprising two types: one is called referential metonymy and the otherpredicative metonymy (see Chapter 2, sections 2.1.2.4 and 2.1.2.5, for details of this).

13. Both forms are morphologically possible.14. Here he means majäz ‘aqlï, because the whole discussion is devoted to this type

of majäz and the examples are illustrations of it.15. ‘Cognitive’ in the sense of rational as delineated in the definition of this type of

transfer above.16. I mean here al-Qazwïni, who showed some sort of systematicity in his discussion

of majäz.17. According to one interpretation, the verse suggests that had the fingers been of equal

length, man wouldn’t have been able to perform skillful actions with his hands.

2 Metonymy in Modern Figurative Theory

1. This is perhaps why for quite a long time the paper remained almost unnoticedby linguists or critical theorists. Lodge (1977:73) rightly argues that the paper didnot seem very inviting to literary critics because its contents did not sound rele-vant to their concerns.

2. Bansloben (1996:8) argues that ‘critics such as Burke and White consider the two-fold model far too limiting. Instead, they advocate the more traditional four tropeapproach to rhetoric’. I think Jakobson revolutionised the theory of tropes, andhis work benefited the theory of metonymy in particular in many ways. In otherwords, Jakobson put an end to a long tradition of trope enumeration. He tried toreveal the discursive and signifying power of both metonymy and metaphor.

3. Terms invented by I.A. Richards in his The Philosophy of Rhetoric. Goatly uses‘topic’ instead of ‘tenor’ and uses the term ‘ground’ (G) (1997).

4. As we have seen in Chapter 1, this is the name Rosiene uses for an unknownancient Latin rhetorician to whom Rhetorica ad Herennium is credited. Sometimesthe treatise is ascribed to Cicero and as a result the author is sometimes referredto using the name Cicero, and followed by in square brackets ‘the Auctor’. Thisis the way the author is referred to in the bibliography of this book.

5. In semiotics and semantics, the term ‘seme’ as Wales asserts is ‘from Greek sema“sign”, and by analogy with PHONEME and MORPHEME, seme is used inEuropean STRUCTURALISM and semantics (e.g. Greimas 1966; Coseriu 1967) todescribe a minimal DISTINCTIVE FEATURE of meaning or COMPONENT.’ Semesdefine the essential denotations of different lexical items within a lexical field interms of binary oppositions: e.g. items of clothing can be marked as being ‘with[�] or without [�] sleeves’ (e.g. jacket vs. waistcoat) (Wales 1989:415).

6. On p. 109 the Groupe maintain that to construct a metaphor, we must coupletwo complementary synecdoches that function in a precisely inverse way andthat fix an intersection between the terms S and R.

7. This is the symbol the Groupe use to designate the ‘semantic plane’.8. This is the symbol the Groupe use to designate the ‘material plane’.9. Osterwalder cites George S. Klein, who has labelled two basic attitudes – levelling

and sharpening. The former ‘tends towards the obliteration of differences to facil-itate categorisation’ while the latter ‘is a tendency to be hypersensitive to minu-tiae, to respond excessively to fine nuances and small differences’ (1978:17).

10. This is of course indicative of a cyclical metonymic signification process whichbegins with the cloth moving to the desk and then to the whole office.

Notes 185

1403_913323_09_not.qxd 11/19/03 8:43 PM Page 185

11. This situation is usually referred in sociolinguistic studies as ‘diglossia’. The firstoccurrence of the term ‘diglossia’, according to Hary (1993), was by Krumbacherin the very earliest years of the twentieth century, in which the author discussedboth Greek and Arabic diglossia. Charles Ferguson reintroduced the term in hisinfluential paper bearing the name ‘Diglossia’ as its title. He defines ‘diglossia’ as‘a relatively stable language situation in which, in addition to the primary dialectsof the language (which may include a standard or regional standards), there is avery divergent, highly codified (often grammatically more complex) superposedvariety, the vehicle of a large and respected body of written literature, either ofan earlier period or in another speech community, which is learned largely by for-mal education and is used for most written and formal spoken purposes but is notused by any sector of the community for ordinary conversation’ (1964:435).

12. See for these terms Fauconnier (1985:3–34).13. I am using the term ‘indexical’ in a semiotic sense. See the discussion of this point

in the next chapter, section 3.4.2. 14. The notion of schema is developed in Chapter 5 to account for the role of

metonymy as a process of schematic coherence in text.15. I think the reason for this is that there is a shift in theoretical perspectives in which

modern accounts of tropes tend to be more intellectually driven to find out gen-eral rules and principles governing the cognitive processes involved in both ofthese two tropes. This focus is obviously different from that which is characteris-tic of traditional accounts of the tropes and which aims to provide taxonomies offigures and strives to enumerate types, examples and species without actually pro-viding general principles. See for further discussion on this Cooper (1986:12–20).

16. See the interesting discussion of the pragmatic aspects of tautologies in Wierzbicka(1991:392–452).

17. Panther and Thornburg (1999) treat this type of metonymy as potential foractual.

18. Here ‘functional’ has no terminological load. It refers to the fact that Gibbs ismore interested in the function of metonymy in discourse.

19. Compare also ‘Nixon, who was a staunch anti-communist, bombed Hanoi almostimmediately after taking office,’ which is surely predicative.

3 Metonymy and Semiotics

1. Keller (1998:47) coins this term to refer to Plato’s mystery of how is it possible‘when I utter this sound, [and] I have that thing in mind … [that] you know thatI have it in mind’?

2. In this connection, Fiske and Hartley maintain that ‘the central concerns of semi-otics … are: … the relationship between a sign and its meaning; and the way signsare combined into codes’ (1978:37).

3. For a good survey of this vastness see Eco (1976:9–14), who refers to the numer-ous areas that are related in some way or another to the semiotic field as the polit-ical boundaries.

4. It should be noted that even in linguistics the level of substance is usuallyregarded as to fall outside the scope of linguistics. See Aitchison (1995:8), whereshe illustrates the scope of linguistics by a diagram and maintains that ‘in thecentre is phonetics, the study of human speech sounds. A good knowledge ofphonetics is useful for a linguist. Yet it is a basic background knowledge ratherthan part of linguistics itself.’

186 Notes

1403_913323_09_not.qxd 11/19/03 8:43 PM Page 186

5. It should be noted here that ‘sign’ language which is a non-verbal system ofcommunication has acquired quite a lot of conventionality and systematicity.This would make it a proper communication system, albeit a restricted one.Hawkes shows the validity of this assumption when he argues that

a language, … does not construct its formations of words by reference to thepatterns of ‘reality’, but on the basis of its own internal and self-sufficientrules. The word ‘dog’ exists and functions within the structure of the Englishlanguage, without reference to any four-legged barking creature’s real exis-tence. The word’s behaviour derives from its inherent structural status as anoun rather than its referent’s actual status as an animal. (Hawkes 1977:16–17)

6. Elsewhere (1983:16) Saussure asks ‘why is it that semiology is not yet recognisedas an autonomous science with its own object of study, like other sciences?’Saussure provides an answer to this, stating that ‘nothing is more appropriatethan the study of languages to bring out the nature of the semiological problem’(Saussure 1983:16).

7. This quotation is taken from ‘some aspects of structuralism and its application toliterary theory’ in a website belonging to the author at the following address:http://www.brocku.ca/english/courses/4F70/struct.html

8. Halliday is a pioneer in this respect. His theory of language as social-semiotic andhis concern for the functions of language in society have helped him to offera detailed model of language in society which attempts to explain how languageis actually used by people in real-life communicative situations.

9. See Lyons (1977b:ch. 3), who argues that prevarication is one of the main prop-erties of human language because human language is perhaps the only system ofcommunication that allows for misinforming.

10. In this connection he argues that

the individual’s receptive and coordinating faculties build up a stock ofimprints which turn out to be for all practical purposes the same as the nextperson’s. How must we envisage this social product, so that the language itselfcan be seen to be clearly distinct from the rest? If we could collect the total-ity of word patterns stored in all those individuals, we should have the socialbond which constitutes their language. It is a fund accumulated by the mem-bers of the community through the practice of speech, a grammatical systemexisting potentially in every brain, or more exactly in the brains of a group ofindividuals; for language is never complete in any single individual, but existsperfectly only in the collectivity. (Saussure 1983:13).

11. See his example of the symbol of justice, ‘the scales’, which he wittily questionswhether we can replace with a chariot (Saussure 1983:68).

12. As will be shown later in this chapter, Peirce’s conception of sign is more relevantto the course of this book because of its consideration of an ontological dimen-sion of sign in addition to the cognitive and linguistic dimensions.

13. Sampson (1980:47), after a lengthy discussion of Durkheim’s concept of ‘collec-tive mind’ and the clash between ‘methodological collectivism’ and ‘method-ological individualism’ as two ways of thinking about the subject-matter ofsociology, argues that this was ‘very much a live issue in the intellectual milieuwithin which Saussure’s views on language were formed’. Compare also Eagleton,who refers to the concept as the ‘collective consciousness’ (1983:110). See alsoRicoeur (1976:3) for a similar view.

Notes 187

1403_913323_09_not.qxd 11/19/03 8:43 PM Page 187

14. Fauconnier (1985) in this regard comments that

language, then, is not merely interpreted with respect to worlds, models,contexts, situations, and so forth. Rather, it is involved in constructions of itsown. It builds up mental spaces, relations between them, and relations betweenelements within them. To the extent that two of us build up similar space con-figurations from the same linguistic and pragmatic data, we may ‘communicate’;communication is a possible corollary of the construction process. (Ibid.:2)

15. This view is known as the Sapir–Whorf hypothesis of linguistic relativity.16. As we have seen above, Hawkes echoes this view also.17. See Eco (1976), who provides a number of classification systems which have been

used to classify signs.18. For a detailed discussion of the various types of Peirce’s sign classification see

Hervey (1982:23–32).19. Interesting discussions regarding the notion of nature vs. convention occupied

much of Greek philosophy. Different views were provided by the two famousGreek schools of linguistic philosophy, namely, the Stoics and the Alexandrians.The former advocated the analogist perspective of language as being essentiallyregular owing to the system of symmetries provided by convention. The latteradvocated a view of language as anomalous and maintained that language lacksregularity because of the inherent irregularities in nature. See for further detailsLyons (1968:4–12).

20. It is to be noted here that the English translation of Frege’s Bedeutung is ‘refer-ence’, which captures the ontological features that Frege himself intends.

21. Compare Hervey (1982:136) for similar ideas expressed by Barthes’s notion ofconnotation, which itself draws on ideas from Hjelmslev.

22. This can be related to the three types of speech acts which speech act theory pos-tulates. The first is the locutionary level, which refers to the speech activity,which is here the signification activity. The second is the illocutionary act, whichrefers to the type of act the utterance is enforcing and which can be linked withthe notion of warning of potential fire in our example here. The third level is theperlocutionary act, which is the response of the receivers to the warning. See forfurther detail on this Lyons (1977:ch. 16); Levinson (1983:ch. 5).

23. This argument is founded on the translation of the definitions of metonymy bythese authors. The translations are those of Rosiene in his thesis.

24. See the definition and the discussion of this definition in Chapter 1, section 1.4.1.25. Nunberg (1993:2). Nunberg argues that the meaning of indexicals like ‘we’, for

example, has three components. The first is the ‘deictic component’, which is‘a function from occurrences or utterances of an expression to elements of the con-text of utterance’. The second component Nunberg calls the ‘classificatory compo-nent’, which is associated with the interpretation of an expression rather than withits index. This tells us about features of plurality and animacy of the expression.The third component is that which Nunberg calls the ‘relational component’,which ‘constrains the correspondence that has to hold between the index and theinterpretation’ (ibid.: 8–9). Nunberg further argues that the main difference betweena description and an indexical is that the former characterises its interpretationwhereas the latter provides an object that corresponds to it. The object is an indexaccording to Nunberg, who maintains that this index

stands in a ‘relation of contiguity’ to its object, as a rolling gait to a sailor, arap on the door to a caller, a symptom to a disease. This is the characteristic

188 Notes

1403_913323_09_not.qxd 11/19/03 8:43 PM Page 188

and most remarkable feature of these expressions. They enable us to turn thecontext itself into an auxiliary means of expression, so that contextual featuresare made to serve as pointers to the content of the utterance. (Ibid.:19–20)

26. These are the knowledge of words, the knowledge of concepts and the knowledgeof things that I have outlined in various places in this chapter. This interdomainconception of metonymy has been integrated with a parallel three-dimensional con-ception of sign and also with a parallel three-dimensional conception of sign modes.

27. As discussed in major sources such as those of Schank and Abelson (1977), Minsky(1975), Bartlett (1932), Rumelhart (1977), de Beaugrande and Dressler (1981) andUngerer and Schmid (1996).

28. See the discussion of ‘procedural knowledge’ in de Beaugrande and Dressler(1981:87–8).

4 Metonymy and Text Cohesion

1. Indexical here is taken to mean the type of indexicality which is a pervasive fea-ture of discourse. Edwards in this respect argues that ‘all words are indexical inusage’, because every time we use a word we use it for that particular situation toserve a purpose at that particular time.

2. Here I attend to the tradition which distinguishes sentence meaning from utter-ance meaning on the ground that the former is the meaning that is conveyed bya sentence regardless of context and the latter is primarily the meaning thespeaker intended to convey regardless of the form. See for this view Blakemore(1992:5–6). From a semiotic perspective, ‘sentence meaning’ would correspond to‘meaning potential’, while ‘utterance meaning’ would correspond to ‘meaningactual’ whereby the signification is reduced to a particular context or intendedsignification.

3. See also Brown and Yule, who propose a communicative definition of text whenthey state: ‘[T]ext … is the verbal record of a communicative event’ (1983:190).See also Halliday and Matthiessen (1999), who echo this in their definition of textgeneration as a logogenetic process which creates meaning in the course ofinstantiation as the text unfolds (p. 384).

4. See Halliday and Hasan’s statement that ‘it is the underlying semantic relation …that actually has the cohesive power’ (ibid., p. 229). However, as Brown and Yuleargue, ‘[T]hey insist that it is the presence of the cohesive marker which consti-tutes “textness” ’ (1983:192). Brown and Yule further argue that

[A]ny adequate model of discourse description must be able to accommodatethe various connections which do exist in texts … The ‘cohesion’ model doesnot. It is, however, only fair to point out that Halliday & Hasan are not con-cerned to produce a description which accounts for how texts are understood.They are, rather, concerned to examine the linguistic resources available to thespeaker/writer to mark cohesive relationships. (p. 204)

Moreover, Brown and Yule hold that Halliday and Hasan’s model of cohesion doesnot make a distinction between ‘the meaning relations which hold between itemsin a text and the explicit expression of those ‘meaning relations’ within a text’.This distinction, as Brown and Yule state, is ‘an important distinction to be drawn,which many students adopting Halliday & Hasan’s approach have failed to drawand which Halliday & Hasan themselves are somewhat ambivalent about’ (p. 195).

Notes 189

1403_913323_09_not.qxd 11/19/03 8:43 PM Page 189

5. There is actually an approach to the analysis of text which is known as ‘substitu-tional text linguistics’ which perceives of the relations underlying textness as sub-stitutions. See for example Harweg (1977).

6. It should be pointed out here also that the word ‘gents’ is itself metonymic stand-ing in a formal part–whole relationship with the full form of the word ‘gentle-men’ of the type form for form exactly like the form UN stands in a part–wholerelationship with the full form United Nations.

7. Langacker argues for a view of reference as cognitive and that language formhelps to realise basic cognitive processes. He talks about reference point phe-nomena which he defines as ‘our capacity to invoke the conception of one entityas a cognitive reference point for purposes of establishing mental contact withanother’ (1993:1). This conception is essentially identical with the metonymicrepresentation processes which provide reference points between entities such asbetween parts and wholes.

8. The unmarked structure would involve a repetition of the word fa’s ‘axe’.9. This is to say that the metonymic relation might be representational, i.e a

stand-for relation realising a cause for effect or the effect for the cause, or the partfor the whole or the whole for the part. See the relational model of metonymy inFigure 3.5.

10. The example is taken from the third epistle of ‘ikhwän al-safä’ the ‘Brethern ofPurity’ cited in Dickins and Watson (1999:540).

11. Perhaps the terms ‘central’ and ‘peripheral’ can be correlated with their semioticcounterparts of ‘denotative’ and ‘connotative’ respectively. So the central seman-tic traits are those which pertain to the denotative and basic dimension of signi-fication, whereas the peripheral traits are those concerned with the behaviour oflexical items in texts and social contexts, which is a value added to the basic coremeaning.

12. For further detail on this issue see the discussion of ‘absolute synonymy’ in Cruse,who argues that if we conceive that ‘two lexical units would be absolute syn-onyms if and only if all their contextual relations were identical’, then

it would, of course, be quite impracticable to prove that two items wereabsolute synonyms by this definition, because that would mean checkingtheir relations in all conceivable contexts (it would be theoretically impossi-ble, if, as is probably the case, the number of possible contexts is infinite).

13. See the example and discussion of reiteration above.14. This is a common feature of lexical relations in text which are mainly used to

impose an ideology.15. This example is an extract from a translation of a message addressed by the late

Ayatollah Khomeini to the instructors and students of religious seminaries inIran.

16. It should be pointed out in this connection that the terms ‘denotation’ and ‘con-notation’ have acquired a high degree of diversity in treatment. Garza-Cuarónprovides a six-way distinction between the two terms:

The first is the distinction between denotation, understood as the direct ref-erence of a sign to an object, and connotation, understood as adjacent mean-ing which is added to primary meaning. The second is the relation betweensuppositio and significatio. The former here is the referential realisation of asign in a given proposition or the property of a term to stand for something

190 Notes

1403_913323_09_not.qxd 11/19/03 8:43 PM Page 190

(or the ‘actual’ representation of an object in a concrete sign). The latter is theassignation of a meaning to a form or the convention by means of which aform signifies a class of objects. The third point concerns the senses modernlogicians associate with the terms denotation and connotation. Denotation isused to refer to those subjects of which a term can be predicated, or simplythe object to which a term refers. Connotation, in turn, is the set of proper-ties that determine membership in a given class. This pair of terms are [sic]generally associated with other pairs such as extension–intension and com-prehension–extension. The fourth is the distinction between denotation, as areferential meaning, and connotation as an emotive meaning (the associationof images, experiences and values). The fifth is the opposition between thedenotation of a word, as its principal, primary, fixed or literal meaning, andits variable and ambiguous connotations, as secondary meanings addedfigurative or metaphoric senses and affective values. The sixth is the point inwhich Urban suggests that there are three different kinds of connotation:conceptual connotation, which is the abstraction that most words imply, intu-itive connotation, which is the conjuring up of meaning itself, the intuitionof the world or of objects through meaning, and emotional connotation.(Garza-Cuarón 1991:119–20)

17. Compare the example of ‘dog’ and ‘animal’ by Dickins (1998:301–2) whichaddresses a similar issue of the role of denotation and connotation in euphemisticexpressions. Cruse gives the example of ‘I’m going to take the animal for a walk’,which is used instead of ‘I’m going to take the dog for a walk’. Cruse then arguesthat this substitution is ‘no longer neutral – to produce a neutral utterance, thespeaker must use dog. The effect of avoiding the generic term in this way is oftento add negative emotive overtones to the utterance’ (1986:155).

18. This point will be further discussed in the next chapter when I discuss the notionof schematic knowledge and text organisation and interpretation. There I shallshow that part–whole movement seems to be logical and therefore necessary,whereas whole–part movement amounts to expectation by inferencing. Thismight or might not be satisfied by the progression of the text. And perhaps thisis what creates the dynamic interactive process of text interpretation.

5 Metonymy and Text Coherence

1. See the definition and discussion of this theory below. This term has been treatedas the general area of conceptual representation of knowledge in the mind. Seefor example the list of references in de Beaugrande and Dressler where they dis-cuss the elements of this theory (1981:90).

2. This hypothesis has been demonstrated in Chapter 2 of this book.3. Other theories also exist, for example frame theory, theory of higher-level knowl-

edge structures and script theory; see for details Schank and Abelson (1977), deBeaugrande and Dressler (1981) and Cook (1989).

4. This obviously contrasts with Rosch’s (1975) model of categorisation, which isessentially interactionist. There is interaction both ways, up and down. From thebasic level people generalise to reach superordinates and at the same time theyalso make discriminations to make subordinates.

5. This view was propounded by the Genevan psychologist Jean Piaget (1896–1980)in relation to cognitive development in children.

Notes 191

1403_913323_09_not.qxd 11/19/03 8:43 PM Page 191

6. See Langacker’s argument that metonymy is ‘basically a reference-point phenomenon’ (1993:30).

7. The two terms ‘words’ and ‘rules’ are generally based on a public lecture by StevenPinker in Newcastle, England. The terms also appear in his book Words and Rules(1999).

8. Lakoff (1987) argues in this regard that ‘stereotypes are used in certain situationsto define expectations, make judgments and draw inferences’ (p. 86).

9. De Beaugrande and Dressler’s notion of ‘inheritance’ is very closely related to thenotion of ‘prototypes’ discussed by Rosch (1975) and the notion of ‘best exam-ple’ discussed by Ungerer and Schmid (1996).

10. This is also called al-’ahad al-dhikrï; see for example Hasan (n.d. p.420).11. Compare the concept of al-’ahad al-dhihnï in Ibn Hishäm (n.d., p.50).12. Compare the concept of al-’ahad al-hudhürï in Ibn Hishäm (n.d., p.50).13. The concept of defamiliarisation is a central notion in formalist criticism pro-

pounded by the Russian school of linguistics and poetics. For details of the schooland the concept see Havranek (1964 [1932] ), Mukarovsky (1964 [1932]),Shklovssky (1917) and Fowler (1996).

14. See for a critique of Saussure’s theory of arbitrariness Keller (1998:130–40).15. See also the discussion of the notion of motivatedness of linguistic signification

that I provide in Chapter 3, section 3.2.1, above.16. This is shown by the use of double-headed arrows in the model in Figure 5.1.17. A column entitled fikrah ‘idea’ written by Mustafa Amïn when Rabin the Israeli

prime minister was assassinated.

192 Notes

1403_913323_09_not.qxd 11/19/03 8:43 PM Page 192

193

‘Abbäs, I. (1981) Ta’rïkh al-Naqd al-’Adabi ‘ind al-‘Arab. Beirut: Där al-Thaqäfah.‘Abd al-Muttalib, M. (1984) al-Baläghah wa al-‘uslübiyyah. Cairo: Al-Hay’ah al-Misriyyah

lil Kitäb.Abu Deeb, K. (1979) Al-Jurjäni’s Theory of Poetic Imagery. Warminister: Aris & Phillips.Abu Zayd, N.H. (1998) Mafhüm al-Nass: Diräsah fi ‘Ulüm al-Qur’än. Beirut: Al-Markaz

al-Thaqäfi al-’Arabi.Aitchison, J. (1995) Linguistics: An Introduction. London: Hodder & Stoughton.Al-‘Alawi, Y.B.H. (1914) Kitäb At-Tiräz al-Mutadhamin li Asrär al-Baläghah wa ‘Ulüm

Haqä’iq al’I‘jäz, 3 vols. Cairo: Där al-Kutub al-Khudaywiyyah.Al-Ali, M. (1999) Genre, Schema and the Academic Writing Process: An Inquiry into the

Generalisability of Generic Structure and Its Relationship to Schematic Knowledge.Unpublished PhD thesis, University of Durham.

Al-‘Askari, A.H. (1952) Kitäb al-Sinä‘tayn: al-Kitäbati wa al-Shi‘r, Eds ‘Ali al-Bajäwi andMuhammad Ibrahïm. Cairo: Al-Malabi.

Al-Jähiz, A.A. (1948) Al-Bayän wa al-Tabyïn, ed. ‘Abd al-Saläm Harün, 4 vols. Cairo:Matba‘at al-Ta’lïf wa al-Tarjamah.

Al-Järim, A. and M. Amïn (1951) Al-Baläghah al-Wädhihah. Cairo: Ministry of Education.Al-Jurjäni, A. (1946) Dalä’il al’I‘jäz f ï ‘Ilm al-Ma‘äni. Ridha, M. (ed.) Cairo: Al-Manar.Al-Jurjäni, A. (1954) Asrär al-Baläghah, ed. Ritter, H. Istanbul: Ministry of Education

Printing.Al-Mubarrad, M.Y. (1993) At-Kamel, ed. Al-Dali, M.A. Beirut: Risalah.Allwood, J. and P. Gärdenfors (1999) Cognitive Semantics: Meaning and Cognition.

Amsterdam: Benjamins.Al-Qayrawäni, I.R. (1907) Al-‘Umdah fi Sinä‘at al-Shi‘r wa Naqdih. Cairo: Maktabat

al-Khänji.Al-Qazwïni, al-K. (n.d.) al-’Ïdhäh fi ‘Ulüm al-Baläghah, al-Ma‘äni wa al-Bayän wa

al-Badï‘. Mukhtasar Talkhïs al-Miftäh. Beirut: Där al-Jïl.Al-Sakkäki, Y.M.A (1937) Miftäh al-‘Ulüm. Cairo: Matba‘at Mustafa al-Bäbi al-Halabi.Al-Sharafi, A. (1997) Argumentative Discourse in its Poetic Aspect: An Investigation into the

Poetic Aspects in the Phonological Structure of Argumentative Discourse. UnpublishedMA dissertation, University of Durham.

Al-Shäyib, A. (1996) Al-’Uslüb, 10th edn. Cairo: Maktabat al-Nahdhah al-Misriyyah.Al-Taftazäni, S.M. (1910) Talkhïs al-Miftäh. Cairo: Al-Maktabah al-’Amïriyyah, Buläq.Anderson, L. and P. Trudgill (1992) Bad Language. Harmondsworth: Penguin.Aristotle (1926) The Art of Rhetoric, Engl. transl. J. H. Freese. London: Heinemann.Aristotle (1927) The Poetics, Engl. transl. W.H. Fyee. London: Heinemann.Aristotle (1984) De Interpretatione, transl. J.L. Ackrill. The Complete Works of Aristotle:

The Revised Oxford Translation, ed. Jonathan Barnes, vol. 1. Princeton UniversityPress, 25–38.

‘Atïq, A. (1985) ‘Ilm al-Ma’äni. Beirut: Där al-Nahdhah al-’Arabiyyah.‘Atïq, A. (n.d.) ‘Ilm al-Bayän. Beirut: Där al-Nahdhah al-’Arabiyyah.Austin, J.L. (1962) How to Do Things with Words. Oxford University Press.Baker, M. (1992) In Other Words: A Coursebook on Translation. London: Routledge.Bakhtin, M.M. and P.N. Medvedev (1985) The Formal Method in Literary Scholarship,

trans. A.J. Wehrle. Cambridge: Harvard University Press.

Bibliography

1403_913323_10_bib.qxd 11/19/03 8:44 PM Page 193

Bansloben, E.P. (1996) Jakobson Revisited: The Use of Metaphor in the German Realist Novella.Unpublished PhD dissertation, University of Northern Carolina at Chapel Hill.

Barcelona, A. (2000) Metaphor and Metonymy at the Crossroads: A Cognitive Perspective.Berlin: Mouton de Gruyter.

Bartlett, F. (1932) Remembering: A Study in Experimental and Social Psychology. CambridgeUniversity Press.

Beeston, A.F.L. (1970) The Arabic Language Today. London: Hutchinson.Berlin, B. and P. Kay (1969) Basic Colour Terms: Their Universality and Evolution.

Berkeley: University of California Press.Bickerton, D. (1990) Language and Species. University of Chicago Press.Birch, D. (1993) ‘Drama Praxis and the Dialogic Imperative’. In Ghadessy.Blakemore, D. (1992) Understanding Utterances: An Introduction to Pragmatics. Oxford:

Blackwell.Bohas, G., J.-P. Guillaume and D.E. Kouloughli (1990) The Arabic Linguistic Tradition.

London: Routledge.Bohn, W. (1984) ‘Roman Jakobson’s Theory of Metaphor and Metonymy: An Annotated

Bibliography’, Style 18:4, 534–50.Bredin, H. (1984) ‘Metonymy’, Poetics Today 5:1, 45–58.Brown, P. and S. Levinson (1978) ‘Universals in Language Usage: Politeness

Phenomena’, in Goody, E. (ed.), Questions and Politeness: Strategies in Social Interaction.Cambridge University Press.

Brown, G. and G. Yule (1983) Discourse Analysis. Cambridge University Press.Burke, K. (1945) A Grammar of Motives. New York: Prentice Hall.Callow, K. (1974) Discourse Considerations in Translating the World of God, Michigan:

ZondervamCameron, D. (1995) Verbal Hygiene. London: Routledge.Cameron, L. and L. Graham (1999) Researching and Applying Metaphor. Cambridge

University Press.Carrel, P. (ed.) (1988) Interactive Approaches to Second Language Reading. Cambridge

University Press.Carter, D. (1986) Interpreting Anaphors in Natural Language Texts. Chichester: Ellis

Horwood.Carter, R. (1982) Language and Literature: An Introductory Reader in Stylistics. London:

Allen & Unwin.Charolles, M. (1983) ‘Coherence as a Principle in the Interpretation of Discourse’, Text

3:1, 71–97.Chierchia, G. and S. McConneil-Ginet (1990) Meaning and Grammar: An Introduction

to Semantics. Cambridge, MA: MIT Press.Chomsky, N. (1995) ‘Language and Nature’, Mind 104:413, 1–62.Cicero (1939) Brutus, Engl. trans. G.L. Hendrickson. Cambridge, MA: Harvard

University Press.Cicero (1939) Orator, Engl. trans. H.M. Hubbell. Cambridge, MA: Harvard University

Press.Cicero [The Auctor] (1954) Rhetorica ad Herennium, Engl. trans. H. Caplan. Cambridge,

MA: Harvard University Press.Cicero (1967) De Oratore, Engl. trans. E.W. Sutton, 3 vols. Cambridge, MA: Harvard

University Press.Collins English Dictionary and Thesaurus (1993) Glasgow: HarperCollins Publishers.Conely, T. (1990) Rhetoric in the European Tradition. Chicago: University of Chicago

Press.

194 Bibliography

1403_913323_10_bib.qxd 11/19/03 8:44 PM Page 194

Conte, M-E., J.S. Petfi and E. Szer (eds) (1989) Text and Discourse Connectedness: Proceedingsof the Conference on Connexity and Coherence, Urbino, 16–21 July, 1984. Amsterdam:Benjamins.

Cook, G. (1989) Discourse. Oxford University Press.Cook, G. (1990) A Theory of Discourse Deviation: The Application of Schema Theory to the

Analysis of Literary Discourse. PhD thesis, School of English, University of Leeds.Cook, V. and Newson, M. (1996) Chomsky’s Universal Grammar: An Introduction, 2nd

edn. London: Blackwell.Cooper, D. (1986) Metaphor. Oxford: Blackwell.Coseriu, E. (1967) Lexicalische Solidaritäten, Poetica 1,293–303.Coulthard, M. (1977) Introduction to Discourse Analysis. London: Longman.Croft, W. (1993) ‘The role of domains in the Interpretation of Metaphors and Meto-

nymies’, Cognitive Linguistics 4:4, 335–70.Cruse, D.A. (1986) Lexical Semantics. Cambridge University Press.Crystal, D. (1987) Cambridge Encyclopedia of Language. Cambridge University Press.Crystal, D. (1991) A Dictionary of Linguistics and Phonetics. London: Blackwell.Crystal, D. (1992) Dictionary of Language and Languages. London: Penguin.Crystal, D. and D. Davy (1969) Investigating English Style. London: Longman.Culler, J. (1975) Structuralist Poetics: Structuralism, Linguistics and the Study of Literature.

London: Routledge.Culler, J. (1986) Saussure. London: Fontana.Curme, G. (1964) English Grammar. New York: Barnes & Noble.Deacon, D., M. Pickering, P. Golding and G. Murdock (1999) Researching Communications:

A Practical Guide to Methods in Media and Cultural Analysis. London: Arnold.de Beaugrande, R. (1980) Text, Discourse and Process. London: Longman.de Beaugrande, R. and W. Dressler (1981) Introduction to Text Linguistics. London:

Longman.de Joia, A. and A. Stenton (1980) Terms in Systemic Linguistics: A Guide to Halliday.

London: Batsford Academic and Educational Ltd.Demetrius (1927) On Style, Engl. trans. W.R. Roberts. London: Heinemann Dhayf, S. (1965) Al-Baläghah Tatawwur wa Tärïkh. Beirut: Där al-Ma’ärif.Dickins, J. (1998) Extended Axiomatic Linguistics. Trends in Linguistics, Studies and

Monographs No. 111. Berlin: Mouton de Gruyter.Dickins, J. and J. Watson (1999) Standard Arabic: An Advanced Course. Cambridge

University Press.Dirven, R. (1999) ‘Conversion as a Conceptual Metonymy of Event Schemata’, in

Panther and Radden.Dirven, R, R. Frank and C. Ilie (eds) (2001) Language and Ideology, vol. II: Descriptive

Cognitive Approaches. Amsterdam: Benjamins.Dixon, P. (1971) Rhetoric. London: Methuen.Dressler, W. (ed.) (1977) Current Trends in Textlinguistics. Berlin: Walter de Gruyter.Dyer, M., R. Cullingford and S. Alvarado (1987) ‘Scripts’, in Schapiro, S-C. (ed.),

Encyclopedia of Artificial Intelligence. New York: Benjamins.Eagleton, T. (1983) Literary Theory: An Introduction. Oxford: Blackwell.Eco, U. (1976) A Theory of Semiotics. Bloomington: Indiana University Press.Eco, U. (1979) The Role of the Reader. Bloomington: Indiana University Press.Eco, U. (1984) Semiotics and the Philosophy of Language. London: Macmillan.Edwards, D. (1997) Discourse and Cognition. London: Sage.Ehlich, K. (1989) ‘Deictic Expressions and the Connexity of Text’, in Conte et al.Encyclopaedia of Islam (1960) Leiden: Brill.

Bibliography 195

1403_913323_10_bib.qxd 11/19/03 8:44 PM Page 195

Enkvist, N.E. (1973) Linguistic Stylistics. The Hague: Mouton.Enkvist, N.E. (1978) ‘Coherence, Pseudo-Coherence and Non-Coherence’, in

Östman, J.O. (ed.), Cohesion and Semantics. Abo: Abo Akademi Foundation: Finland.Enkvist, N.E. (1985) ‘Text and Discourse Linguistics, Rhetoric and Stylistics’, in van

Dijk, T.V. (ed.), Discourse and Literature. Amsterdam: Benjamins.Esquivel, L. (1992) Like Water for Chocolate. London: Black Swan.Fabb, N., D. Attridge, A. Durant and C. MacCabe (1987) The Linguistics of Writing:

Arguments Between Language and Literatrue. Manchester University Press.Fadhl, S. (1991) Shafarät al-Nass. Cairo: Där al-Fikr.Fadhl, S. (1992) Baläghat al-Khitäb wa ‘Ilm al-Nass. Kuwait: ‘Älam al-Ma‘rifah Series,

No. 164. National Council for Culture and Arts.Fairclough, N. (1989) Language and Power. London: Longman.Fass, D. (1997) Processing Metaphor and Metonymy. London: Ablex.Fauconnier, G. (1985) Mental Spaces: Aspects of Meaning Construction in Natural

Language. Cambridge, MA: MIT Press.Fauconnier, G. and M. Turner (1999) ‘Metonymy and Conceptual Integration’, in

Panther and Radden.Fauconnier, G. and M. Turner (2002) The Way We Think: Conceptual Blending and the

Mind’s Hidden Complexities. New York: Basic Books.Ferguson, C. (1964) ‘Diglossia’, in Hymes, D. (ed.) Language in Culture and Society: A

Reader in Linguistics and Anthropology. New York: Harper & Row, 429–39.Fitth, J. (1951) ‘Modes of Meaning’, in Papers in Linguistics 1934–51. London: Oxford

University Press.Firth, J. (1957) ‘A Synopsis of Linguistic Theory 1930–55’, in Firth.Firth, J.R. (ed.) (1957) Studies in Linguistic Analysis, Special volume of the Philological Society.

Oxford: Blackwell.Fiske, J. and J. Hartley (1978) Reading Television. London: Methuen.Fowler, R. (1996) Linguistic Criticism. Oxford University Press.Francis, G. (1986) Anaphoric Nouns. Discourse Analysis Monographs No. 11.

Department of English, University of Brimingham.Frege, G. (1966) ‘On Sense and Reference’, in Geach, P. and M. Black (eds and

trans), Translations from the Philosophical Writings of Gottlob Frege. Oxford: Blackwell,56–78.

Garza-Cuarón, B. (1991) Connotation and Meaning. Berlin: Mouton de Gruyter.Gelder, G.J.H. van (1982) Beyond the Line: Classical Arabic Literary Critics on the

Coherence and Unity of the Poem. Leiden: Brill.Ghadessy, M. (ed.) (1988) Registers of Written English: Situational Factors and Linguistic

Features. London: Printer.Ghadessy, M. (ed.) (1993) Register Analysis: Theory and Practice. London: Pinter.Gibbs, R. (1994) The Poetics of Mind: Figurative Thought, Language and Understanding.

Cambridge University Press.Gibbs, R. (1999) ‘Speaking and Thinking with Metonymy’, in Panther and Radden.Goatly, A. (1997) The Language of Metaphor. London: Routledge.Goffman, E. (1974) Frame Analysis. New York: Harper & Row.Goldberg, A.E. (1996) Conceptual Structure, Discourse and Language. Stanford,

California: CSLI.Goodman, N. (1968) Languages of Art. Indianapolis: Bobbs-Merril.Goossens, L. (1990) ‘Metaphtonymy: the Interaction of Metaphor and Metonymy in

Expressions for Linguistic Action’, Cognitive Linguistics 1:3, 323–40.Green, G. (1996) Pragmatics and Natural Language Understanding, 2nd edn. Hillsdale,

NJ: Erlbaum.

196 Bibliography

1403_913323_10_bib.qxd 11/19/03 8:44 PM Page 196

Greenlee, D. (1973) Peirce’s Concept of Sign. The Hague: Mouton.Greimas, A.J. (1966) Sémantique structurale: recherche de méthode. Paris: Larousse.Grice, P. (1975) ‘Logic and Conversation’, in P. Cole and J. Morgan (eds), Syntax and

Semantics, vol. 3: Speech Acts. New York: Academic Press.Grosz, B., D. Appelt, P. Martin, and F. Pereira (1987) TEAM: An experiment in the design

of transportable natural-language interfaces. Artificial Intelligence, 3:22, 173–243.Group mu (1970) Rhétorique générale. Paris: Larousse.Groupe mu (1981) A General Rhetoric, trans. P. Burrell and E. Slotkin. Baltimore, MD:

John Hopkins University Press.Haddärah, M. (1989) ‘Ilm al-Bayän. Beirut: Där al-’Ulüm al-’Arabiyyah.Hall, S. (1982) ‘The Rediscovering of “Ideology”: Return of the Repressed in Media

Studies’, in Gurevitch, M., J. Curren and J. Woollacott (eds), Culture, Society and theMedia. London: Routledge.

Hall, S. (1980) ‘Encoding/Decoding’, in Centre for Contemporary Cultural Studies (ed.),Culture, Media, Language: Working Papers in Cultural Studies. London: Hutchinson.

Halle, M. (1983) Roman Jakobson: What He Taught Us. Columbus, Ohio: Slavica Publishers.Halliday, M.A.K. (1957) ‘Some Aspects of Systemic Description and Comparison in

Grammatical Analysis’, in Firth (ed.).Halliday, M.A.K. (1978) Language as Social Semiotic: The Social Interpretation of Language

and Meaning. London: Arnold.Halliday, M.A.K. (1985) An Introduction to Functional Grammar. London: Arnold.Halliday, M.A.K. (1989) Spoken and Written Language, 2nd edn. Oxford University Press.Halliday, M.A.K. and R. Hasan, (1976) Cohesion in English. London: Longman.Halliday, M.A.K. and R. Hasan (1989) Language, Context and Text: Aspects of Language

in a Social-Semiotic Perspective, 2nd edn. Oxford University Press.Halliday, M.A.K. and C.M.I.M. Matthiessen (1999) Construing Experience Through

Meaning: A Language-based Approach to Cognition. Open Linguistics Series. London:Continuum.

Harris, R. and T.J. Taylor (1997) Landmarks in Linguistic Thought, 1: The WesternTradition from Socrates to Saussure. 2nd edn. London: Routledge.

Hartley, J. (1982) Understanding News. London: Methuen.Havranek, B. (1964 [1932]) ‘The Functional Differentiation of the Standard Language’

(in Czech); Engl. trans. in Garvin, P.L. (ed.), A Prague School Reader on Aesthetics,Literary Structure and Style. Washington, DC: Georgetown University Press.

Harweg, R. (1977) ‘Substitutional Textlinguistics’, in Dressler.Hary, B. (1993) Multiglossia in Jeudo-Arabic. Leiden: Brill.Hasan, A. (n.d.) Al-Nahw al-Wäf ï, vol. 1. Cairo: Där al-Ma’ärif.Hasan, R. (1984) ‘Coherence and Cohesive Harmony’, in Flood, J. (ed.), Understanding

Reading Comprehension. Newark, DE:IRA. International Reading Association.Hasan, R. (1989) ‘The Structure of a Text’, in Halliday, M.A.K. and R. Hasan.Hasan, R. (1989) Linguistics, Language and Verbal Art, 2nd edn. Oxford University Press.Hatch, E. (1991) Discourse and Language Education. Cambridge University Press.Hatim, B. and I. Mason (1997) The Translator as Communicator. London: Routledge.Hawkes, T. (1977) Structuralism and Semiotics. London: Methuen.Hawkins, J.A. (1978) Definiteness and Indefiniteness. London: Croom Helm.Heritage, J. and D. Watson (1979) ‘Formulations as Conversational Objects’, in

Psathas, G. (ed.), Everyday Language: Studies in Ethnomethodology. New York: Irvington.Hervey, S. (1982) Semiotic Perspectives. London: Allen & Unwin.Hjelmslev, L. (1961) Prolegomena to a Theory of Language. Madison: University of

Wisconsin Press.Hodge, R. and G. Kress (1988) Social Semiotics. Cambridge University Press.

Bibliography 197

1403_913323_10_bib.qxd 11/19/03 8:44 PM Page 197

Hoey, M. (1986) ‘The Discourse Colony: A Preliminary Study of a Neglected DiscourseType’, in Coulthard, M. (ed.), Talking About Text: Studies Presented to David Brazil onhis Retirement, Discourse Analysis Monographs No. 13. English Language Research,University of Birmingham.

Hoey, M. (1991) Patterns of Lexis in Text. Oxford University Press.Holes, C. (1995) Modern Arabic: Structures, Functions and Varieties. London: Longman.Hope, T.E., T.B. Reid, R. Harris, and G. Price (1981) Language Meaning and Style: Essays

in Memory of Stephen Ullmann. Leeds University Press.Householder, F. (1995) ‘Plato and His Predecessors’, in Koerner and Asher.Hutchings, S. (1990) A Semiotic Analysis of the Short Stories of Leonid Andreev 1900–1909.

London: Modern Humanities Research Association.Ibn al-Athïr (1983) Al-Mathal al-Sä’ir f i ‘Adab al-Kätib wa al-Shä’ir, 3 vols. (eds) al-

Hawfi, A. and Tabbänah, B. Riyadh: Där al-Rifä’i.Ibn Hishäm, A. (n.d.) Mughani Al-Labïb ‘an kutub al’a’ärïb, vol. 1. Edited by Muhammad

Abd al-Hamïd. Cairo.Isocrates (1930) Isocrates, Engl. trans. George Norlin, 3 vols. London: Heinemann.Jackendoff, R. (1993) Patterns in the Mind: Language and Human Nature. London: Harvester

Wheatsheaf.Jakobson, R. (1960) ‘Closing Statement: Linguistics and Poetics’, in Lodge.Jakobson, R. (1968) ‘Poetry of Grammar and Grammar of Poetry’, Selected Writings

III, 87–97. The Hague: Mouton.Jakobson, R. (1971) ‘Two Aspects of Language and Two Types of Aphasic Disturbances’,

in Jakobson, R. and M. Halle (eds) Fundamentals of Language. The Hague: Mouton.Jakobson, R. and L. Waugh (1979) The Sound Shape of Language. Bloomington: Indiana

University Press.Johns, A.M. (1997) Text, Role, and Context. Cambridge University Press.Johnstone, B. (1991) Repetition in Arabic Discourse. Amsterdam: Benjamins.Jordan, M. (1984) Rhetoric of Everyday English Texts. London: George Allen & Unwin.Kanazi, G. (1989) Studies in the Kitäb As-Sinä’atayn of Abu Hilal al-’Askari. Leiden: Brill.Keller, R. (1998) A Theory of Linguistic Signs. Oxford University Press.Kennedy, G. (1963) The Art of Persuasion in Greece. Princeton University Press.Kennedy, G. (1969) Quintilian. New York: Twayne.Kennedy, G. (1972) The Art of Rhetoric in the Roman World. Princeton University Press.Kennedy, G. (1980) Classical Rhetoric and its Christian and Secular Tradition from Ancient

to Modern Times. London: Croom Helm.Kennedy, G. (1983) Greek Rhetoric Under Christian Emperors. Princeton University Press.Kennedy, G. (1984) New Testament Interpretation through Rhetorical Criticism. Chapel

Hill: University of Northern Carolina Press.Kennedy, G. (1994) A New History of Classical Rhetoric. Princeton University Press.Khattäbi, M. (1991) Lisäniyyät al-Nass: Madkhal ‘ila ‘Insijäm al-Khitäb. Beirut: Al-

Markaz al-Thaqäfi al-’Arabi.Kintsch, W. (1977) ‘On Comprehending Stories’, in Just, M. and P. Carpenter (eds),

Cognitive Processes in Comprehension. Hillsdale, NJ: Erlbaum.Kittay, E.F. (1987) Metaphor: Its Cognitive Force and Linguistic Structure. Oxford:

Clarendon Press.Koch, P. (1999) ‘Frame and Contiguity: On the Cognitive Bases of Metonymy and

Certain Types of Word Formation’, in Panther and Radden.Koerner, E.F.K. and R.E. Asher (1995) Concise History of Language Sciences from the

Sumerians to the Cognitivists. London: Pergamon.Koffka, K. (1935) Principles of Gestalt Psychology. London: Routledge & Kegan Paul.

198 Bibliography

1403_913323_10_bib.qxd 11/19/03 8:44 PM Page 198

Kress, G. (1985) Linguistic Processes in Sociocultural Practice. Victoria: Deakin UniversityPress.

Kress, G. and Hodge, B. (1993) Language as Ideology. London: Routledge.Lakoff, G. and M. Johnson (1980) Metaphors We Live By. University of Chicago Press.Lakoff, G. (1987) Women, Fire and Dangerous Things: What Categories Reveal about the

Mind. University of Chicago Press.Lakoff, G. and M. Johnson (1999) Philosophy in the Flesh: The Embodied Mind and Its

Challenge to Western Thought. New York: Basic Books.Lakoff, G. and M. Turner (1989) More than Cool Reason: A Field Guide to Poetic Metaphor.

University of Chicago Press.Langacker, R. (1987) Foundations of Cognitive Grammar, vol. 1: Theoretical Prerequisites.

Stanford University Press.Langacker, R. (1993) ‘Reference-Points Constructions’, Cognitive Linguistics 4:1, 1–38.Leech, G. (1969) A Linguistic Guide to English Poetry. London: Longman.Leech, G. and M.H. Short (1981) Style in Fiction: A Linguistic Introduction to English

Fictional Prose. London: Longman.Levin, S.R. (1977) The Semantics of Metaphor. London: Johns Hopkins University Press.Levinson, S. (1983) Pragmatics. Cambridge University Press.Lewis, J. (1991) The Ideological Octopus: An Exploration of Television and its Audience.

New York: Routledge.Lodge, D. (1977) The Modes of Modern Writing. London: Arnold.Lodge, D. (1988) Modern Criticism and Theory. London: Longman.Longinus (1927) On the Sublime, Engl. trans. W.H. Fyee. Harvard University Press.Lönngren, L. (1989) ‘How Can the Meaning of a Text be Represented?’, in

Conte et al.Lucy, J. (1991) Language Diversity and Thought: A Reformulation of the Linguistic

Relativity Hypothesis. Cambridge University Press.Lyons, J. (1968) Introduction to Theoretical Linguistics. Cambridge University Press.Lyons, J. (1970) New Horizons in Linguistics. London: Penguin.Lyons, J. (1977) Semantics, vols I and II. Cambridge University Press.Lyons, J. (1981) Language, Meaning and Context. London: Fontana.Lyons, J. (1991) Natural Language and Universal Grammar: Essays in Linguistic Theory,

vol. 1. Cambridge University Press.Lyons, J. (1995) Linguistic Semantics: An Introduction. Cambridge University Press.McCarthy, M.J. (1998) Spoken Language and Applied Linguistics. Cambridge University

Press.McCullers, C. (1961) The Heart is a Lonely Hunter. Harmandsworth: Penguin.Maida, A.S. (1987) ‘Frame Theory’, in Stuart, C.S. (ed.), Encyclopedia of Artificial

Intelligence. New York: Wiley, 302–12.Malinowski, B. (1923) ‘The Problem of Meaning in Primitive Languages’, in Ogden, C.K.

and I.A. Richards (eds), The Meaning of Meaning. London: Kegan Paul.Malotki, E. (1983) Hopi Time: A Linguistic Analysis of the Temporal Concepts in the Hopi

Language. Berlin: Mouton.Mandler, G. (1985) Cognitive Psychology: An Essay in Cognitive Science. Hillsdale, NJ:

Lawrence Erlbaum.Matejka, L. and K. Pomorska (1971) Readings in Russian Poetics: Formalist and Structuralist

Views. Cambridge: Massachusetts Institute of Technology.Matlüb, A. (1978) Diräsät Baläghiyyah wa Naqdiyyah. Baghdad: Där al-Rashïd.Miftäh, M. (1985) Tahlïl al-Khitäb al-Shi‘rï: ’Isträtïjiyät al-Tanäs. Casablanca: Al-Markaz

al-Thaqäfi al-‘Arabi.

Bibliography 199

1403_913323_10_bib.qxd 11/19/03 8:44 PM Page 199

Minsky, M. (1975) ‘A Framework for Representing Knowledge’, in Winston, P.H. (ed.),The Psychology of Computer Vision. London: McGraw-Hill Book Company.

Morris, C. (1938) ‘Foundations of the Theory of Signs’, International Encyclopedia ofScience, vol. 1, no. 2, University of Chicago Press.

Mukarovsky, J. (1964 [1932]) ‘Standard Language and Poetic Language’, Engl. trans.in Garvin, P.L. (ed.), A Prague School Reader on Aesthetics, Literary Structure and Style.Georgetown University Press.

Mulder, J.W.F. and S.G. Hervey (1972) Theory of the Linguistic Sign. The Hague:Mouton.

Murphy, J.J. (1974) Rhetoric in the Middle Ages. Berkeley: University of California Press.Murphy, J.J. (1978) Medieval Eloquence: Studies in the Theory and Practice of Medieval

Rhetoric. University of California Press.Nalbantian, S. (1977) The Symbol of the Soul from Holderlin to Yeats: A Study in

Metonymy. London: Macmillan.Nerlich, B., D. Clarke and Z. Todd (1999) ‘Mummy, I Like Being a Sandwich:

Metonymy in Language Acquisition’, in Panther and Radden.Nöth, W. (1977) ‘The Semiotic Framework of Textlinguistics’, in Dressler.Nunan, D. (1993) Introducing Discourse Analysis. London: Penguin.Nunberg, G. (1977) The Pragmatics of Reference. Bloomington, IN: Indiana University

Linguistics Club.Nunberg, G. (1979) ‘The Non-Uniqueness of Semantic Solutions: Polysemy’,

Linguistics and Philosophy 3, 143–84.Nunberg, G. (1993) ‘Indexicality and Deixis’, Linguistics and Philosophy 16, 1–43.Nunberg, G. (1995) ‘Transfers of meaning’. Journal of Semantics 12:2, 109–32.Ohman, R. (1971) ‘Speech Acts Theory and the Definition of Literature’, Philosophy

and Rhetoric 10, 1–19.Osterwalder, H. (1978) T.S. Eliot: Between Metaphor and Metonymy – A Study of His Essays

and Plays in Terms of Jakobson’s Typology. Bern: Francke.Pankhurst, A. (1997) ‘Interpreting Unknown Worlds: Functions of Metonymic

Conceptualisation in William Golding’s The Sea Trilogy’, Language and Literature 6:2,121–31.

Pankhurst, A. (1998) Metonymy: Semantic, Pragmatic, Cognitive and Stylistic Perspectives.Unpublished PhD thesis, University of Edinburgh.

Pankhurst, A. (1999) ‘Recontextualisation of Metonymy in Narrative and the Case ofMorrison’s Song of Solomon’, in Panther and Radden.

Panther, K-U. and G. Radden (eds) (1999) Metonymy in Language and Thought.Amsterdam: Benjamins.

Panther, K-U. and L. Thornburg (1999) ‘The Potentiality for Actuality Metonymy inEnglish and Hungarian’, in Panther and Radden.

Peirce, C.S. (1955) ‘Logic as Semiotic: The Theory of Signs’, in Buchler, J. (ed.),Philosophical Writings of Peirce. New York: Dover.

Peirce, C.S. (1960) Collected Papers of Charles Sanders Peirce, vols 1–8, eds Hartshorne,C. and P. Weiss. Cambridge, MA: Harvard University Press.

Perrine, L. (1992) Figures of speech. In the academic American encyclopedia.Danbury, CN: Grolier Electronic Publishing, Inc.

Petöfi, J. (1977) ‘A Formal Semiotic Text Theory as an Integrated Theory of NaturalLanguage’, in Dressler.

Phillips, M. (1985) Aspects of Text Structure: An Investigation of the Lexical Organisationof Text. Amsterdam: North-Holland.

Pinker, S. (1999) Words and Rules: The Ingredients of Language. London: Weidenfeld &Nicolson.

200 Bibliography

1403_913323_10_bib.qxd 11/19/03 8:44 PM Page 200

Plett, H. (1985) ‘Rhetoric’, in van Dijk, T.V., Amsterdam:Pomorska, K. (1992) Jakobsonian Poetics and Slavic Narrative. Duke University Press.Pomorska, K. and S. Rudy (eds) (1985) Roman Jakobson: Verbal Art, Verbal Sign, Verbal

Time. Oxford: Blackwell.Pratt, Mary. L. (1977) Toward A Speech Act Theory of Literary Discourse. Bloomington:

Indiana University Press.Pullum, G.K. (1991) ‘The Great Eskimo Vocabulary Hoax’, in Pulluman, G. The Great

Eskimo Vocabulary and Other Irreverent Essays on the Study of Language. University ofChicago Press.

Pustejovsky, J. (1991) ‘The Generative Lexicon’, Computational Linguistics 17:4, 409–41.Pustejovsky, J. and P. Bouillon (1995) ‘Aspectual Coercion and Logical Polysemy’,

Journal of Semantics 12:2, 133–62.Qudämah ibn Ja‘far (1958) Naqd al-Shi‘r, ed. ‘Ïsa Mikhä’ïl Saba. Beirut:Quillian, M.R. (1969) The teachable language emprehender: A simulation program

and theory of language. Communications of the ACM, 12(8), 459–475.Quintilian (1979) The Institutio Oratoria of Quintilian, Engl. trans. H.E. Butler, 4 vols.

Cambridge, MA: Harvard University Press.Quirk, R. and S. Greenbaum (1973) A University Grammar of English. London: Longman.Race, W.H. (1990) Style and Rhetoric in Pindar’s Odes. London: Scholars Press.Radden, G. and Z. Kövecses (1999) ‘Towards a Theory of Metonymy’, in Panther and

Radden.Reddy, M. (1979) ‘The Conduit Metaphor’, in Ortony, A. (ed), Metaphor and Thought.

Cambridge University Press.Reisbeck, C.K. and R.C. Schank (1978) ‘Comprehension by Computer: Expectation-

Based Analysis of Sentences in Context’, in Levelt, W.J.M. and G.B. Flores d’Arcais(eds), Studies in the Perception of Language. New York: Wiley.

Reiser, H. (1977) ‘On the Development of Text Grammar’, in Dressler.Reuland, E. and W. Abraham (eds) (1993) Knowledge and Language. Kluwer.Richards, I.A. (1936) The Philosophy of Rhetoric. Oxford University Press.Richardson, K. (1998) Models of Cognitive Development. East Essex Hove: Psychology Press.Ricoeur, P. (1976) Interpretation Theory: Discourse and the Surplus of Meaning. Fort

Worth, Texas: Texas Christian University Press.Ricoeur, P. (1978) The Rule of Metaphor, trans. R. Czerny with Kathleen Mclaughlin and

John Costello et al. London: Routledge & Kegan Paul.Ringbom, H. (ed.) (1975) Style and Text. Studies Presented to Nils Enkvist. Stockholm:

Sprakforlaget Skriptor AB and Abo Akademi in collaboration.Robinson, N. (1996) Discovering the Qur’än: A Contemporary Approach to a Veiled Text.

London: SCM Press.Rosch, E. (1973) ‘Natural Categories’, Cognitive Psychology 4, 328–50.Rosch, E. (1975) ‘Cognitive Reference Points’, Cognitive Psychology 7, 532–47.Rosch, E. (1977) ‘Human Categorisation’, in Warren N. (ed.), Studies in Cross-Cultural

Psychology. London: Academic Press.Rosch, E. (1978) ‘Principles of Categorisation’, in Rosch, E and Lloyd, B.B. Cognition

and Categorisation. Hillsdale, NJ: Erlbaum.Rosch, E. (1988) ‘Coherence and Categorisation: A Historical View’, in Kessel, F.S. (ed.),

The Development of Language and Language Researchers. Essays in Honour of RogerBrown. Hillsdale, NJ: Lawrence Erlbaum Associates.

Rosiene, A.M. (1992) Classical and Medieval Latin Metonymy in Relation to ContemporaryFigurative Theory. Unpublished PhD thesis. Northwestern University.

Rouchota, V. and A.H. Jucker (1998) Current Issues in Relevance Theory. Amsterdam:Benjamins.

Bibliography 201

1403_913323_10_bib.qxd 11/19/03 8:44 PM Page 201

Rumelhart, D. (1975) ‘Notes on a Schema for Stories’, in Bobrow, D. and A. Collins (eds),Representation and Understanding: Studies in Cognitive Science. New York: AcademicPress.

Rumelhart, D. (1977) Introduction to Human Information Processing. New York: Wiley.Rumelhart, D. (1977) ‘Understanding and Summarising Brief Stories’, in La Berge, D.

and J. Samuels (eds), Basic Processes in Reading: Perception and Comprehension. Hillsdale,NJ: Erlbaum.

Sampson, G. (1980) Schools of Linguistics: Competition and Evolution. London:Hutchinson University Library.

Sanford, A.J. and S.C. Garrod (1981) Understanding Written Language. Chichester: Wiley.Sapir, E. (1949) Language: An Introduction to the Study of Speech. New York: Harcourt Brace.Saussure, F. de (1983) Course in General Linguistics, trans. and annotated by Roy Harris.

London: Duckworth.Schank, R.C. and R.P. Abelson (1977) Scripts, Plans, Goals and Understanding: An inquiry

into Human Knowledge Structures. Hillsdale, NJ: Erlbaum.Schiffrin, D. (1994) Approaches to Discourse. London: Blackwell.Searle, J. (1975) ‘Indirect Speech Acts’, in Cole, P. and J. Morgan (eds), Syntax and

Semantics, vol. 3: Speech Acts. New York: Academic Press.Searle, J. (1979) ‘Metaphor’, in Ortony, A. (ed.), Metaphor and Thought. Cambridge

University Press.Searle, J., F. Kiefer, and M. Bierwisch (eds) (1980) Speech Act Theory and Pragmatics.

Dordrecht; London: Reidel.Sebeok, T. (1991) A Sign is Just a Sign. Bloomington: Indiana University Press.Shaykhün, M.A. (1978) Al-’Uslüb al-Kinä’ï: Nash’atuh – Tatawwuruh – Baläghatuh.

Cairo: Maktabat al-Kulliyät al-Azhariyyah.Shklovsky, V. (1985 [1917]) ‘Art as Technique’, trans. in Leman, L. and M.J. Reis (eds),

Russian Formalist Criticism. Lincoln: University of Nebraska Press.Sinclair, J.M. and R.M. Coulthard (1975) Towards an Analysis of Discourse. Oxford

University Press.Soskice, J.M. (1985) Metaphor and Religious Language. Oxford: Clarendon Press.Sowa, J. (1984) Conceptual Structures: Information Processing in Mind and Machine.

Reading, MA: Addison-Wesley.Sperber, D. and D. Wilson (1986) Relevance: Communication and Cognition. Oxford:

Blackwell.Stallard, D. (1993) ‘Two kinds of Metonymy’, in Proceedings of the Annual Meeting of

the Association for Computational Linguistics (ACL ’93). Ohio State University,Columbus, OH, 87–94.

Stam, R., R. Burgoyne, S. Flitterman-Lewis (1992) New Vocabularies in Film Semiotics:Structuralism, Post-Structuralism and Beyond. London: Routledge.

Stanford, W.B. (1936) Greek Metaphor: Studies in Theory and Practice. Oxford: Blackwell.Steiner, G. (1992) After Babel, 2nd edition. Oxford University Press.Stirling, L. (1996) ‘Metonymy and Anaphora’, Belgian Journal of Linguistics 10, 69–88.Süfiyyah, M.M. (1984) Al-Mabähith Al-Bayäniyyah bayna Ibn Al-Athïr wa Al-‘Alawi.

Tripoli, Libya: Al-Munsha’ah al-‘Ämmah lin Nashr wa al-Tawzï‘ wa al-’I‘län.Swales, J. (1990) Genre Analysis: English in Academic and Research Settings. Cambridge

University Press.Tabbänah, B. (1981) Abu Hiläl al-‘Askari wa Maqäyïsuhu al-Baläghiyyah wa al-

Naqdiyyah. Beirut: Där al-Thaqäfah.Tadros, A. (1985) Prediction in Text. Discourse Analysis Monographs. Department of

English, University of Birmingham.

202 Bibliography

1403_913323_10_bib.qxd 11/19/03 8:44 PM Page 202

Talmon, R. (1986) ‘Who was the First Arab Grammarian? A New Approach to an OldProblem’, Journal of Arabic Linguistics 15, 128–45.

Tannen, D. (1993) Framing in Discourse. Oxford University Press.Thwaites, T., L. Davis and W. Mules (1994) Tools for Cultural Studies. An Introduction.

South Melbourne: Macmillan.Toolan, M. (ed.) (1992) Language, Text, and Context: Essays in Stylistics. London:

Routledge.Turner, G. (1992) British Cultural Studies: An Introduction. New York: Routledge.Turner, M. and G. Fauconnier (2000) ‘Metaphor, Metonymy, and Binding’, in Barcelona.Ullmann, S. (1951) The Principles of Semantics. Glasgow: Jackson.Ullmann, S. (1962) Semantics: An Introduction to the Science of Meaning. Oxford: Blackwell.Ungerer, F. and H-J. Schmid (1996) An Introduction to Cognitive Linguistics. London:

Longman.van Dijk, T.A. (1977) Text and Context. London: Longman.van Dijk. T.A. (ed.) (1985) Discourse and Literature. Amsterdam: Benjamins.van Dijk, T.A. and W. Kintsch (1977) ‘Cognitive Psychology and Discourse: Recalling

and Summarising Stories’, in Dressler.van Dijk, T.A. and W. Kintsch (1983) Strategies of Discourse Comprehension. London:

Academic Press.Ventola, E. (1987) The Structure of Social Action. A Systematic Approach to the Semiotics

of Service Encounters. London: Frances Pinter.Ventola, E. (ed.) (1991) Approaches to the Analysis of Literary Discourse. Abo: Abo

Akademis Forlag–Abo Academy Press.Versteegh, K. (1984) Pidginisation and Creolisation: The Case of Arabic. Amsterdam:

Benjamins.Versteegh, K. (1997) Landmarks in Linguistic Thought, III: The Arabic Linguistic Tradition.

London: Routledge.Vickers, B. (1988) In Defense of Rhetoric. Oxford: Clarendon Press.Wahbah, M. and K. al-Muhandis (1979) Mu‘jam al-Mustalahät al-‘Arabiyyah f i al-

Lughah wa al-’Adab. Beirut: Maktabat Lubnän.Wahbah, M. (1974) A Dictionary of Literary Terms. English – French – Arabic. Beirut:

Librairie du Liban.Wakao, T. and S. Helmreich (1993) ‘Translation of Metonymy in an Interlingual MT

System’, in Proceedings of the First Conference of the Pacific Association for ComputationalLinguistics (PACLING’93), Vancouver, BC, Canada, 288–96.

Wales, K. (1989) A Dictionary of Stylistics. London: Longman.Waltereit, R. (1999) ‘Grammatical Constraints on Metonymy’, in Panther and Radden.Warsen, B. (1999) ‘Aspects of Referential Metonymy’, in Panther and Radden (eds.)Waugh, L. (1976) Roman Jakobson’s Science of Language. Lisse: De Ridder.Wells, R. (1977) ‘Criteria for Semiosis’, in Sebeok, T. (ed.), A Perfusion of Signs.

Bloomington: Indiana University Press.Wertheimer, M. (1958) ‘Principles of Perceptual Organization’, in Beardslee, D.C. and

M. Wertheimer (eds), Readings in Perception. Princeton: NJ: Van Nostrand, 115–35.Whitelock, P. and K. Kelby (1995) Linguistic and Computational Techniques in Machine

Translation System Designs. London: University of California Press.Whorf, B.L. (1956) Language, Thought and Reality: Selected Writing of Benjamin Lee

Whorf. Cambridge, MA: MIT Press.Widdowson, H.G. (1978) Teaching Language as Communication. Oxford University Press.Wierzbicka, A. (1991) Cross-Cultural Pragmatics: The Semantics of Human Interaction.

Trends in Linguistics, Studies and Monograph No. 53. Berlin: Mouton de Greuyter.

Bibliography 203

1403_913323_10_bib.qxd 11/19/03 8:44 PM Page 203

Winston, P. (1970) Learning Structural Descriptions from Examples. PhD thesis,Massachusetts Institute of Technology.

Winston, P. (1975) The Psychology of Computer Vision. London: McGraw-Hill.Winter, E. (1986) ‘Clause Relations as Information Structure: Two Basic Text Structures

in English’, in: Coulthard, M. (ed.), Talking about Text: Studies Presented to DavidBrazil on His Retirement. Discourse Analysis Monograph No. 3. Birmingham EnglishLanguage Research, University of Birmingham.

Wittgenstein, L. (1968) Philosophical Investigations, trans. G.E.M. Anscombe, 3rd edn.Oxford: Blackwell.

Yamanashi, M.-A. (1987) ‘Metonymic Interpretation and Associative Processes inNatural Language’, in Nagao, M. (ed.), Language and Artificial Intelligence: Proceedingsof an International Symposium on Language and Artificial Intelligence held in Kyoto,Japan, 16–21 March 1986. Amsterdam: Elsevier, 77–86.

Ziff, P. (1984) Epistemic Analysis: A Coherence Theory of Knowledge. Dordrecht: Reidel.

204 Bibliography

1403_913323_10_bib.qxd 11/19/03 8:44 PM Page 204

Abelson, R.P. 4, 7, 8, 137, 139,142, 160–1, 163–6, 176–7, 181,189(n27), 191(n3)

script theory 141, 160Abrams, M.H. 49Abu Bakr 151Abü Hiläl al-‘Askari 24–5, 29Abü ‘Ubaydah 22–3Achard, G. 16Aitchison, J. 80, 186(n4)al-‘Alawi, Y.B.H. 28, 31, 33al-‘Askari, A.H. 23, 25al-Buhturi 24al-Jähiz, A.A. 23, 25al-Jurjäni, A. 23, 26, 28, 29–30,

32–3, 102, 188(n24)al-Mäliki (rhetorician) 28al-Mubarrad 23, 184(n10)Al-Mutanabbï (‘best Arab poet’)

125al-Qayrawäni, I.R. 25–6al-Qazwïni, al-K. 27, 31, 33,

185(n16)al-Sakkäki, Y.M.A. 23, 27, 29, 30–1Allah 30, 33, 125, 130, 150, 171,

174Ampère, A-M. 47, 62Anderson, L. 89Antonapoulos, S. 156–60Aristotle 12–14, 41, 42, 48, 96–7,

99, 101‘four types of metaphor’ 13representational theory of signs

96‘Atïq, A. 124–5Auctor see Cicero [The Auctor]Austin, J.L. 184(n2)

Baker, M. 116, 120Bansloben, E.P. 185(n2)Barthes, R. 188(n21)

Bartlett, F. 137, 139, 145–6,189(n27)

schema theory 141Bede, Venerable 14, 15, 19, 49Beeston, A.F.L. 127Berlin, B. 97Bickerton, D. 92Birch, D. 94Bishr ibn al-Mu‘tamir 25Blakemore, D. 189(n2)Boas, F. 97Bredin, H. 35, 41, 60

linguistic theory of metonymy48–53

Brown, G. 189(n3–4)Brown, P. 184(n2)Burke, K. 35, 185(n2), 194

Caesar, J. 43Callow, K. 116Cameron, D. 94Caplan, H. 16Carroll, L. 80Cassiodorus 15, 18–19Charisius 15, 18, 19, 102Cicero (The Auctor) 15, 16, 17, 19,

40, 48, 185(n4), 194Cicero 15, 16–17, 18, 19, 48, 49,

185(n4)Cleopatra 78Cook, G. 85, 139, 141, 146, 153,

191(n3)Cooper, D. 186(n15)Cruse, D.A. 129–30, 134–5,

190(n12), 191(n17)Crystal, D. 5–6, 55, 86, 130Curme, G. 61

de Beaugrande, R. 115, 139,142–4, 146, 147, 161, 177,189(n27–8), 191(n1), 191(n3)

Author Index

Key: f�figure; n�note; t�table; bold�extended discussion or headingemphasized in main text.

——————

1403_913323_11_auin.qxd 11/19/03 9:24 PM Page 205

Deacon, D. et al. 84, 92, 195Golding, P. 195Murdock, G. 195Pickering, M. 195

Demetrius 14Democritus 12Dickins, J. x, 83, 190(n10),

191(n17)Diomedes 15, 18, 19, 102Dionysus 13Dirven, R. 136Dixon, P. 184(n1)Donatus 15, 18Doyle, F.C. 49Dressler, W. 115, 139, 142–4,

146–7, 161, 177, 189(n27–8),191(n1), 191(n3)

Du Marsais 42, 52Dubois, J. 40Durkheim, É. 89Dyer, M. et al. 160–1, 195

Alvarado, S. 195Cullingford, R. 195

Eagleton, T. 187(n13)Eberhard 15Eco, U. 14, 83–5, 90, 186(n3),

188(n17)Edeline, F. 40Edwards, D. 161, 184(n3), 189(n1)Enkvist, N.E. 6–7Esnault, G. 39, 43, 45Esquivel, L. 155

Fairclough, N. 93, 94Fass, D. 74

cognitive theory of metonymy58–64

Fauconnier, G. x, 59, 78, 154,186(n12), 188(n14)


Ferguson, C. 186(n11)Fillmore, C. 57Fiske, J. 186(n2)Fowler, R. 192(n13)Frege, G. 49–50, 97–101, 188(n20)Fraunce, A. 49

Garza-Cuarón, B. 190–1(n16)Geoffrey of Vinsauf 15, 19

Gervase of Melkley 15, 19Ghadessy, M. 94Gibbs, R. 5, 131, 137, 177


‘functional theory’ 68–9,186(n18)

metonymy as cohesive device64

Gray, T. 32Grice, P. 63, 74, 85, 184(n2)Grosz, B. 62

Halliday, M.A.K. 4, 7, 8, 110–12,115–20, 122–4, 126–7, 137,187(n8), 189(n3–4)

Hartley, J. 186(n2)Harweg, R. 190(n5)Hary, B. 186(n11)Hasan, A. 149, 151–2, 192(n10)Hasan, R. 4, 7, 8, 110–12, 115,

117–20, 122–4, 126–7, 137,189(n4)

Hatim, B. 94, 129Havranek, B. 192(n13)Hawkes, T. 83, 92, 187(n5),

188(n16)Henry, A. 50–1Herodotus 11Hervey, S. 188(n18, n21)Hjelmslev, L. 90, 188(n21)Hodge, B. 94Hoey, M. 133Homer 78Householder, F. 12

Ibn al-Athïr 23, 27–8, 30Ibn al-Mu’tazz 23Ibn Hishäm, A. 149–50, 152,

192(n11–12)’Imru’u al-Qays 25, 26Isidore Junior 15, 19Isidore of Seville 15, 19, 102Isocrates 11–12

Jakobson, R. 10, 13, 46, 48, 49, 60,64, 104, 109, 184(n5–6)

linguistic theory of metonymy36–40, 185(n1–2)

metonymy ‘defies interpretation’60

206 Author Index

1403_913323_11_auin.qxd 11/19/03 9:24 PM Page 206

Jakobson – continuedmodern figurative and stylistic

theory 11polarity of metaphor and

metonymy 39problem 39

Jesus Christ 36John of Garland 15, 19John/hungry, 163–4, 169, 175Johnson, M. 51, 53–5, 57, 58, 62,

64Julian of Toledo 15, 19

Kanazi, G. 24Kant, I. 138Kay, P. 97Keller, R. 96, 186(n1), 192(n14)Kennedy, G. 184(n1)Khafäji 24Khomeini, Ayatollah 129–30,

190(n15)Kintsch, W. 139Klein, G.S. 185(n9)Klinkenberg, J.-M. 40Koch, P. 14Koffka, K. 138Kövecses, Z. 3, 38, 106

cognitive theory of metonymy69–72, 113

Kress, G. 94Krumbacher, K. 186(n11)

Lakoff, G. 51, 62, 64–5, 69, 97,106, 116, 137, 141–2, 192(n8)

linguistic theory of metonymy53–8

Lamb, M. 83Langacker, R. 57, 141, 184(n4),

190(n7), 192(n6)cognitive explanation of

metonymy 70Leech, G. 41Levinson, S. 184(n2)Lewis, J. 89Lodge, D. 13, 48, 185(n1)Longinus 14Lyons, J. 129, 130–1, 187(n9),

188(n19)

McCullers, C. 156Maida, A.S. 142

Malinowski, B. 184(n2)Mason, I. 94, 129Matthew of Vendôme 15, 19Matthiessen, C.M.I.M. 111,

115–16, 137, 189(n3)Miftäh, M. 184(n7)Minguet, P. 40Minsky, M. 4, 7, 8, 137, 142, 153,

175–6, 189(n27)frame theory 141, 146–7

Mohammed, Prophet 151Morris, C. 82, 86Moses 121Mukarovsky, J. 192(n13)Mustafa Amïn 192(n17)

Naguib Mahfouz 181Nerlich, B. xNerlich, B. et al. 39, 200

Clarke, D. 200Todd, Z. 200

Nixon, R.M. 74, 186(n19)Nunberg, G. 56, 59, 62, 75–7,

102–3, 112, 188–9(n25)

Ohm, G. 47, 62Orwell, G. 10Osterwalder, H. 43, 44, 185(n9)

Panther, K.-U. 186(n17)Peirce, C.S. 81–2, 85–6, 88,

90–2, 94–5, 102, 104, 127, 187(n12), 188(n18)

notion of sign 91fPerrine, L. 58–9Pharaoh 121, 150Piaget, J. 191(n5)Pinker, S. 147, 192(n7)Pire, F. 40Plato 11, 12, 78Plett, H. 6Pompeius 15, 18Preminger, A. 49Proclus 12

Qamar, Sheikh 182Qudämah ibn Ja‘far 24, 25Quillian, M.R. 61Quintilian 15, 17, 19, 49

Author Index 207

1403_913323_11_auin.qxd 11/19/03 9:24 PM Page 207

Rabin, Y. 170–4, 192(n17)Radden, G. x, 3, 38, 106

cognitive theory of metonymy69–72, 113

Reddy, M. 59–60Reisbeck, C.K. 161Richards, I.A. 35, 185(n3)Ricoeur, P. 187(n13)Rosch, E. 97, 141, 191(n4),

192(n9)Rosiene, A.M. 15, 16, 40, 185(n4),

188(n23)Rumelhart, D. 4, 137, 139, 146,

189(n27)Russell, C. (fictional protagonist)

153–4

Sacerdos 15, 17–18, 19, 104Sadat, M.A. El- 174Saddam Hussein 178Sampson, G. 187(n13)Sapir, E. 84, 92, 97, 188(n15)Saussure, F. de 81, 82–3, 85,

87–93, 127, 169, 187(n6,n10–11, n13), 192(n14)

‘arbitrariness’ in theory of semiology challenged 88

‘narrow conception of semiology’88

paradigmatic relations 130‘structural’ concept of sign 91–2,

92–3syntagmatic relations 131

Schank, R.C. 4, 7, 8, 137, 139, 142,160–1, 163–6, 176–7, 181,189(n27), 191(n3)

script theory 141, 160Schmid, H-J. 137, 189(n27),

192(n9)Searle, J. 85Sergius 15Shakespeare, W. 55Shklovsky, V. 192(n13)Singer, J. 156–60Soskice, J.M. 101Sowa, J. 142

Stallard, D. 112cognitive theory of metonymy

56, 62, 63, 72–5Stam, R. et al. (1992) 81, 202

Burgoyne, R. 202Flitterman-Lewis, S. 202

Stanford, W.B. 11–12Stirling, L. 109Swales, J. 137

Tabbänah, B. 25Taylor, E. 78Tesauro, E. 14Thornburg, L. x, 186(n17)Thucydides 11Tita (baby) 155–6Trinton, H. 40Trudgill, P. 89Turner, G. 86Turner, M. 55–6, 64, 78, 79, 154

Ullmann, S.‘does not distinguish between

metonymy and synecdoche’46

linguistic theory of metonymy40–1, 45–8, 49, 82, 114

Ungerer, F. 137, 189(n27), 192(n9)Urban 191(n16)

van Dijk, T.A. 139, 153Volta, A. 47, 62

Wales, K. 6, 32, 86, 185(n5)Warren, B. 109Watson, J. x, 190(n10)Wells, R. 89Wertheimer, M. 138White, H. 185(n2)Whorf, B.L. 92, 97, 188(n15)Wierzbicka, A. 186(n16)Winston, P. 175

Yeats, W.B. 56Yule, G. 189(n3–4)

208 Author Index

1403_913323_11_auin.qxd 11/19/03 9:24 PM Page 208

��o��́ (metaphor) 11��o��́�� 11��́� 11

action 97, 143, 160, 164, 166, 180, 182

global 140stereotypical sequence 162

activationconceptual knowledge 171general goal of text 182scenario 172, 177script and schema 172

actual things (pragmata) 96, 97, 99actual for virtual (metonymic

principle) 133adjacency 21, 22, 24, 34, 38, 145adjectival phrase 74adjective-noun metonymies 61adjectives 158affections in soul ( pathemata) 96,

97, 99airlines 63–4, 72–3, 74, 77al (Arabic definite article) 149, 150

generic al 149knowledge-based al 149

al al-‘ahdiyyah (al ofknowledge/familiarity) 149

knowledge-based al 149three types 149–52

al al-dhihniyyah (domain ofconcepts) 152

al al-dhikriyyah (domain oflanguage) 152

al al-hudhüriyyah (domain of thingsand objects) 152

al al-ma‘hüdatu dhihniyyan(cognitive al) 150–1, 192(n11)

al al-ma‘hüdatu dhikriyyan (textualal) 149–50, 151, 192(n10)

al al-ma‘hüdatu hudhüriyyan(existential al, ontological al,exophorical al) 152, 192(n12)

al-’ahad al-dhihnï 192(n11)al-‘ahad al-dhikrï 192(n10)al-’ahad al-hudhürï 192(n12)al-Bayän wa al-Tabyïn (Al-Jähiz) 23al-Kämil (al-Mubarrad) 23Al-Sharq al-’Awsat (newspaper)

170–4, 178–9, 192(n17)al-Sinä‘atayn (Abü Hiläl al-‘Askari)

29al’Ïdhäh fi ‘Ulüm al-Baläghah, al-

Ma’äni wa al-Bayän wa al-Badï’.Mukhtasar Talkhïs al-Miftäh(al-Qazwïni) 27, 31, 33

Alexandrians 188(n19)allotexts (variants) 7allusion (ta‘rïdh) 24, 25, 28analogy 16anaphora 65, 73–4anaphoric reference 112, 120, 126Animal Farm (Orwell) 10animals 132, 134f, 191(n17)antecedent 65–6, 112–13anthropology 92, 97, 110antiphrasis 44antonomasia 58approximative nature of

communication 8, 83,184(n3)

Arab grammarians 23, 150–1Arabic 120

Classical 120–1diglossia 46, 186(n11)lexical reiteration 127–8,

190(n10)modern 120–1seems to lack ‘auxiliary

verbs’ 120

Subject Index

Key: f�figure; n�note; t�table; bold�extended discussion or headingemphasized in main text.

——————

1403_913323_12_subin.qxd 11/21/03 6:41 PM Page 209

Arabic – continuedtranscription xiverb-subject-object (VSO) word

order 171Arabic rhetoric 20–34

three dimensions of contiguoustransference 22f, 22

types of transference 20, 21fArabic-English dictionaries 20arbitrariness 88, 94, 96, 104, 144,

167–8, 169, 192(n14)linguistic sign 93

arbitraryknowledge 145, 170knowledge structures 4sequenced events 144

argument 95argumentative discourse 173artist for artform metonymy 62ash under cauldron (generosity)

26, 47, 60Asrär al-baläghah (al-Jurjäni)

29, 32–3association 15, 26–7, 41, 45, 47,

139, 144, 148, 176cognitive 32contiguous 97–8

At-Tiräz (Al-‘Alawi) 28, 31attribution (of actions) 30audience 23, 148, 149axe/blunt 118, 120

badï‘ (embellishment) 23, 25baläghah (rhetoric) 24bayän (‘eloquence’) 31Bedouin Arabs 10behaviourism 145–6best example (Ungerer and Schmid)

192(n9)bias 94, 169, 171, 174binary opposition 185(n5)borrowing 93Boston 74boundaries 42, 49, 52, 59, 60,

186(n15)brain 139, 141, 177Brethren of Purity (ikhwän al-safä)

190(n10)

Cabinet/they (indirect anaphoricreference) 112

Caliph 85cataphoric reference 158categories 79, 137categorisation 56–7, 63, 65, 70,

97, 142, 148interactionist model (Rosch)

191(n4)causal chain 176causality 1, 4, 52, 53, 79, 82, 95,

102–3, 105–6, 107f, 109, 137,142, 144, 145, 167–9, 172–5,178–80

contextually bound 167–8localised 174–5personal 167–8specific 168

cause-effect metonymy 4, 66, 79,105f, 106, 140, 163, 166, 170,175, 178–9, 181–3, 190(n9)

chain of metonymic interpretation59–60

chair 95, 132–3, 175children (cognitive development)

191(n5)Chinese language 116cinema 148classical metonymy 11–20classical rhetoric 2, 7, 11–12, 40,

102, 108, 110clausal ellipsis 123, 124clausal substitution 119–20clauses 122

Halliday and Hasan’s definition 123

modal element 123propositional element 123

‘Clear over Obscure’(communicative principle) 71–2

closure principle 138codes 186(n2)coercion (Grosz) 62cognition 1, 5, 7, 9, 10, 35, 39, 43,

48, 57, 81, 90, 92–3, 97, 101,103, 106–7, 111f, 112–14, 116,127, 139, 145, 146, 156, 166,169, 177–9

analytic and syntheticdimensions (Bredin) 51

holistic 139individual role 96

210 Subject Index

1403_913323_12_subin.qxd 11/21/03 6:42 PM Page 210

cognition – continuedlink with discourse through

metonymic processes 137motivatedness 169‘overall orientation of author’s

position’ 7social 121

cognitiveal 150–1, 192(n11)causality 183grammar (Langacker) 57hyperbole 55linguistics 2model 8, 163–6, 175–6,

189(n27)principles 180process 154processes (general rules)

186(n15)psychologists 72psychology 8, 64, 65, 110,

184(n2)reasoning 105science 14semantic studies 2transfer 30, 185(n15)understanding 103

cognitivists 37, 38, 69coherence, see text coherencecohesive devices 110, 112, 113‘collective consciousness’ (Eagleton)

187(n13)‘collective mind’ (Durkheim) 89,

90, 187(n13)collectivity 99Collins English Dictionary and

Thesaurus (1995) 138collocational relations 131combination (Jakobson) 37communication 6, 35, 81, 83, 84,

85, 89, 106, 136–7, 139, 170,179, 188(n14)

approximative nature 8, 83,184(n3)

effectiveness 7, 36maxim 184(n2)philosophical and

anthropological theories 8,184(n2)

physical stage 9production stage 9

real-life phenomenon 89systems 82, 187(n5)

communicative acts 93components 185(n5)

deictic, classificatory, relational(Nunberg) 188(n25)

comprehension-extension191(n16)

computational linguistics 141concatenation 37–8concept for concept metonymy 4,

70, 107, 107f, 128, 134–6, 140,156, 162

concept entity 51concept for form metonymy 107,

107f, 110, 114, 116, 117, 121–4,125, 136

concept for thing metonymy 4,71, 107, 107f, 136

concepts 3, 8, 9, 22, 34, 51, 80, 87,89–91f, 96, 106, 118, 126, 139,145, 146, 148, 152, 175, 176,184(n8), 189(n26)

conceptualblending 78, 79, 154integration 79, 154relationships 70representation 8, 152, 191

connectedness 64cognitive 106formal 106semantic 33

connectivity 59, 114, 161, 177causal 10, 179cognitive 158formal 127

connectors (Fauconnier) 77, 78connotation 43–4, 50, 132, 133,

190(n11), 190–1(n16–17)Barthes 188(n21)

contained-for-container metonymy 13

container-for-content metonymy24, 34, 60

context 1, 3, 7, 8, 24, 28, 33, 44,106, 111f, 112, 113, 115, 127,130, 133, 141, 148, 167–8,184(n2), 185(n11), 188(n14),189(n25), 189(n2)

social 190(n11)socio-cultural 89

Subject Index 211

1403_913323_12_subin.qxd 11/21/03 12:20 PM Page 211

contextual al 152contiguity 1, 4, 14, 24, 34, 37–8,

40–2, 45–9, 52–4, 58–9, 61, 69,79, 102–6, 107f, 109, 137, 142,144–5, 148, 167, 169–70, 173,176–7, 179, 184(n10),188–9(n25)

cognitive perspective 38concatenative 37function in discourse 43Jakobson 37names 47, 114senses 47, 48spatial relations (Ullmann) 46, 47spatio-temporal 44syntagmatic 44temporal relations (Ullmann)

46, 47control centres (de Beaugrande and

Dressler) 142–3controller for controlled

metonymy 178convention 47–8, 84, 87, 96, 97,

104, 184(n10)conventional

knowledge 145, 170knowledge-structures 4links 47–8metonymy 52, 62motivatedness 48sequenced events 144, 168

conventionality 52–3, 72, 85, 93,96, 103, 160, 187(n5)

conversation 148conversation analysis 184(n3)conversational implicature 65cooperative principle (Grice)

63, 184(n2)‘covering’ 28Cratylus (Plato) 12culture 10, 47, 60, 69, 84, 85,

92, 93, 97, 107, 132–3, 141,160–3, 167

Dalä’il al-‘i’jäz (al-Jurjäni) 26, 30De Bello Gallico (Julius Caesar) 43De Interpretatione (Aristotle) 96defamiliarisation 160, 192(n13)default mode (of

understanding) 176

default plan interpretation 180deferred reference (Nunberg) 75–7definite article 164–5, 173definiteness (three Arabic

modes) 149deixis 55–6, 102, 103, 152, 156,

188(n25)deletion 38, 45, 114demetaphorisation 44demetonymisation 44demonstrative reference 117denominatio (Latin) 16, 18, 19denomination 16, 40denotation 132, 133, 190(n11),

190–1(n16–17)descriptive knowledge 106,

145, 146‘deviation’ 175dialects 186(n11)dichotomies

nature-convention 96, 188(n19)personal-communal 97subjective-objective 97, 98

diction 18, 34‘differentia’ (Bredin) 49diglossia 46, 186(n11)directives (Halliday and Hasan)

115discourse 6, 37, 46, 64, 68, 70, 96,

109, 110, 140, 151, 186(n18)argumentative 173link with cognition through

metonymic processes 137pragmatic interpretation 78

discourse analysis 36, 148discourse-processing: metonymic

reasoning ‘fundamental’ 177domains 38, 39, 50, 78, 105, 106,

139–40, 149, 150, 153, 180lexical 176

double-unit signification 11dysphemism 58

effect for cause metonymy 4, 140,105f, 106, 166, 183, 190(n9)

ellipsis 115, 117, 122–5, 181Arabic rhetorical tradition 124clausal 123, 124lexical 122–3nominal 122

212 Subject Index

1403_913323_12_subin.qxd 11/21/03 12:20 PM Page 212

ellipsis – continuedoperator 123Ullmann 47, 110, 114–15verbal 122–3

elocution 7eloquence 31, 148encoding 93, 100, 105, 111, 160Encyclopaedic Dictionary of Language

and Languages (Crystal) 86endophoric reference 117English language 20, 45, 87, 91,

116, 120, 123, 127, 138, 165entities 9, 144, 145, 146, 148epistemology 1, 42, 54, 91, 97,

98, 99, 100, 111f, 132, 140,147, 149

two levels of experience 98Euagoras (Isocrates) 11euphemism 21–2, 25, 58, 71–2,

114–15, 121, 167, 184(n10),190(n6), 191(n17)

event-for-time metonymy 58events 70, 143, 160, 161, 164, 180existential al 152, 192(n12)existentiality 21, 184(n8)exophoric reference 23, 111,

117, 156expectation 10, 112, 137, 146,

153, 155–6, 160, 175, 181,191(n18), 192(n8)

experience 106, 139, 140, 142,144, 146, 153

explicitness 23, 184(n9)expression 16, 98, 189(n25)extension-intension 191(n16)external binding agreement

condition 73

face 67, 71, 184(n2)factuality 101figuration 30, 31, 33, 34, 44, 71figurative

language 35, 85, 89, 99–100theory 2, 35–79, 100, 185–6treatment 8understanding 9

figure-ground relations 103fingertips 33, 185(n17)fire alarm 100food 85

form 3, 25, 94, 70, 106–8, 111,111f, 189(n2)

form for concept metonymy 70,107, 107f, 110, 116–21, 125,127–9, 130f, 136

form for form metonymy 107,107f, 110, 113–14, 116–21, 125, 127, 129, 136

‘same form repeated’ 120form for thing metonymy 107,

107f, 110, 116, 117, 125,127, 136

form-concept for form-conceptmetonymy 71

form-concept for thing metonymy 71

formalist criticism 192(n13)frame (Fillmore) 57frame activation 153frame theory (Minsky) 141, 142,

146–7, 191(n3)frames 4, 137, 142–8, 160, 161,

164, 168, 177same as ‘schemata’ 152–3

functional linguistics 111furniture 132–3

gender test (Italian, Arabic) 76‘general style’ (Cicero) 16generalization 142‘gents’ 114–15, 190(n6)genus 57, 106, 132genus-species (Aristotle) 41gestalt psychology 57,

138–9, 148global patterns (de Beaugrande and

Dressler) 142, 146, 147,161, 177

goals 4, 137, 141, 142, 144, 145,156, 181–3

grammar:Arabic 149, 165English 165late antique 15

grammarians 18Arab 23, 150–1

grammatical cohesion 115–21grammatical methods 7Greece 11–15, 81, 96Greek language 6, 186(n11)

Subject Index 213

1403_913323_12_subin.qxd 11/21/03 12:20 PM Page 213

ground (G) 38, 39, 185(n3)Groupe mu 15, 40–5, 48–52,

185(n6–8)

hadhf (‘ellipsis’): three types 124hadhf of musnad ’ilayhi 124‘ham sandwich’ 54–5, 72–4, 103,

129, 162–3Hanoi 74haqïqah (versus majäz) 29Hebrew language 116hierarchies: linguistic and

extralinguistic 135homonymy 12house: cultural expectations 10Householder, F. 12humour 72hunger 163, 167, 168, 169, 175hypallage 17, 18hyperbole 85hyperdomains 78hyperspace 78hyponymy 130–2, 134

cognitive 131–2, 131fmetonymic model 132fphysical 131–2, 131f

icon 95iconicity 95, 116idealised cognitive models (ICMs)

56–8, 65, 69, 77, 106, 116, 118four fundamental questions 70parts 71of the week 57whole 71

ideas 108, 110Frege 97–101

ideational metafunction 111identification principle (IP)

(Fauconnier) 77identity 79ideology 93, 94, 129–30, 169, 171,

174, 178, 190(n14)idiom 185(n11)‘ilm al-kaläm (Muslim scholastic

debates) 29image 139imagery 100imitation 97implicitness 22, 23, 184(n9)

inclusioncognitive 135linguistic and extra-linguistic

135semantic 135

Index 95Indexical basis of metonymy 3,

102–4, 109, 189(n1)Indexicality 102–4, 109, 188(n25),

189(n1)versus symbolicity 104

inferences 192(n8)inferencing 122, 138, 139,

191(n18)inflation/savings 53–4, 63, 75information 10, 74, 111, 122, 146,

148, 176, 177informativeness 6informativity 54inheritance (de Beaugrande and

Dressler) 147, 148f, 192(n9)instrumental header (IH) 163–4instrumental relationship 33intellection 40intelligence 148intentionality 79, 85intentions 145, 169interaction 3, 7, 8, 9, 92, 93, 96,

106, 107, 112, 144, 150interactionist models 7, 153,

191(n4)intermediary term (I) 43internal conceptualisation header

(ICH) 165–6interpersonal metafunction 111interpretant (Peirce) 90, 91, 91f,

95, 102interpretation 39, 88, 99, 112,

188(n25)cyclical process 100

intertextuality 121invention 7, 107f, 108inventor-for-invention relation

13, 47’irdäf (contiguity) 24, 25irony 37, 85’isnäd (predication) 29isorrophy 12isti‘ärah (‘borrowing’/‘metaphor’)

20, 21f, 29, 33

214 Subject Index

1403_913323_12_subin.qxd 11/21/03 7:20 PM Page 214

Japanese language 116jargon 72

keel/deep 38, 52–3, 114kinäyah (implicitness) 2, 21–2, 22f,

22–8, 34, 184(n7, n9–10)‘can mean the literal

meaning’ 27linguistic root 28, 185(n13)‘metonymy’ 20a principal feature 26–7social aspect 23, 184(n10)

king/waiting/arsenic 66, 177Kitäb al-Badï‘ (Ibn al-Mu’tazz)

23–4knowledge 88

arbitrary 143f, 145, 170argumentative 170classification 142commonsense 147conceptual 171conceptual representation

191(n1)conventional 143f, 145, 160, 170cultural 163, 166culture 167, 168descriptive 106, 143f, 145, 146domains 105, 189(n26)general 115, 146, 147, 178procedural 106, 143f, 189(n28)prototypical 144, 154–5schematic 149stereotypical 144, 160, 164, 169three modes 80universal 163, 166, 167, 168

knowledge structures 4, 106, 138,141, 144, 146, 147, 152–3, 168,176, 189(n27)

arbitrary 4conceptual 171conventional 4higher-level 147, 191(n3)theories 142, 191(n3)three main types 175unified model 143f

knowledge transfer 147kunayah 184(n10)

‘labelled part-whole hierarchy’(Cruse) 135

Laborintus (Eberhard) 15lack of mice (no food) 85‘ladies’ 47–8, 118language 22, 43, 51, 57, 70, 82–3,

85, 90, 92–3, 103–4, 106–7,111, 111f, 133, 139–40, 145,149, 152, 159, 169, 176,179–80, 187(n5–6, n13),188(n14)

generalallows for misinforming

187(n9)analogist perspective 188(n19)everyday 34‘exists perfectly only in the

collectivity’ 87, 187(n10)figurative 8–11figurative and literal (no clear

dividing line) 8functions 187(n8)‘lacks regularity’ 188(n19)literal 100literary 100, 160, 172‘loose fit’ between descriptions

and referents 184(n3)nature and role of metonymy

64–9‘not innocent’ 94, 97‘real-life communicative

situations’ 187(n8)reflection of reality

(assumption) 8regulative function 68‘representation of a

representation’ 97‘social-semiotic’ (Halliday)

187(n8)written 58

specificAmerican Indian 97Arabic 20, 23, 46, 76, 116, 117,

120–1, 127–8, 171, 186(n11),190(n10)

Chinese 116English 20, 45, 87, 91, 116,

120, 123, 127, 138, 165Greek 6, 186(n11)Hebrew 116Italian 76Japanese 116

Subject Index 215

1403_913323_12_subin.qxd 11/21/03 12:20 PM Page 215

language disorders 37language form 190(n7)language system 92language use 30, 67, 101

philosophical andanthropological theories 8

language-processing 153langue 87, 89Latin rhetorical tradition 2, 15,

40, 102läzim (entailing) 27levelling 43, 185(n9)lexical cohesion 110, 115, 125–35lexicalisation 44lexis 70Like Water for Chocolate (Esquivel)

155linearisation 7linguistic

categories 10concepts 89linguistic philosophy 188(n19)reasoning 70linguistic signs 37, 39, 88transference see transference

linguistics 81, 82, 91, 186(n4)anthropological 92critical 84functional 111relationship with semiotics 82–3scope 186(n4)theories of metonymy 36–58,

185–6linguists 38, 69, 72, 90, 97literary critics 25, 37, 185(n1)literature 6, 36, 72, 159, 186(n11)lobster/he 113locale header (LH) 164–5locality 34logic 7–8, 51, 81, 99, 114, 132,

138, 166, 191(n18)logical connectedness 32logical consistency 6logos (Greek) 16‘long sword’ (tall man) 26, 27loose transference see majäz mursallying 84–5, 187(n9)

ma‘äni (‘notions’) 31ma’dhün (‘authorised marriage

contracts conductor’) 166

majäz (transference) 20, 21f, 29,31, 33, 185(n14, n16)

al-Jurjäni’s definition 33fifteen types (al-’Alawi) 33–4‘metaphor’ 27

majäz al-‘alä’iq al-’ukhrä 21fmajäz al-muläzamah (‘contiguity’)

relations 20, 21fmajäz al-mushäbahah (‘similarity’)

relations 20, 21fMajäz al-Qur’an (Abü ‘Ubaydah)

22majäz ‘aqlï (cognitive or rational

transference) 2, 20, 21f, 21,22f, 22, 28–31, 32, 34

al-Jurjäni’s definition 30al-Sakkäki’s definition 30–1differentiated from

‘personification’ 32intentionality 29–30

majäz ’irtibätï (contiguoustransference) 22f

majäz ’isnädï (predication transfer) 31

majäz lughawi (linguistictransference) 21f, 29, 31, 32, 33

majäz mufrad (transference ofindividual words) 31

majäz murakkab (constructionaltransference) 31

majäz mursal (loose or non-similarity transference) 2,20, 21f, 184(n7), 21, 22f, 22,31–4, 184(n8)

al-Sakkäki’s discussion 33malzüm (entailed) 27mapping

cognitive 58forms onto concepts and things

156metaphoric 57metonymic 57metonymy and metaphor

contrasted 59part-whole 79of source onto target 79

material plane (�) 42, 43, 185(n8)

maxim of quality (Grice) 85maxim of quantity (Grice) 63, 74

216 Subject Index

1403_913323_12_subin.qxd 11/21/03 6:42 PM Page 216

maxims of quantity and relevance 148

meaning 18, 48, 84, 92, 94, 98,103, 110, 111, 126, 137

cognitive theory 51direct literal 98extensional mode 51figurative 101Frege’s theory 99, 100indirect 24intensional mode 50–1literal 39, 101personalised 100pragmatic aspect 136rhetorical 28virtual and actual 114

meaning creation 77, 89, 93, 112,189(n3)

personal subjective side 99meaning relations 189(n4)meaning transfer (Nunberg) 75–7medieval period 40memory 138, 142, 144, 152, 172mental binding (Turner and

Fauconnier) 78same as ‘conceptual blending’

78mental processes 9mental space 57mental spaces (Fauconnier) 77, 78,

116, 188(n14)‘blended’ 78–9

merism 58meronomy 134–5meronymy 130, 134–5, 135fmetafunction

ideational 111interpersonal 111textual 111

metalanguage 38, 77, 184(n6)metaphor 10, 11–12, 14, 16, 35,

42, 43, 59, 85, 101, 109,184(n6)

Aristotle’s definition 41boundary (with metonymy) 49,

52Bredin 48, 52cognitive approach 38–9‘creates relation between objects’

(Bredin) 52‘denotative’ (Groupe mu) 44

discursive and signifying power185(n2)

domains 38Fauconnier 77figurative nature 44four types (Aristotle) 13function 101Groupe mu 41‘infinite multiplicity of

meanings’ 44isti‘ärah 29Jakobson’s theory 37majäz 27‘product of two synecdoches’

41, 185(n6)‘understanding’ function 54violation of selection

restrictions 63Metaphors We Live By (Lakoff and

Johnson, 1980) 53metasememe 44methodology: ‘collectivism’

versus ‘individualism’187(n13)

metonymia (Greek) 19metonymic

expressions (Gibbs) 66extension 46interpretation 62models 64‘objects’ (Curme) 61processes (semiotic) 18, 19projection constraint (Turner

and Fauconnier) 79reasoning 183recognition 62relations 38, 50, 53representation 176signification (three-dimensional

model) 90understanding (four fundamental

processes) 40metonymisation 41metonymy

characteristicsarbitrariness 52, 60, 62discursive and signifying power

185(n2)‘essentially combinatorial

aspect of language andthought’ 176

Subject Index 217

1403_913323_12_subin.qxd 11/21/03 12:20 PM Page 217

metonymy – continued‘essentially knowledge-based’,

142‘fundamentally indexical’ 103interdomain nature 139‘lacks creative depth of

metaphor’ (Bredin) 52‘more obvious’ than metaphor

55‘not banal’ 52‘presupposes relation between

objects’ (Bredin) 52‘realistic, referential,

contextually relevant’ 39‘reference-point phenomenon’

(Langacker) 70‘schematic representation of

knowledge’ 149similarity with Peircean notion

of sign 91, 91f‘systematic character’ (Lakoff

and Johnson) 51‘twice-true’ (Fass) 59‘western concept’ 20

definitions 1, 2, 36, 40,188(n23–4)

‘four fundamental shortcomingsof traditional view’ 69–70

Lakoff and Johnson 53functions 59, 68–9, 101

‘cost-effective communication’39

‘fundamental principle of ourcognitive structure’ 137,191(n2)

‘meaning-creating process, not aproduct’ 77

‘mode of contiguous and causalsignification’ 80

nature and role in text 7‘process of meaning creation’

12process of signification 80, 105‘provides mental access that

cuts across domains oflanguage, mind and reality’180

‘shortens distances ofinterpretation’ 115

as signification 102

‘text-binding process’ 78thought organisation 167

miscellaneousactive zones 184(n4)‘essential observation’ 34indexical basis 103, 109,

189(n1)indexical sign 3

theories and approaches36–79, 185–6

‘diversity not a problem forinterpretation’ 39

figurative accounts 177‘four distinct processes’ (the

Auctor) 16hypotheses 1, 19, 110, 180,

191(n2)interdomain conception

189(n26)linguistic approach ‘not

sufficient’ 39linguistic theories 36–58,

185–6‘new way of looking at’

(Gibbs) 64relational model 3, 105f,

190(n9)semiotic approach 1, 2–3, 8,

22, 88, 95, 110structure of book 1–4textual approach 10–11, 88textual model 1, 10–11,

22, 80, 101, 105, 105f, 106–8, 109, 133, 139, 149

textual theory 40, 58, 95–6,104, 110

theoretical framework andhistorical background 5–34,184–5

‘theory of linguisticsignification’ 12

theory of motivatedsignification 94

twofold theory (Groupe mu)42

typescognitive 19, 38–9, 51, 58–79,

97, 186(n15–19)combinational 40conventional 52, 62

218 Subject Index

1403_913323_12_subin.qxd 11/21/03 12:20 PM Page 218

metonymy – continuedpredicative 62, 63–4, 72–5,

185(n12), 186(n19)referential 54, 55, 62, 63–4,

72–5, 101, 185(n12)representational 141, 152, 167,

180, 192(n6)textual 64

‘Metonymy’ (Bredin, 1984) 48Middle Ages 15, 36mind 9, 93, 106, 107, 139, 141,

180, 191(n1)misinforming 187(n9)modules

entities 144rules of sequencing and

combination 144words 144

monarch/crown 75morpheme 185(n5)morphology 70motivatedness 53, 80, 88, 89, 93,

94, 103, 130, 169cognitive 48conventional 48in linguistic signification

192(n15)musnad (predicate) 29, 124, 125musnad ’ilayh (subject) 29musnad ’ilayhi (subject) 124myth 94, 100

na‘jah/na‘jatan (ewe/s) 27–8,184(n10)

names 45, 47, 59, 96, 98proper 159

natural language data 68nature 93, 96, 97nature-convention 188(n19)negotiation 8, 112Newcastle 192(n7)nominal ellipsis 122nominal substitution 21, 119, 120non-literal understanding 9, 73,

85, 101non-similarity transference 20nonymy 12notions (ma‘äni) 31noun phrase (NP) 73, 74, 76,

112, 149

Arabic 32deep 76, 112surface 76, 112

nouns 60, 149common 116, 158indefinite 152proper 158

novelists 64

object (grammar) 51, 125objectivity 96, 97, 98, 99, 101, 174objects 9, 22, 34, 42, 50, 52, 80,

91, 91f, 92, 95, 106, 143, 145,146, 152, 188(n25)

and linguistic categories, ‘not aperfect correspondence’ 10

obstacle hypothesis 68omission 47–8On Style (Demetrius) 14On the Sublime (Longinus) 14onomatopoeia 87ontology 1, 19, 22f, 34, 41–2, 50,

54, 69–70, 93, 97–8, 100, 105,111f, 127, 131–2, 135, 140, 149,170, 187(n12), 188(n20)

oratio (Latin) 16orations 11orators 17ornament 17

paradigmaticlinguistic signs (Saussure) 37, 88metonymic representation 176relations (Saussure) 130structural notions 145

part-part metonymy 71, 107part for whole relation 4, 33, 47,

51, 53–4, 58, 60, 68, 79, 103–5,105f, 113–14, 131, 131f, 133–4,140–1, 145–6, 148–9, 153, 155,157, 160, 162, 166, 170, 179,183, 190(n6–7), 191(n18)

partiality 34perception 97, 138–9, 153

holistic 148periphrasis 58personal intentions 93personal reference 17

see also referencepersonification 32, 53, 63, 75

Subject Index 219

1403_913323_12_subin.qxd 11/21/03 12:20 PM Page 219

persuasion 6, 72phenomena

linguistic, epistemological,ontological dimensions97, 98

three levels 97, 98philosophers 69, 72, 81, 96,

188(n19)ethical (Plato) 12idealistic (Plato) 12

phonemes 12, 185(n5)phonetics 186(n4)‘place’ for ‘people working in place’

relation 65–6‘plans’ 4, 137, 141–2, 144–5, 147,

156, 161, 176–81, 183definition 176

Plato’s idealistic philosophy 12Plato’s puzzle (Keller) 81, 186(n1)playwrights 64poetic

devices 12, 20language 8, 11, 12, 32, 34, 172licence 12

poetics 15Poetics and Rhetoric (Aristotle)

12–13poetry/poets 11, 14, 23–4, 64, 72politeness 184(n2)polyonymy 12polysemy 12, 28, 89positionality 22, 34, 184(n8)potential-for-actual metonymy

186(n17)pragmatic theory 139pragmatics (use and interpretation

of signs) 82, 133pragmatism 5, 67precondition header (PH) 163predicates 98

coerced 73predication 7, 30, 31, 32, 34

figurative 21prediction 10, 112, 137, 161,

169, 180prejudice 169, 174preposition-noun metonymies 61prevarication 187(n9)Principles of Semantics (Ullmann,

1951) 45

pro-forms 115, 116procedural knowledge 106,

189(n28)producer-for-product

relation 47product-for-process relation 58pronouns 22, 23, 116, 118,

126, 158demonstrative 102

proper names 98, 99proper noun 62proper name-product relation 62property for whole metonymy

61–2proposition 95prose 12, 14

realist 10, 11, 48, 184(n5)prototypes 137, 142, 143f, 144,

147, 192(n9)prototypicality 132proximity 17, 18, 39, 138,

184(n10)psychology 8, 57, 64, 65, 68, 80,

81, 87, 72, 110, 138–9, 148,184(n2–3)

Qur’an 21–5, 27, 29, 30, 121, 124,150, 151, 184(n10)

radial category (Lakoff) 58radial category construction

(Lakoff), 141rasül (messenger) 150rationality 102reactions 146real world 91, 92, 95, 97, 147realist prose 10, 11, 48, 184(n5)reality 9, 10, 69, 90, 92–3,

96, 99, 104, 106, 107, 139, 180, 188(n16)

‘constructed and created bylanguage’ 97

realization 111reductionism 1, 2–3, 17, 18,

19, 37, 40, 41, 44, 48–9, 54, 176, 185(n2)

practical 5theoretical 5

redundancy 38, 45, 47–8, 64,114–15

220 Subject Index

1403_913323_12_subin.qxd 11/21/03 12:20 PM Page 220

reference 101, 115–18actual/potential 116cognitive 113‘cognitive’ (Langacker) 190(n7)comparative 117deferred 112, 118demonstrative 117endophoric 117exophoric 23, 111, 117, 156extralinguistic 110–11, 135Frege 49–50, 98, 99, 100,

188(n20)Halliday and Hasan 117indirect anaphoric 112ontological 50, 70, 135,

188(n20)personal 117‘semantic’ 120textual 113universal model 116

reference items 115reference point 136, 142, 190(n7),

192(n6)cognitive 97, 141

referencing 109, 112–13referential relations 157freferentiality 54, 55, 76, 101, 109,

112, 129indirect 112

register: systemic functionallinguistics theory 137

reiteration 126–35, 190(n13)anaphoric referential function

126‘relation’ 16relativism 97relevant over irrelevant (cognitive

principle) 72Renaissance 15, 36, 37repetition 127, 133

morphological 133representamen (Peirce), 91–2, 102

same as ‘signifier’ (Saussure) 90representation 1, 2, 3, 4, 50, 79,

80, 97, 99, 101, 105, 105f, 106,137, 190(n9), 191(n16)

‘crucial relation that underliesprocess of metonymy’ 105

Fauconnier 77imprecise 8

linguistic 97nine types 3‘sign’ and ‘metonymy’ 91, 91f

resemblance 16response 146, 188(n21)restaurant 55, 161–4, 167, 181restaurant script 163resulting term (R) 43rheme 95, 111rhetoric 2, 5–8, 107

Arabic 2, 11, 23, 34, 60, 102‘arrangement and elocution’

107f, 108classical 7, 40, 102, 108, 110cognitive view 14definition 6Greek 2, 11–12Groupe mu 40Indian 23Latin 2link with textlinguistics 7modern 35pejorative sense 36western 11–20, 29, 34, 102

Rhetorica ad Herennium (the Auctor)15, 185(n4)

rhetoricaldevices 77effects 72figures 75, 99

rhetoricians 11, 14, 17, 18, 46, 57,69, 72

Arab 21, 23, 28, 29, 31, 148,184(n10)

classical 49western 148

roles 161, 164–6Rome 15–20rules 6, 48, 147, 192(n7)Russia

communist revolution 10school of linguistics and poetics

192(n13)

Saigon 74saliency 103Sana’a/Yemeni government

(contiguous association) 97–8Sapir-Whorf hypothesis of linguistic

relativity 92, 97, 188(n15)

Subject Index 221

1403_913323_12_subin.qxd 11/21/03 12:20 PM Page 221

scenarios 4, 61, 65, 137, 142–5,160, 167, 168–76, 177, 180, 183

conflict of 174differentiated from ‘scripts’

171–2multiple 173

schema theory 136, 137, 138–45,191(n1)

schemas 56, 152, 168, 177,186(n14)

Antonapoulos and Singer 159fblends 154fcity 148, 153definitions 138, 145–6differentiated from a ‘scripts’

162getting up, going to work 140–1hospital 172kitchen 10, 155–6levels of delicacy/detail 159office 153, 154f, 154–5persona 153, 154frestaurant 166school 155studying 151taxi 149town 158, 158ftrain 149washing 161–2

schemata 4, 137, 139, 142–5,145–60, 175, 176

interdomain nature 139schematic knowledge 191(n18)script headers (Schank and Abelson)

163–6script theory 142, 191(n3)scripts 4, 137, 141–4, 147, 148,

156, 160–8, 169, 175–7differentiated from ‘scenario’

171–2differentiated from a ‘schema’

162hospital 172

‘Second Coming’ (Yeats) 56selection (Jakobson) 37semantic

change 45, 46, 114coercion 72cognitive world 147fields 39, 42

plane (�) 42, 43, 185(n7)relations 148structure 79traits 129, 190(n11)

semantics (meaning of signs) 82,112, 133, 185(n5), 189(n4)

Semantics: Introduction to Science ofMeaning (Ullmann, 1962) 45

semes 41, 42, 185(n5)‘semic fields’ 51semiology 81, 83, 90, 91,

187(n6)semiosis 93, 181semiotic systems 83semiotics 1, 3, 80–108, 186–9,

190(n11)‘actually a theory of meaning’

81, 186(n2)Eco’s definition 84–5general perspective 80–5lacks ‘predictive power’ 82‘largely qualitative’ 82representational theory

96–101‘science of signs’ 81–2scope 88triadic representations of notion

of sign 3‘vastness of phenomena’ 82,

186(n3)‘sense’ (Frege) 49–50, 97–100sense creation (Gibbs) 66–7‘On Sense and Reference’ (Frege)

98sense selection (Gibbs) 66sense-contiguity 41senses 45, 47, 48, 50sensory data 139sentence construction 23, 31, 32sentences 96, 98, 111, 122

components 124meaning 85, 189(n2)‘not value-free’ 112truth-value 98unmarked Arabic 171

sequenced events 144, 148,160, 166

arbitrary 144conventional 144, 168

sharpening 185(n9)

222 Subject Index

1403_913323_12_subin.qxd 11/21/03 12:20 PM Page 222

sign 81–2, 98, 106, 116, 170creation and interpretation 101Frege 100‘index and icon’ 88integrated conceptions 89linguistic 101–4nature 86–94, 187(n12)types and modes 94–6,

188(n17–18)see also semiotics

sign combinations 98‘sign’ language 187(n5)sign-function (Hjelmslev) 90significatio 190(n16)signification 1, 18, 19, 37, 52, 54,

81, 82, 85–96, 101, 113, 116,118, 188(n21), 189(n2)

cyclical process 100, 185(n10)extralinguistic 110figurative 27–8iconic 95indexical mode 53internal 120linguistic 80, 169, 192(n15)literal 27–8literal objective 100metonymic (relational model)

104–6metonymy as 102motivated 104, 130multiple 60nature of interpretant (rheme,

proposition, argument) 95nature of signs 86–94principle of interpretation 86rules and principles 81Saussurean 169semiotic approach 111symbolic/iconic/indexical 116syntagmatic and combinational

dimensions 109textual 120three elements (sign, signifier,

signified) 89three modes (icon, symbol,

index) 95, 188(n16),189(n26)

triadic model (Radden andKövecses) 106

triadic scheme 34, 131

two-level process 95types and modes of sign 94–6,

188(n17–18)verbal and non-verbal 83

signifier/signified 87, 89–91, 91f,93, 95, 102–5, 169

similarity 16, 32, 37–9, 41, 44–5,79, 95

similarity principle 138simile 28situational factors 7situationality 184(n8)situations 107, 113, 115, 143

real-life 6stereotypical 146, 176

slang 72smoke/fire 88, 100, 103social

behaviour 84bond 187(n10)experience 105processes 94sciences 82struggle 93

socio-culturalcontext 89plot 172

sociolinguistics 92sociology 8, 67, 110, 184(n2),

187(n13)solidarity 71sophists 12sound 87, 97, 133, 139space 79, 137spatial relations 34, 46spatio-temporal contiguity 44speakers 64specialisation 142species 57, 60, 106, 132species to genus relation 41, 131species to species relation 131specificity 51speech 97

locutionary, illocutionary,perlocutionary 188(n21)

speech act theory 85, 184(n2),188(n21)

speech community 87, 90, 93,104, 187(n10)

socio-cultural practice 121

Subject Index 223

1403_913323_12_subin.qxd 11/21/03 12:20 PM Page 223

SPGU (scripts, plans, goals,understanding) model (Schankand Abelson) 141, 142

spoken sounds (phonai) 96, 97stand-for relations 91, 92,

99, 102, 113, 120, 126, 127, 135, 140, 141, 161,167, 190(n9)

starting term (S) 42–3stereotypes 137, 139, 142, 143f,

144, 192(n8)cultural association 141

Stoic linguistic philosophy 16Stoics 188(n19)structural

coherence 6integration 183notions (paradigmatic and

syntagmatic) 145psychology 138relations 51–2, 175

structuralism 35, 83–4, 92, 138,185(n5), 187(n7)

definition 83Saussurean 169

structuralist project 40, 48structuralists 43structure

image-schematic 57prepositional 57

style 7, 14, 46, 100stylistics 6, 110sub-actions 140sub-schema 156, 158, 158fsubject 51subjectivity 97, 98, 99subjects (grammar) 116, 124subscript/superscript 162substance 186(n4)substitution 1, 5, 17, 19–20, 50,

59, 70–1, 78, 101, 110, 112–15,118–21, 131

anaphoric in nature 120Arabic 120–1, 190(n8)clausal 119–20concepts 2‘essentially grammatical’ 120grammatical function 118–19nominal 21, 119, 120things for things 19

verbal 119words 2, 19

substitutional text linguistics190(n5)

superordinate 127, 131superordinateness (‘converse of

hyponymy’) 130, 132–4superschema 156suppositio 190(n16)symbol 87, 88, 95, 96symbol of justice 187(n11)symbolicity 53, 103

versus indexicality 104symbolism 95symbols (symbola) 96symbols 184(n6)synecdoche 13, 37, 41, 42, 45,

48, 59boundary with metonymy 42,

49, 52, 59, 186(n15)Bredin 51figurative nature 44function 68–9, 101‘inseparable’ from metonymy 43

synonymy 28, 89, 127, 128–30,190(n12)

syntactic parallelism 133syntactic structures 158syntatics (relations between

signs) 82syntagmatic

axis 114contiguity 44linguistic signs (Saussure) 37, 88metonymic representation 176relations 40, 131signification 109structural notions 145

systematicity 187(n5)

ta‘rïdh (allusion) 24, 25, 28taboo 121tanker 59tasrïh (explicitness) 184(n9)tatbï‘ (same as kinäyah) 25tautology 67, 186(n16)tawïlu al-nijäd see ‘long sword’taxonomy 186(n15)telescope 98tenor (T) 38, 39, 137, 185(n3)

224 Subject Index

1403_913323_12_subin.qxd 11/21/03 6:42 PM Page 224

terminological confusion 141, 142text 6, 40, 58, 93, 101, 105, 106,

137, 190(n11)argumentative 170Brown and Yule’s definition

189(n3)creation and organisation 103descriptive 170formal connectedness 114Halliday and Hasan 111–12‘new way of analysing’ 126potential power of metonymy 5semiotic and cognitive

approaches 84semiotic model 111f, 189(n2)spoken 6written 6, 84

text analysis 84, 133, 187(n8)text atoms (Enkvist) 7text coherence 4, 6, 10, 58, 80,

106–8, 136–83, 191–2‘basic thrust of book’ 64textual interaction 79

text cohesion 3–4, 80, 106, 107f,108, 109–35

Halliday and Hasan’s theory8, 110

new insight and new perspective 110

‘semantic concept’ 110text connectedness (semantic

theory) 113text generation 136, 148, 168text interpretation 94, 133, 136,

148, 168, 176, 191(n18)text organisation 107f, 108, 136,

148, 168, 191(n18)text-processing 58, 141, 168,

169–70, 176cognitive 111human 166–7machine 166

texteme 7textlinguistics 2, 5–8, 36, 84,

94, 110cognitive model 7definition 5–6Enkvist’s theories 6–7figurative language 8–11interaction-based model 7

link with rhetoric 7predication-based model 7sentence-based model 6–7

textness 112, 190(n5)textual

al 149–50, 151, 192(n10)cohesive devices 110, 112, 113communication 84interaction 6tension 112

textuality 3, 6texture 6, 58, 112, 114thing for concept relation 4, 107,

107f, 136thing for form relation 107, 107f,

116, 136‘thing ordered’ for ‘person ordering

that thing’ relation 162thing for thing relation 4, 107,

107f, 136things 8, 9, 19, 22, 41, 49, 59, 70,

80, 118, 144–5, 152, 175, 176,189(n26)

thinking 102behaviouristic 145modes 34, 51

thought 9, 89, 92, 96, 97, 144,169, 176, 177

figurative mode 72nature and role of metonymy

64–9time 46, 56, 58, 68, 79, 148, 149,

189(n1)titles (sign of respect) 28toponymy 58totality 34Traditions of the Prophet 24transfer 50, 53transference 25, 31, 49

cognitive 21cognitive (Arabic rhetoric) 20,

21flinguistic (Arabic rhetoric) 20,

21fthree forms (Arabic) 2

‘Transfers of Meaning’ (Nunberg,1995) 75

transmutatio (Latin) 18, 19transnominatio (Latin) 18, 19transnomination 18, 19

Subject Index 225

1403_913323_12_subin.qxd 11/21/03 12:20 PM Page 225

trigger/target 38, 49, 77, 141,186(n12)

trope enumeration 48–9, 51,185(n2)

tropes 1, 5, 15, 32, 34, 37, 38, 41,43, 48, 49, 52, 68, 185(n2),186(n15)

contiguity-based 58differences 51semantic theory 46

tropical understanding 18truth-value 98‘Two Aspects of Language and Two

Types of Aphasic Disturbances’(Jakobson, 1971) 36,185(n1–2)

types 60typicality 175–6

UN/United Nations (part–wholerelationship) 190(n6)

understanding 40, 54, 88, 103,139, 141, 148, 176, 180

uniqueness 79United Nations 71universality 96, 97, 99universe of discourse 148universities xutterance meaning 85, 189(n2)utterances 112

variants (allotexts) 7vehicle (V) 38, 39

verbal ellipsis 122–3verbal substitution 119verbs: ‘can function metonymically’

60–1violation of selection restrictions

62–3, 73vocative 152

war 13warning 100, 188(n21)water (hydrogen/oxygen) 88–9Webster’s Third New International

Dictionary 41whole-part metonymy 4, 9, 51,

53, 68, 71, 105, 105f, 106–7,133, 140, 146, 148, 153, 157,160, 166, 183, 186(n17),191(n18)

whole-whole metonymy 107women 26, 27–8, 184(n10)word patterns 187(n10)words 32, 34, 80, 91, 91f, 98, 106,

144, 145, 147, 175, 176,184(n8), 189(n1), 189(n26),192(n7)

Words and Rules (Pinker, 1999),192(n7)

worldview 92, 188(n14)

yümi’u (‘indicates’, ‘signifies’) 102

Za‘baläwi (Naguib Mahfouz)181–3

226 Subject Index

1403_913323_12_subin.qxd 11/21/03 12:20 PM Page 226

Documents

Al-Sharafi - Textual Metonym.pdf